INT WINAPI WinMain( HINSTANCE hInst, HINSTANCE, LPSTR, INT )
{
WNDCLASSEX wc = { sizeof(WNDCLASSEX), CS_CLASSDC, MsgProc, 0L, 0L,
GetModuleHandle(NULL), NULL, NULL, NULL, NULL,
"D3D Tutorial", NULL };
RegisterClassEx( &wc );
HWND hWnd = CreateWindow( "D3D Tutorial", "D3D Tutorial 04",
WS_OVERLAPPEDWINDOW,
100, 100, 300, 300,
GetDesktopWindow(), NULL, wc.hInstance, NULL );
if (FAILED( InitD3D(hWnd) )) goto MAIN_END;
if (FAILED( init_geometry() )) goto MAIN_END;
ShowWindow( hWnd, SW_SHOWDEFAULT );
UpdateWindow( hWnd );
MSG msg;
ZeroMemory( &msg, sizeof(msg) );
while ( msg.message != WM_QUIT ) {
if ( PeekMessage( &msg, NULL, 0, 0, PM_REMOVE ) ) {
TranslateMessage( &msg );
DispatchMessage( &msg );
}
else {
Render();
}
}
MAIN_END:
UnregisterClass( "D3D Tutorial", wc.hInstance );
return 0;
}
int main()
{
init_geometry();
shape_t *s = new_taurus(30, 15, 30, 30);
flushOBJ(stdout);
free_shape(s);
finalize_geometry();
return 0;
}
void to_revolution()
{
init_geometry();
shape_t *s = new_shape();
for(unsigned i = 0; i < numpoints; ++i) {
add_vertex(s, new_vertex(pt[i].x, WINDOW_HEIGHT-pt[i].y, 0));
}
new_revolution(s, DIVISION_NUMBER);
flushOBJ(file_out);
finalize_geometry();
}
int main()
{
init_geometry();
shape_t *s = new_shape();
add_vertex(s, new_vertex(0, 300, 0));
add_vertex(s, new_vertex(20, 300, 0));
add_vertex(s, new_vertex(20, 290, 0));
add_vertex(s, new_vertex(300, 220, 0));
add_vertex(s, new_vertex(295, 218, 0));
add_vertex(s, new_vertex(60, 260, 0));
add_vertex(s, new_vertex(5, 220, 0));
add_vertex(s, new_vertex(5, 0, 0));
add_vertex(s, new_vertex(0, 0, 0));
shape_t *u = new_revolution(s, 60);
flushOBJ(stdout);
free_shape(s);
free_shape(u);
finalize_geometry();
return 0;
}
int main()
{
init_geometry();
shape_t *planet = new_sphere(50, 20, 20);
shape_t *sp1 = new_sphere(10, 16, 16);
shape_t *sp2 = new_sphere(10, 16, 16);
shape_t *t1 = new_taurus(30, 4, 50, 10);
shape_t *t2 = new_taurus(30, 4, 40, 10);
shape_translate(sp1, 80, 70, 40);
shape_translate(sp2, -90, 10, -60);
shape_scale(t1, 5, 1.2, 5);
shape_scale(t2, 3.5, 1.2, 3.5);
shape_rotate(t1, -20, 0, 0, 1);
shape_rotate(t2, 20, 0, 0, 1);
flushOBJ(stdout);
free_shape(planet);
free_shape(t1);
free_shape(t2);
finalize_geometry();
return 0;
}
int main(int argc, char **argv) {
int c, i, mu, nu;
int filename_set = 0;
int dims[4] = {0,0,0,0};
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, iix;
int xx0, xx1, xx2, xx3;
int y0min, y0max, y1min, y1max, y2min, y2max, y3min, y3max;
int y0, y1, y2, y3, iy;
int z0, z1, z2, z3, iz;
int gid, status;
int model_type = -1;
double *disc = (double*)NULL;
double *disc2 = (double*)NULL;
double *work = (double*)NULL;
double q[4], fnorm;
char filename[100], contype[200];
double ratime, retime;
double rmin2, rmax2, rsqr;
complex w, w1;
FILE *ofs;
fftw_complex *in=(fftw_complex*)NULL;
#ifdef MPI
fftwnd_mpi_plan plan_p, plan_m;
#else
fftwnd_plan plan_p, plan_m;
#endif
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "h?f:t:")) != -1) {
switch (c) {
case 't':
model_type = atoi(optarg);
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
fprintf(stdout, "\n**************************************************\n");
fprintf(stdout, "* vp_disc_ft\n");
fprintf(stdout, "**************************************************\n\n");
#ifdef MPI
if(g_cart_id==0) fprintf(stdout, "# Warning: MPI-version not yet available; exit\n");
exit(200);
#endif
/*********************************
* initialize MPI parameters
*********************************/
mpi_init(argc, argv);
/* initialize fftw */
dims[0]=T_global; dims[1]=LX; dims[2]=LY; dims[3]=LZ;
#ifdef MPI
plan_p = fftwnd_mpi_create_plan(g_cart_grid, 4, dims, FFTW_BACKWARD, FFTW_MEASURE);
plan_m = fftwnd_mpi_create_plan(g_cart_grid, 4, dims, FFTW_FORWARD, FFTW_MEASURE);
fftwnd_mpi_local_sizes(plan_p, &T, &Tstart, &l_LX_at, &l_LXstart_at, &FFTW_LOC_VOLUME);
#else
plan_p = fftwnd_create_plan(4, dims, FFTW_BACKWARD, FFTW_MEASURE | FFTW_IN_PLACE);
plan_m = fftwnd_create_plan(4, dims, FFTW_FORWARD, FFTW_MEASURE | FFTW_IN_PLACE);
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
#endif
fprintf(stdout, "# [%2d] fftw parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
#ifdef MPI
if(T==0) {
fprintf(stderr, "[%2d] local T is zero; exit\n", g_cart_id);
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
exit(2);
}
#endif
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(1);
}
geometry();
/****************************************
* allocate memory for the contractions
****************************************/
disc = (double*)calloc( 8*VOLUME, sizeof(double));
if( disc == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(3);
}
disc2 = (double*)calloc( 32*VOLUME, sizeof(double));
if( disc2 == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc2\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(3);
}
for(ix=0; ix<32*VOLUME; ix++) disc2[ix] = 0.;
work = (double*)calloc(32*VOLUME, sizeof(double));
if( work == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for work\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(3);
}
/****************************************
* prepare Fourier transformation arrays
****************************************/
in = (fftw_complex*)malloc(FFTW_LOC_VOLUME*sizeof(fftw_complex));
if(in==(fftw_complex*)NULL) {
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(4);
}
/***************************************
* set model type function
***************************************/
switch (model_type) {
case 0:
model_type_function = pidisc_model;
fprintf(stdout, "# function pointer set to type pidisc_model\n");
case 1:
model_type_function = pidisc_model1;
fprintf(stdout, "# function pointer set to type pidisc_model1\n");
break;
case 2:
model_type_function = pidisc_model2;
fprintf(stdout, "# function pointer set to type pidisc_model2\n");
break;
case 3:
model_type_function = pidisc_model3;
fprintf(stdout, "# function pointer set to type pidisc_model3\n");
break;
default:
model_type_function = NULL;
fprintf(stdout, "# no model function selected; will add zero\n");
break;
}
/****************************************
* prepare the model for pidisc
* - same for all gauge configurations
****************************************/
rmin2 = g_rmin * g_rmin;
rmax2 = g_rmax * g_rmax;
if(model_type > -1) {
for(mu=0; mu<16; mu++) {
model_type_function(model_mrho, model_dcoeff_re, model_dcoeff_im, work, plan_m, mu);
for(x0=-(T-1); x0<T; x0++) {
y0 = (x0 + T_global) % T_global;
for(x1=-(LX-1); x1<LX; x1++) {
y1 = (x1 + LX) % LX;
for(x2=-(LY-1); x2<LY; x2++) {
y2 = (x2 + LY) % LY;
for(x3=-(LZ-1); x3<LZ; x3++) {
y3 = (x3 + LZ) % LZ;
iy = g_ipt[y0][y1][y2][y3];
rsqr = (double)(x1*x1) + (double)(x2*x2) + (double)(x3*x3);
if(rmin2-rsqr<=_Q2EPS && rsqr-rmax2<=_Q2EPS) continue; /* radius in range for data usage, so continue */
disc2[_GWI(mu,iy,VOLUME) ] += work[2*iy ];
disc2[_GWI(mu,iy,VOLUME)+1] += work[2*iy+1];
}}}}
memcpy((void*)in, (void*)(disc2+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_p, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_p, in, NULL);
#endif
memcpy((void*)(disc2+_GWI(mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
}
} else {
for(ix=0; ix<32*VOLUME; ix++) disc2[ix] = 0.;
}
/***********************************************
* start loop on gauge id.s
***********************************************/
for(gid=g_gaugeid; gid<=g_gaugeid2; gid+=g_gauge_step) {
if(g_cart_id==0) fprintf(stdout, "# Start working on gauge id %d\n", gid);
/* read the new contractions */
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
sprintf(filename, "%s.%.4d.%.4d", filename_prefix, gid, Nsave);
if(g_cart_id==0) fprintf(stdout, "# Reading contraction data from file %s\n", filename);
if(read_lime_contraction(disc, filename, 4, 0) == 106) {
if(g_cart_id==0) fprintf(stderr, "Error, could not read from file %s, continue\n", filename);
continue;
}
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time to read contraction: %e seconds\n", retime-ratime);
/************************************************
* prepare \Pi_\mu\nu (x,y)
************************************************/
# ifdef MPI
ratime = MPI_Wtime();
# else
ratime = (double)clock() / CLOCKS_PER_SEC;
# endif
for(x0=-T+1; x0<T; x0++) {
y0min = x0<0 ? -x0 : 0;
y0max = x0<0 ? T : T-x0;
for(x1=-LX+1; x1<LX; x1++) {
y1min = x1<0 ? -x1 : 0;
y1max = x1<0 ? LX : LX-x1;
for(x2=-LY+1; x2<LY; x2++) {
y2min = x2<0 ? -x2 : 0;
y2max = x2<0 ? LY : LY-x2;
for(x3=-LZ+1; x3<LZ; x3++) {
y3min = x3<0 ? -x3 : 0;
y3max = x3<0 ? LZ : LZ-x3;
xx0 = (x0+T ) % T;
xx1 = (x1+LX) % LX;
xx2 = (x2+LX) % LY;
xx3 = (x3+LX) % LZ;
ix = g_ipt[xx0][xx1][xx2][xx3];
rsqr = (double)(x1*x1) + (double)(x2*x2) + (double)(x3*x3);
if(rmin2-rsqr>_Q2EPS || rsqr-rmax2>_Q2EPS) continue;
for(y0=y0min; y0<y0max; y0++) {
z0 = y0 + x0;
for(y1=y1min; y1<y1max; y1++) {
z1 = y1 + x1;
for(y2=y2min; y2<y2max; y2++) {
z2 = y2 + x2;
for(y3=y3min; y3<y3max; y3++) {
z3 = y3 + x3;
iy = g_ipt[y0][y1][y2][y3];
iz = g_ipt[z0][z1][z2][z3];
i=0;
for(mu=0; mu<4; mu++) {
for(nu=0; nu<4; nu++) {
iix = _GWI(i,ix,VOLUME);
_co_eq_co_ti_co(&w, (complex*)(disc+_GWI(mu,iz,VOLUME)), (complex*)(disc+_GWI(nu,iy,VOLUME)));
work[iix ] += w.re;
work[iix+1] += w.im;
i++;
}}
}}}}
}}}}
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time to calculate \\Pi_\\mu\\nu in position space: %e seconds\n", retime-ratime);
/***********************************************
* Fourier transform
***********************************************/
for(mu=0; mu<16; mu++) {
memcpy((void*)in, (void*)(work+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_p, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_p, in, NULL);
#endif
memcpy((void*)(work+_GWI(mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
}
fnorm = 1. / ((double)T_global * (double)(LX*LY*LZ));
if(g_cart_id==0) fprintf(stdout, "# P-fnorm = %16.5e\n", fnorm);
for(x0=0; x0<T; x0++) {
q[0] = (double)(x0+Tstart) / (double)T_global;
for(x1=0; x1<LX; x1++) {
q[1] = (double)x1 / (double)LX;
for(x2=0; x2<LY; x2++) {
q[2] = (double)x2 / (double)LY;
for(x3=0; x3<LZ; x3++) {
q[3] = (double)x3 / (double)LZ;
ix = g_ipt[x0][x1][x2][x3];
i=0;
for(mu=0; mu<4; mu++) {
for(nu=0; nu<4; nu++) {
iix = _GWI(i,ix,VOLUME);
w.re = cos(M_PI * (q[mu] - q[nu]));
w.im = sin(M_PI * (q[mu] - q[nu]));
work[iix ] = work[iix ] * fnorm + disc2[iix ];
work[iix+1] = work[iix+1] * fnorm + disc2[iix+1];
_co_eq_co_ti_co(&w1, (complex*)(work+iix), &w);
work[iix ] = w1.re;
work[iix+1] = w1.im;
i++;
}}
}}}}
/***********************************************
* save results
***********************************************/
sprintf(filename, "%s.%.4d.%.4d", filename_prefix2, gid, Nsave);
if(g_cart_id==0) fprintf(stdout, "# Saving results to file %s\n", filename);
sprintf(contype, "cvc-disc-P");
write_lime_contraction(work, filename, 64, 16, contype, gid, Nsave);
/*
sprintf(filename, "%sascii.%.4d.%.4d", filename_prefix2, gid, Nsave);
write_contraction(work, NULL, filename, 16, 2, 0);
*/
if(g_cart_id==0) fprintf(stdout, "# Finished working on gauge id %d\n", gid);
} /* of loop on gid */
/***********************************************
* free the allocated memory, finalize
***********************************************/
free_geometry();
fftw_free(in);
free(disc);
free(disc2);
free(work);
#ifdef MPI
fftwnd_mpi_destroy_plan(plan_p);
fftwnd_mpi_destroy_plan(plan_m);
MPI_Finalize();
#else
fftwnd_destroy_plan(plan_p);
fftwnd_destroy_plan(plan_m);
#endif
return(0);
}
int main(int argc, char **argv) {
int c, mu, status;
int filename_set = 0;
int mode = 0;
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, iix, iiy, gid, iclass;
int Thp1, nclass;
int *picount;
double *conn = (double*)NULL;
double *conn2 = (double*)NULL;
double q[4], qsqr;
int verbose = 0;
char filename[800];
double ratime, retime;
int *qid=NULL, *qcount=NULL, **qrep=NULL, **qmap=NULL;
double **qlist=NULL, qmax=0.;
int VOL3;
FILE *ofs;
fftw_complex *corrt=NULL;
fftw_complex *pi00=(fftw_complex*)NULL, *pijj=(fftw_complex*)NULL, *piavg=(fftw_complex*)NULL;
fftw_plan plan_m;
while ((c = getopt(argc, argv, "h?vf:m:q:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'm':
mode = atoi(optarg);
break;
case 'q':
qmax = atof(optarg);
fprintf(stdout, "\n# [] qmax set to %e\n", qmax);
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
/* initialize fftw, create plan with FFTW_FORWARD --- in contrast to
* FFTW_BACKWARD in e.g. avc_exact */
plan_m = fftw_create_plan(T_global, FFTW_FORWARD, FFTW_MEASURE | FFTW_IN_PLACE);
if(plan_m==NULL) {
fprintf(stderr, "Error, could not create fftw plan\n");
return(1);
}
T = T_global;
Thp1 = T/2 + 1;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
fprintf(stdout, "# [%2d] fftw parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
VOL3 = LX*LY*LZ;
status = make_qlatt_orbits_3d_parity_avg(&qid, &qcount, &qlist, &nclass, &qrep, &qmap);
if(status != 0) {
fprintf(stderr, "\n[] Error while creating h4-lists\n");
exit(4);
}
fprintf(stdout, "# [] number of classes = %d\n", nclass);
// exit(255);
/****************************************
* allocate memory for the contractions *
****************************************/
conn = (double*)calloc(32*VOLUME, sizeof(double));
if( (conn==(double*)NULL) ) {
fprintf(stderr, "could not allocate memory for contr. fields\n");
exit(3);
}
/*
conn2 = (double*)calloc(32*VOLUME, sizeof(double));
if( (conn2==(double*)NULL) ) {
fprintf(stderr, "could not allocate memory for contr. fields\n");
exit(4);
}
pi00 = (fftw_complex*)malloc(VOLUME*sizeof(fftw_complex));
if( (pi00==(fftw_complex*)NULL) ) {
fprintf(stderr, "could not allocate memory for pi00\n");
exit(2);
}
pijj = (fftw_complex*)fftw_malloc(VOLUME*sizeof(fftw_complex));
if( (pijj==(fftw_complex*)NULL) ) {
fprintf(stderr, "could not allocate memory for pijj\n");
exit(2);
}
*/
corrt = fftw_malloc(T*sizeof(fftw_complex));
if(corrt == NULL) {
fprintf(stderr, "\nError, could not alloc corrt\n");
exit(3);
}
for(gid=g_gaugeid; gid<=g_gaugeid2; gid+=g_gauge_step) {
// for(ix=0; ix<VOLUME; ix++) {pi00[ix].re = 0.; pi00[ix].im = 0.;}
// for(ix=0; ix<VOLUME; ix++) {pijj[ix].re = 0.; pijj[ix].im = 0.;}
/***********************
* read contractions *
***********************/
ratime = (double)clock() / CLOCKS_PER_SEC;
sprintf(filename, "%s.%.4d", filename_prefix, gid);
fprintf(stdout, "# Reading data from file %s\n", filename);
if(format==2) {
status = read_contraction(conn, NULL, filename, 16);
} else {
status = read_lime_contraction(conn, filename, 16, 0);
}
if(status != 0) {
fprintf(stderr, "Error: could not read from file %s; status was %d\n", filename, status);
continue;
}
/*
sprintf(filename, "%s.%.4d.%.4d", filename_prefix2, gid);
fprintf(stdout, "# Reading data from file %s\n", filename);
status = read_lime_contraction(conn2, filename, 16, 0);
if(status == 106) {
fprintf(stderr, "Error: could not read from file %s; status was %d\n", filename, status);
continue;
}
*/
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# time to read contractions %e seconds\n", retime-ratime);
/***********************
* fill the correlator *
***********************/
ratime = (double)clock() / CLOCKS_PER_SEC;
/*
for(x1=0; x1<LX; x1++) {
for(x2=0; x2<LY; x2++) {
for(x3=0; x3<LZ; x3++) {
for(x0=0; x0<T; x0++) {
iix = g_ipt[0][x1][x2][x3]*T+x0;
for(mu=1; mu<4; mu++) {
ix = _GWI(5*mu,g_ipt[x0][x1][x2][x3],VOLUME);
pijj[iix].re += ( conn[ix ] - conn2[ix ] ) * (double)Nsave / (double)(Nsave-1);
pijj[iix].im += ( conn[ix+1] - conn2[ix+1] ) * (double)Nsave / (double)(Nsave-1);
}
ix = 2*g_ipt[x0][x1][x2][x3];
pi00[iix].re += ( conn[ix ] - conn2[ix ] ) * (double)Nsave / (double)(Nsave-1);
pi00[iix].im += ( conn[ix+1] - conn2[ix+1] ) * (double)Nsave / (double)(Nsave-1);
}
}}}
*/
for(iclass=0;iclass<nclass;iclass++) {
if(qlist[iclass][0] >= qmax) {
// fprintf(stdout, "\n# [] will skip class %d, momentum squared = %f is too large\n", iclass, qlist[iclass][0]);
continue;
// } else {
// fprintf(stdout, "\n# [] processing class %d, momentum squared = %f\n", iclass, qlist[iclass][0]);
}
for(x0=0; x0<T; x0++) {
corrt[x0].re = 0.;
corrt[x0].im = 0.;
}
/*
for(x1=0;x1<VOL3;x1++) {
if(qid[x1]==iclass) {
fprintf(stdout, "# using mom %d ---> (%d, %d, %d)\n", x1, qrep[iclass][1], qrep[iclass][2], qrep[iclass][3]);
for(x0=0; x0<T; x0++) {
ix = x0*VOL3 + x1;
corrt[x0].re += conn[_GWI(5,ix,VOLUME) ] + conn[_GWI(10,ix,VOLUME) ] + conn[_GWI(15,ix,VOLUME) ];
corrt[x0].im += conn[_GWI(5,ix,VOLUME)+1] + conn[_GWI(10,ix,VOLUME)+1] + conn[_GWI(15,ix,VOLUME)+1];
}
}
}
*/
for(x0=0; x0<T; x0++) {
for(x1=0;x1<qcount[iclass];x1++) {
x2 = qmap[iclass][x1];
// if(x0==0) fprintf(stdout, "# using mom %d ---> (%d, %d, %d)\n", x2, qrep[iclass][1], qrep[iclass][2], qrep[iclass][3]);
ix = x0*VOL3 + x2;
corrt[x0].re += conn[_GWI(5,ix,VOLUME) ] + conn[_GWI(10,ix,VOLUME) ] + conn[_GWI(15,ix,VOLUME) ];
corrt[x0].im += conn[_GWI(5,ix,VOLUME)+1] + conn[_GWI(10,ix,VOLUME)+1] + conn[_GWI(15,ix,VOLUME)+1];
}
}
// fprintf(stdout, "\n\n# ------------------------------\n");
for(x0=0; x0<T; x0++) {
corrt[x0].re /= (double)T * qcount[iclass];
corrt[x0].im /= (double)T * qcount[iclass];
}
/* fftw(plan_m, 1, corrt, 1, T, (fftw_complex*)NULL, 0, 0); */
fftw_one(plan_m, corrt, NULL);
sprintf(filename, "rho.%.4d.x%.2dy%.2dz%.2d", gid, qrep[iclass][1], qrep[iclass][2], qrep[iclass][3]);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing VKVK data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f%21.12f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu, qlist[iclass][0]);
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 0, 0, 0, corrt[0].re, 0., gid);
for(x0=1; x0<(T/2); x0++) {
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 0, 0, x0,
corrt[x0].re, corrt[T-x0].re, gid);
}
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 0, 0, (T/2), corrt[T/2].re, 0., gid);
fflush(ofs);
fclose(ofs);
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# time to fill correlator %e seconds\n", retime-ratime);
} // of loop on classes
} // end of loop on gauge id
/***************************************
* free the allocated memory, finalize *
***************************************/
if(corrt != NULL) free(corrt);
free_geometry();
if(pi00 != NULL) free(pi00);
if(pijj != NULL) free(pijj);
fftw_destroy_plan(plan_m);
finalize_q_orbits(&qid, &qcount, &qlist, &qrep);
if(qmap != NULL) {
free(qmap[0]);
free(qmap);
}
if(g_cart_id == 0) {
g_the_time = time(NULL);
fprintf(stdout, "\n# [] %s# [] end of run\n", ctime(&g_the_time));
fprintf(stderr, "\n# [] %s# [] end of run\n", ctime(&g_the_time));
}
return(0);
}
int main(int argc, char **argv) {
int c, mu, nu, status, gid;
int filename_set = 0;
int l_LX_at, l_LXstart_at;
int source_location, have_source_flag = 0;
int x0, x1, x2, x3, ix;
int sx0, sx1, sx2, sx3;
int tsize = 0;
double *conn = NULL;
double *conn2 = (double*)NULL;
int verbose = 0;
char filename[800];
double ratime, retime;
FILE *ofs;
int ivec[4], idx[4], imu;
double q[4], wre, wim;
fftw_complex *inT=NULL, *outT=NULL, *inL=NULL, *outL=NULL;
fftw_plan plan_m_T, plan_m_L;
while ((c = getopt(argc, argv, "h?vf:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
// set the default values
set_default_input_values();
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# [get_corr_v2] reading input parameters from file %s\n", filename);
read_input_parser(filename);
// some checks on the input data
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
fprintf(stdout, "# [get_corr_v2] T=%d, LX=%d, LY=%d, LZ=%d\n", T_global, LX, LY, LZ);
if(g_proc_id==0) fprintf(stderr, "[get_corr_v2] Error, T and L's must be set\n");
usage();
}
// initialize MPI parameters
mpi_init(argc, argv);
/* initialize fftw, create plan with FFTW_FORWARD --- in contrast to
* FFTW_BACKWARD in e.g. avc_exact */
plan_m_T = fftw_create_plan(T_global, FFTW_FORWARD, FFTW_MEASURE);
plan_m_L = fftw_create_plan(LX, FFTW_FORWARD, FFTW_MEASURE);
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
fprintf(stdout, "# [%2d] fftw parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "[get_corr_v2] Error from init_geometry\n");
EXIT(1);
}
geometry();
/****************************************
* allocate memory for the contractions *
****************************************/
conn = (double*)calloc(32 * VOLUME, sizeof(double));
if( (conn==NULL) ) {
fprintf(stderr, "[get_corr_v2] Error, could not allocate memory for contr. fields\n");
EXIT(2);
}
conn2= (double*)calloc(8 * T, sizeof(double));
if( (conn2==NULL) ) {
fprintf(stderr, "[get_corr_v2] Error, could not allocate memory for corr.\n");
EXIT(3);
}
/*****************************************
* prepare Fourier transformation arrays *
*****************************************/
inT = (fftw_complex*)malloc(T * sizeof(fftw_complex));
inL = (fftw_complex*)malloc(LX * sizeof(fftw_complex));
outT = (fftw_complex*)malloc(T * sizeof(fftw_complex));
outL = (fftw_complex*)malloc(LX * sizeof(fftw_complex));
if( inT==NULL || inL==NULL || outT==NULL || outL==NULL ) {
fprintf(stderr, "[get_corr_v2] Error, could not allocate fftw fields\n");
EXIT(4);
}
/********************************
* determine source coordinates *
********************************/
/*
have_source_flag = (int)(g_source_location/(LX*LY*LZ)>=Tstart && g_source_location/(LX*LY*LZ)<(Tstart+T));
if(have_source_flag==1) fprintf(stdout, "# [get_corr_v2] process %2d has source location\n", g_cart_id);
sx0 = g_source_location/(LX*LY*LZ)-Tstart;
sx1 = (g_source_location%(LX*LY*LZ)) / (LY*LZ);
sx2 = (g_source_location%(LY*LZ)) / LZ;
sx3 = (g_source_location%LZ);
if(have_source_flag==1) {
fprintf(stdout, "# [get_corr_v2] local source coordinates: (%3d,%3d,%3d,%3d)\n", sx0, sx1, sx2, sx3);
source_location = g_ipt[sx0][sx1][sx2][sx3];
}
have_source_flag = 0;
*/
for(gid=g_gaugeid; gid<=g_gaugeid2; gid+=g_gauge_step) {
memset(conn, 0, 32*VOLUME*sizeof(double));
memset(conn2, 0, 8*T*sizeof(double));
/***********************
* read contractions *
***********************/
ratime = (double)clock() / CLOCKS_PER_SEC;
sprintf(filename, "%s.%.4d", filename_prefix, gid);
if(format==2 || format==3) {
status = read_contraction(conn, NULL, filename, 16);
} else if( format==0) {
status = read_lime_contraction(conn, filename, 16, 0);
}
if(status != 0) {
// fprintf(stderr, "[get_corr_v2] Error from read_contractions, status was %d\n", status);
// EXIT(5);
fprintf(stderr, "[get_corr_v2] Warning, could not read contractions for gid %d, status was %d\n", gid, status);
continue;
}
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# [get_corr_v2] time to read contractions %e seconds\n", retime-ratime);
// TEST Pi_mm
/*
fprintf(stdout, "# [get_corr_v2] Pi_mm\n");
for(x0=0; x0<T; x0++) {
for(x1=0; x1<LX; x1++) {
for(x2=0; x2<LY; x2++) {
for(x3=0; x3<LZ; x3++) {
ix = g_ipt[x0][x1][x2][x3];
for(nu=0;nu<4;nu++) {
wre = conn[_GWI(5*nu,ix,VOLUME)];
wim = conn[_GWI(5*nu,ix,VOLUME)+1];
fprintf(stdout, "\t%3d%3d%3d%3d%3d%16.7e%16.7e\n", nu, x0, x1, x2, x3, wre, wim);
}
}}}}
*/
// TEST Ward Identity
/*
fprintf(stdout, "# [get_corr_v2] Ward identity\n");
for(x0=0; x0<T; x0++) {
q[0] = 2. * sin(M_PI * (double)x0 / (double)T);
for(x1=0; x1<LX; x1++) {
q[1] = 2. * sin(M_PI * (double)x1 / (double)LX);
for(x2=0; x2<LY; x2++) {
q[2] = 2. * sin(M_PI * (double)x2 / (double)LY);
for(x3=0; x3<LZ; x3++) {
q[3] = 2. * sin(M_PI * (double)x3 / (double)LZ);
ix = g_ipt[x0][x1][x2][x3];
for(nu=0;nu<4;nu++) {
wre = q[0] * conn[_GWI(4*0+nu,ix,VOLUME)] + q[1] * conn[_GWI(4*1+nu,ix,VOLUME)] \
+ q[2] * conn[_GWI(4*2+nu,ix,VOLUME)] + q[3] * conn[_GWI(4*3+nu,ix,VOLUME)];
wim = q[0] * conn[_GWI(4*0+nu,ix,VOLUME)+1] + q[1] * conn[_GWI(4*1+nu,ix,VOLUME)+1] \
+ q[2] * conn[_GWI(4*2+nu,ix,VOLUME)+1] + q[3] * conn[_GWI(4*3+nu,ix,VOLUME)+1];
fprintf(stdout, "\t%3d%3d%3d%3d%3d%16.7e%16.7e\n", nu, x0, x1, x2, x3, wre, wim);
}
}}}}
*/
/***********************
* fill the correlator *
***********************/
ratime = (double)clock() / CLOCKS_PER_SEC;
for(mu=0; mu<4; mu++) {
ivec[0] = (0 + mu)%4;
ivec[1] = (1 + mu)%4;
ivec[2] = (2 + mu)%4;
ivec[3] = (3 + mu)%4;
idx[ivec[1]] = 0;
idx[ivec[2]] = 0;
idx[ivec[3]] = 0;
tsize = (mu==0) ? T : LX;
for(x0=0; x0<tsize; x0++) {
idx[ivec[0]] = x0;
for(nu=1; nu<4; nu++) {
imu = (mu+nu) % 4;
// ix = get_indexf(idx[0],idx[1],idx[2],idx[3],imu,imu);
ix = _GWI(5*imu, g_ipt[idx[0]][idx[1]][idx[2]][idx[3]], VOLUME);
// TEST
//fprintf(stdout, "\tPi_%d_%d x0=%3d mu=%3d\tix=%8d\n", mu, mu, x0, imu, ix);
conn2[2*(mu*T+x0) ] += conn[ix ];
conn2[2*(mu*T+x0)+1] += conn[ix+1];
}
}
}
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# [get_corr_v2] time to fill correlator %e seconds\n", retime-ratime);
// TEST
/*
fprintf(stdout, "# [get_corr_v2] correlators\n");
for(mu=0;mu<4;mu++) {
for(x0=0; x0<T; x0++) {
fprintf(stdout, "\t%3d%3d%25.16e%25.16e\n", mu, x0, conn2[2*(mu*T+x0)], conn2[2*(mu*T+x0)+1]);
}}
*/
/*****************************************
* reverse Fourier transformation
*****************************************/
ratime = (double)clock() / CLOCKS_PER_SEC;
memcpy((void*)inT, (void*)conn2, 2*T*sizeof(double));
fftw_one(plan_m_T, inT, outT);
for(ix=0; ix<T; ix++) {
conn2[2*ix ] = outT[ix].re / (double)T;
conn2[2*ix+1] = outT[ix].im / (double)T;
}
for(mu=1; mu<4; mu++) {
memcpy((void*)inL, (void*)(conn2+2*mu*T), 2*LX*sizeof(double));
fftw_one(plan_m_L, inL, outL);
for(ix=0; ix<LX; ix++) {
conn2[2*(mu*T+ix) ] = outL[ix].re / (double)LX;
conn2[2*(mu*T+ix)+1] = outL[ix].im / (double)LX;
}
}
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# [get_corr_v2] time for Fourier transform %e seconds\n", retime-ratime);
ratime = (double)clock() / CLOCKS_PER_SEC;
sprintf(filename, "v0v0_corr.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "[get_corr_v2] Error, could not open file %s for writing\n", filename);
EXIT(6);
}
x0 = 0;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*x0], 0., gid);
for(x0=1; x0<T/2; x0++) {
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*x0], conn2[2*(T-x0)], gid);
}
x0 = T / 2;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*x0], 0., gid);
fclose(ofs);
for(mu=1; mu<4; mu++) {
sprintf(filename, "v%dv%d_corr.%.4d", mu, mu, gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "[get_corr_v2] Error, could not open file %s for writing\n", filename);
EXIT(7);
}
x0 = 0;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*(mu*T+x0)], 0., gid);
for(x0=1; x0<LX/2; x0++) {
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*(mu*T+x0)], conn2[2*(mu*T+ LX-x0)], gid);
}
x0 = LX / 2;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*(mu*T+x0)], 0., gid);
fclose(ofs);
}
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# [get_corr_v2] time to write correlator %e seconds\n", retime-ratime);
} // of loop on gid
/***************************************
* free the allocated memory, finalize *
***************************************/
free_geometry();
fftw_free(inT);
fftw_free(outT);
fftw_free(inL);
fftw_free(outL);
free(conn);
free(conn2);
fftw_destroy_plan(plan_m_T);
fftw_destroy_plan(plan_m_L);
fprintf(stdout, "# [get_corr_v2] %s# [get_corr_v2] end of run\n", ctime(&g_the_time));
fflush(stdout);
fprintf(stderr, "[get_corr_v2] %s[get_corr_v2] end of run\n", ctime(&g_the_time));
fflush(stderr);
return(0);
}
int main(int argc, char **argv) {
int c, mu;
int filename_set = 0;
int l_LX_at, l_LXstart_at;
int source_location, have_source_flag = 0;
int x0, ix;
int sx0, sx1, sx2, sx3;
int check_WI=0;
double *conn = (double*)NULL;
double *conn2 = (double*)NULL;
int verbose = 0;
char filename[800];
double ratime, retime;
FILE *ofs;
/**************************
* variables for WI check */
int x1, x2, x3, nu;
double wre, wim, q[4];
/**************************/
fftw_complex *in=(fftw_complex*)NULL, *out=(fftw_complex*)NULL;
fftw_plan plan_m;
while ((c = getopt(argc, argv, "wh?vf:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'w':
check_WI = 1;
fprintf(stdout, "# [get_rho_corr] check WI in momentum space\n");
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
set_default_input_values();
if(filename_set==0) strcpy(filename, "cvc.input");
// set the default values
set_default_input_values();
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# [get_rho_corr] reading input parameters from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
/* initialize fftw, create plan with FFTW_FORWARD --- in contrast to
* FFTW_BACKWARD in e.g. avc_exact */
plan_m = fftw_create_plan(T_global, FFTW_FORWARD, FFTW_MEASURE);
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
fprintf(stdout, "# [%2d] fftw parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
/****************************************
* allocate memory for the contractions *
****************************************/
conn = (double*)calloc(2 * 16 * VOLUME, sizeof(double));
if( (conn==(double*)NULL) ) {
fprintf(stderr, "could not allocate memory for contr. fields\n");
exit(3);
}
for(ix=0; ix<32*VOLUME; ix++) conn[ix] = 0.;
conn2= (double*)calloc(2 * T, sizeof(double));
if( (conn2==(double*)NULL) ) {
fprintf(stderr, "could not allocate memory for corr.\n");
exit(2);
}
for(ix=0; ix<2*T; ix++) conn2[ix] = 0.;
/*****************************************
* prepare Fourier transformation arrays *
*****************************************/
in = (fftw_complex*)malloc(T*sizeof(fftw_complex));
out = (fftw_complex*)malloc(T*sizeof(fftw_complex));
if( (in==(fftw_complex*)NULL) || (out==(fftw_complex*)NULL) ) exit(4);
/********************************
* determine source coordinates *
********************************/
have_source_flag = (int)(g_source_location/(LX*LY*LZ)>=Tstart && g_source_location/(LX*LY*LZ)<(Tstart+T));
if(have_source_flag==1) fprintf(stdout, "process %2d has source location\n", g_cart_id);
sx0 = g_source_location/(LX*LY*LZ)-Tstart;
sx1 = (g_source_location%(LX*LY*LZ)) / (LY*LZ);
sx2 = (g_source_location%(LY*LZ)) / LZ;
sx3 = (g_source_location%LZ);
if(have_source_flag==1) {
fprintf(stdout, "local source coordinates: (%3d,%3d,%3d,%3d)\n", sx0, sx1, sx2, sx3);
source_location = g_ipt[sx0][sx1][sx2][sx3];
}
have_source_flag = 0;
/***********************
* read contractions *
***********************/
ratime = (double)clock() / CLOCKS_PER_SEC;
// read_contraction(conn, (int*)NULL, filename_prefix, 16);
read_lime_contraction(conn, filename_prefix, 16, 0);
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "time to read contractions %e seconds\n", retime-ratime);
// TEST Ward Identity
if(check_WI) {
fprintf(stdout, "# [get_corr_v5] Ward identity\n");
sprintf(filename, "WI.%.4d", Nconf);
ofs = fopen(filename, "w");
if(ofs == NULL) exit(32);
for(x0=0; x0<T; x0++) {
q[0] = 2. * sin(M_PI * (double)x0 / (double)T);
for(x1=0; x1<LX; x1++) {
q[1] = 2. * sin(M_PI * (double)x1 / (double)LX);
for(x2=0; x2<LY; x2++) {
q[2] = 2. * sin(M_PI * (double)x2 / (double)LY);
for(x3=0; x3<LZ; x3++) {
q[3] = 2. * sin(M_PI * (double)x3 / (double)LZ);
ix = g_ipt[x0][x1][x2][x3];
for(nu=0;nu<4;nu++) {
wre = q[0] * conn[_GWI(4*0+nu,ix,VOLUME)] + q[1] * conn[_GWI(4*1+nu,ix,VOLUME)] \
+ q[2] * conn[_GWI(4*2+nu,ix,VOLUME)] + q[3] * conn[_GWI(4*3+nu,ix,VOLUME)];
wim = q[0] * conn[_GWI(4*0+nu,ix,VOLUME)+1] + q[1] * conn[_GWI(4*1+nu,ix,VOLUME)+1] \
+ q[2] * conn[_GWI(4*2+nu,ix,VOLUME)+1] + q[3] * conn[_GWI(4*3+nu,ix,VOLUME)+1];
fprintf(ofs, "\t%3d%3d%3d%3d%3d%16.7e%16.7e\n", nu, x0, x1, x2, x3, wre, wim);
}
}}}}
fclose(ofs);
}
/***********************
* fill the correlator *
***********************/
ratime = (double)clock() / CLOCKS_PER_SEC;
for(x0=0; x0<T; x0++) {
for(mu=1; mu<4; mu++) {
ix = get_indexf(x0,0,0,0,mu,mu);
fprintf(stdout, "x0=%3d, mu=%3d\tix=%8d\n", x0, mu, ix);
conn2[2*x0 ] += conn[ix ];
conn2[2*x0+1] += conn[ix+1];
}
}
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "time to fill correlator %e seconds\n", retime-ratime);
/********************************
* test: print correl to stdout *
********************************/
for(x0=0; x0<T; x0++) {
fprintf(stdout, "%3d%25.16e%25.16e\n", x0, conn2[2*x0], conn[2*x0+1]);
}
/*****************************************
* do the reverse Fourier transformation *
*****************************************/
ratime = (double)clock() / CLOCKS_PER_SEC;
memcpy((void*)in, (void*)conn2, 2*T*sizeof(double));
fftw_one(plan_m, in, out);
for(ix=0; ix<T; ix++) {
conn2[2*ix ] = out[ix].re / (double)T;
conn2[2*ix+1] = out[ix].im / (double)T;
}
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "time for Fourier transform %e seconds\n", retime-ratime);
ratime = (double)clock() / CLOCKS_PER_SEC;
sprintf(filename, "rho_corr.%.4d", Nconf);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "could not open file %s for writing\n", filename);
exit(5);
}
//for(x0=0; x0<T; x0++) {
// fprintf(ofs, "%3d%25.16e%25.16e\n", x0, conn2[2*x0], conn2[2*x0+1]);
//}
x0 = 0;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*x0], 0., Nconf);
for(x0=1; x0<T/2; x0++) {
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*x0], conn2[2*(T-x0)], Nconf);
}
x0 = T/2;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*x0], 0., Nconf);
fclose(ofs);
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "time to write correlator %e seconds\n", retime-ratime);
/***************************************
* free the allocated memory, finalize *
***************************************/
free_geometry();
fftw_free(in);
fftw_free(out);
free(conn);
free(conn2);
fftw_destroy_plan(plan_m);
return(0);
}
int main(int argc, char **argv) {
int c, i, mu, nu;
int count = 0;
int filename_set = 0;
//int use_real_part = 1;
int ix, iix;
int sid, status, gid, it, ir, it2;
double *disc = (double*)NULL;
double *work = (double*)NULL;
double *bias = (double*)NULL;
//double fnorm;
int verbose = 0;
unsigned int VOL3;
char filename[100];
double ratime, retime;
double *tmp = NULL;
complex w;
FILE *ofs = NULL;
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "h?vf:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
g_the_time = time(NULL);
fprintf(stdout, "# [pi_ud_tp0] using global time stamp %s", ctime(&g_the_time));
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
fprintf(stdout, "# [pi_ud_tp0] **************************************************\n");
fprintf(stdout, "# [pi_ud_tp0] pi_ud_p\n");
fprintf(stdout, "# [pi_ud_tp0] **************************************************\n\n");
/*********************************
* initialize MPI parameters
*********************************/
mpi_init(argc, argv);
#ifdef MPI
if(T==0) {
fprintf(stderr, "[%2d] local T is zero; exit\n", g_cart_id);
EXIT(2);
}
#endif
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
EXIT(1);
}
geometry();
/****************************************
* allocate memory for the contractions
****************************************/
disc = (double*)calloc(16*VOLUME, sizeof(double));
if( disc == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc\n");
EXIT(3);
}
work = (double*)calloc(2*T_global, sizeof(double));
if( work == (double*)NULL ) {
fprintf(stderr, "[pi_ud_tp0] could not allocate memory for work\n");
EXIT(5);
}
bias = (double*)calloc(2*T_global, sizeof(double));
if( bias == (double*)NULL ) {
fprintf(stderr, "[pi_ud_tp0] could not allocate memory for bias\n");
EXIT(6);
}
tmp = (double*)calloc(2*T_global, sizeof(double));
if( tmp == (double*)NULL ) {
fprintf(stderr, "[pi_ud_tp0] could not allocate memory for tmp\n");
EXIT(8);
}
/***********************************************
* start loop on gauge id.s
***********************************************/
for(gid=g_gaugeid; gid<=g_gaugeid2; gid+=g_gauge_step) {
memset(work, 0, 2*T_global*sizeof(double));
memset(bias, 0, 2*T_global*sizeof(double));
count = 0;
/***********************************************
* start loop on source id.s
***********************************************/
for(sid=g_sourceid; sid<=g_sourceid2; sid+=g_sourceid_step) {
memset(disc, 0, 16*VOLUME*sizeof(double));
ratime = CLOCK;
sprintf(filename, "jc_ud_x.%.4d.%.4d", gid, sid);
status = read_lime_contraction(disc, filename, 4, 0);
if(status!=0) {
fprintf(stderr, "Error, could not read contraction data from file %s\n", filename);
EXIT(7);
}
retime = CLOCK;
if(g_cart_id==0) fprintf(stdout, "# time to read contractions: %e seconds\n", retime-ratime);
count++;
ratime = CLOCK;
// add current to sum
for(it=0; it<T; it++) {
tmp[2*it ] = 0.;
tmp[2*it+1] = 0.;
for(iix=0; iix<VOL3; iix++) {
ix = it * VOL3 + iix;
tmp[2*it ] += disc[_GWI(1,ix,VOLUME) ] + disc[_GWI(2,ix,VOLUME) ] + disc[_GWI(3,ix,VOLUME) ];
tmp[2*it+1] += disc[_GWI(1,ix,VOLUME)+1] + disc[_GWI(2,ix,VOLUME)+1] + disc[_GWI(3,ix,VOLUME)+1];
}
}
for(it=0; it<2*T_global; it++) { work[it] += tmp[it]; }
// add to bias
for(it=0; it<T_global; it++) {
for(ir=0; ir<T_global; ir++) {
it2 = (it + ir ) % T_global;
_co_eq_co_ti_co( &w, (complex*)&(tmp[2*it2]), (complex*)&(tmp[2*it]) );
bias[2*it ] += w.re;
bias[2*it+1] += w.im;
}}
retime = CLOCK;
if(g_cart_id==0) fprintf(stdout, "# [pi_ud_tp0] time to calculate contractions: %e seconds\n", retime-ratime);
if(count==Nsave) {
memset(disc, 0, 2*T_global*sizeof(double));
for(it=0; it<T_global; it++) {
for(ir=0; ir<T_global; ir++) {
it2 = (it + ir ) % T_global;
_co_eq_co_ti_co( &w, (complex*)&(work[2*it2]), (complex*)&(work[2*it]) );
disc[2*it ] += w.re;
disc[2*it+1] += w.im;
}}
for(it=0; it<2*T_global; it++) {
disc[it] -= bias[it];
}
sprintf(filename, "pi_ud_t.%.4d.%.4d", gid, count);
ofs = fopen(filename, "w");
if(ofs == NULL) {
fprintf(stderr, "[pi_ud_tp0] Error, could not open file %s for writing\n", filename);
EXIT(8);
}
fprintf(ofs, "# [pi_ud_tp0] results for disc. t-dependent correlator at zero spatial momentum\n# %s", ctime(&g_the_time));
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 11, 1, 0, disc[0], 0., Nconf);
for(it=1; it<T_global/2; it++) {
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 11, 1, it, disc[it], disc[2*(T_global-it)], Nconf);
}
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 11, 1, T_global/2, disc[T_global/2], 0., Nconf);
fclose(ofs);
retime = CLOCK;
if(g_cart_id==0) fprintf(stdout, "# [pi_ud_tp0] time to save cvc results: %e seconds\n", retime-ratime);
} // of count % Nsave == 0
} // of loop on sid
} // of loop on gid
/***********************************************
* free the allocated memory, finalize
***********************************************/
free_geometry();
if(disc != NULL) free(disc);
if(work != NULL) free(work);
if(bias != NULL) free(bias);
if(tmp != NULL) free(tmp);
if(g_cart_id == 0) {
fprintf(stdout, "# [pi_ud_tp0] %s# [pi_ud_tp0] end of run\n", ctime(&g_the_time));
fflush(stdout);
fprintf(stderr, "# [pi_ud_tp0] %s# [pi_ud_tp0] end of run\n", ctime(&g_the_time));
fflush(stderr);
}
#ifdef MPI
MPI_Finalize();
#endif
return(0);
}
int main(int argc, char **argv) {
int Thm1;
int c, i, mu, nthreads;
int count = 0;
int filename_set = 0;
int l_LX_at, l_LXstart_at;
int x0, x1, y0;
int ix, iy, idx1, idx2;
int VOL3;
int sid1, sid2, status, gid;
size_t nprop=0;
double *data=NULL, *data2=NULL, *data3=NULL;
double fnorm;
double *mom2=NULL, *mom4=NULL;
char filename[100];
double ratime, retime;
FILE *ofs=NULL;
/****************************************
* initialize the distance vectors
****************************************/
while ((c = getopt(argc, argv, "h?f:")) != -1) {
switch (c) {
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
fprintf(stdout, "\n**************************************************\n");
fprintf(stdout, "* jc_corr\n");
fprintf(stdout, "**************************************************\n\n");
T = T_global;
Thm1 = T / 2 - 1;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
VOL3 = LX*LY*LZ;
fprintf(stdout, "# [%2d] parameters:\n"\
"# T = %3d\n"\
"# Tstart = %3d\n"\
"# l_LX_at = %3d\n"\
"# l_LXstart_at = %3d\n"\
"# FFTW_LOC_VOLUME = %3d\n",
g_cart_id, T, Tstart, l_LX_at, l_LXstart_at, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
/****************************************
* allocate memory for the contractions
****************************************/
nprop = (size_t)(g_sourceid2 - g_sourceid) / (size_t)g_sourceid_step + 1;
fprintf(stdout, "\n# [jc_corr] number of stoch. propagators = %lu\n", nprop);
data = (double*)calloc(8*FFTW_LOC_VOLUME, sizeof(double));
if( data==NULL ) {
fprintf(stderr, "could not allocate memory for data\n");
exit(3);
}
/* nprop * T * 3(i=1,2,3) * 2(real and imaginary part) */
data2 = (double*)calloc(nprop*T*6, sizeof(double));
if( data2==NULL ) {
fprintf(stderr, "could not allocate memory for data2\n");
exit(3);
}
data3 = (double*)calloc(2*T, sizeof(double));
if( data3==NULL ) {
fprintf(stderr, "could not allocate memory for data3\n");
exit(3);
}
fnorm = 1. / ( (double)nprop * (double)(nprop-1) * (double)(LX*LY*LZ) );
fprintf(stdout, "\n# [jc_corr] fnorm = %25.16e\n", fnorm);
for(ix=0; ix<nprop*T; ix++) data2[ix] = 0.;
/***********************************************
* start loop on gauge id.s
***********************************************/
for(gid=g_gaugeid; gid<=g_gaugeid2; gid++) {
/* calculate the t-dependent current at zero spatial momentum */
for(sid1=0; sid1<nprop; sid1++) {
sprintf(filename, "jc_ud_x.%.4d.%.4d", gid, g_sourceid + sid1*g_sourceid_step);
if(read_lime_contraction(data, filename, 4, 0) != 0) {
fprintf(stderr, "\n[jc_corr] Error, could not read field no. %d\n", sid1);
exit(15);
}
for(mu=0;mu<3;mu++) {
for(x0=0;x0<T;x0++) {
ix = g_ipt[x0][0][0][0];
ix = _GWI(5*(mu+1), ix, VOLUME);
for(iy=0;iy<VOL3;iy++) {
data2[2*(sid1*3*T + mu*T + x0) ] += data[ix + 2*iy ];
data2[2*(sid1*3*T + mu*T + x0)+1] += data[ix + 2*iy+1];
}
}
}
}
/***********************************************
* calculate the correlator
* - remember: x1 is the time difference of the correlator,
* x0 and y0 are the time coordinates of the currents
***********************************************/
ratime = (double)clock() / CLOCKS_PER_SEC;
for(i=0;i<2*T; i++) data3[i] = 0.;
for(sid1=0; sid1<nprop-1; sid1++) {
for(sid2=sid1+1; sid2<nprop; sid2++) {
for(y0=0;y0<T;y0++) {
for(x1=0;x1<T;x1++) {
x0 = (y0 + x1) % T;
// first component
idx1 = 2 * ( sid1*3*T + 0*T + y0 );
idx2 = 2 * ( sid2*3*T + 0*T + x0 );
// real part of the product
data3[2*x1 ] += data2[idx1 ] * data2[idx2 ] - data2[idx1+1]*data2[idx2+1];
// imaginary part of the product
data3[2*x1+1] += data2[idx1+1] * data2[idx2 ] + data2[idx1 ]*data2[idx2+1];
// second component
idx1 = 2 * ( sid1*3*T + 1*T + y0 );
idx2 = 2 * ( sid2*3*T + 1*T + x0 );
// real part of the product
data3[2*x1 ] += data2[idx1 ] * data2[idx2 ] - data2[idx1+1]*data2[idx2+1];
// imaginary part of the product
data3[2*x1+1] += data2[idx1+1] * data2[idx2 ] + data2[idx1 ]*data2[idx2+1];
// third component
idx1 = 2 * ( sid1*3*T + 2*T + y0 );
idx2 = 2 * ( sid2*3*T + 2*T + x0 );
// real part of the product
data3[2*x1 ] += data2[idx1 ] * data2[idx2 ] - data2[idx1+1]*data2[idx2+1];
// imaginary part of the product
data3[2*x1+1] += data2[idx1+1] * data2[idx2 ] + data2[idx1 ]*data2[idx2+1];
}
}
}} // of sid2 and sid1
// normalization
for(x0=0;x0<2*T;x0++) { data3[x0] *= fnorm; }
for(x0=0;x0<T/2-1;x0++) {
mom2[x0] = 0.;
mom4[x0] = 0.;
}
for(x0=1;x0<T/2;x0++) {
if(x0==1) {
mom2[0] = ( data3[2*x0] + data3[2*(T-x0)] ) * (double)(x0*x0);
mom4[0] = ( data3[2*x0] + data3[2*(T-x0)] ) * (double)(x0*x0*x0*x0);
} else {
mom2[x0-1] = mom2[x0-2] + ( data3[2*x0] + data3[2*(T-x0)] ) * (double)(x0*x0);
mom4[x0-1] = mom4[x0-2] + ( data3[2*x0] + data3[2*(T-x0)] ) * (double)(x0*x0*x0*x0);
}
}
for(i=0;i<Thm1;i++) mom2[i] /= 6.;
for(i=0;i<Thm1;i++) mom4[i] /= 72.;
/************************************************
* save results in position space
************************************************/
sprintf(filename, "pi_ud_tp0.%4d.%.4d", gid, nprop);
ofs = fopen(filename, "w");
if (ofs==NULL) {
fprintf(stderr, "\n[jc_corr] Error, could not open file %s for writing\n", filename);
exit(9);
}
fprintf(ofs, "0 1 0%25.16e%25.16e%d\n", data3[0], 0., gid);
for(x0=1;x0<=Thm1;x0++)
fprintf(ofs, "0 1 %2d%25.16e%25.16e%d\n", x0, data3[x0], data3[T-x0], gid);
fprintf(ofs, "0 1 %2d%25.16e%25.16e%d\n", x0, data3[x0], 0., gid);
fclose(ofs);
sprintf(filename, "pi_ud_mom.%4d.%.4d", gid, nprop);
ofs = fopen(filename, "w");
if (ofs==NULL) {
fprintf(stderr, "\n[jc_corr] Error, could not open file %s for writing\n", filename);
exit(9);
}
for(i=0;i<Thm1;i++)
fprintf(ofs, "%2d%25.16e%25.16e\n", i, mom2[i], mom4[i]);
fclose(ofs);
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id == 0) fprintf(stdout, "# time for building correl.: %e seconds\n", retime-ratime);
} /* of loop on gid */
/***********************************************
* free the allocated memory, finalize
***********************************************/
free_geometry();
free(data);
free(data2);
free(data3);
free(mom2);
free(mom4);
return(0);
}
int main(int argc, char **argv) {
int c, mu, status;
int filename_set = 0;
int mode = 0;
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, iix, iiy, gid;
int Thp1, nclass;
int *oh_count=(int*)NULL, *oh_id=(int*)NULL, oh_nc;
int *picount;
double *conn = (double*)NULL;
double *conn2 = (double*)NULL;
double **oh_val=(double**)NULL;
double q[4], qsqr;
int verbose = 0;
char filename[800];
double ratime, retime;
FILE *ofs;
fftw_complex *corrt=NULL;
fftw_complex *pi00=(fftw_complex*)NULL, *pijj=(fftw_complex*)NULL, *piavg=(fftw_complex*)NULL;
fftw_plan plan_m;
while ((c = getopt(argc, argv, "h?vf:m:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'm':
mode = atoi(optarg);
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
/* initialize fftw, create plan with FFTW_FORWARD --- in contrast to
* FFTW_BACKWARD in e.g. avc_exact */
plan_m = fftw_create_plan(T_global, FFTW_FORWARD, FFTW_MEASURE | FFTW_IN_PLACE);
if(plan_m==NULL) {
fprintf(stderr, "Error, could not create fftw plan\n");
return(1);
}
T = T_global;
Thp1 = T/2 + 1;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
fprintf(stdout, "# [%2d] fftw parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
/****************************************
* allocate memory for the contractions *
****************************************/
conn = (double*)calloc(32*VOLUME, sizeof(double));
if( (conn==(double*)NULL) ) {
fprintf(stderr, "could not allocate memory for contr. fields\n");
exit(3);
}
/*
conn2 = (double*)calloc(32*VOLUME, sizeof(double));
if( (conn2==(double*)NULL) ) {
fprintf(stderr, "could not allocate memory for contr. fields\n");
exit(4);
}
pi00 = (fftw_complex*)malloc(VOLUME*sizeof(fftw_complex));
if( (pi00==(fftw_complex*)NULL) ) {
fprintf(stderr, "could not allocate memory for pi00\n");
exit(2);
}
pijj = (fftw_complex*)fftw_malloc(VOLUME*sizeof(fftw_complex));
if( (pijj==(fftw_complex*)NULL) ) {
fprintf(stderr, "could not allocate memory for pijj\n");
exit(2);
}
*/
corrt = fftw_malloc(T*sizeof(fftw_complex));
for(gid=g_gaugeid; gid<=g_gaugeid2; gid++) {
// for(ix=0; ix<VOLUME; ix++) {pi00[ix].re = 0.; pi00[ix].im = 0.;}
// for(ix=0; ix<VOLUME; ix++) {pijj[ix].re = 0.; pijj[ix].im = 0.;}
/***********************
* read contractions *
***********************/
ratime = (double)clock() / CLOCKS_PER_SEC;
sprintf(filename, "%s", filename_prefix);
fprintf(stdout, "# Reading data from file %s\n", filename);
status = read_lime_contraction(conn, filename, 16, 0);
if(status == 106) {
fprintf(stderr, "Error: could not read from file %s; status was %d\n", filename, status);
continue;
}
/*
sprintf(filename, "%s.%.4d.%.4d", filename_prefix2, gid);
fprintf(stdout, "# Reading data from file %s\n", filename);
status = read_lime_contraction(conn2, filename, 16, 0);
if(status == 106) {
fprintf(stderr, "Error: could not read from file %s; status was %d\n", filename, status);
continue;
}
*/
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# time to read contractions %e seconds\n", retime-ratime);
/***********************
* fill the correlator *
***********************/
ratime = (double)clock() / CLOCKS_PER_SEC;
/*
for(x1=0; x1<LX; x1++) {
for(x2=0; x2<LY; x2++) {
for(x3=0; x3<LZ; x3++) {
for(x0=0; x0<T; x0++) {
iix = g_ipt[0][x1][x2][x3]*T+x0;
for(mu=1; mu<4; mu++) {
ix = _GWI(5*mu,g_ipt[x0][x1][x2][x3],VOLUME);
pijj[iix].re += ( conn[ix ] - conn2[ix ] ) * (double)Nsave / (double)(Nsave-1);
pijj[iix].im += ( conn[ix+1] - conn2[ix+1] ) * (double)Nsave / (double)(Nsave-1);
}
ix = 2*g_ipt[x0][x1][x2][x3];
pi00[iix].re += ( conn[ix ] - conn2[ix ] ) * (double)Nsave / (double)(Nsave-1);
pi00[iix].im += ( conn[ix+1] - conn2[ix+1] ) * (double)Nsave / (double)(Nsave-1);
}
}}}
*/
for(x0=0; x0<T; x0++) {
ix = g_ipt[x0][0][0][0];
corrt[x0].re = conn[_GWI(5,ix,VOLUME) ] + conn[_GWI(10,ix,VOLUME) ] + conn[_GWI(15,ix,VOLUME) ];
corrt[x0].im = conn[_GWI(5,ix,VOLUME)+1] + conn[_GWI(10,ix,VOLUME)+1] + conn[_GWI(15,ix,VOLUME)+1];
corrt[x0].re /= (double)T;
corrt[x0].im /= (double)T;
}
/* fftw(plan_m, 1, corrt, 1, T, (fftw_complex*)NULL, 0, 0); */
fftw_one(plan_m, corrt, NULL);
sprintf(filename, "rho.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing VKVK data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu);
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 0, 0, 0, corrt[0].re, 0., gid);
for(x0=1; x0<(T/2); x0++) {
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 0, 0, x0,
corrt[x0].re, corrt[T-x0].re, gid);
}
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 0, 0, (T/2), corrt[T/2].re, 0., gid);
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# time to fill correlator %e seconds\n", retime-ratime);
#ifdef _UNDEF
free(conn);
/* free(conn2); */
/********************************
* test: print correl to stdout *
********************************/
/*
fprintf(stdout, "\n\n# ***************** pijj *****************\n");
for(ix=0; ix<LX*LY*LZ; ix++) {
iix = ix*T;
for(x0=0; x0<T; x0++) {
fprintf(stdout, "%6d%3d%25.16e%25.16e\n", ix, x0, pijj[iix+x0].re, pijj[iix+x0].im);
}
}
fprintf(stdout, "\n\n# ***************** pi00 *****************\n");
for(ix=0; ix<LX*LY*LZ; ix++) {
iix = ix*T;
for(x0=0; x0<T; x0++) {
fprintf(stdout, "%6d%3d%25.16e%25.16e\n", ix, x0, pi00[iix+x0].re, pi00[iix+x0].im);
}
}
*/
/*****************************************
* do the reverse Fourier transformation *
*****************************************/
ratime = (double)clock() / CLOCKS_PER_SEC;
fftw(plan_m, LX*LY*LZ, pi00, 1, T, (fftw_complex*)NULL, 0, 0);
fftw(plan_m, LX*LY*LZ, pijj, 1, T, (fftw_complex*)NULL, 0, 0);
for(ix=0; ix<VOLUME; ix++) {
pi00[ix].re /= (double)T; pi00[ix].im /= (double)T;
pijj[ix].re /= 3.*(double)T; pijj[ix].im /= 3.*(double)T;
}
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# time for Fourier transform %e seconds\n", retime-ratime);
/*****************************************
* write to file
*****************************************/
ratime = (double)clock() / CLOCKS_PER_SEC;
sprintf(filename, "pi00.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing pi00-data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu);
for(x1=0; x1<LX; x1++) {
for(x2=0; x2<LY; x2++) {
for(x3=0; x3<LZ; x3++) {
ix = g_ipt[0][x1][x2][x3]*T;
/* fprintf(ofs, "# px=%3d, py=%3d, pz=%3d\n", x1, x2, x3); */
for(x0=0; x0<T; x0++) {
/* fprintf(ofs, "%3d%25.16e%25.16e\n", x0, pi00[ix+x0].re, pi00[ix+x0].im); */
fprintf(ofs, "%3d%3d%3d%3d%25.16e%25.16e\n", x1, x2, x3, x0, pi00[ix+x0].re, pi00[ix+x0].im);
}
}}}
fclose(ofs);
sprintf(filename, "pijj.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing pijj-data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu);
for(x1=0; x1<LX; x1++) {
for(x2=0; x2<LY; x2++) {
for(x3=0; x3<LZ; x3++) {
ix = g_ipt[0][x1][x2][x3]*T;
/* fprintf(ofs, "# px=%3d, py=%3d, pz=%3d\n", x1, x2, x3); */
for(x0=0; x0<T; x0++) {
/* fprintf(ofs, "%3d%25.16e%25.16e\n", x0, pijj[ix+x0].re, pijj[ix+x0].im); */
fprintf(ofs, "%3d%3d%3d%3d%25.16e%25.16e\n", x1, x2, x3, x0, pijj[ix+x0].re, pijj[ix+x0].im);
}
}}}
fclose(ofs);
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# time to write correlator %e seconds\n", retime-ratime);
/*
if(mode==0) {
ratime = (double)clock() / CLOCKS_PER_SEC;
if( (picount = (int*)malloc(VOLUME*sizeof(int))) == (int*)NULL) exit(110);
sprintf(filename, "corr.00.mom");
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
for(ix=0; ix<VOLUME; ix++) picount[ix] = 0;
for(x1=0; x1<LX; x1++) {
q[1] = 2. * sin(M_PI * (double)x1 / (double)LX);
for(x2=0; x2<LY; x2++) {
q[2] = 2. * sin(M_PI * (double)x2 / (double)LY);
for(x3=0; x3<LZ; x3++) {
q[3] = 2. * sin(M_PI * (double)x3 / (double)LZ);
qsqr = q[1]*q[1] + q[2]*q[2] + q[3]*q[3];
if( qsqr>=g_qhatsqr_min-_Q2EPS && qsqr<= g_qhatsqr_max+_Q2EPS ) {
ix = g_ipt[0][x1][x2][x3];
picount[ix] = 1;
fprintf(ofs, "%3d%3d%3d%6d%25.16e\n", x1, x2, x3, ix, qsqr);
}
}}}
fclose(ofs);
sprintf(filename, "corr_00.00.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing corr_00-data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu);
for(ix=0; ix<VOLUME; ix++) {
if(picount[ix]>0) {
for(x0=0; x0<T; x0++) {
fprintf(ofs, "%3d%3d%25.16e%25.16e\n", ix, x0, pi00[ix*T+x0].re, pi00[ix*T+x0].im);
}
}
}
fclose(ofs);
sprintf(filename, "corr_jj.00.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing corr_jj-data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu);
for(ix=0; ix<VOLUME; ix++) {
if(picount[ix]>0) {
for(x0=0; x0<T; x0++) {
fprintf(ofs, "%3d%3d%25.16e%25.16e\n", ix, x0, pijj[ix*T+x0].re, pijj[ix*T+x0].im);
}
}
}
fclose(ofs);
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# time for O_h averaging %e seconds\n", retime-ratime);
free(picount);
} else if(mode==1) {
ratime = (double)clock() / CLOCKS_PER_SEC;
if( (picount = (int*)malloc(VOLUME*sizeof(int))) == (int*)NULL) exit(110);
sprintf(filename, "corr.01.mom");
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
if( (picount = (int*)malloc(VOLUME*sizeof(int))) == (int*)NULL) exit(110);
for(ix=0; ix<VOLUME; ix++) picount[ix] = 0;
for(x1=0; x1<LX; x1++) {
q[1] = 2. * M_PI * (double)x1 / (double)LX;
for(x2=0; x2<LY; x2++) {
q[2] = 2. * M_PI * (double)x2 / (double)LY;
for(x3=0; x3<LZ; x3++) {
q[3] = 2. * M_PI * (double)x3 / (double)LZ;
qsqr = q[1]*q[1] + q[2]*q[2] + q[3]*q[3];
if( qsqr>=g_qhatsqr_min-_Q2EPS && qsqr<= g_qhatsqr_max+_Q2EPS ) {
ix = g_ipt[0][x1][x2][x3];
picount[ix] = 1;
fprintf(ofs, "%3d%3d%3d%6d%25.16e\n", x1, x2, x3, ix, qsqr);
}
}}}
fclose(ofs);
sprintf(filename, "corr_00.01.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing corr_01-data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu);
for(ix=0; ix<VOLUME; ix++) {
if(picount[ix]>0) {
for(x0=0; x0<T; x0++) {
fprintf(ofs, "%3d%3d%25.16e%25.16e\n", ix, x0, pi00[ix*T+x0].re, pi00[ix*T+x0].im);
}
}
}
fclose(ofs);
sprintf(filename, "corr_jj.01.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing corr_jj-data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu);
for(ix=0; ix<VOLUME; ix++) {
if(picount[ix]>0) {
for(x0=0; x0<T; x0++) {
fprintf(ofs, "%3d%3d%25.16e%25.16e\n", ix, x0, pijj[ix*T+x0].re, pijj[ix*T+x0].im);
}
}
}
fclose(ofs);
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# time for writing: %e seconds\n", retime-ratime);
free(picount);
} else if(mode==2) {
if(make_H3orbits(&oh_id, &oh_count, &oh_val, &oh_nc) != 0) return(123);
ratime = (double)clock() / CLOCKS_PER_SEC;
nclass = oh_nc / Thp1;
if( (piavg = (fftw_complex*)malloc(oh_nc*sizeof(fftw_complex))) == (fftw_complex*)NULL) exit(110);
if( (picount = (int*)malloc(oh_nc*sizeof(int))) == (int*)NULL) exit(110);
for(ix=0; ix<oh_nc; ix++) {
piavg[ix].re = 0.;
piavg[ix].im = 0.;
picount[ix] = 0;
}
for(ix=0; ix<LX*LY*LZ; ix++) {
for(x0=0; x0<Thp1; x0++) {
iix = ix*T+x0;
iiy = oh_id[ix]*Thp1+x0;
piavg[iiy].re += pi00[iix].re;
piavg[iiy].im += pi00[iix].im;
if(x0>0 && x0<T/2) {
iix = ix*T+(T-x0);
piavg[iiy].re += pi00[iix].re;
piavg[iiy].im += pi00[iix].im;
}
}
picount[oh_id[ix]]++;
}
for(ix=0; ix<nclass; ix++) {
for(x0=0; x0<Thp1; x0++) {
iix = ix*Thp1+x0;
if(picount[ix]>0) {
piavg[iix].re /= (double)picount[ix];
piavg[iix].im /= (double)picount[ix];
if(x0>0 && x0<T/2) {
piavg[iix].re /= 2.;
piavg[iix].im /= 2.;
}
}
}
}
sprintf(filename, "corr02_00.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing corr-00-data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu);
for(x1=0; x1<nclass; x1++) {
if(oh_val[0][x1]>=g_qhatsqr_min-_Q2EPS && oh_val[0][x1]<=g_qhatsqr_max+_Q2EPS) {
ix = x1*Thp1;
for(x0=0; x0<Thp1; x0++) {
fprintf(ofs, "%25.16e%3d%25.16e%25.16e%5d\n", oh_val[0][x1], x0, piavg[ix+x0].re, piavg[ix+x0].im,
picount[x1]);
}
}
}
fclose(ofs);
for(ix=0; ix<oh_nc; ix++) {
piavg[ix].re = 0.;
piavg[ix].im = 0.;
picount[ix] = 0;
}
for(ix=0; ix<LX*LY*LZ; ix++) {
for(x0=0; x0<Thp1; x0++) {
iix = ix*T+x0;
iiy = oh_id[ix]*Thp1+x0;
piavg[iiy].re += pijj[iix].re;
piavg[iiy].im += pijj[iix].im;
if(x0>0 && x0<T/2) {
iix = ix*T+(T-x0);
piavg[iiy].re += pijj[iix].re;
piavg[iiy].im += pijj[iix].im;
}
}
picount[oh_id[ix]]++;
}
for(ix=0; ix<nclass; ix++) {
for(x0=0; x0<Thp1; x0++) {
iix = ix*Thp1+x0;
if(picount[ix]>0) {
piavg[iix].re /= (double)picount[ix];
piavg[iix].im /= (double)picount[ix];
if(x0>0 && x0<T/2) {
piavg[iix].re /= 2.;
piavg[iix].im /= 2.;
}
}
}}
sprintf(filename, "corr02_jj.%.4d", gid);
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
fprintf(stdout, "# writing corr-jj-data to file %s\n", filename);
fprintf(ofs, "# %6d%3d%3d%3d%3d%12.7f%12.7f\n", gid, T_global, LX, LY, LZ, g_kappa, g_mu);
for(x1=0; x1<nclass; x1++) {
ix = x1*Thp1;
for(x0=0; x0<Thp1; x0++) {
fprintf(ofs, "%25.16e%3d%25.16e%25.16e%5d\n", oh_val[0][x1], x0, piavg[ix+x0].re, piavg[ix+x0].im,
picount[x1]);
}
}
fclose(ofs);
sprintf(filename, "corr.02.mom");
if( (ofs=fopen(filename, "w")) == (FILE*)NULL ) {
fprintf(stderr, "Error: could not open file %s for writing\n", filename);
exit(5);
}
for(ix=0; ix<VOLUME; ix++) fprintf(ofs, "%5d%25.16e%5d", ix, oh_val[0][ix], picount[ix]);
fclose(ofs);
retime = (double)clock() / CLOCKS_PER_SEC;
fprintf(stdout, "# time for O_h averaging %e seconds\n", retime-ratime);
free(piavg); free(picount);
}
*/
#endif
}
/***************************************
* free the allocated memory, finalize *
***************************************/
free(corrt);
free_geometry();
/*
free(pi00);
free(pijj);
*/
fftw_destroy_plan(plan_m);
return(0);
}
FX_ENTRY GrContext_t FX_CALL
grSstWinOpen(
GrScreenResolution_t screen_resolution,
GrScreenRefresh_t refresh_rate,
GrColorFormat_t color_format,
GrOriginLocation_t origin_location,
int nColBuffers,
int nAuxBuffers)
{
uint32_t screen_width, screen_height;
struct retro_variable var = { "mupen64-screensize", 0 };
// ZIGGY
// allocate static texture names
// the initial value should be big enough to support the maximal resolution
free_texture = 32*1024*1024;
default_texture = free_texture++;
color_texture = free_texture++;
depth_texture = free_texture++;
LOG("grSstWinOpen(%d, %d, %d, %d, %d %d)\r\n", screen_resolution&~0x80000000, refresh_rate, color_format, origin_location, nColBuffers, nAuxBuffers);
width = 640;
height = 480;
bool ret = environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var);
if (ret && var.value)
{
if (sscanf(var.value ? var.value : "640x480", "%dx%d", &width, &height) != 2)
{
width = 640;
height = 480;
}
}
glViewport(0, 0, width, height);
lfb_color_fmt = color_format;
if (origin_location != GR_ORIGIN_UPPER_LEFT) DISPLAY_WARNING("origin must be in upper left corner");
if (nColBuffers != 2) DISPLAY_WARNING("number of color buffer is not 2");
if (nAuxBuffers != 1) DISPLAY_WARNING("number of auxiliary buffer is not 1");
if (isExtensionSupported("GL_ARB_texture_env_combine") == 0 &&
isExtensionSupported("GL_EXT_texture_env_combine") == 0)
DISPLAY_WARNING("Your video card doesn't support GL_ARB_texture_env_combine extension");
if (isExtensionSupported("GL_ARB_multitexture") == 0)
DISPLAY_WARNING("Your video card doesn't support GL_ARB_multitexture extension");
if (isExtensionSupported("GL_ARB_texture_mirrored_repeat") == 0)
DISPLAY_WARNING("Your video card doesn't support GL_ARB_texture_mirrored_repeat extension");
nbAuxBuffers = 4;
//glGetIntegerv(GL_AUX_BUFFERS, &nbAuxBuffers);
if (nbAuxBuffers > 0)
printf("Congratulations, you have %d auxilliary buffers, we'll use them wisely !\n", nbAuxBuffers);
blend_func_separate_support = 1;
packed_pixels_support = 0;
if (isExtensionSupported("GL_EXT_blend_func_separate") == 0)
{
DISPLAY_WARNING("GL_EXT_blend_func_separate not supported.\n");
blend_func_separate_support = 0;
}
else
{
printf("GL_EXT_blend_func_separate supported.\n");
blend_func_separate_support = 1;
}
// we can assume that non-GLES has GL_EXT_packed_pixels
// support -it's included since OpenGL 1.2
#ifdef GLES
if (isExtensionSupported("GL_EXT_packed_pixels") != 0)
#endif
packed_pixels_support = 1;
if (isExtensionSupported("GL_ARB_texture_non_power_of_two") == 0)
{
DISPLAY_WARNING("GL_ARB_texture_non_power_of_two supported.\n");
npot_support = 0;
}
else
{
printf("GL_ARB_texture_non_power_of_two supported.\n");
npot_support = 1;
}
if (isExtensionSupported("GL_EXT_fog_coord") == 0)
{
DISPLAY_WARNING("GL_EXT_fog_coord not supported.\n");
fog_coord_support = 0;
}
else
{
printf("GL_EXT_fog_coord supported.\n");
fog_coord_support = 1;
}
if (isExtensionSupported("GL_ARB_shading_language_100") &&
isExtensionSupported("GL_ARB_shader_objects") &&
isExtensionSupported("GL_ARB_fragment_shader") &&
isExtensionSupported("GL_ARB_vertex_shader"))
{}
#ifdef GLES
if (isExtensionSupported("GL_EXT_texture_format_BGRA8888"))
{
printf("GL_EXT_texture_format_BGRA8888 supported.\n");
bgra8888_support = 1;
}
else
{
DISPLAY_WARNING("GL_EXT_texture_format_BGRA8888 not supported.\n");
bgra8888_support = 0;
}
#endif
glViewport(0, 0, width, height);
viewport_width = width;
viewport_height = height;
// VP try to resolve z precision issues
// glMatrixMode(GL_MODELVIEW);
// glLoadIdentity();
// glTranslatef(0, 0, 1-zscale);
// glScalef(1, 1, zscale);
widtho = width/2;
heighto = height/2;
pBufferWidth = pBufferHeight = -1;
current_buffer = GL_BACK;
texture_unit = GL_TEXTURE0;
{
int i;
for (i=0; i<NB_TEXBUFS; i++)
texbufs[i].start = texbufs[i].end = 0xffffffff;
}
FindBestDepthBias();
init_geometry();
init_textures();
init_combiner();
return 1;
}
int main(int argc, char **argv) {
int c, i, j, mu, nu;
int count = 0;
int filename_set = 0;
int dims[4] = {0,0,0,0};
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, iix, it;
int sid, status, gid;
double **corr=NULL, **corr2=NULL;
double *tcorr=NULL, *tcorr2=NULL;
double *work = (double*)NULL;
double q[4], fnorm;
int verbose = 0;
int do_gt = 0;
int nsource=0;
char filename[100], contype[200];
double ratime, retime;
double plaq;
double spinor1[24], spinor2[24], U_[18];
double *gauge_trafo=(double*)NULL;
double mom2, mom4;
complex w, w1, *cp1, *cp2, *cp3;
FILE *ofs;
#ifdef MPI
// MPI_Init(&argc, &argv);
fprintf(stderr, "[jc_ud_x] Error, only non-mpi version implemented\n");
exit(1);
#endif
while ((c = getopt(argc, argv, "h?f:")) != -1) {
switch (c) {
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
fprintf(stdout, "\n**************************************************\n");
fprintf(stdout, "* jc_ud_x\n");
fprintf(stdout, "**************************************************\n\n");
/*********************************
* initialize MPI parameters
*********************************/
// mpi_init(argc, argv);
/* initialize */
dims[0]=T_global; dims[1]=LX; dims[2]=LY; dims[3]=LZ;
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
/*************************************************
* allocate mem for gauge field and spinor fields
*************************************************/
alloc_gauge_field(&g_gauge_field, VOLUMEPLUSRAND);
no_fields = 2;
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields; i++) alloc_spinor_field(&g_spinor_field[i], VOLUMEPLUSRAND);
/****************************************
* allocate memory for the contractions
****************************************/
nsource = (g_sourceid2 - g_sourceid + 1) / g_sourceid_step;
if(g_cart_id==0) fprintf(stdout, "# nsource = %d\n", nsource);
corr = (double**)calloc( nsource, sizeof(double*));
corr[0] = (double*)calloc( nsource*T*8, sizeof(double));
for(i=1;i<nsource;i++) corr[i] = corr[i-1] + 8*T;
corr2 = (double**)calloc( nsource, sizeof(double*));
corr2[0] = (double*)calloc( nsource*8*T, sizeof(double));
for(i=1;i<nsource;i++) corr2[i] = corr2[i-1] + 8*T;
tcorr = (double*)calloc(T*8, sizeof(double));
tcorr2 = (double*)calloc(T*8, sizeof(double));
/***********************************************
* start loop on gauge id.s
***********************************************/
for(gid=g_gaugeid; gid<=g_gaugeid2; gid++) {
sprintf(filename, "%s.%.4d", gaugefilename_prefix, gid);
if(g_cart_id==0) fprintf(stdout, "# reading gauge field from file %s\n", filename);
read_lime_gauge_field_doubleprec(filename);
xchange_gauge();
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "# measured plaquette value: %25.16e\n", plaq);
/* reset disc to zero */
for(ix=0; ix<nsource*8*T; ix++) corr[0][ix] = 0.;
for(ix=0; ix<nsource*8*T; ix++) corr2[0][ix] = 0.;
count=0;
/***********************************************
* start loop on source id.s
***********************************************/
for(sid=g_sourceid; sid<=g_sourceid2; sid+=g_sourceid_step) {
/* read the new propagator to g_spinor_field[0] */
ratime = (double)clock() / CLOCKS_PER_SEC;
if(format==0) {
sprintf(filename, "%s.%.4d.%.2d.inverted", filename_prefix, gid, sid);
if(read_lime_spinor(g_spinor_field[0], filename, 0) != 0) break;
}
else if(format==1) {
sprintf(filename, "%s.%.4d.%.5d.inverted", filename_prefix, gid, sid);
if(read_cmi(g_spinor_field[0], filename) != 0) {
fprintf(stderr, "\nError from read_cmi\n");
break;
}
}
xchange_field(g_spinor_field[0]);
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id==0) fprintf(stdout, "# time to read prop.: %e seconds\n", retime-ratime);
ratime = (double)clock() / CLOCKS_PER_SEC;
/* apply [1] = D_tm [0] */
Q_phi_tbc(g_spinor_field[1], g_spinor_field[0]);
xchange_field(g_spinor_field[1]);
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id==0) fprintf(stdout, "# time to apply D_W: %e seconds\n", retime-ratime);
ratime = (double)clock() / CLOCKS_PER_SEC;
/* calculate real and imaginary part */
for(mu=0; mu<4; mu++) {
for(x0=0; x0<T; x0++) {
for(x1=0; x1<LX; x1++) {
for(x2=0; x2<LY; x2++) {
for(x3=0; x3<LZ; x3++) {
ix = g_ipt[x0][x1][x2][x3];
_cm_eq_cm_ti_co(U_, g_gauge_field+_GGI(ix,mu), &(co_phase_up[mu]));
_fv_eq_cm_ti_fv(spinor1, U_, &g_spinor_field[0][_GSI(g_iup[ix][mu])]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_mi_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[1][_GSI(ix)], spinor2);
corr[count][2*(mu*T+x0) ] -= 0.5*w.re;
corr[count][2*(mu*T+x0)+1] -= 0.5*w.im;
_fv_eq_cm_dag_ti_fv(spinor1, U_, &g_spinor_field[0][_GSI(ix)]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_pl_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[1][_GSI(g_iup[ix][mu])], spinor2);
corr[count][2*(mu*T+x0) ] -= 0.5*w.re;
corr[count][2*(mu*T+x0)+1] -= 0.5*w.im;
_fv_eq_gamma_ti_fv(spinor1, mu, &g_spinor_field[0][_GSI(ix)]);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[1][_GSI(ix)], spinor1);
corr2[count][2*(mu*T+x0) ] -= w.re;
corr2[count][2*(mu*T+x0)+1] -= w.im;
}}}
}
} // of mu
count++;
} // of sid
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id==0) fprintf(stdout, "# time to calculate contractions: %e seconds\n", retime-ratime);
for(ix=0;ix<8*T;ix++) tcorr[ix] = 0.;
for(ix=0;ix<8*T;ix++) tcorr2[ix] = 0.;
for(i=0;i<nsource-1;i++) {
for(j=i+1;j<nsource;j++) {
for(mu=0;mu<4;mu++) {
for(x0=0;x0<T;x0++) { // times at source
for(x1=0;x1<T;x1++) { // times at sink
it = (x1 - x0 + T) % T;
// conserved current
tcorr[2*(mu*T+it) ] += corr[i][2*(mu*T+x1)] * corr[j][2*(mu*T+x0) ] - corr[i][2*(mu*T+x1)+1] * corr[j][2*(mu*T+x0)+1];
tcorr[2*(mu*T+it)+1] += corr[i][2*(mu*T+x1)] * corr[j][2*(mu*T+x0)+1] + corr[i][2*(mu*T+x1)+1] * corr[j][2*(mu*T+x0) ];
tcorr[2*(mu*T+it) ] += corr[j][2*(mu*T+x1)] * corr[i][2*(mu*T+x0) ] - corr[j][2*(mu*T+x1)+1] * corr[i][2*(mu*T+x0)+1];
tcorr[2*(mu*T+it)+1] += corr[j][2*(mu*T+x1)] * corr[i][2*(mu*T+x0)+1] + corr[j][2*(mu*T+x1)+1] * corr[i][2*(mu*T+x0) ];
// local current
tcorr2[2*(mu*T+it) ] += corr2[i][2*(mu*T+x1)] * corr2[j][2*(mu*T+x0) ] - corr2[i][2*(mu*T+x1)+1] * corr2[j][2*(mu*T+x0)+1];
tcorr2[2*(mu*T+it)+1] += corr2[i][2*(mu*T+x1)] * corr2[j][2*(mu*T+x0)+1] + corr2[i][2*(mu*T+x1)+1] * corr2[j][2*(mu*T+x0) ];
tcorr2[2*(mu*T+it) ] += corr2[j][2*(mu*T+x1)] * corr2[i][2*(mu*T+x0) ] - corr2[j][2*(mu*T+x1)+1] * corr2[i][2*(mu*T+x0)+1];
tcorr2[2*(mu*T+it)+1] += corr2[j][2*(mu*T+x1)] * corr2[i][2*(mu*T+x0)+1] + corr2[j][2*(mu*T+x1)+1] * corr2[i][2*(mu*T+x0) ];
}}
}
}}
fnorm = 1. / ( g_prop_normsqr * g_prop_normsqr * (double)(LX*LY*LZ) * (double)(LX*LY*LZ) * nsource * (nsource-1));
if(g_cart_id==0) fprintf(stdout, "X-fnorm = %e\n", fnorm);
for(ix=0;ix<8*T;ix++) tcorr[ix] *= fnorm;
for(ix=0;ix<8*T;ix++) tcorr2[ix] *= fnorm;
/************************************************
* save results
************************************************/
if(g_cart_id == 0) fprintf(stdout, "# save results for gauge id %d and sid %d\n", gid, sid);
/* save the result in position space */
sprintf(filename, "jc_u_tp0.%.4d.%.4d", gid, sid);
ofs = fopen(filename, "w");
for(x0=0;x0<T;x0++) fprintf(ofs, "%d%25.16e%25.16e%25.16e%25.16e%25.16e%25.16e%25.16e%25.16e\n", x0,
tcorr[2*(0*T+x0)], tcorr[2*(0*T+x0)+1],
tcorr[2*(1*T+x0)], tcorr[2*(1*T+x0)+1],
tcorr[2*(2*T+x0)], tcorr[2*(2*T+x0)+1],
tcorr[2*(3*T+x0)], tcorr[2*(3*T+x0)+1]);
fclose(ofs);
} /* of loop on gid */
/***********************************************
* free the allocated memory, finalize
***********************************************/
free(g_gauge_field);
for(i=0; i<no_fields; i++) free(g_spinor_field[i]);
free(g_spinor_field);
free_geometry();
free(corr);
free(corr2);
free(tcorr);
free(tcorr2);
return(0);
}
int main(int argc, char **argv) {
int c, mu, nu, status;
int i, j, ncon=-1, ir, is, ic, id;
int filename_set = 0;
int x0, x1, x2, x3, ix, iix;
int y0, y1, y2, y3, iy, iiy;
int start_valuet=0, start_valuex=0, start_valuey=0;
int num_threads=1, threadid, nthreads;
int seed, seed_set=0;
double diff1, diff2;
/* double *chi=NULL, *psi=NULL; */
double plaq=0., pl_ts, pl_xs, pl_global;
double *gauge_field_smeared = NULL;
double s[18], t[18], u[18], pl_loc;
double spinor1[24], spinor2[24];
double *pl_gather=NULL;
double dtmp;
complex prod, w, w2;
int verbose = 0;
char filename[200];
char file1[200];
char file2[200];
FILE *ofs=NULL;
double norm, norm2;
fermion_propagator_type *prop=NULL, prop2=NULL, seq_prop=NULL, seq_prop2=NULL, prop_aux=NULL, prop_aux2=NULL;
int idx, eoflag, shift;
float *buffer = NULL;
unsigned int VOL3;
size_t items, bytes;
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "h?vf:N:c:C:t:s:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'N':
ncon = atoi(optarg);
break;
case 'c':
strcpy(file1, optarg);
break;
case 'C':
strcpy(file2, optarg);
break;
case 't':
num_threads = atoi(optarg);
break;
case 's':
seed = atoi(optarg);
fprintf(stdout, "# [] use seed value %d\n", seed);
seed_set = 1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
if(g_cart_id==0) fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
/* initialize T etc. */
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] T_global = %3d\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] LX_global = %3d\n"\
"# [%2d] LX = %3d\n"\
"# [%2d] LXstart = %3d\n"\
"# [%2d] LY_global = %3d\n"\
"# [%2d] LY = %3d\n"\
"# [%2d] LYstart = %3d\n",\
g_cart_id, g_cart_id, T_global, g_cart_id, T, g_cart_id, Tstart,
g_cart_id, LX_global, g_cart_id, LX, g_cart_id, LXstart,
g_cart_id, LY_global, g_cart_id, LY, g_cart_id, LYstart);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(101);
}
geometry();
if(init_geometry_5d() != 0) {
fprintf(stderr, "ERROR from init_geometry_5d\n");
exit(102);
}
geometry_5d();
VOL3 = LX*LY*LZ;
/* read the gauge field */
alloc_gauge_field(&g_gauge_field, VOLUMEPLUSRAND);
sprintf(filename, "%s.%.4d", gaugefilename_prefix, Nconf);
if(g_cart_id==0) fprintf(stdout, "# reading gauge field from file %s\n", filename);
if(strcmp(gaugefilename_prefix, "identity")==0) {
status = unit_gauge_field(g_gauge_field, VOLUME);
} else {
// status = read_nersc_gauge_field_3x3(g_gauge_field, filename, &plaq);
// status = read_ildg_nersc_gauge_field(g_gauge_field, filename);
status = read_lime_gauge_field_doubleprec(filename);
// status = read_nersc_gauge_field(g_gauge_field, filename, &plaq);
// status = 0;
}
if(status != 0) {
fprintf(stderr, "[apply_Dtm] Error, could not read gauge field\n");
exit(11);
}
xchange_gauge();
// measure the plaquette
if(g_cart_id==0) fprintf(stdout, "# read plaquette value 1st field: %25.16e\n", plaq);
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "# measured plaquette value 1st field: %25.16e\n", plaq);
g_kappa5d = 0.5 / (5. + g_m0);
fprintf(stdout, "# [] g_kappa5d = %e\n", g_kappa5d);
no_fields=4;
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields; i++) alloc_spinor_field(&g_spinor_field[i], L5*VOLUMEPLUSRAND);
/*
items = VOL3 * 288;
bytes = items * sizeof(float);
if( (buffer = (float*)malloc( bytes ) ) == NULL ) {
fprintf(stderr, "[] Error, could not allocate buffer\n");
exit(20);
}
*/
/****************************************
* read read the spinor fields
****************************************/
/*
prop = create_fp_field(VOL3);
create_fp(&prop2);
create_fp(&prop_aux);
create_fp(&prop_aux2);
create_fp(&seq_prop);
create_fp(&seq_prop2);
*/
#ifdef MPI
if(!seed_set) { seed = g_seed; }
srand(seed+g_cart_id);
for(ix=0;ix<VOLUME*L5;ix++) {
for(i=0;i<24;i++) {
spinor1[i] = 2* (double)rand() / (double)RAND_MAX - 1.;
}
_fv_eq_fv(g_spinor_field[0]+_GSI(ix), spinor1 );
}
for(i=0;i<g_nproc;i++) {
if(g_cart_id==i) {
if(i==0) ofs = fopen("source", "w");
else ofs = fopen("source", "a");
for(is=0;is<L5;is++) {
for(x0=0;x0<T; x0++) {
for(x1=0;x1<LX; x1++) {
for(x2=0;x2<LX; x2++) {
for(x3=0;x3<LX; x3++) {
iix = is*VOLUME*g_nproc + (((x0+g_proc_coords[0]*T)*LX*g_nproc_x+ x1+g_proc_coords[1]*LX )*LY*g_nproc_y + x2+g_proc_coords[2]*LY )*LZ*g_nproc_z + x3+g_proc_coords[3]*LZ;
ix = g_ipt_5d[is][x0][x1][x2][x3];
for(c=0;c<24;c++) {
fprintf(ofs, "%8d%8d%3d%25.16e\n", iix, ix, c, g_spinor_field[0][_GSI(ix)+c]);
}
}}}}
}
fclose(ofs);
}
#ifdef MPI
MPI_Barrier(g_cart_grid);
#endif
}
#else
ofs = fopen("source", "r");
for(ix=0;ix<24*VOLUME*L5;ix++) {
fscanf(ofs, "%d%d%d%lf", &x1,&x2,&x3, &dtmp);
g_spinor_field[0][_GSI(x1)+x3] = dtmp;
}
fclose(ofs);
#endif
xchange_field_5d(g_spinor_field[0]);
Q_DW_Wilson_dag_phi(g_spinor_field[1], g_spinor_field[0]);
xchange_field_5d(g_spinor_field[1]);
Q_DW_Wilson_phi(g_spinor_field[2], g_spinor_field[1]);
sprintf(filename, "prop_%.2d.%.2d", g_nproc, g_cart_id);
ofs = fopen(filename, "w");
printf_spinor_field_5d(g_spinor_field[2], ofs);
fclose(ofs);
// for(ix=0;ix<VOLUME*L5;ix++) {
// for(i=0;i<24;i++) {
// spinor1[i] = 2* (double)rand() / (double)RAND_MAX - 1.;
// }
// _fv_eq_fv(g_spinor_field[1]+_GSI(ix), spinor1 );
// }
/*
xchange_field_5d(g_spinor_field[0]);
sprintf(filename, "spinor.%.2d", g_cart_id);
ofs = fopen(filename, "w");
printf_spinor_field_5d(g_spinor_field[0], ofs);
fclose(ofs);
*/
/*
// 2 = D 0
Q_DW_Wilson_phi(g_spinor_field[2], g_spinor_field[0]);
// 3 = D^dagger 1
Q_DW_Wilson_dag_phi(g_spinor_field[3], g_spinor_field[1]);
// <1, 2> = <1, D 0 >
spinor_scalar_product_co(&w, g_spinor_field[1], g_spinor_field[2], VOLUME*L5);
// <3, 0> = < D^dagger 1, 0 >
spinor_scalar_product_co(&w2, g_spinor_field[3], g_spinor_field[0], VOLUME*L5);
fprintf(stdout, "# [] w = %e + %e*1.i\n", w.re, w.im);
fprintf(stdout, "# [] w2 = %e + %e*1.i\n", w2.re, w2.im);
fprintf(stdout, "# [] abs difference = %e \n", sqrt(_SQR(w2.re-w.re)+_SQR(w2.im-w.im)) );
*/
/*
for(i=0;i<12;i++) {
fprintf(stdout, "s1[%2d] <- %25.16e + %25.16e*1.i\n", i+1, spinor1[2*i], spinor1[2*i+1]);
}
for(i=0;i<24;i++) {
spinor2[i] = 2* (double)rand() / (double)RAND_MAX - 1.;
}
for(i=0;i<12;i++) {
fprintf(stdout, "s2[%2d] <- %25.16e + %25.16e*1.i\n", i+1, spinor2[2*i], spinor2[2*i+1]);
}
_fv_mi_eq_PRe_fv(spinor2, spinor1);
for(i=0;i<12;i++) {
fprintf(stdout, "s3[%2d] <- %25.16e + %25.16e*1.i\n", i+1, spinor2[2*i], spinor2[2*i+1]);
}
*/
/*
ofs = fopen("dw_spinor", "w");
Q_DW_Wilson_phi(g_spinor_field[1], g_spinor_field[0]);
printf_spinor_field(g_spinor_field[1], ofs);
fclose(ofs);
g_kappa = g_kappa5d;
ofs = fopen("wilson_spinor", "w");
Q_Wilson_phi(g_spinor_field[2], g_spinor_field[0]);
printf_spinor_field(g_spinor_field[2], ofs);
fclose(ofs);
*/
#ifdef _UNDEF
/*******************************************************************
* propagators
*******************************************************************/
// for(i=0; i<12;i++)
for(i=0; i<1;i++)
{
//sprintf(file1, "source.%.4d.t00x00y00z00.%.2d.inverted", Nconf, i);
sprintf(file1, "/home/mpetschlies/quda-0.3.2/tests/prop");
if(g_cart_id==0) fprintf(stdout, "# Reading prop. from file %s\n", file1);
fflush(stdout);
//if( read_lime_spinor(g_spinor_field[0], file1, 0) != 0 ) {
ofs = fopen(file1, "rb");
if( fread(g_spinor_field[0], sizeof(double), 24*L5*VOLUME, ofs) != 24*L5*VOLUME) {
fprintf(stderr, "Error, could not read proper amount of data from file %s\n", file1);
exit(100);
}
fclose(ofs);
for(ix=0;ix<VOLUME*L5;ix++) {
_fv_ti_eq_re(g_spinor_field[0]+_GSI(ix), 2.*g_kappa5d);
}
/*
if( (ofs = fopen("prop_full", "w")) == NULL ) exit(22);
for(ix=0;ix<L5;ix++) {
fprintf(ofs, "# [] s = %d\n", ix);
printf_spinor_field(g_spinor_field[0]+_GSI(ix*VOLUME), ofs);
}
fclose(ofs);
*/
// reorder, multiply with g2
for(is=0,iix=0; is<L5; is++) {
for(ix=0; ix<VOLUME; ix++) {
iiy = lexic2eot_5d (is, ix);
_fv_eq_fv(spinor1, g_spinor_field[0]+_GSI(iiy));
_fv_eq_gamma_ti_fv(g_spinor_field[1]+_GSI(iix), 2, spinor1 );
iix++;
}}
Q_DW_Wilson_phi(g_spinor_field[2], g_spinor_field[1]);
// Q_DW_Wilson_dag_phi(g_spinor_field[2], g_spinor_field[1]);
fprintf(stdout, "# [] finished application of Dirac operator\n");
fflush(stdout);
// reorder, multiply with g2
for(is=0, iix=0;is<L5;is++) {
for(ix=0; ix<VOLUME; ix++) {
iiy = lexic2eot_5d(is, ix);
_fv_eq_fv(spinor1, g_spinor_field[2]+_GSI(iix));
_fv_eq_gamma_ti_fv(g_spinor_field[1]+_GSI(iiy), 2, spinor1 );
iix++;
}}
if( (ofs = fopen("my_out", "w")) == NULL ) exit(23);
for(ix=0;ix<L5;ix++) {
fprintf(ofs, "# [] s = %d\n", ix);
printf_spinor_field(g_spinor_field[1]+_GSI(ix*VOLUME), ofs);
}
fclose(ofs);
sprintf(file1, "/home/mpetschlies/quda-0.3.2/tests/source");
if(g_cart_id==0) fprintf(stdout, "# Reading prop. from file %s\n", file1);
fflush(stdout);
//if( read_lime_spinor(g_spinor_field[0], file1, 0) != 0 ) {
ofs = fopen(file1, "rb");
if( fread(g_spinor_field[2], sizeof(double), 24*L5*VOLUME, ofs) != 24*L5*VOLUME) {
fprintf(stderr, "Error, could not read proper amount of data from file %s\n", file1);
exit(100);
}
fclose(ofs);
/*
if( (ofs = fopen("v_out", "w")) == NULL ) exit(23);
for(ix=0;ix<L5;ix++) {
fprintf(ofs, "# [] s = %d\n", ix);
printf_spinor_field(g_spinor_field[2]+_GSI(ix*VOLUME), ofs);
}
fclose(ofs);
*/
spinor_scalar_product_re(&norm2, g_spinor_field[2], g_spinor_field[2], VOLUME*L5);
for(ix=0;ix<VOLUME*L5;ix++) {
_fv_mi_eq_fv(g_spinor_field[1]+_GSI(ix), g_spinor_field[2]+_GSI(ix));
}
spinor_scalar_product_re(&norm, g_spinor_field[1], g_spinor_field[1], VOLUME*L5);
fprintf(stdout, "\n# [] absolut residuum squared: %e; relative residuum %e\n", norm, sqrt(norm/norm2) );
} // of loop on spin color indices
#endif
/***********************************************
* free the allocated memory, finalize
***********************************************/
free(g_gauge_field);
free_geometry();
if(gauge_field_smeared != NULL) free(gauge_field_smeared);
if(g_spinor_field != NULL) {
for(i=0; i<no_fields; i++) free(g_spinor_field[i]);
free(g_spinor_field);
}
free(buffer);
free_fp_field(&prop);
free_fp(&prop2);
free_fp(&prop_aux);
free_fp(&prop_aux2);
free_fp(&seq_prop);
free_fp(&seq_prop2);
g_the_time = time(NULL);
fprintf(stdout, "# [] %s# [] end fo run\n", ctime(&g_the_time));
fflush(stdout);
fprintf(stderr, "# [] %s# [] end fo run\n", ctime(&g_the_time));
fflush(stderr);
#ifdef MPI
MPI_Finalize();
#endif
return(0);
}
int main(int argc, char *argv[])
{
Parameters *parameters; // user defined parameters
Geometry *geometry; // homogenous cube geometry
Material *material; // problem material
Bank *source_bank; // array for particle source sites
Tally *tally; // scalar flux tally
double *keff; // effective multiplication factor
double t1, t2; // timers
#ifdef _OPENMP
unsigned long counter = 0; //counter to decide the start pos of master bank copy from sub banks
Bank *g_fission_bank; //global fission bank
#endif
// Get inputs: set parameters to default values, parse parameter file,
// override with any command line inputs, and print parameters
parameters = init_parameters();
parse_parameters(parameters);
read_CLI(argc, argv, parameters);
print_parameters(parameters);
// Set initial RNG seed
set_initial_seed(parameters->seed);
set_stream(STREAM_INIT);
// Create files for writing results to
init_output(parameters);
// Set up geometry
geometry = init_geometry(parameters);
// Set up material
material = init_material(parameters);
// Set up tallies
tally = init_tally(parameters);
// Create source bank and initial source distribution
source_bank = init_source_bank(parameters, geometry);
// Create fission bank
#ifdef _OPENMP
omp_set_num_threads(parameters->n_threads); // Set number of openmp threads
printf("threads num: %d\n", parameters->n_threads);
// Allocate one master fission bank
g_fission_bank = init_bank(2*parameters->n_particles);
#endif
// Set up array for k effective
keff = calloc(parameters->n_active, sizeof(double));
center_print("SIMULATION", 79);
border_print();
printf("%-15s %-15s %-15s\n", "BATCH", "KEFF", "MEAN KEFF");
#ifdef _OPENMP
// Start time
t1 = omp_get_wtime();
run_eigenvalue(counter, g_fission_bank, parameters, geometry, material, source_bank, fission_bank, tally, keff);
// Stop time
t2 = omp_get_wtime();
#endif
printf("Simulation time: %f secs\n", t2-t1);
// Free memory
#ifdef _OPENMP
free_bank(g_fission_bank);
#endif
free(keff);
free_tally(tally);
free_bank(source_bank);
free_material(material);
free(geometry);
free(parameters);
return 0;
}
int main(int argc, char **argv) {
int c, i, mu;
int count = 0;
int filename_set = 0;
int l_LX_at, l_LXstart_at;
int x0, x1, ix, idx;
int VOL3;
int sid;
double *disc = (double*)NULL;
int verbose = 0;
char filename[100];
double ratime, retime;
double plaq;
double spinor1[24], spinor2[24];
double _2kappamu;
double *gauge_field_f=NULL, *gauge_field_timeslice=NULL;
double v4norm = 0., vvnorm = 0.;
complex w;
FILE *ofs1, *ofs2;
/* double sign_adj5[] = {-1., -1., -1., -1., +1., +1., +1., +1., +1., +1., -1., -1., -1., 1., -1., -1.}; */
double hopexp_coeff[8], addreal, addimag;
int gindex[] = { 5 , 1 , 2 , 3 , 6 ,10 ,11 ,12 , 4 , 7 , 8 , 9 , 0 ,15 , 14 ,13 };
int isimag[] = { 0 , 0 , 0 , 0 , 1 , 1 , 1 , 1 , 0 , 1 , 1 , 1 , 0 , 1 , 1 , 1 };
double gsign[] = {-1., 1., 1., 1., -1., 1., 1., 1., 1., 1., 1., 1., 1., 1., -1., 1.};
#ifdef MPI
MPI_Status status;
#endif
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "h?vgf:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
#ifdef MPI
T = T_global / g_nproc;
Tstart = g_cart_id * T;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
VOL3 = LX*LY*LZ;
#else
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
VOL3 = LX*LY*LZ;
#endif
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(1);
}
geometry();
/* read the gauge field */
alloc_gauge_field(&g_gauge_field, VOLUMEPLUSRAND);
sprintf(filename, "%s.%.4d", gaugefilename_prefix, Nconf);
if(g_cart_id==0) fprintf(stdout, "reading gauge field from file %s\n", filename);
read_lime_gauge_field_doubleprec(filename);
xchange_gauge();
/* measure the plaquette */
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "# measured plaquette value: %25.16e\n", plaq);
if(Nlong > -1) {
/* N_ape = 5; */
alpha_ape = 0.4;
if(g_cart_id==0) fprintf(stdout, "# apply fuzzing of gauge field and propagators with parameters:\n"\
"# Nlong = %d\n# N_ape = %d\n# alpha_ape = %f\n", Nlong, N_ape, alpha_ape);
alloc_gauge_field(&gauge_field_f, VOLUMEPLUSRAND);
if( (gauge_field_timeslice = (double*)malloc(72*VOL3*sizeof(double))) == (double*)NULL ) {
fprintf(stderr, "Error, could not allocate mem for gauge_field_timeslice\n");
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(2);
}
for(x0=0; x0<T; x0++) {
memcpy((void*)gauge_field_timeslice, (void*)(g_gauge_field+_GGI(g_ipt[x0][0][0][0],0)), 72*VOL3*sizeof(double));
for(i=0; i<N_ape; i++) {
APE_Smearing_Step_Timeslice(gauge_field_timeslice, alpha_ape);
}
fuzzed_links_Timeslice(gauge_field_f, gauge_field_timeslice, Nlong, x0);
}
free(gauge_field_timeslice);
}
/* test: print the fuzzed APE smeared gauge field to stdout */
/*
for(ix=0; ix<36*VOLUME; ix++) {
fprintf(stdout, "%6d%25.16e%25.16e%25.16e%25.16e\n", ix, gauge_field_f[2*ix], gauge_field_f[2*ix+1], g_gauge_field[2*ix], g_gauge_field[2*ix+1]);
}
*/
/* allocate memory for the spinor fields */
no_fields = 4;
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields; i++) alloc_spinor_field(&g_spinor_field[i], VOLUMEPLUSRAND);
/* allocate memory for the contractions */
disc = (double*)calloc(4*16*T*2, sizeof(double));
if( disc==(double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc\n");
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(3);
}
for(ix=0; ix<4*32*T; ix++) disc[ix] = 0.;
if(g_cart_id==0) {
sprintf(filename, "cvc_2pt_disc_vv.%.4d", Nconf);
ofs1 = fopen(filename, "w");
sprintf(filename, "cvc_2pt_disc_v4.%.4d", Nconf);
ofs2 = fopen(filename, "w");
if(ofs1==(FILE*)NULL || ofs2==(FILE*)NULL) {
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(5);
}
}
/* add the HPE coefficients */
if(format==1) {
addimag = 2*g_kappa*g_mu/sqrt(1 + 4*g_kappa*g_kappa*g_mu*g_mu)* LX*LY*LZ*3*4*2.*g_kappa*g_kappa*4;
addreal = 1./sqrt(1 + 4*g_kappa*g_kappa*g_mu*g_mu)*LX*LY*LZ*3*4*2.*g_kappa*g_kappa*4;
v4norm = 1. / ( 8. * g_kappa * g_kappa );
vvnorm = g_mu / ( 4. * g_kappa );
} else {
addimag = 2*g_kappa*g_mu/sqrt(1 + 4*g_kappa*g_kappa*g_mu*g_mu)* LX*LY*LZ*3*4*2.*g_kappa*2;
addreal = 1./sqrt(1 + 4*g_kappa*g_kappa*g_mu*g_mu)*LX*LY*LZ*3*4*2.*g_kappa*2;
v4norm = 1. / ( 4. * g_kappa );
vvnorm = g_mu / ( 4. * g_kappa );
}
/* calculate additional contributions for 1 and gamma_5 */
_2kappamu = 2.*g_kappa*g_mu;
hopexp_coeff[0] = 24. * g_kappa * LX*LY*LZ / (1. + _2kappamu*_2kappamu);
hopexp_coeff[1] = 0.;
hopexp_coeff[2] = -768. * g_kappa*g_kappa*g_kappa * LX*LY*LZ * _2kappamu*_2kappamu /
( (1.+_2kappamu*_2kappamu)*(1.+_2kappamu*_2kappamu)*(1.+_2kappamu*_2kappamu) );
hopexp_coeff[3] = 0.;
hopexp_coeff[4] = 0.;
hopexp_coeff[5] = -24.*g_kappa * LX*LY*LZ * _2kappamu / (1. + _2kappamu*_2kappamu);
hopexp_coeff[6] = 0.;
hopexp_coeff[7] = -384. * g_kappa*g_kappa*g_kappa * LX*LY*LZ *
(1.-_2kappamu*_2kappamu)*_2kappamu /
( (1.+_2kappamu*_2kappamu)*(1.+_2kappamu*_2kappamu)*(1.+_2kappamu*_2kappamu) );
/* start loop on source id.s */
for(sid=g_sourceid; sid<=g_sourceid2; sid+=g_sourceid_step) {
for(ix=0; ix<4*32*T; ix++) disc[ix] = 0.;
/* read the new propagator */
sprintf(filename, "%s.%.4d.%.5d.inverted", filename_prefix, Nconf, sid);
/* sprintf(filename, "%s.%.4d.%.2d.inverted", filename_prefix, Nconf, sid); */
if(read_lime_spinor(g_spinor_field[1], filename, 0) != 0) {
fprintf(stderr, "[%2d] Error, could not read from file %s\n", g_cart_id, filename);
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(4);
}
count++;
xchange_field(g_spinor_field[1]);
/* calculate the source: apply Q_phi_tbc */
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
Q_phi_tbc(g_spinor_field[0], g_spinor_field[1]);
xchange_field(g_spinor_field[0]);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time to apply Q_tm %e seconds\n", retime-ratime);
/* apply gamma5_BdagH4_gamma5 */
gamma5_BdagH4_gamma5(g_spinor_field[2], g_spinor_field[0], g_spinor_field[3]);
/* attention: additional factor 2kappa because of CMI format */
/*
if(format==1) {
for(ix=0; ix<VOLUME; ix++) {
_fv_ti_eq_re(&g_spinor_field[2][_GSI(ix)], 2.*g_kappa);
}
}
*/
if(Nlong>-1) {
if(g_cart_id==0) fprintf(stdout, "# fuzzing propagator with Nlong = %d\n", Nlong);
memcpy((void*)g_spinor_field[3], (void*)g_spinor_field[1], 24*VOLUMEPLUSRAND*sizeof(double));
Fuzz_prop(gauge_field_f, g_spinor_field[3], Nlong);
}
/* add new contractions to disc */
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
for(x0=0; x0<T; x0++) { /* loop on time */
for(x1=0; x1<VOL3; x1++) { /* loop on sites in timeslice */
ix = x0*VOL3 + x1;
for(mu=0; mu<16; mu++) { /* loop on index of gamma matrix */
_fv_eq_gamma_ti_fv(spinor1, mu, &g_spinor_field[1][_GSI(ix)]);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[2][_GSI(ix)], spinor1);
disc[2*( x0*16+mu) ] += w.re;
disc[2*( x0*16+mu)+1] += w.im;
_fv_eq_gamma_ti_fv(spinor1, 5, &g_spinor_field[1][_GSI(ix)]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[1][_GSI(ix)], spinor2);
disc[2*(16*T + x0*16+mu) ] += w.re;
disc[2*(16*T + x0*16+mu)+1] += w.im;
if(Nlong>-1) {
_fv_eq_gamma_ti_fv(spinor1, mu, &g_spinor_field[3][_GSI(ix)]);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[2][_GSI(ix)], spinor1);
disc[2*(32*T + x0*16+mu) ] += w.re;
disc[2*(32*T + x0*16+mu)+1] += w.im;
_fv_eq_gamma_ti_fv(spinor1, 5, &g_spinor_field[3][_GSI(ix)]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[1][_GSI(ix)], spinor2);
disc[2*(48*T + x0*16+mu) ] += w.re;
disc[2*(48*T + x0*16+mu)+1] += w.im;
}
}
}
}
if(g_cart_id==0) fprintf(stdout, "# addimag = %25.16e\n", addimag);
if(g_cart_id==0) fprintf(stdout, "# addreal = %25.16e\n", addreal);
for(x0=0; x0<T; x0++) {
disc[2*( x0*16+4) ] += addreal;
disc[2*( x0*16+5)+1] -= addimag;
/*
if(Nlong>-1) {
disc[2*(32*T + x0*16+4) ] += addreal;
disc[2*(32*T + x0*16+5)+1] -= addimag;
}
*/
}
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# contractions in %e seconds\n", retime-ratime);
/* write current disc to file */
if(g_cart_id==0) {
if(sid==g_sourceid) fprintf(ofs1, "#%6d%3d%3d%3d%3d\t%f\t%f\n", Nconf, T, LX, LY, LZ, g_kappa, g_mu);
if(sid==g_sourceid) fprintf(ofs2, "#%6d%3d%3d%3d%3d\t%f\t%f\n", Nconf, T, LX, LY, LZ, g_kappa, g_mu);
for(x0=0; x0<T; x0++) {
for(mu=0; mu<16; mu++) {
idx = gindex[mu];
ix = 16*x0 + idx;
if(isimag[mu]==0) {
fprintf(ofs2, "%6d%3d%4d%4d%25.16e%25.16e%25.16e%25.16e\n",
Nconf, mu, x0, sid,
gsign[mu]*disc[2* ix ]*v4norm, gsign[mu]*disc[2* ix +1]*v4norm,
gsign[mu]*disc[2*(32*T+ix)]*v4norm, gsign[mu]*disc[2*(32*T+ix)+1]*v4norm);
} else {
fprintf(ofs2, "%6d%3d%4d%4d%25.16e%25.16e%25.16e%25.16e\n",
Nconf, mu, x0, sid,
gsign[mu]*disc[2*( ix)+1]*v4norm, -gsign[mu]*disc[2* ix ]*v4norm,
gsign[mu]*disc[2*(32*T+ix)+1]*v4norm, -gsign[mu]*disc[2*(32*T+ix)]*v4norm);
}
}
}
for(x0=0; x0<T; x0++) {
for(mu=0; mu<16; mu++) {
idx = gindex[mu];
ix = 16*x0 + idx;
if(isimag[mu]==0) {
fprintf(ofs1, "%6d%3d%4d%4d%25.16e%25.16e%25.16e%25.16e\n",
Nconf, mu, x0, sid,
gsign[mu]*disc[2*(16*T+ix)+1]*vvnorm, -gsign[mu]*disc[2*(16*T+ix)]*vvnorm,
gsign[mu]*disc[2*(48*T+ix)+1]*vvnorm, -gsign[mu]*disc[2*(48*T+ix)]*vvnorm);
} else {
fprintf(ofs1, "%6d%3d%4d%4d%25.16e%25.16e%25.16e%25.16e\n",
Nconf, mu, x0, sid,
-gsign[mu]*disc[2*(16*T+ix)]*vvnorm, -gsign[mu]*disc[2*(16*T+ix)+1]*vvnorm,
-gsign[mu]*disc[2*(48*T+ix)]*vvnorm, -gsign[mu]*disc[2*(48*T+ix)+1]*vvnorm);
}
}
}
#ifdef MPI
for(c=1; c<g_nproc; c++) {
MPI_Recv(disc, 128*T, MPI_DOUBLE, c, 100+c, g_cart_grid, &status);
for(x0=0; x0<T; x0++) {
for(mu=0; mu<16; mu++) {
idx=gindex[mu];
ix = 16*x0 + idx;
if(isimag[mu]==0) {
fprintf(ofs2, "%6d%3d%4d%4d%25.16e%25.16e%25.16e%25.16e\n",
Nconf, mu, c*T+x0, sid,
gsign[mu]*disc[2* ix ]*v4norm, gsign[mu]*disc[2* ix +1]*v4norm,
gsign[mu]*disc[2*(32*T+ix)]*v4norm, gsign[mu]*disc[2*(32*T+ix)+1]*v4norm);
} else {
fprintf(ofs2, "%6d%3d%4d%4d%25.16e%25.16e%25.16e%25.16e\n",
Nconf, mu, c*T+x0, sid,
gsign[mu]*disc[2*( ix)+1]*v4norm, -gsign[mu]*disc[2* ix ]*v4norm,
gsign[mu]*disc[2*(32*T+ix)+1]*v4norm, -gsign[mu]*disc[2*(32*T+ix)]*v4norm);
}
}
}
for(x0=0; x0<T; x0++) {
for(mu=0; mu<16; mu++) {
idx = gindex[mu];
ix = 16*x0 + idx;
if(isimag[mu]==0) {
fprintf(ofs1, "%6d%3d%4d%4d%25.16e%25.16e%25.16e%25.16e\n",
Nconf, mu, c*T+x0, sid,
gsign[mu]*disc[2*(16*T+ix)+1]*vvnorm, -gsign[mu]*disc[2*(16*T+ix)]*vvnorm,
gsign[mu]*disc[2*(48*T+ix)+1]*vvnorm, -gsign[mu]*disc[2*(48*T+ix)]*vvnorm);
} else {
fprintf(ofs1, "%6d%3d%4d%4d%25.16e%25.16e%25.16e%25.16e\n",
Nconf, mu, c*T+x0, sid,
-gsign[mu]*disc[2*(16*T+ix)]*vvnorm, -gsign[mu]*disc[2*(16*T+ix)+1]*vvnorm,
-gsign[mu]*disc[2*(48*T+ix)]*vvnorm, -gsign[mu]*disc[2*(48*T+ix)+1]*vvnorm);
}
}
}
}
#endif
}
#ifdef MPI
else {
for(c=1; c<g_nproc; c++) {
if(g_cart_id==c) {
MPI_Send(disc, 128*T, MPI_DOUBLE, 0, 100+c, g_cart_grid);
}
}
}
#endif
} /* of loop on sid */
if(g_cart_id==0) { fclose(ofs1); fclose(ofs2); }
if(g_cart_id==0) {
fprintf(stdout, "# contributions from HPE:\n");
fprintf(stdout, "(1) X = id\t%25.16e%25.16e\n"\
" \t%25.16e%25.16e\n"\
"(2) X = 5\t%25.16e%25.16e\n"\
" \t%25.16e%25.16e\n",
hopexp_coeff[0], hopexp_coeff[1], hopexp_coeff[2], hopexp_coeff[3],
hopexp_coeff[4], hopexp_coeff[5], hopexp_coeff[6], hopexp_coeff[7]);
}
/* free the allocated memory, finalize */
free(g_gauge_field); g_gauge_field=(double*)NULL;
for(i=0; i<no_fields; i++) free(g_spinor_field[i]);
free(g_spinor_field); g_spinor_field=(double**)NULL;
free_geometry();
free(disc);
if(Nlong>-1) free(gauge_field_f);
#ifdef MPI
MPI_Finalize();
#endif
return(0);
}
int main(int argc, char **argv) {
int c, i, mu, nu;
int count = 0;
int filename_set = 0;
int dims[4] = {0,0,0,0};
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, iix;
int sid, status;
double *disc = (double*)NULL;
double *data = (double*)NULL;
double *bias = (double*)NULL;
double *work = (double*)NULL;
double q[4], fnorm;
int verbose = 0;
char filename[100], contype[200];
double ratime, retime;
double plaq;
double spinor1[24], spinor2[24], U_[18];
complex w, w1, *cp1, *cp2, *cp3, *cp4;
fftw_complex *in=(fftw_complex*)NULL;
#ifdef MPI
fftwnd_mpi_plan plan_p, plan_m;
#else
fftwnd_plan plan_p, plan_m;
#endif
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "h?vf:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa <= 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.\n");
usage();
}
if(hpe_order%2==0 && hpe_order>0) {
if(g_proc_id==0) fprintf(stdout, "HPE order should be odd\n");
usage();
}
fprintf(stdout, "\n**************************************************\n"\
"* vp_disc_hpe_stoch_subtract with HPE of order %d\n"\
"**************************************************\n\n", hpe_order);
/*********************************
* initialize MPI parameters
*********************************/
mpi_init(argc, argv);
/* initialize fftw */
dims[0]=T_global; dims[1]=LX; dims[2]=LY; dims[3]=LZ;
#ifdef MPI
plan_p = fftwnd_mpi_create_plan(g_cart_grid, 4, dims, FFTW_BACKWARD, FFTW_MEASURE);
plan_m = fftwnd_mpi_create_plan(g_cart_grid, 4, dims, FFTW_FORWARD, FFTW_MEASURE);
fftwnd_mpi_local_sizes(plan_p, &T, &Tstart, &l_LX_at, &l_LXstart_at, &FFTW_LOC_VOLUME);
#else
plan_p = fftwnd_create_plan(4, dims, FFTW_BACKWARD, FFTW_MEASURE | FFTW_IN_PLACE);
plan_m = fftwnd_create_plan(4, dims, FFTW_FORWARD, FFTW_MEASURE | FFTW_IN_PLACE);
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
#endif
fprintf(stdout, "# [%2d] fftw parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
#ifdef MPI
if(T==0) {
fprintf(stderr, "[%2d] local T is zero; exit\n", g_cart_id);
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
exit(101);
}
#endif
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(102);
}
geometry();
/************************************************
* read the gauge field, measure the plaquette
************************************************/
alloc_gauge_field(&g_gauge_field, VOLUMEPLUSRAND);
sprintf(filename, "%s.%.4d", gaugefilename_prefix, Nconf);
if(g_cart_id==0) fprintf(stdout, "# reading gauge field from file %s\n", filename);
read_lime_gauge_field_doubleprec(filename);
xchange_gauge();
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "# measured plaquette value: %25.16e\n", plaq);
/****************************************
* allocate memory for the spinor fields
****************************************/
no_fields = 3;
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields; i++) alloc_spinor_field(&g_spinor_field[i], VOLUMEPLUSRAND);
/****************************************
* allocate memory for the contractions
****************************************/
disc = (double*)calloc(16*VOLUME, sizeof(double));
if( disc == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(103);
}
data = (double*)calloc(16*VOLUME, sizeof(double));
if( data== (double*)NULL ) {
fprintf(stderr, "could not allocate memory for data\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(104);
}
for(ix=0; ix<16*VOLUME; ix++) data[ix] = 0.;
work = (double*)calloc(32*VOLUME, sizeof(double));
if( work == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for work\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(105);
}
bias = (double*)calloc(32*VOLUME, sizeof(double));
if( bias == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for bias\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(106);
}
for(ix=0; ix<32*VOLUME; ix++) bias[ix] = 0.;
/****************************************
* prepare Fourier transformation arrays
****************************************/
in = (fftw_complex*)malloc(FFTW_LOC_VOLUME*sizeof(fftw_complex));
if(in==(fftw_complex*)NULL) {
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(107);
}
/***********************************************
* start loop on source id.s
***********************************************/
for(sid=g_sourceid; sid<=g_sourceid2; sid+=g_sourceid_step) {
for(ix=0; ix<16*VOLUME; ix++) disc[ix] = 0.;
/* read the new propagator */
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(format==0) {
sprintf(filename, "%s.%.4d.%.2d.inverted", filename_prefix, Nconf, sid);
if(read_lime_spinor(g_spinor_field[2], filename, 0) != 0) break;
}
else if(format==1) {
sprintf(filename, "%s.%.4d.%.5d.inverted", filename_prefix, Nconf, sid);
if(read_cmi(g_spinor_field[2], filename) != 0) break;
}
xchange_field(g_spinor_field[2]);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time to read prop.: %e seconds\n", retime-ratime);
count++;
/************************************************
* calculate the source: apply Q_phi_tbc
************************************************/
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
Q_phi_tbc(g_spinor_field[0], g_spinor_field[2]);
xchange_field(g_spinor_field[0]);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time to calculate source: %e seconds\n", retime-ratime);
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
/************************************************
* HPE: apply BH to order hpe_order+2
************************************************/
if(hpe_order>0) {
BHn(g_spinor_field[1], g_spinor_field[2], hpe_order+2);
} else {
memcpy((void*)g_spinor_field[1], (void*)g_spinor_field[2], 24*VOLUMEPLUSRAND*sizeof(double));
}
/************************************************
* add new contractions to (existing) disc
************************************************/
for(mu=0; mu<4; mu++) {
iix = _GWI(mu,0,VOLUME);
for(ix=0; ix<VOLUME; ix++) {
_cm_eq_cm_ti_co(U_, &g_gauge_field[_GGI(ix, mu)], &co_phase_up[mu]);
_fv_eq_cm_ti_fv(spinor1, U_, &g_spinor_field[1][_GSI(g_iup[ix][mu])]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_mi_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[0][_GSI(ix)], spinor2);
disc[iix ] = -0.5 * w.re;
disc[iix+1] = -0.5 * w.im;
data[iix ] -= 0.5 * w.re;
data[iix+1] -= 0.5 * w.im;
_fv_eq_cm_dag_ti_fv(spinor1, U_, &g_spinor_field[1][_GSI(ix)]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_pl_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[0][_GSI(g_iup[ix][mu])], spinor2);
disc[iix ] -= 0.5 * w.re;
disc[iix+1] -= 0.5 * w.im;
data[iix ] -= 0.5 * w.re;
data[iix+1] -= 0.5 * w.im;
iix += 2;
}
}
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time to contract cvc: %e seconds\n", retime-ratime);
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
for(mu=0; mu<4; mu++) {
memcpy((void*)in, (void*)(disc+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_m, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_m, in, NULL);
#endif
memcpy((void*)(disc+_GWI(4+mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
memcpy((void*)in, (void*)(disc+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_p, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_p, in, NULL);
#endif
memcpy((void*)(disc+_GWI(mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
} /* of mu =0 ,..., 3*/
for(mu=0; mu<4; mu++) {
for(nu=0; nu<4; nu++) {
cp1 = (complex*)(disc+_GWI(mu, 0,VOLUME));
cp2 = (complex*)(disc+_GWI(4+nu, 0,VOLUME));
cp3 = (complex*)(bias+_GWI(4*mu+nu,0,VOLUME));
for(ix=0; ix<VOLUME; ix++) {
_co_eq_co_ti_co(&w1, cp1, cp2);
cp3->re += w1.re;
cp3->im += w1.im;
cp1++; cp2++; cp3++;
}
}}
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time for Fourier trafo and adding to bias: %e seconds\n",
retime-ratime);
} /* of loop on sid */
/************************************************
* save results for count == Nsave
************************************************/
if(count==Nsave) {
if(g_cart_id == 0) fprintf(stdout, "# save results for count = %d\n", count);
for(ix=0; ix<16*VOLUME; ix++) disc[ix] = 0.;
if(hpe_order>0) {
sprintf(filename, "vp_disc_hpe%.2d_loops_X.%.4d", hpe_order, Nconf);
if(g_cart_id==0) fprintf(stdout, "# reading loop part from file %s\n", filename);
if( (status = read_lime_contraction(disc, filename, 4, 0)) != 0 ) {
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(108);
}
}
/* save the result in position space */
fnorm = 1. / ( (double)count * g_prop_normsqr );
if(g_cart_id==0) fprintf(stdout, "# X-fnorm = %e\n", fnorm);
for(mu=0; mu<4; mu++) {
for(ix=0; ix<VOLUME; ix++) {
work[_GWI(mu,ix,VOLUME) ] = data[_GWI(mu,ix,VOLUME) ] * fnorm + disc[_GWI(mu,ix,VOLUME) ];
work[_GWI(mu,ix,VOLUME)+1] = data[_GWI(mu,ix,VOLUME)+1] * fnorm + disc[_GWI(mu,ix,VOLUME)+1];
}
}
sprintf(filename, "vp_disc_hpe%.2d_subtracted_X.%.4d.%.4d", hpe_order, Nconf, count);
sprintf(contype, "cvc-disc-hpe-loops-%2d-to-%2d-stoch-subtracted-X", hpe_order, hpe_order+2);
write_lime_contraction(work, filename, 64, 4, contype, Nconf, count);
/*
sprintf(filename, "vp_disc_hpe%.2d_subtracted_X.%.4d.%.4d.ascii", hpe_order, Nconf, count);
write_contraction(work, NULL, filename, 4, 2, 0);
*/
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
for(mu=0; mu<4; mu++) {
memcpy((void*)in, (void*)(data+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_m, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_m, in, NULL);
#endif
memcpy((void*)(data+_GWI(4+mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
memcpy((void*)in, (void*)(data+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_p, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_p, in, NULL);
#endif
memcpy((void*)(data+_GWI(mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
}
fnorm = 1. / ( g_prop_normsqr*g_prop_normsqr * (double)count * (double)(count-1) );
if(g_cart_id==0) fprintf(stdout, "# P-fnorm for purely stochastic part = %e\n", fnorm);
for(mu=0; mu<4; mu++) {
for(nu=0; nu<4; nu++) {
cp1 = (complex*)(data+_GWI(mu, 0,VOLUME));
cp2 = (complex*)(data+_GWI(4+nu, 0,VOLUME));
cp3 = (complex*)(work+_GWI(4*mu+nu,0,VOLUME));
cp4 = (complex*)(bias+_GWI(4*mu+nu,0,VOLUME));
for(ix=0; ix<VOLUME; ix++) {
_co_eq_co_ti_co(&w1, cp1, cp2);
cp3->re = ( w1.re - cp4->re ) * fnorm;
cp3->im = ( w1.im - cp4->im ) * fnorm;
cp1++; cp2++; cp3++; cp4++;
}
}}
for(mu=0; mu<4; mu++) {
memcpy((void*)in, (void*)(disc+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_m, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_m, in, NULL);
#endif
memcpy((void*)(disc+_GWI(4+mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
memcpy((void*)in, (void*)(disc+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_p, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_p, in, NULL);
#endif
memcpy((void*)(disc+_GWI(mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
}
fnorm = 1. / ( g_prop_normsqr * (double)count );
if(g_cart_id==0) fprintf(stdout, "# P-fnorm for mixed stochastic-loop part = %e\n", fnorm);
for(mu=0; mu<4; mu++) {
for(nu=0; nu<4; nu++) {
cp1 = (complex*)(data + _GWI(mu, 0,VOLUME));
cp2 = (complex*)(disc + _GWI(4+nu, 0,VOLUME));
cp3 = (complex*)(work + _GWI(4*mu+nu,0,VOLUME));
for(ix=0; ix<VOLUME; ix++) {
_co_eq_co_ti_co(&w1, cp1, cp2);
cp3->re += w1.re * fnorm;
cp3->im += w1.im * fnorm;
cp1++; cp2++; cp3++;
}
cp1 = (complex*)(disc + _GWI(mu, 0,VOLUME));
cp2 = (complex*)(data + _GWI(4+nu, 0,VOLUME));
cp3 = (complex*)(work + _GWI(4*mu+nu,0,VOLUME));
for(ix=0; ix<VOLUME; ix++) {
_co_eq_co_ti_co(&w1, cp1, cp2);
cp3->re += w1.re * fnorm;
cp3->im += w1.im * fnorm;
cp1++; cp2++; cp3++;
}
}}
fnorm = 1. / ( (double)T_global * (double)(LX*LY*LZ) );
if(g_cart_id==0) fprintf(stdout, "# P-fnorm for final estimator (1/T/V) = %e\n", fnorm);
for(mu=0; mu<4; mu++) {
for(nu=0; nu<4; nu++) {
cp1 = (complex*)(disc + _GWI(mu, 0,VOLUME));
cp2 = (complex*)(disc + _GWI(4+nu, 0,VOLUME));
cp3 = (complex*)(work + _GWI(4*mu+nu,0,VOLUME));
for(x0=0; x0<T; x0++) {
q[0] = (double)(x0+Tstart) / (double)T_global;
for(x1=0; x1<LX; x1++) {
q[1] = (double)x1 / (double)LX;
for(x2=0; x2<LY; x2++) {
q[2] = (double)x2 / (double)LY;
for(x3=0; x3<LZ; x3++) {
q[3] = (double)x3 / (double)LZ;
ix = g_ipt[x0][x1][x2][x3];
w.re = cos(M_PI * ( q[mu] - q[nu] ) );
w.im = sin(M_PI * ( q[mu] - q[nu] ) );
_co_eq_co_ti_co(&w1, cp1, cp2);
cp3->re += w1.re;
cp3->im += w1.im;
_co_eq_co_ti_co(&w1, cp3, &w);
cp3->re = w1.re * fnorm;
cp3->im = w1.im * fnorm;
cp1++; cp2++; cp3++;
}}}}
}}
sprintf(filename, "vp_disc_hpe%.2d_subtracted_P.%.4d.%.4d", hpe_order, Nconf, count);
sprintf(contype, "cvc-disc-hpe-loops-%2d-to-%2d-stoch-subtracted-P", hpe_order, hpe_order+2);
write_lime_contraction(work, filename, 64, 16, contype, Nconf, count);
/*
sprintf(filename, "vp_disc_hpe%.2d_subtracted_P.%.4d.%.4d.ascii", hpe_order, Nconf, count);
write_contraction(work, NULL, filename, 16, 2, 0);
*/
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time to save cvc results: %e seconds\n", retime-ratime);
} /* of if count == Nsave */
/***********************************************
* free the allocated memory, finalize
***********************************************/
free(g_gauge_field);
for(i=0; i<no_fields; i++) free(g_spinor_field[i]);
free(g_spinor_field);
free_geometry();
fftw_free(in);
free(disc);
free(bias);
free(data);
free(work);
#ifdef MPI
fftwnd_mpi_destroy_plan(plan_p);
fftwnd_mpi_destroy_plan(plan_m);
MPI_Finalize();
#else
fftwnd_destroy_plan(plan_p);
fftwnd_destroy_plan(plan_m);
#endif
return(0);
}
int main(int argc, char **argv) {
int c;
int count, ncon=-1;
int filename_set = 0;
int ix;
double *disc = (double*)NULL;
double *disc2 = (double*)NULL;
double adiffre, adiffim, mdiffre, mdiffim, Mdiffre, Mdiffim, hre, him;
int verbose = 0;
char filename[200];
char file1[200];
char file2[200];
while ((c = getopt(argc, argv, "h?vf:N:c:C:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'N':
ncon = atoi(optarg);
break;
case 'c':
strcpy(file1, optarg);
break;
case 'C':
strcpy(file2, optarg);
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
if(g_cart_id==0) fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
/* initialize */
T = T_global;
Tstart = 0;
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n",\
g_cart_id, g_cart_id, T, g_cart_id, Tstart);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(101);
}
geometry();
/****************************************
* allocate memory for the contractions
****************************************/
if(ncon<=0) {
fprintf(stderr, "Error, incompatible contraction type specified; exit\n");
exit(102);
} else {
fprintf(stdout, "# Using contraction type %d\n", ncon);
}
disc = (double*)calloc(2*ncon*VOLUME, sizeof(double));
if( disc == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc\n");
exit(103);
}
disc2 = (double*)calloc(2*ncon*VOLUME, sizeof(double));
if( disc2 == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc2\n");
exit(104);
}
/****************************************
* read contractions
****************************************/
if( read_lime_contraction(disc, file1, 64, ncon, 0) != 0 ) {
fprintf(stderr, "Error, could not read from file %s; exit\n", file1);
exit(105);
}
if( read_lime_contraction(disc2, file2, 64, ncon, 0) != 0 ) {
fprintf(stderr, "Error, could not read from file %s; exit\n", file2);
exit(106);
}
/****************************************
* calculate difference
****************************************/
mdiffre = fabs(disc[0] - disc2[0]);
mdiffim = fabs(disc[1] - disc2[1]);
Mdiffre = 0.;
Mdiffim = 0.;
adiffre = 0.;
adiffim = 0.;
for(ix=0; ix<ncon*VOLUME; ix++) {
adiffre += disc[2*ix ] - disc2[2*ix ];
adiffim += disc[2*ix+1] - disc2[2*ix+1];
hre = fabs(disc[2*ix ] - disc2[2*ix ]);
him = fabs(disc[2*ix+1] - disc2[2*ix+1]);
if(hre<mdiffre) mdiffre = hre;
if(hre>Mdiffre) Mdiffre = hre;
if(him<mdiffim) mdiffim = him;
if(him>Mdiffim) Mdiffim = him;
}
adiffre /= (double)VOLUME * (double)ncon;
adiffim /= (double)VOLUME * (double)ncon;
fprintf(stdout, "# Results for files %s and %s:\n", file1, file2);
fprintf(stdout, "average difference\t%25.16e\t%25.16e\n", adiffre, adiffim);
fprintf(stdout, "minimal abs. difference\t%25.16e\t%25.16e\n", mdiffre, mdiffim);
fprintf(stdout, "maximal abs. difference\t%25.16e\t%25.16e\n", Mdiffre, Mdiffim);
/***********************************************
* free the allocated memory, finalize
***********************************************/
free_geometry();
free(disc);
free(disc2);
return(0);
}
int main(int argc, char **argv) {
const int n_c=3;
const int n_s=4;
const char outfile_prefix[] = "delta_pp_2pt_v3";
int c, i, icomp;
int filename_set = 0;
int append, status;
int l_LX_at, l_LXstart_at;
int ix, it, iix, x1,x2,x3;
int ir, ir2, is;
int VOL3;
int do_gt=0;
int dims[3];
double *connt=NULL;
spinor_propagator_type *connq=NULL;
int verbose = 0;
int sx0, sx1, sx2, sx3;
int write_ascii=0;
int fermion_type = 1; // Wilson fermion type
int num_threads=1;
int pos;
char filename[200], contype[200], gauge_field_filename[200];
double ratime, retime;
//double plaq_m, plaq_r;
double *work=NULL;
fermion_propagator_type fp1=NULL, fp2=NULL, fp3=NULL, fp4=NULL, fpaux=NULL, uprop=NULL, dprop=NULL, *stochastic_fp=NULL;
spinor_propagator_type sp1, sp2;
double q[3], phase, *gauge_trafo=NULL;
double *stochastic_source=NULL, *stochastic_prop=NULL;
complex w, w1;
size_t items, bytes;
FILE *ofs;
int timeslice;
DML_Checksum ildg_gauge_field_checksum, *spinor_field_checksum=NULL, connq_checksum;
uint32_t nersc_gauge_field_checksum;
/***********************************************************/
int *qlatt_id=NULL, *qlatt_count=NULL, **qlatt_rep=NULL, **qlatt_map=NULL, qlatt_nclass=0;
int use_lattice_momenta = 0;
double **qlatt_list=NULL;
/***********************************************************/
/***********************************************************/
int rel_momentum_filename_set = 0, rel_momentum_no=0;
int **rel_momentum_list=NULL;
char rel_momentum_filename[200];
/***********************************************************/
/***********************************************************/
int snk_momentum_no = 1;
int **snk_momentum_list = NULL;
int snk_momentum_filename_set = 0;
char snk_momentum_filename[200];
/***********************************************************/
/*******************************************************************
* Gamma components for the Delta:
*/
//const int num_component = 16;
//int gamma_component[2][16] = { {0,0,0,0,1,1,1,1,2,2,2,2,3,3,3,3}, \
// {0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3}};
//double gamma_component_sign[16] = {1., 1.,-1., 1., 1., 1.,-1., 1.,-1.,-1., 1.,-1., 1., 1.,-1., 1.};
const int num_component = 4;
int gamma_component[2][4] = { {0, 1, 2, 3},
{0, 1, 2, 3} };
double gamma_component_sign[4] = {+1.,+1.,+1.,+1.};
/*
*******************************************************************/
fftw_complex *in=NULL;
#ifdef MPI
fftwnd_mpi_plan plan_p;
#else
fftwnd_plan plan_p;
#endif
#ifdef MPI
MPI_Status status;
#endif
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "ah?vgf:t:F:p:P:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'a':
write_ascii = 1;
fprintf(stdout, "# [] will write in ascii format\n");
break;
case 'F':
if(strcmp(optarg, "Wilson") == 0) {
fermion_type = _WILSON_FERMION;
} else if(strcmp(optarg, "tm") == 0) {
fermion_type = _TM_FERMION;
} else {
fprintf(stderr, "[] Error, unrecognized fermion type\n");
exit(145);
}
fprintf(stdout, "# [] will use fermion type %s ---> no. %d\n", optarg, fermion_type);
break;
case 't':
num_threads = atoi(optarg);
fprintf(stdout, "# [] number of threads set to %d\n", num_threads);
break;
case 's':
use_lattice_momenta = 1;
fprintf(stdout, "# [] will use lattice momenta\n");
break;
case 'p':
rel_momentum_filename_set = 1;
strcpy(rel_momentum_filename, optarg);
fprintf(stdout, "# [] will use current momentum file %s\n", rel_momentum_filename);
break;
case 'P':
snk_momentum_filename_set = 1;
strcpy(snk_momentum_filename, optarg);
fprintf(stdout, "# [] will use nucleon momentum file %s\n", snk_momentum_filename);
break;
case 'g':
do_gt = 1;
fprintf(stdout, "# [] will perform gauge transform\n");
break;
case 'h':
case '?':
default:
usage();
break;
}
}
#ifdef OPENMP
omp_set_num_threads(num_threads);
#endif
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
#ifdef OPENMP
status = fftw_threads_init();
if(status != 0) {
fprintf(stderr, "\n[] Error from fftw_init_threads; status was %d\n", status);
exit(120);
}
#endif
/******************************************************
*
******************************************************/
VOL3 = LX*LY*LZ;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
if(N_Jacobi>0) {
// alloc the gauge field
alloc_gauge_field(&g_gauge_field, VOL3);
switch(g_gauge_file_format) {
case 0:
sprintf(gauge_field_filename, "%s.%.4d", gaugefilename_prefix, Nconf);
break;
case 1:
sprintf(gauge_field_filename, "%s.%.5d", gaugefilename_prefix, Nconf);
break;
}
} else {
g_gauge_field = NULL;
}
/*********************************************************************
* gauge transformation
*********************************************************************/
if(do_gt) { init_gauge_trafo(&gauge_trafo, 1.); }
// determine the source location
sx0 = g_source_location/(LX*LY*LZ)-Tstart;
sx1 = (g_source_location%(LX*LY*LZ)) / (LY*LZ);
sx2 = (g_source_location%(LY*LZ)) / LZ;
sx3 = (g_source_location%LZ);
// g_source_time_slice = sx0;
fprintf(stdout, "# [] source location %d = (%d,%d,%d,%d)\n", g_source_location, sx0, sx1, sx2, sx3);
source_timeslice = sx0;
if(!use_lattice_momenta) {
status = make_qcont_orbits_3d_parity_avg(&qlatt_id, &qlatt_count, &qlatt_list, &qlatt_nclass, &qlatt_rep, &qlatt_map);
} else {
status = make_qlatt_orbits_3d_parity_avg(&qlatt_id, &qlatt_count, &qlatt_list, &qlatt_nclass, &qlatt_rep, &qlatt_map);
}
if(status != 0) {
fprintf(stderr, "\n[] Error while creating h4-lists\n");
exit(4);
}
fprintf(stdout, "# [] number of classes = %d\n", qlatt_nclass);
/***************************************************************************
* read the relative momenta q to be used
***************************************************************************/
/*
ofs = fopen(rel_momentum_filename, "r");
if(ofs == NULL) {
fprintf(stderr, "[] Error, could not open file %s for reading\n", rel_momentum_filename);
exit(6);
}
rel_momentum_no = 0;
while( fgets(line, 199, ofs) != NULL) {
if(line[0] != '#') {
rel_momentum_no++;
}
}
if(rel_momentum_no == 0) {
fprintf(stderr, "[] Error, number of momenta is zero\n");
exit(7);
} else {
fprintf(stdout, "# [] number of current momenta = %d\n", rel_momentum_no);
}
rewind(ofs);
rel_momentum_list = (int**)malloc(rel_momentum_no * sizeof(int*));
rel_momentum_list[0] = (int*)malloc(3*rel_momentum_no * sizeof(int));
for(i=1;i<rel_momentum_no;i++) { rel_momentum_list[i] = rel_momentum_list[i-1] + 3; }
count=0;
while( fgets(line, 199, ofs) != NULL) {
if(line[0] != '#') {
sscanf(line, "%d%d%d", rel_momentum_list[count], rel_momentum_list[count]+1, rel_momentum_list[count]+2);
count++;
}
}
fclose(ofs);
fprintf(stdout, "# [] current momentum list:\n");
for(i=0;i<rel_momentum_no;i++) {
fprintf(stdout, "\t%3d%3d%3d%3d\n", i, rel_momentum_list[i][0], rel_momentum_list[i][1], rel_momentum_list[i][2]);
}
*/
/***************************************************************************
* read the nucleon final momenta to be used
***************************************************************************/
ofs = fopen(snk_momentum_filename, "r");
if(ofs == NULL) {
fprintf(stderr, "[] Error, could not open file %s for reading\n", snk_momentum_filename);
exit(6);
}
snk_momentum_no = 0;
while( fgets(line, 199, ofs) != NULL) {
if(line[0] != '#') {
snk_momentum_no++;
}
}
if(snk_momentum_no == 0) {
fprintf(stderr, "[] Error, number of momenta is zero\n");
exit(7);
} else {
fprintf(stdout, "# [] number of nucleon final momenta = %d\n", snk_momentum_no);
}
rewind(ofs);
snk_momentum_list = (int**)malloc(snk_momentum_no * sizeof(int*));
snk_momentum_list[0] = (int*)malloc(3*snk_momentum_no * sizeof(int));
for(i=1;i<snk_momentum_no;i++) { snk_momentum_list[i] = snk_momentum_list[i-1] + 3; }
count=0;
while( fgets(line, 199, ofs) != NULL) {
if(line[0] != '#') {
sscanf(line, "%d%d%d", snk_momentum_list[count], snk_momentum_list[count]+1, snk_momentum_list[count]+2);
count++;
}
}
fclose(ofs);
fprintf(stdout, "# [] the nucleon final momentum list:\n");
for(i=0;i<snk_momentum_no;i++) {
fprintf(stdout, "\t%3d%3d%3d%3d\n", i, snk_momentum_list[i][0], snk_momentum_list[i][1], snk_momentum_list[i][1], snk_momentum_list[i][2]);
}
/***********************************************************
* allocate memory for the spinor fields
***********************************************************/
g_spinor_field = NULL;
if(fermion_type == _TM_FERMION) {
no_fields = 2*n_s*n_c+3;
} else {
no_fields = n_s*n_c+3;
}
if(N_Jacobi>0) no_fields++;
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields-2; i++) alloc_spinor_field(&g_spinor_field[i], VOL3);
// work
if(N_Jacobi>0) work = g_spinor_field[no_fields-4];
// stochastic_fv
stochastic_fv = g_spinor_field[no_fields-3];
// stochastic source and propagator
alloc_spinor_field(&g_spinor_field[no_fields-2], VOLUME);
stochastic_source = g_spinor_field[no_fields-2];
alloc_spinor_field(&g_spinor_field[no_fields-1], VOLUME);
stochastic_prop = g_spinor_field[no_fields-1];
spinor_field_checksum = (DML_Checksum*)malloc(no_fields * sizeof(DML_Checksum) );
if(spinor_field_checksum == NULL ) {
fprintf(stderr, "[] Error, could not alloc checksums for spinor fields\n");
exit(73);
}
/*************************************************
* allocate memory for the contractions
*************************************************/
items = 4* num_component*T;
bytes = sizeof(double);
connt = (double*)malloc(items*bytes);
if(connt == NULL) {
fprintf(stderr, "\n[] Error, could not alloc connt\n");
exit(2);
}
for(ix=0; ix<items; ix++) connt[ix] = 0.;
items = num_component * (size_t)VOL3;
connq = create_sp_field( items );
if(connq == NULL) {
fprintf(stderr, "\n[] Error, could not alloc connq\n");
exit(2);
}
items = (size_t)VOL3;
stochastic_fp = create_sp_field( items );
if(stochastic_fp== NULL) {
fprintf(stderr, "\n[] Error, could not alloc stochastic_fp\n");
exit(22);
}
/******************************************************
* initialize FFTW
******************************************************/
items = g_fv_dim * (size_t)VOL3;
bytes = sizeof(fftw_complex);
in = (fftw_complex*)malloc( items * bytes );
if(in == NULL) {
fprintf(stderr, "[] Error, could not malloc in for FFTW\n");
exit(155);
}
dims[0]=LX; dims[1]=LY; dims[2]=LZ;
//plan_p = fftwnd_create_plan(3, dims, FFTW_FORWARD, FFTW_MEASURE | FFTW_IN_PLACE);
plan_p = fftwnd_create_plan_specific(3, dims, FFTW_FORWARD, FFTW_MEASURE, in, g_fv_dim, (fftw_complex*)( stochastic_fv ), g_fv_dim);
// create the fermion propagator points
create_fp(&uprop);
create_fp(&dprop);
create_fp(&fp1);
create_fp(&fp2);
create_fp(&fp3);
create_fp(&stochastic_fp);
create_sp(&sp1);
create_sp(&sp2);
// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
// !! implement twisting for _TM_FERMION
// !!
// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
#ifdef OPENMP
#pragma omp parallel for private(ix) shared(stochastic_prop)
#endif
for(ix=0;ix<VOLUME;ix++) { _fv_eq_zero(stochastic_prop+_GSI(ix)); }
for(sid=g_sourceid; sid<=g_sourceid2;sid+=g_sourceid_step) {
switch(g_soruce_type) {
case 2: // timeslice source
sprintf(filename, "%s.%.4d.%.2d.%.5d.inverted", filename_prefix, Nconf, source_timeslice, sid);
break;
default:
fprintf(stderr, "# [] source type %d not implented; exit\n", g_source_type);
exit(100);
}
fprintf(stdout, "# [] trying to read sample up-prop. from file %s\n", filename);
read_lime_spinor(stochastic_source, filename, 0);
#ifdef OPENMP
#pragma omp parallel for private(ix) shared(stochastic_prop, stochastic_source)
#endif
for(ix=0;ix<VOLUME;ix++) { _fv_pl_eq_fv(stochastic_prop+_GSI(ix), stochastic_source+_GSI(ix)); }
}
#ifdef OPENMP
#pragma omp parallel for private(ix) shared(stochastic_prop, stochastic_source)
#endif
fnorm = 1. / ( (double)(g_sourceid2 - g_sourceid + 1) * g_prop_normsqr );
for(ix=0;ix<VOLUME;ix++) { _fv_ti_eq_re(stochastic_prop+_GSI(ix), fnorm); }
// calculate the source
if(fermion_type && g_propagator_bc_type == 1) {
Q_Wilson_phi(stochastic_source, stochastic_prop);
} else {
Q_phi_tbc(stochastic_source, stochastic_prop);
}
/******************************************************
* prepare the stochastic fermion field
******************************************************/
// read timeslice of the gauge field
if( N_Jacobi>0) {
switch(g_gauge_file_format) {
case 0:
status = read_lime_gauge_field_doubleprec_timeslice(g_gauge_field, gauge_field_filename, source_timeslice, &ildg_gauge_field_checksum);
break;
case 1:
status = read_nersc_gauge_field_timeslice(g_gauge_field, gauge_field_filename, source_timeslice, &nersc_gauge_field_checksum);
break;
}
if(status != 0) {
fprintf(stderr, "[] Error, could not read gauge field\n");
exit(21);
}
for(i=0; i<N_ape; i++) {
#ifdef OPENMP
status = APE_Smearing_Step_Timeslice_threads(g_gauge_field, alpha_ape);
#else
status = APE_Smearing_Step_Timeslice(g_gauge_field, alpha_ape);
#endif
}
}
// read timeslice of the 12 up-type propagators and smear them
//
// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
// !! implement twisting for _TM_FERMION
// !!
// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
for(is=0;is<n_s*n_c;is++) {
if(fermion_type != _TM_FERMION) {
sprintf(filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.%.2d.inverted", filename_prefix, Nconf, sx0, sx1, sx2, sx3, is);
} else {
sprintf(filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.%.2d.inverted", filename_prefix2, Nconf, sx0, sx1, sx2, sx3, is);
}
status = read_lime_spinor_timeslice(g_spinor_field[is], source_timeslice, filename, 0, spinor_field_checksum+is);
if(status != 0) {
fprintf(stderr, "[] Error, could not read propagator from file %s\n", filename);
exit(102);
}
if(N_Jacobi > 0) {
fprintf(stdout, "# [] Jacobi smearing propagator no. %d with paramters N_Jacobi=%d, kappa_Jacobi=%f\n",
is, N_Jacobi, kappa_Jacobi);
for(c=0; c<N_Jacobi; c++) {
#ifdef OPENMP
Jacobi_Smearing_Step_one_Timeslice_threads(g_gauge_field, g_spinor_field[is], work, kappa_Jacobi);
#else
Jacobi_Smearing_Step_one_Timeslice(g_gauge_field, g_spinor_field[is], work, kappa_Jacobi);
#endif
}
}
}
for(is=0;is<g_fv_dim;is++) {
for(ix=0;ix<VOL3;ix++) {
iix = source_timeslice * VOL3 + ix;
_fv_eq_gamma_ti_fv(spinor1, 5, g_spinor_field[is]+_GSI(iix));
_co_eq_fv_dagger_ti_fv(&w, stochastic_source+_GSI(ix), spinor1);
stochastic_fv[_GSI(ix)+2*is ] = w.re;
stochastic_fv[_GSI(ix)+2*is+1] = w.im;
}
}
// Fourier transform
items = g_fv_dim * (size_t)VOL3;
bytes = sizeof(double);
memcpy(in, stochastic_fv, items*bytes );
#ifdef OPENMP
fftwnd_threads(num_threads, plan_p, g_fv_dim, in, g_fv_dim, 1, (fftw_complex*)(stochastic_fv), g_fv_dim, 1);
#else
fftwnd(plan_p, g_fv_dim, in, g_fv_dim, 1, (fftw_complex*)(stochastic_fv), g_fv_dim, 1);
#endif
/******************************************************
* loop on sink momenta (most likely only one: Q=(0,0,0))
******************************************************/
for(imom_snk=0;imom_snk<snk_momentum_no; imom_snk++) {
// create Phi_tilde
_fv_eq_zero( spinor2 );
for(ix=0;ix<LX;ix++) {
for(iy=0;iy<LY;iy++) {
for(iz=0;iz<LZ;iz++) {
iix = timeslice * VOL3 + ix;
phase = -2.*M_PI*( (ix-sx1) * snk_momentum_list[imom_snk][0] / (double)LX
+ (iy-sx2) * snk_momentum_list[imom_snk][1] / (double)LY
+ (iz-sx3) * snk_momentum_list[imom_snk][2] / (double)LZ);
w.re = cos(phase);
w.im = sin(phase);
_fv_eq_fv_ti_co(spinor1, stochastic_prop + _GSI(iix), &w);
_fv_pl_eq_fv(spinor2, spinor);
}}}
// create Theta
for(ir=0;ir<g_fv_dim;ir++) {
for(is=0;is<g_fv_dim;is++) {
_co_eq_co_ti_co( &(stochastic_fp[ix][ir][2*is]), &(spinor2[2*ir]), &(stochastic_fv[_GSI(ix)+2*is]) );
}}
/******************************************************
* loop on timeslices
******************************************************/
for(timeslice=0; timeslice<T; timeslice++) {
append = (int)( timeslice != 0 );
// read timeslice of the gauge field
if( N_Jacobi>0) {
switch(g_gauge_file_format) {
case 0:
status = read_lime_gauge_field_doubleprec_timeslice(g_gauge_field, gauge_field_filename, timeslice, &ildg_gauge_field_checksum);
break;
case 1:
status = read_nersc_gauge_field_timeslice(g_gauge_field, gauge_field_filename, timeslice, &nersc_gauge_field_checksum);
break;
}
if(status != 0) {
fprintf(stderr, "[] Error, could not read gauge field\n");
exit(21);
}
for(i=0; i<N_ape; i++) {
#ifdef OPENMP
status = APE_Smearing_Step_Timeslice_threads(g_gauge_field, alpha_ape);
#else
status = APE_Smearing_Step_Timeslice(g_gauge_field, alpha_ape);
#endif
}
}
// read timeslice of the 12 up-type propagators and smear them
for(is=0;is<n_s*n_c;is++) {
sprintf(filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.%.2d.inverted", filename_prefix, Nconf, sx0, sx1, sx2, sx3, is);
status = read_lime_spinor_timeslice(g_spinor_field[is], timeslice, filename, 0, spinor_field_checksum+is);
if(status != 0) {
fprintf(stderr, "[] Error, could not read propagator from file %s\n", filename);
exit(102);
}
if(N_Jacobi > 0) {
fprintf(stdout, "# [] Jacobi smearing propagator no. %d with paramters N_Jacobi=%d, kappa_Jacobi=%f\n",
is, N_Jacobi, kappa_Jacobi);
for(c=0; c<N_Jacobi; c++) {
#ifdef OPENMP
Jacobi_Smearing_Step_one_Timeslice_threads(g_gauge_field, g_spinor_field[is], work, kappa_Jacobi);
#else
Jacobi_Smearing_Step_one_Timeslice(g_gauge_field, g_spinor_field[is], work, kappa_Jacobi);
#endif
}
}
}
if(fermion_type == _TM_FERMION) {
// read timeslice of the 12 down-type propagators, smear them
for(is=0;is<n_s*n_c;is++) {
if(do_gt == 0) {
sprintf(filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.%.2d.inverted", filename_prefix2, Nconf, sx0, sx1, sx2, sx3, is);
status = read_lime_spinor_timeslice(g_spinor_field[n_s*n_c+is], timeslice, filename, 0, spinor_field_checksum+n_s*n_c+is);
if(status != 0) {
fprintf(stderr, "[] Error, could not read propagator from file %s\n", filename);
exit(102);
}
if(N_Jacobi > 0) {
fprintf(stdout, "# [] Jacobi smearing propagator no. %d with paramters N_Jacobi=%d, kappa_Jacobi=%f\n",
is, N_Jacobi, kappa_Jacobi);
for(c=0; c<N_Jacobi; c++) {
#ifdef OPENMP
Jacobi_Smearing_Step_one_Timeslice_threads(g_gauge_field, g_spinor_field[n_s*n_c+is], work, kappa_Jacobi);
#else
Jacobi_Smearing_Step_one_Timeslice(g_gauge_field, g_spinor_field[n_s*n_c+is], work, kappa_Jacobi);
#endif
}
}
}
}
/******************************************************
* contractions
******************************************************/
for(ix=0;ix<VOL3;ix++)
//for(ix=0;ix<1;ix++)
{
// assign the propagators
_assign_fp_point_from_field(uprop, g_spinor_field, ix);
if(fermion_type==_TM_FERMION) {
_assign_fp_point_from_field(dprop, g_spinor_field+n_s*n_c, ix);
} else {
_fp_eq_fp(dprop, uprop);
}
flavor rotation for twisted mass fermions
if(fermion_type == _TM_FERMION) {
_fp_eq_rot_ti_fp(fp1, uprop, +1, fermion_type, fp2);
_fp_eq_fp_ti_rot(uprop, fp1, +1, fermion_type, fp2);
// _fp_eq_rot_ti_fp(fp1, dprop, -1, fermion_type, fp2);
// _fp_eq_fp_ti_rot(dprop, fp1, -1, fermion_type, fp2);
}
// test: print fermion propagator point
//printf_fp(uprop, stdout);
for(icomp=0; icomp<num_component; icomp++) {
_sp_eq_zero( connq[ix*num_component+icomp]);
/******************************************************
* first contribution
******************************************************/
_fp_eq_zero(fp1);
_fp_eq_zero(fp2);
_fp_eq_zero(fp3);
// C Gamma_1 x S_u = g0 g2 Gamma_1 S_u
_fp_eq_gamma_ti_fp(fp1, gamma_component[0][icomp], uprop);
_fp_eq_gamma_ti_fp(fp3, 2, fp1);
_fp_eq_gamma_ti_fp(fp1, 0, fp3);
// S_u x C Gamma_2 = S_u x g0 g2 Gamma_2
_fp_eq_fp_ti_gamma(fp2, 0, uprop);
_fp_eq_fp_ti_gamma(fp3, 2, fp2);
_fp_eq_fp_ti_gamma(fp2, gamma_component[1][icomp], fp3);
// first part
// reduce
_fp_eq_zero(fp3);
_fp_eq_fp_eps_contract13_fp(fp3, fp1, uprop);
// reduce to spin propagator
_sp_eq_zero( sp1 );
_sp_eq_fp_del_contract23_fp(sp1, fp2, fp3);
// second part
// reduce to spin propagator
_sp_eq_zero( sp2 );
_sp_eq_fp_del_contract24_fp(sp2, fp2, fp3);
// add and assign
_sp_pl_eq_sp(sp1, sp2);
_sp_eq_sp_ti_re(sp2, sp1, -gamma_component_sign[icomp]);
_sp_eq_sp( connq[ix*num_component+icomp], sp2);
/******************************************************
* second contribution
******************************************************/
_fp_eq_zero(fp1);
_fp_eq_zero(fp2);
_fp_eq_zero(fp3);
// first part
// C Gamma_1 x S_u = g0 g2 Gamma_1 S_u
_fp_eq_gamma_ti_fp(fp1, gamma_component[0][icomp], uprop);
_fp_eq_gamma_ti_fp(fp3, 2, fp1);
_fp_eq_gamma_ti_fp(fp1, 0, fp3);
// S_u x C Gamma_2 = S_u g0 g2 Gamma_2 (same S_u as above)
_fp_eq_fp_ti_gamma(fp2, 0, fp1);
_fp_eq_fp_ti_gamma(fp3, 2, fp2);
_fp_eq_fp_ti_gamma(fp1, gamma_component[1][icomp], fp3);
// reduce
_fp_eq_zero(fp3);
_fp_eq_fp_eps_contract13_fp(fp3, fp1, uprop);
// reduce to spin propagator
_sp_eq_zero( sp1 );
_sp_eq_fp_del_contract23_fp(sp1, uprop, fp3);
// second part
// C Gamma_1 x S_u = g0 g2 Gamma_1 S_u
_fp_eq_gamma_ti_fp(fp1, gamma_component[0][icomp], uprop);
_fp_eq_gamma_ti_fp(fp3, 2, fp1);
_fp_eq_gamma_ti_fp(fp1, 0, fp3);
// S_u x C Gamma_2 = S_u g0 g2 Gamma_2
_fp_eq_fp_ti_gamma(fp2, 0, uprop);
_fp_eq_fp_ti_gamma(fp3, 2, fp2);
_fp_eq_fp_ti_gamma(fp2, gamma_component[1][icomp], fp3);
// reduce
_fp_eq_zero(fp3);
_fp_eq_fp_eps_contract13_fp(fp3, fp1, fp2);
// reduce to spin propagator
_sp_eq_zero( sp2 );
_sp_eq_fp_del_contract24_fp(sp2, uprop, fp3);
// add and assign
_sp_pl_eq_sp(sp1, sp2);
_sp_eq_sp_ti_re(sp2, sp1, -gamma_component_sign[icomp]);
_sp_pl_eq_sp( connq[ix*num_component+icomp], sp2);
/******************************************************
* third contribution
******************************************************/
_fp_eq_zero(fp1);
_fp_eq_zero(fp2);
_fp_eq_zero(fp3);
// first part
// C Gamma_1 x S_u = g0 g2 Gamma_1 S_u
_fp_eq_gamma_ti_fp(fp1, gamma_component[0][icomp], uprop);
_fp_eq_gamma_ti_fp(fp3, 2, fp1);
_fp_eq_gamma_ti_fp(fp1, 0, fp3);
// S_u x C Gamma_2 = S_u g0 g2 Gamma_2
_fp_eq_fp_ti_gamma(fp2, 0, fp1);
_fp_eq_fp_ti_gamma(fp3, 2, fp2);
_fp_eq_fp_ti_gamma(fp1, gamma_component[1][icomp], fp3);
// reduce
_fp_eq_zero(fp3);
_fp_eq_fp_eps_contract13_fp(fp3, fp1, uprop);
// reduce to spin propagator
_sp_eq_zero( sp1 );
_sp_eq_fp_del_contract34_fp(sp1, uprop, fp3);
// second part
// C Gamma_1 x S_u = g0 g2 Gamma_1 S_u
_fp_eq_gamma_ti_fp(fp1, gamma_component[0][icomp], uprop);
_fp_eq_gamma_ti_fp(fp3, 2, fp1);
_fp_eq_gamma_ti_fp(fp1, 0, fp3);
// S_u x C Gamma_2 = S_u g0 g2 Gamma_2
_fp_eq_fp_ti_gamma(fp2, 0, uprop);
_fp_eq_fp_ti_gamma(fp3, 2, fp2);
_fp_eq_fp_ti_gamma(fp2, gamma_component[1][icomp], fp3);
// reduce
_fp_eq_zero(fp3);
_fp_eq_fp_eps_contract13_fp(fp3, fp1, fp2);
// reduce to spin propagator
_sp_eq_zero( sp2 );
_sp_eq_fp_del_contract34_fp(sp2, uprop, fp3);
// add and assign
_sp_pl_eq_sp(sp1, sp2);
_sp_eq_sp_ti_re(sp2, sp1, -gamma_component_sign[icomp]);
_sp_pl_eq_sp( connq[ix*num_component+icomp], sp2);
} // of icomp
} // of ix
/***********************************************
* finish calculation of connq
***********************************************/
if(g_propagator_bc_type == 0) {
// multiply with phase factor
fprintf(stdout, "# [] multiplying timeslice %d with boundary phase factor\n", timeslice);
ir = (timeslice - sx0 + T_global) % T_global;
w1.re = cos( 3. * M_PI*(double)ir / (double)T_global );
w1.im = sin( 3. * M_PI*(double)ir / (double)T_global );
for(ix=0;ix<num_component*VOL3;ix++) {
_sp_eq_sp(sp1, connq[ix] );
_sp_eq_sp_ti_co( connq[ix], sp1, w1);
}
} else if (g_propagator_bc_type == 1) {
// multiply with step function
if(timeslice < sx0) {
fprintf(stdout, "# [] multiplying timeslice %d with boundary step function\n", timeslice);
for(ix=0;ix<num_component*VOL3;ix++) {
_sp_eq_sp(sp1, connq[ix] );
_sp_eq_sp_ti_re( connq[ix], sp1, -1.);
}
}
}
if(write_ascii) {
sprintf(filename, "%s_x.%.4d.t%.2dx%.2dy%.2dz%.2d.ascii", outfile_prefix, Nconf, sx0, sx1, sx2, sx3);
write_contraction2( connq[0][0], filename, num_component*g_sv_dim*g_sv_dim, VOL3, 1, append);
}
/******************************************************************
* Fourier transform
******************************************************************/
items = 2 * num_component * g_sv_dim * g_sv_dim * VOL3;
bytes = sizeof(double);
memcpy(in, connq[0][0], items * bytes);
ir = num_component * g_sv_dim * g_sv_dim;
#ifdef OPENMP
fftwnd_threads(num_threads, plan_p, ir, in, ir, 1, (fftw_complex*)(connq[0][0]), ir, 1);
#else
fftwnd(plan_p, ir, in, ir, 1, (fftw_complex*)(connq[0][0]), ir, 1);
#endif
// add phase factor from the source location
iix = 0;
for(x1=0;x1<LX;x1++) {
q[0] = (double)x1 / (double)LX;
for(x2=0;x2<LY;x2++) {
q[1] = (double)x2 / (double)LY;
for(x3=0;x3<LZ;x3++) {
q[2] = (double)x3 / (double)LZ;
phase = 2. * M_PI * ( q[0]*sx1 + q[1]*sx2 + q[2]*sx3 );
w1.re = cos(phase);
w1.im = sin(phase);
for(icomp=0; icomp<num_component; icomp++) {
_sp_eq_sp(sp1, connq[iix] );
_sp_eq_sp_ti_co( connq[iix], sp1, w1) ;
iix++;
}
}}} // of x3, x2, x1
// write to file
sprintf(filename, "%s_q.%.4d.t%.2dx%.2dy%.2dz%.2d.Qx%.2dQy%.2dQz%.2d.%.5d", outfile_prefix, Nconf, sx0, sx1, sx2, sx3,
qlatt_rep[snk_momentum_list[imom_snk]][1],qlatt_rep[snk_momentum_list[imom_snk]][2],qlatt_rep[snk_momentum_list[imom_snk]][3],
g_sourceid2-g_sourceid+1);
sprintf(contype, "2-pt. function, (t,q_1,q_2,q_3)-dependent, source_timeslice = %d", sx0);
write_lime_contraction_timeslice(connq[0][0], filename, 64, num_component*g_sv_dim*g_sv_dim, contype, Nconf, 0, &connq_checksum, timeslice);
if(write_ascii) {
strcat(filename, ".ascii");
write_contraction2(connq[0][0],filename, num_component*g_sv_dim*g_sv_dim, VOL3, 1, append);
}
/***********************************************
* calculate connt
***********************************************/
for(icomp=0;icomp<num_component; icomp++) {
// fwd
_sp_eq_sp(sp1, connq[icomp]);
_sp_eq_gamma_ti_sp(sp2, 0, sp1);
_sp_pl_eq_sp(sp1, sp2);
_co_eq_tr_sp(&w, sp1);
connt[2*(icomp*T + timeslice) ] = w.re * 0.25;
connt[2*(icomp*T + timeslice)+1] = w.im * 0.25;
// bwd
_sp_eq_sp(sp1, connq[icomp]);
_sp_eq_gamma_ti_sp(sp2, 0, sp1);
_sp_mi_eq_sp(sp1, sp2);
_co_eq_tr_sp(&w, sp1);
connt[2*(icomp*T+timeslice + num_component*T) ] = w.re * 0.25;
connt[2*(icomp*T+timeslice + num_component*T)+1] = w.im * 0.25;
}
} // of loop on timeslice
// write connt
sprintf(filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.fw", outfile_prefix, Nconf, sx0, sx1, sx2, sx3);
ofs = fopen(filename, "w");
if(ofs == NULL) {
fprintf(stderr, "[] Error, could not open file %s for writing\n", filename);
exit(3);
}
fprintf(ofs, "#%12.8f%3d%3d%3d%3d%8.4f%6d\n", g_kappa, T_global, LX, LY, LZ, g_mu, Nconf);
for(icomp=0; icomp<num_component; icomp++) {
ir = sx0;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], 0, connt[2*(icomp*T+ir)], 0., Nconf);
for(it=1;it<T/2;it++) {
ir = ( it + sx0 ) % T_global;
ir2 = ( (T_global - it) + sx0 ) % T_global;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], it, connt[2*(icomp*T+ir)], connt[2*(icomp*T+ir2)], Nconf);
}
ir = ( it + sx0 ) % T_global;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], it, connt[2*(icomp*T+ir)], 0., Nconf);
}
fclose(ofs);
sprintf(filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.bw", outfile_prefix, Nconf, sx0, sx1, sx2, sx3);
ofs = fopen(filename, "w");
if(ofs == NULL) {
fprintf(stderr, "[] Error, could not open file %s for writing\n", filename);
exit(3);
}
fprintf(ofs, "#%12.8f%3d%3d%3d%3d%8.4f%6d\n", g_kappa, T_global, LX, LY, LZ, g_mu, Nconf);
for(icomp=0; icomp<num_component; icomp++) {
ir = sx0;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], 0, connt[2*(num_component*T+icomp*T+ir)], 0., Nconf);
for(it=1;it<T/2;it++) {
ir = ( it + sx0 ) % T_global;
ir2 = ( (T_global - it) + sx0 ) % T_global;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], it, connt[2*(num_component*T+icomp*T+ir)], connt[2*(num_component*T+icomp*T+ir2)], Nconf);
}
ir = ( it + sx0 ) % T_global;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], it, connt[2*(num_component*T+icomp*T+ir)], 0., Nconf);
}
fclose(ofs);
} // of loop on sink momentum ( = Delta^++ momentum, Qvec)
/***********************************************
* free the allocated memory, finalize
***********************************************/
free_geometry();
if(connt!= NULL) free(connt);
if(connq!= NULL) free(connq);
if(gauge_trafo != NULL) free(gauge_trafo);
if(g_spinor_field!=NULL) {
for(i=0; i<no_fields; i++) free(g_spinor_field[i]);
free(g_spinor_field); g_spinor_field=(double**)NULL;
}
if(spinor_field_checksum !=NULL) free(spinor_field_checksum);
if(g_gauge_field != NULL) free(g_gauge_field);
if(snk_momemtum_list != NULL) {
if(snk_momentum_list[0] != NULL) free(snk_momentum_list[0]);
free(snk_momentum_list);
}
if(rel_momemtum_list != NULL) {
if(rel_momentum_list[0] != NULL) free(rel_momentum_list[0]);
free(rel_momentum_list);
}
// free the fermion propagator points
free_fp( &uprop );
free_fp( &dprop );
free_fp( &fp1 );
free_fp( &fp2 );
free_fp( &fp3 );
free_sp( &sp1 );
free_sp( &sp2 );
free(in);
fftwnd_destroy_plan(plan_p);
g_the_time = time(NULL);
fprintf(stdout, "# [] %s# [] end fo run\n", ctime(&g_the_time));
fflush(stdout);
fprintf(stderr, "# [] %s# [] end fo run\n", ctime(&g_the_time));
fflush(stderr);
#ifdef MPI
MPI_Finalize();
#endif
return(0);
}
int main(int argc, char **argv) {
int c, i, mu, status;
int ispin, icol, isc;
int n_c = 3;
int n_s = 4;
int count = 0;
int filename_set = 0;
int dims[4] = {0,0,0,0};
int grid_size[4];
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, iix, iy, is, it, i3;
int sl0, sl1, sl2, sl3, have_source_flag=0;
int source_proc_coords[4], lsl0, lsl1, lsl2, lsl3;
int check_residuum = 0;
unsigned int VOL3, V5;
int do_gt = 0;
int full_orbit = 0;
int smear_source = 0;
char filename[200], source_filename[200], source_filename_write[200];
double ratime, retime;
double plaq_r=0., plaq_m=0., norm, norm2;
double spinor1[24];
double *gauge_qdp[4], *gauge_field_timeslice=NULL, *gauge_field_smeared=NULL;
double _1_2_kappa, _2_kappa, phase;
FILE *ofs;
int mu_trans[4] = {3, 0, 1, 2};
int threadid, nthreads;
int timeslice, source_timeslice;
char rng_file_in[100], rng_file_out[100];
int *source_momentum=NULL;
int source_momentum_class = -1;
int source_momentum_no = 0;
int source_momentum_runs = 1;
int imom;
int num_gpu_on_node=0, rank;
int source_location_5d_iseven;
int convert_sign=0;
#ifdef HAVE_QUDA
int rotate_gamma_basis = 1;
#else
int rotate_gamma_basis = 0;
#endif
omp_lock_t *lck = NULL, gen_lck[1];
int key = 0;
/****************************************************************************/
/* for smearing parallel to inversion */
double *smearing_spinor_field[] = {NULL,NULL};
int dummy_flag = 0;
/****************************************************************************/
/****************************************************************************/
#if (defined HAVE_QUDA) && (defined MULTI_GPU)
int x_face_size, y_face_size, z_face_size, t_face_size, pad_size;
#endif
/****************************************************************************/
/************************************************/
int qlatt_nclass;
int *qlatt_id=NULL, *qlatt_count=NULL, **qlatt_rep=NULL, **qlatt_map=NULL;
double **qlatt_list=NULL;
/************************************************/
/************************************************/
double boundary_condition_factor;
int boundary_condition_factor_set = 0;
/************************************************/
//#ifdef MPI
// kernelPackT = true;
//#endif
/***********************************************
* QUDA parameters
***********************************************/
#ifdef HAVE_QUDA
QudaPrecision cpu_prec = QUDA_DOUBLE_PRECISION;
QudaPrecision cuda_prec = QUDA_DOUBLE_PRECISION;
QudaPrecision cuda_prec_sloppy = QUDA_SINGLE_PRECISION;
QudaGaugeParam gauge_param = newQudaGaugeParam();
QudaInvertParam inv_param = newQudaInvertParam();
#endif
while ((c = getopt(argc, argv, "soch?vgf:p:b:S:R:")) != -1) {
switch (c) {
case 'v':
g_verbose = 1;
break;
case 'g':
do_gt = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'c':
check_residuum = 1;
fprintf(stdout, "# [invert_dw_quda] will check residuum again\n");
break;
case 'p':
n_c = atoi(optarg);
fprintf(stdout, "# [invert_dw_quda] will use number of colors = %d\n", n_c);
break;
case 'o':
full_orbit = 1;
fprintf(stdout, "# [invert_dw_quda] will invert for full orbit, if source momentum set\n");
case 's':
smear_source = 1;
fprintf(stdout, "# [invert_dw_quda] will smear the sources if they are read from file\n");
break;
case 'b':
boundary_condition_factor = atof(optarg);
boundary_condition_factor_set = 1;
fprintf(stdout, "# [invert_dw_quda] const. boundary condition factor set to %e\n", boundary_condition_factor);
break;
case 'S':
convert_sign = atoi(optarg);
fprintf(stdout, "# [invert_dw_quda] using convert sign %d\n", convert_sign);
break;
case 'R':
rotate_gamma_basis = atoi(optarg);
fprintf(stdout, "# [invert_dw_quda] rotate gamma basis %d\n", rotate_gamma_basis);
break;
case 'h':
case '?':
default:
usage();
break;
}
}
// get the time stamp
g_the_time = time(NULL);
/**************************************
* set the default values, read input
**************************************/
if(filename_set==0) strcpy(filename, "cvc.input");
if(g_proc_id==0) fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
#ifdef MPI
#ifdef HAVE_QUDA
grid_size[0] = g_nproc_x;
grid_size[1] = g_nproc_y;
grid_size[2] = g_nproc_z;
grid_size[3] = g_nproc_t;
fprintf(stdout, "# [] g_nproc = (%d,%d,%d,%d)\n", g_nproc_x, g_nproc_y, g_nproc_z, g_nproc_t);
initCommsQuda(argc, argv, grid_size, 4);
#else
MPI_Init(&argc, &argv);
#endif
#endif
#if (defined PARALLELTX) || (defined PARALLELTXY)
EXIT_WITH_MSG(1, "[] Error, 2-dim./3-dim. MPI-Version not yet implemented");
#endif
// some checks on the input data
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stderr, "[invert_dw_quda] Error, T and L's must be set\n");
usage();
}
// set number of openmp threads
// initialize MPI parameters
mpi_init(argc, argv);
// the volume of a timeslice
VOL3 = LX*LY*LZ;
V5 = T*LX*LY*LZ*L5;
g_kappa5d = 0.5 / (5. + g_m5);
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] kappa5d = %e\n", g_kappa5d);
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] L5 = %3d\n",\
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, L5);
#ifdef MPI
if(T==0) {
fprintf(stderr, "[%2d] local T is zero; exit\n", g_cart_id);
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
exit(2);
}
#endif
if(init_geometry() != 0) {
fprintf(stderr, "[invert_dw_quda] Error from init_geometry\n");
EXIT(1);
}
geometry();
if( init_geometry_5d() != 0 ) {
fprintf(stderr, "[invert_dw_quda] Error from init_geometry_5d\n");
EXIT(2);
}
geometry_5d();
/**************************************
* initialize the QUDA library
**************************************/
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] initializing quda\n");
#ifdef HAVE_QUDA
// cudaGetDeviceCount(&num_gpu_on_node);
if(g_gpu_per_node<0) {
if(g_cart_id==0) fprintf(stderr, "[] Error, number of GPUs per node not set\n");
EXIT(106);
} else {
num_gpu_on_node = g_gpu_per_node;
}
#ifdef MPI
rank = comm_rank();
#else
rank = 0;
#endif
g_gpu_device_number = rank % num_gpu_on_node;
fprintf(stdout, "# [] process %d/%d uses device %d\n", rank, g_cart_id, g_gpu_device_number);
initQuda(g_gpu_device_number);
#endif
/**************************************
* prepare the gauge field
**************************************/
// read the gauge field from file
alloc_gauge_field(&g_gauge_field, VOLUMEPLUSRAND);
if(strcmp( gaugefilename_prefix, "identity")==0 ) {
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] Setting up unit gauge field\n");
for(ix=0;ix<VOLUME; ix++) {
for(mu=0;mu<4;mu++) {
_cm_eq_id(g_gauge_field+_GGI(ix,mu));
}
}
} else if(strcmp( gaugefilename_prefix, "random")==0 ) {
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] Setting up random gauge field with seed = %d\n", g_seed);
init_rng_state(g_seed, &g_rng_state);
random_gauge_field(g_gauge_field, 1.);
plaquette(&plaq_m);
sprintf(filename, "%s.%.4d", gaugefilename_prefix, Nconf);
check_error(write_lime_gauge_field(filename, plaq_m, Nconf, 64), "write_lime_gauge_field", NULL, 12);
} else {
if(g_gauge_file_format == 0) {
// ILDG
sprintf(filename, "%s.%.4d", gaugefilename_prefix, Nconf);
if(g_cart_id==0) fprintf(stdout, "# Reading gauge field from file %s\n", filename);
status = read_lime_gauge_field_doubleprec(filename);
} else if(g_gauge_file_format == 1) {
// NERSC
sprintf(filename, "%s.%.5d", gaugefilename_prefix, Nconf);
if(g_cart_id==0) fprintf(stdout, "# Reading gauge field from file %s\n", filename);
status = read_nersc_gauge_field(g_gauge_field, filename, &plaq_r);
//status = read_nersc_gauge_field_3x3(g_gauge_field, filename, &plaq_r);
}
if(status != 0) {
fprintf(stderr, "[invert_dw_quda] Error, could not read gauge field");
EXIT(12);
}
}
#ifdef MPI
xchange_gauge();
#endif
// measure the plaquette
plaquette(&plaq_m);
if(g_cart_id==0) fprintf(stdout, "# Measured plaquette value: %25.16e\n", plaq_m);
if(g_cart_id==0) fprintf(stdout, "# Read plaquette value : %25.16e\n", plaq_r);
#ifndef HAVE_QUDA
if(N_Jacobi>0) {
#endif
// allocate the smeared / qdp ordered gauge field
alloc_gauge_field(&gauge_field_smeared, VOLUMEPLUSRAND);
for(i=0;i<4;i++) {
gauge_qdp[i] = gauge_field_smeared + i*18*VOLUME;
}
#ifndef HAVE_QUDA
}
#endif
#ifdef HAVE_QUDA
// transcribe the gauge field
omp_set_num_threads(g_num_threads);
#pragma omp parallel for private(ix,iy,mu)
for(ix=0;ix<VOLUME;ix++) {
iy = g_lexic2eot[ix];
for(mu=0;mu<4;mu++) {
_cm_eq_cm(gauge_qdp[mu_trans[mu]]+18*iy, g_gauge_field+_GGI(ix,mu));
}
}
// multiply timeslice T-1 with factor of -1 (antiperiodic boundary condition)
if(g_proc_coords[0]==g_nproc_t-1) {
if(!boundary_condition_factor_set) boundary_condition_factor = -1.;
fprintf(stdout, "# [] process %d multiplies gauge-field timeslice T_global-1 with boundary condition factor %e\n", g_cart_id,
boundary_condition_factor);
omp_set_num_threads(g_num_threads);
#pragma omp parallel for private(ix,iy)
for(ix=0;ix<VOL3;ix++) {
iix = (T-1)*VOL3 + ix;
iy = g_lexic2eot[iix];
_cm_ti_eq_re(gauge_qdp[mu_trans[0]]+18*iy, -1.);
}
}
// QUDA precision parameters
switch(g_cpu_prec) {
case 0: cpu_prec = QUDA_HALF_PRECISION; if(g_cart_id==0) fprintf(stdout, "# [] CPU prec = half\n"); break;
case 1: cpu_prec = QUDA_SINGLE_PRECISION; if(g_cart_id==0) fprintf(stdout, "# [] CPU prec = single\n"); break;
case 2: cpu_prec = QUDA_DOUBLE_PRECISION; if(g_cart_id==0) fprintf(stdout, "# [] CPU prec = double\n"); break;
default: cpu_prec = QUDA_DOUBLE_PRECISION; break;
}
switch(g_gpu_prec) {
case 0: cuda_prec = QUDA_HALF_PRECISION; if(g_cart_id==0) fprintf(stdout, "# [] GPU prec = half\n"); break;
case 1: cuda_prec = QUDA_SINGLE_PRECISION; if(g_cart_id==0) fprintf(stdout, "# [] GPU prec = single\n"); break;
case 2: cuda_prec = QUDA_DOUBLE_PRECISION; if(g_cart_id==0) fprintf(stdout, "# [] GPU prec = double\n"); break;
default: cuda_prec = QUDA_DOUBLE_PRECISION; break;
}
switch(g_gpu_prec_sloppy) {
case 0: cuda_prec_sloppy = QUDA_HALF_PRECISION; if(g_cart_id==0) fprintf(stdout, "# [] GPU sloppy prec = half\n"); break;
case 1: cuda_prec_sloppy = QUDA_SINGLE_PRECISION; if(g_cart_id==0) fprintf(stdout, "# [] GPU sloppy prec = single\n"); break;
case 2: cuda_prec_sloppy = QUDA_DOUBLE_PRECISION; if(g_cart_id==0) fprintf(stdout, "# [] GPU sloppy prec = double\n"); break;
default: cuda_prec_sloppy = QUDA_SINGLE_PRECISION; break;
}
// QUDA gauge parameters
gauge_param.X[0] = LX;
gauge_param.X[1] = LY;
gauge_param.X[2] = LZ;
gauge_param.X[3] = T;
inv_param.Ls = L5;
gauge_param.anisotropy = 1.0;
gauge_param.type = QUDA_WILSON_LINKS;
gauge_param.gauge_order = QUDA_QDP_GAUGE_ORDER;
gauge_param.t_boundary = QUDA_ANTI_PERIODIC_T;
gauge_param.cpu_prec = cpu_prec;
gauge_param.cuda_prec = cuda_prec;
gauge_param.reconstruct = QUDA_RECONSTRUCT_12;
gauge_param.cuda_prec_sloppy = cuda_prec_sloppy;
gauge_param.reconstruct_sloppy = QUDA_RECONSTRUCT_12;
gauge_param.gauge_fix = QUDA_GAUGE_FIXED_NO;
gauge_param.ga_pad = 0;
inv_param.sp_pad = 0;
inv_param.cl_pad = 0;
// For multi-GPU, ga_pad must be large enough to store a time-slice
#ifdef MULTI_GPU
x_face_size = inv_param.Ls * gauge_param.X[1]*gauge_param.X[2]*gauge_param.X[3]/2;
y_face_size = inv_param.Ls * gauge_param.X[0]*gauge_param.X[2]*gauge_param.X[3]/2;
z_face_size = inv_param.Ls * gauge_param.X[0]*gauge_param.X[1]*gauge_param.X[3]/2;
t_face_size = inv_param.Ls * gauge_param.X[0]*gauge_param.X[1]*gauge_param.X[2]/2;
pad_size = _MAX(x_face_size, y_face_size);
pad_size = _MAX(pad_size, z_face_size);
pad_size = _MAX(pad_size, t_face_size);
gauge_param.ga_pad = pad_size;
if(g_cart_id==0) printf("# [invert_dw_quda] pad_size = %d\n", pad_size);
#endif
// load the gauge field
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] loading gauge field\n");
loadGaugeQuda((void*)gauge_qdp, &gauge_param);
gauge_qdp[0] = NULL;
gauge_qdp[1] = NULL;
gauge_qdp[2] = NULL;
gauge_qdp[3] = NULL;
#endif
/*********************************************
* APE smear the gauge field
*********************************************/
if(N_Jacobi>0) {
memcpy(gauge_field_smeared, g_gauge_field, 72*VOLUMEPLUSRAND*sizeof(double));
fprintf(stdout, "# [invert_dw_quda] APE smearing gauge field with paramters N_APE=%d, alpha_APE=%e\n", N_ape, alpha_ape);
APE_Smearing_Step_threads(gauge_field_smeared, N_ape, alpha_ape);
xchange_gauge_field(gauge_field_smeared);
}
// allocate memory for the spinor fields
#ifdef HAVE_QUDA
no_fields = 3+2;
#else
no_fields = 6+2;
#endif
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields; i++) alloc_spinor_field(&g_spinor_field[i], VOLUMEPLUSRAND*L5);
smearing_spinor_field[0] = g_spinor_field[no_fields-2];
smearing_spinor_field[1] = g_spinor_field[no_fields-1];
switch(g_source_type) {
case 0:
case 5:
// the source locaton
sl0 = g_source_location / (LX_global*LY_global*LZ);
sl1 = ( g_source_location % (LX_global*LY_global*LZ) ) / ( LY_global*LZ);
sl2 = ( g_source_location % ( LY_global*LZ) ) / ( LZ);
sl3 = g_source_location % LZ;
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] global sl = (%d, %d, %d, %d)\n", sl0, sl1, sl2, sl3);
source_proc_coords[0] = sl0 / T;
source_proc_coords[1] = sl1 / LX;
source_proc_coords[2] = sl2 / LY;
source_proc_coords[3] = sl3 / LZ;
#ifdef MPI
MPI_Cart_rank(g_cart_grid, source_proc_coords, &g_source_proc_id);
#else
g_source_proc_id = 0;
#endif
have_source_flag = g_source_proc_id == g_cart_id;
lsl0 = sl0 % T;
lsl1 = sl1 % LX;
lsl2 = sl2 % LY;
lsl3 = sl3 % LZ;
if(have_source_flag) {
fprintf(stdout, "# [invert_dw_quda] process %d has the source at (%d, %d, %d, %d)\n", g_cart_id, lsl0, lsl1, lsl2, lsl3);
}
break;
case 2:
case 3:
case 4:
// the source timeslice
#ifdef MPI
source_proc_coords[0] = g_source_timeslice / T;
source_proc_coords[1] = 0;
source_proc_coords[2] = 0;
source_proc_coords[3] = 0;
MPI_Cart_rank(g_cart_grid, source_proc_coords, &g_source_proc_id);
have_source_flag = ( g_source_proc_id == g_cart_id );
source_timeslice = have_source_flag ? g_source_timeslice % T : -1;
#else
g_source_proc_id = 0;
have_source_flag = 1;
source_timeslice = g_source_timeslice;
#endif
break;
}
#ifdef HAVE_QUDA
/*************************************************************
* QUDA inverter parameters
*************************************************************/
inv_param.dslash_type = QUDA_DOMAIN_WALL_DSLASH;
if(strcmp(g_inverter_type_name, "cg") == 0) {
inv_param.inv_type = QUDA_CG_INVERTER;
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] using cg inverter\n");
} else if(strcmp(g_inverter_type_name, "bicgstab") == 0) {
inv_param.inv_type = QUDA_BICGSTAB_INVERTER;
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] using bicgstab inverter\n");
#ifdef MULTI_GPU
} else if(strcmp(g_inverter_type_name, "gcr") == 0) {
inv_param.inv_type = QUDA_GCR_INVERTER;
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] using gcr inverter\n");
#endif
} else {
if(g_cart_id==0) fprintf(stderr, "[invert_dw_quda] Error, unrecognized inverter type %s\n", g_inverter_type_name);
EXIT(123);
}
if(inv_param.inv_type == QUDA_CG_INVERTER) {
inv_param.solution_type = QUDA_MAT_SOLUTION;
inv_param.solve_type = QUDA_NORMEQ_PC_SOLVE;
} else if(inv_param.inv_type == QUDA_BICGSTAB_INVERTER) {
inv_param.solution_type = QUDA_MAT_SOLUTION;
inv_param.solve_type = QUDA_DIRECT_PC_SOLVE;
} else {
inv_param.solution_type = QUDA_MATPC_SOLUTION;
inv_param.solve_type = QUDA_DIRECT_PC_SOLVE;
}
inv_param.m5 = g_m5;
inv_param.kappa = 0.5 / (5. + inv_param.m5);
inv_param.mass = g_m0;
inv_param.tol = solver_precision;
inv_param.maxiter = niter_max;
inv_param.reliable_delta = reliable_delta;
#ifdef MPI
// domain decomposition preconditioner parameters
if(inv_param.inv_type == QUDA_GCR_INVERTER) {
if(g_cart_id == 0) printf("# [] settup DD parameters\n");
inv_param.gcrNkrylov = 15;
inv_param.inv_type_precondition = QUDA_MR_INVERTER;
inv_param.tol_precondition = 1e-6;
inv_param.maxiter_precondition = 200;
inv_param.verbosity_precondition = QUDA_VERBOSE;
inv_param.prec_precondition = cuda_prec_sloppy;
inv_param.omega = 0.7;
}
#endif
inv_param.matpc_type = QUDA_MATPC_EVEN_EVEN;
inv_param.dagger = QUDA_DAG_NO;
inv_param.mass_normalization = QUDA_KAPPA_NORMALIZATION; //;QUDA_MASS_NORMALIZATION;
inv_param.cpu_prec = cpu_prec;
inv_param.cuda_prec = cuda_prec;
inv_param.cuda_prec_sloppy = cuda_prec_sloppy;
inv_param.verbosity = QUDA_VERBOSE;
inv_param.preserve_source = QUDA_PRESERVE_SOURCE_NO;
inv_param.dirac_order = QUDA_DIRAC_ORDER;
#ifdef MPI
inv_param.preserve_dirac = QUDA_PRESERVE_DIRAC_YES;
inv_param.prec_precondition = cuda_prec_sloppy;
inv_param.gamma_basis = QUDA_DEGRAND_ROSSI_GAMMA_BASIS;
inv_param.dirac_tune = QUDA_TUNE_NO;
#endif
#endif
/*******************************************
* write initial rng state to file
*******************************************/
if( g_source_type==2 && g_coherent_source==2 ) {
sprintf(rng_file_out, "%s.0", g_rng_filename);
status = init_rng_stat_file (g_seed, rng_file_out);
if( status != 0 ) {
fprintf(stderr, "[invert_dw_quda] Error, could not write rng status\n");
EXIT(210);
}
} else if( (g_source_type==2 /*&& g_coherent_source==1*/) || g_source_type==3 || g_source_type==4) {
if( init_rng_state(g_seed, &g_rng_state) != 0 ) {
fprintf(stderr, "[invert_dw_quda] Error, could initialize rng state\n");
EXIT(211);
}
}
/*******************************************
* prepare locks for openmp
*******************************************/
nthreads = g_num_threads - 1;
lck = (omp_lock_t*)malloc(nthreads * sizeof(omp_lock_t));
if(lck == NULL) {
EXIT_WITH_MSG(97, "[invert_dw_quda] Error, could not allocate lck\n");
}
// init locks
for(i=0;i<nthreads;i++) {
omp_init_lock(lck+i);
}
omp_init_lock(gen_lck);
// check the source momenta
if(g_source_momentum_set) {
source_momentum = (int*)malloc(3*sizeof(int));
if(g_source_momentum[0]<0) g_source_momentum[0] += LX_global;
if(g_source_momentum[1]<0) g_source_momentum[1] += LY_global;
if(g_source_momentum[2]<0) g_source_momentum[2] += LZ_global;
fprintf(stdout, "# [invert_dw_quda] using final source momentum ( %d, %d, %d )\n", g_source_momentum[0], g_source_momentum[1], g_source_momentum[2]);
if(full_orbit) {
status = make_qcont_orbits_3d_parity_avg( &qlatt_id, &qlatt_count, &qlatt_list, &qlatt_nclass, &qlatt_rep, &qlatt_map);
if(status != 0) {
if(g_cart_id==0) fprintf(stderr, "\n[invert_dw_quda] Error while creating O_3-lists\n");
EXIT(4);
}
source_momentum_class = qlatt_id[g_ipt[0][g_source_momentum[0]][g_source_momentum[1]][g_source_momentum[2]]];
source_momentum_no = qlatt_count[source_momentum_class];
source_momentum_runs = source_momentum_class==0 ? 1 : source_momentum_no + 1;
if(g_cart_id==0) fprintf(stdout, "# [] source momentum belongs to class %d with %d members, which means %d runs\n",
source_momentum_class, source_momentum_no, source_momentum_runs);
}
}
if(g_source_type == 5) {
if(g_seq_source_momentum_set) {
if(g_seq_source_momentum[0]<0) g_seq_source_momentum[0] += LX_global;
if(g_seq_source_momentum[1]<0) g_seq_source_momentum[1] += LY_global;
if(g_seq_source_momentum[2]<0) g_seq_source_momentum[2] += LZ_global;
} else if(g_source_momentum_set) {
g_seq_source_momentum[0] = g_source_momentum[0];
g_seq_source_momentum[1] = g_source_momentum[1];
g_seq_source_momentum[2] = g_source_momentum[2];
}
fprintf(stdout, "# [invert_dw_quda] using final sequential source momentum ( %d, %d, %d )\n",
g_seq_source_momentum[0], g_seq_source_momentum[1], g_seq_source_momentum[2]);
}
/***********************************************
* loop on spin-color-index
***********************************************/
for(isc=g_source_index[0]; isc<=g_source_index[1]; isc++)
// for(isc=g_source_index[0]; isc<=g_source_index[0]; isc++)
{
ispin = isc / n_c;
icol = isc % n_c;
for(imom=0; imom<source_momentum_runs; imom++) {
/***********************************************
* set source momentum
***********************************************/
if(g_source_momentum_set) {
if(imom == 0) {
if(full_orbit) {
source_momentum[0] = 0;
source_momentum[1] = 0;
source_momentum[2] = 0;
} else {
source_momentum[0] = g_source_momentum[0];
source_momentum[1] = g_source_momentum[1];
source_momentum[2] = g_source_momentum[2];
}
} else {
source_momentum[0] = qlatt_map[source_momentum_class][imom-1] / (LY_global*LZ_global);
source_momentum[1] = ( qlatt_map[source_momentum_class][imom-1] % (LY_global*LZ_global) ) / LZ_global;
source_momentum[2] = qlatt_map[source_momentum_class][imom-1] % LZ_global;
}
if(g_cart_id==0) fprintf(stdout, "# [] run no. %d, source momentum (%d, %d, %d)\n",
imom, source_momentum[0], source_momentum[1], source_momentum[2]);
}
/***********************************************
* prepare the souce
***********************************************/
if(g_read_source == 0) { // create source
switch(g_source_type) {
case 0:
// point source
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] Creating point source\n");
for(ix=0;ix<L5*VOLUME;ix++) { _fv_eq_zero(g_spinor_field[0]+ix); }
if(have_source_flag) {
if(g_source_momentum_set) {
phase = 2*M_PI*( source_momentum[0]*sl1/(double)LX_global + source_momentum[1]*sl2/(double)LY_global + source_momentum[2]*sl3/(double)LZ_global );
g_spinor_field[0][_GSI(g_ipt[lsl0][lsl1][lsl2][lsl3]) + 2*(n_c*ispin+icol) ] = cos(phase);
g_spinor_field[0][_GSI(g_ipt[lsl0][lsl1][lsl2][lsl3]) + 2*(n_c*ispin+icol)+1] = sin(phase);
} else {
g_spinor_field[0][_GSI(g_ipt[lsl0][lsl1][lsl2][lsl3]) + 2*(n_c*ispin+icol) ] = 1.;
}
}
if(g_source_momentum_set) {
sprintf(source_filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.%.2d.qx%.2dqy%.2dqz%.2d",
filename_prefix, Nconf, sl0, sl1, sl2, sl3, n_c*ispin+icol, source_momentum[0], source_momentum[1], source_momentum[2]);
} else {
sprintf(source_filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.%.2d", filename_prefix, Nconf, sl0, sl1, sl2, sl3, n_c*ispin+icol);
}
#ifdef HAVE_QUDA
// set matpc_tpye
source_location_5d_iseven = ( (g_iseven[g_ipt[lsl0][lsl1][lsl2][lsl3]] && ispin<n_s/2) || (!g_iseven[g_ipt[lsl0][lsl1][lsl2][lsl3]] && ispin>=n_s/2) ) ? 1 : 0;
if(source_location_5d_iseven) {
inv_param.matpc_type = QUDA_MATPC_EVEN_EVEN;
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] matpc type is MATPC_EVEN_EVEN\n");
} else {
inv_param.matpc_type = QUDA_MATPC_ODD_ODD;
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] matpc type is MATPC_ODD_ODD\n");
}
#endif
break;
case 2:
// timeslice source
if(g_coherent_source==1) {
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] Creating coherent timeslice source\n");
status = prepare_coherent_timeslice_source(g_spinor_field[0], gauge_field_smeared, g_coherent_source_base, g_coherent_source_delta, VOLUME, g_rng_state, 1);
if(status != 0) {
fprintf(stderr, "[invert_dw_quda] Error from prepare source, status was %d\n", status);
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 123);
MPI_Finalize();
#endif
exit(123);
}
check_error(prepare_coherent_timeslice_source(g_spinor_field[0], gauge_field_smeared, g_coherent_source_base, g_coherent_source_delta, VOLUME, g_rng_state, 1),
"prepare_coherent_timeslice_source", NULL, 123);
timeslice = g_coherent_source_base;
} else {
if(g_coherent_source==2) {
timeslice = (g_coherent_source_base+isc*g_coherent_source_delta)%T_global;
fprintf(stdout, "# [invert_dw_quda] Creating timeslice source\n");
check_error(prepare_timeslice_source(g_spinor_field[0], gauge_field_smeared, timeslice, VOLUME, g_rng_state, 1),
"prepare_timeslice_source", NULL, 123);
} else {
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] Creating timeslice source\n");
check_error(prepare_timeslice_source(g_spinor_field[0], gauge_field_smeared, g_source_timeslice, VOLUME, g_rng_state, 1),
"prepare_timeslice_source", NULL, 124);
timeslice = g_source_timeslice;
}
}
if(g_source_momentum_set) {
sprintf(source_filename, "%s.%.4d.%.2d.%.5d.qx%.2dqy%.2dqz%.2d", filename_prefix, Nconf,
timeslice, isc, source_momentum[0], source_momentum[1], source_momentum[2]);
} else {
sprintf(source_filename, "%s.%.4d.%.2d.%.5d", filename_prefix, Nconf, timeslice, isc);
}
break;
case 3:
// timeslice sources for one-end trick (spin dilution)
fprintf(stdout, "# [invert_dw_quda] Creating timeslice source for one-end-trick\n");
check_error( prepare_timeslice_source_one_end(g_spinor_field[0], gauge_field_smeared, source_timeslice, source_momentum, isc%n_s, g_rng_state, \
( isc%n_s==(n_s-1) && imom==source_momentum_runs-1 )), "prepare_timeslice_source_one_end", NULL, 125 );
c = N_Jacobi > 0 ? isc%n_s + n_s : isc%n_s;
if(g_source_momentum_set) {
sprintf(source_filename, "%s.%.4d.%.2d.%.2d.qx%.2dqy%.2dqz%.2d", filename_prefix, Nconf,
g_source_timeslice, c, source_momentum[0], source_momentum[1], source_momentum[2]);
} else {
sprintf(source_filename, "%s.%.4d.%.2d.%.2d", filename_prefix, Nconf, g_source_timeslice, c);
}
break;
case 4:
// timeslice sources for one-end trick (spin and color dilution )
fprintf(stdout, "# [invert_dw_quda] Creating timeslice source for one-end-trick\n");
check_error(prepare_timeslice_source_one_end_color(g_spinor_field[0], gauge_field_smeared, source_timeslice, source_momentum,\
isc%(n_s*n_c), g_rng_state, ( isc%(n_s*n_c)==(n_s*n_c-1) && imom==source_momentum_runs-1 )), "prepare_timeslice_source_one_end_color", NULL, 126);
c = N_Jacobi > 0 ? isc%(n_s*n_c) + (n_s*n_c) : isc%(n_s*n_c);
if(g_source_momentum_set) {
sprintf(source_filename, "%s.%.4d.%.2d.%.2d.qx%.2dqy%.2dqz%.2d", filename_prefix, Nconf,
g_source_timeslice, c, source_momentum[0], source_momentum[1], source_momentum[2]);
} else {
sprintf(source_filename, "%s.%.4d.%.2d.%.2d", filename_prefix, Nconf, g_source_timeslice, c);
}
break;
case 5:
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] preparing sequential point source\n");
check_error( prepare_sequential_point_source (g_spinor_field[0], isc, sl0, g_seq_source_momentum,
smear_source, g_spinor_field[1], gauge_field_smeared), "prepare_sequential_point_source", NULL, 33);
sprintf(source_filename, "%s.%.4d.t%.2dx%.2d.y%.2d.z%.2d.%.2d.qx%.2dqy%.2dqz%.2d", filename_prefix2, Nconf,
sl0, sl1, sl2, sl3, isc, g_source_momentum[0], g_source_momentum[1], g_source_momentum[2]);
break;
default:
fprintf(stderr, "\nError, unrecognized source type\n");
exit(32);
break;
}
} else { // read source
switch(g_source_type) {
case 0: // point source
if(g_source_momentum_set) {
sprintf(source_filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.%.2d.qx%.2dqy%.2dqz%.2d", \
filename_prefix2, Nconf, sl0, sl1, sl2, sl3, isc, source_momentum[0], source_momentum[1], source_momentum[2]);
} else {
sprintf(source_filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.%.2d", filename_prefix2, Nconf, sl0, sl1, sl2, sl3, isc);
}
fprintf(stdout, "# [invert_dw_quda] reading source from file %s\n", source_filename);
check_error(read_lime_spinor(g_spinor_field[0], source_filename, 0), "read_lime_spinor", NULL, 115);
break;
case 2: // timeslice source
if(g_source_momentum_set) {
sprintf(source_filename, "%s.%.4d.%.2d.%.5d.qx%.2dqy%.2dqz%.2d", filename_prefix2, Nconf, g_source_timeslice,
isc, source_momentum[0], source_momentum[1], source_momentum[2]);
} else {
sprintf(source_filename, "%s.%.4d.%.2d.%.5d", filename_prefix2, Nconf, g_source_timeslice, isc);
}
fprintf(stdout, "# [invert_dw_quda] reading source from file %s\n", source_filename);
check_error(read_lime_spinor(g_spinor_field[0], source_filename, 0), "read_lime_spinor", NULL, 115);
break;
default:
check_error(1, "source type", NULL, 104);
break;
case -1: // timeslice source
sprintf(source_filename, "%s", filename_prefix2);
fprintf(stdout, "# [invert_dw_quda] reading source from file %s\n", source_filename);
check_error(read_lime_spinor(g_spinor_field[0], source_filename, 0), "read_lime_spinor", NULL, 115);
break;
}
} // of if g_read_source
if(g_write_source) {
check_error(write_propagator(g_spinor_field[0], source_filename, 0, g_propagator_precision), "write_propagator", NULL, 27);
}
/***********************************************************************************************
* here threads split:
***********************************************************************************************/
if(dummy_flag==0) strcpy(source_filename_write, source_filename);
memcpy((void*)(smearing_spinor_field[0]), (void*)(g_spinor_field[0]), 24*VOLUME*sizeof(double));
if(dummy_flag>0) {
// copy only if smearing has been done; otherwise do not copy, do not invert
if(g_cart_id==0) fprintf(stdout, "# [] copy smearing field -> g field\n");
memcpy((void*)(g_spinor_field[0]), (void*)(smearing_spinor_field[1]), 24*VOLUME*sizeof(double));
}
omp_set_num_threads(g_num_threads);
#pragma omp parallel private(threadid, _2_kappa, is, ix, iy, iix, ratime, retime) shared(key,g_read_source, smear_source, N_Jacobi, kappa_Jacobi, smearing_spinor_field, g_spinor_field, nthreads, convert_sign, VOLUME, VOL3, T, L5, isc, rotate_gamma_basis, g_cart_id) firstprivate(inv_param, gauge_param, ofs)
{
threadid = omp_get_thread_num();
if(threadid < nthreads) {
fprintf(stdout, "# [] proc%.2d thread%.2d starting source preparation\n", g_cart_id, threadid);
// smearing
if( ( !g_read_source || (g_read_source && smear_source ) ) && N_Jacobi > 0 ) {
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] smearing source with N_Jacobi=%d, kappa_Jacobi=%e\n", N_Jacobi, kappa_Jacobi);
Jacobi_Smearing_threaded(gauge_field_smeared, smearing_spinor_field[0], smearing_spinor_field[1], kappa_Jacobi, N_Jacobi, threadid, nthreads);
}
/***********************************************
* create the 5-dim. source field
***********************************************/
if(convert_sign == 0) {
spinor_4d_to_5d_threaded(smearing_spinor_field[0], smearing_spinor_field[0], threadid, nthreads);
} else if(convert_sign == 1 || convert_sign == -1) {
spinor_4d_to_5d_sign_threaded(smearing_spinor_field[0], smearing_spinor_field[0], convert_sign, threadid, nthreads);
}
for(is=0; is<L5; is++) {
for(it=threadid; it<T; it+=nthreads) {
memcpy((void*)(g_spinor_field[0]+_GSI(g_ipt_5d[is][it][0][0][0])), (void*)(smearing_spinor_field[0]+_GSI(g_ipt_5d[is][it][0][0][0])), VOL3*24*sizeof(double));
}
}
// reorder, multiply with g2
for(is=0; is<L5; is++) {
for(it=threadid; it<T; it+=nthreads) {
for(i3=0; i3<VOL3; i3++) {
ix = (is*T+it)*VOL3 + i3;
_fv_eq_zero(smearing_spinor_field[1]+_GSI(ix));
}}}
if(rotate_gamma_basis) {
for(it=threadid; it<T; it+=nthreads) {
for(i3=0; i3<VOL3; i3++) {
ix = it * VOL3 + i3;
iy = lexic2eot_5d(0, ix);
_fv_eq_gamma_ti_fv(smearing_spinor_field[1]+_GSI(iy), 2, smearing_spinor_field[0]+_GSI(ix));
}}
for(it=threadid; it<T; it+=nthreads) {
for(i3=0; i3<VOL3; i3++) {
ix = it * VOL3 + i3;
iy = lexic2eot_5d(L5-1, ix);
_fv_eq_gamma_ti_fv(smearing_spinor_field[1]+_GSI(iy), 2, smearing_spinor_field[0]+_GSI(ix+(L5-1)*VOLUME));
}}
} else {
for(it=threadid; it<T; it+=nthreads) {
for(i3=0; i3<VOL3; i3++) {
ix = it * VOL3 + i3;
iy = lexic2eot_5d(0, ix);
_fv_eq_fv(smearing_spinor_field[1]+_GSI(iy), smearing_spinor_field[0]+_GSI(ix));
}}
for(it=threadid; it<T; it+=nthreads) {
for(i3=0; i3<VOL3; i3++) {
ix = it * VOL3 + i3;
iy = lexic2eot_5d(L5-1, ix);
_fv_eq_fv(smearing_spinor_field[1]+_GSI(iy), smearing_spinor_field[0]+_GSI(ix+(L5-1)*VOLUME));
}}
}
fprintf(stdout, "# [] proc%.2d thread%.2d finished source preparation\n", g_cart_id, threadid);
} else if(threadid == g_num_threads-1 && dummy_flag > 0) { // else branch on threadid
fprintf(stdout, "# [] proc%.2d thread%.2d starting inversion for dummy_flag = %d\n", g_cart_id, threadid, dummy_flag);
/***********************************************
* perform the inversion
***********************************************/
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] starting inversion\n");
xchange_field_5d(g_spinor_field[0]);
memset(g_spinor_field[1], 0, (VOLUME+RAND)*L5*24*sizeof(double));
ratime = CLOCK;
#ifdef MPI
if(inv_param.inv_type == QUDA_BICGSTAB_INVERTER || inv_param.inv_type == QUDA_GCR_INVERTER) {
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] calling invertQuda\n");
invertQuda(g_spinor_field[1], g_spinor_field[0], &inv_param);
} else if(inv_param.inv_type == QUDA_CG_INVERTER) {
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] calling testCG\n");
testCG(g_spinor_field[1], g_spinor_field[0], &inv_param);
} else {
if(g_cart_id==0) fprintf(stderr, "# [invert_dw_quda] unrecognized inverter\n");
}
#else
invertQuda(g_spinor_field[1], g_spinor_field[0], &inv_param);
#endif
retime = CLOCK;
if(g_cart_id==0) {
fprintf(stdout, "# [invert_dw_quda] QUDA time: %e seconds\n", inv_param.secs);
fprintf(stdout, "# [invert_dw_quda] QUDA Gflops: %e\n", inv_param.gflops/inv_param.secs);
fprintf(stdout, "# [invert_dw_quda] wall time: %e seconds\n", retime-ratime);
fprintf(stdout, "# [invert_dw_quda] Device memory used:\n\tSpinor: %f GiB\n\tGauge: %f GiB\n",
inv_param.spinorGiB, gauge_param.gaugeGiB);
}
} // of if threadid
// wait till all threads are here
#pragma omp barrier
if(inv_param.mass_normalization == QUDA_KAPPA_NORMALIZATION) {
_2_kappa = 2. * g_kappa5d;
for(ix=threadid; ix<VOLUME*L5;ix+=g_num_threads) {
_fv_ti_eq_re(g_spinor_field[1]+_GSI(ix), _2_kappa );
}
}
#pragma omp barrier
// reorder, multiply with g2
for(is=0;is<L5;is++) {
for(ix=threadid; ix<VOLUME; ix+=g_num_threads) {
iy = lexic2eot_5d(is, ix);
iix = is*VOLUME + ix;
_fv_eq_fv(g_spinor_field[0]+_GSI(iix), g_spinor_field[1]+_GSI(iy));
}}
#pragma omp barrier
if(rotate_gamma_basis) {
for(ix=threadid; ix<VOLUME*L5; ix+=g_num_threads) {
_fv_eq_gamma_ti_fv(g_spinor_field[1]+_GSI(ix), 2, g_spinor_field[0]+_GSI(ix));
}
} else {
for(ix=threadid; ix<VOLUME*L5;ix+=g_num_threads) {
_fv_eq_fv(g_spinor_field[1]+_GSI(ix), g_spinor_field[0]+_GSI(ix));
}
}
if(g_cart_id==0 && threadid==g_num_threads-1) fprintf(stdout, "# [invert_dw_quda] inversion done in %e seconds\n", retime-ratime);
#pragma omp single
{
#ifdef MPI
xchange_field_5d(g_spinor_field[1]);
#endif
/***********************************************
* check residuum
***********************************************/
if(check_residuum && dummy_flag>0) {
// apply the Wilson Dirac operator in the gamma-basis defined in cvc_linalg,
// which uses the tmLQCD conventions (same as in contractions)
// without explicit boundary conditions
#ifdef MPI
xchange_field_5d(g_spinor_field[2]);
xchange_field_5d(g_spinor_field[1]);
#endif
memset(g_spinor_field[0], 0, 24*(VOLUME+RAND)*L5*sizeof(double));
//sprintf(filename, "%s.inverted.ascii.%.2d", source_filename, g_cart_id);
//ofs = fopen(filename, "w");
//printf_spinor_field_5d(g_spinor_field[1], ofs);
//fclose(ofs);
Q_DW_Wilson_phi(g_spinor_field[0], g_spinor_field[1]);
for(ix=0;ix<VOLUME*L5;ix++) {
_fv_mi_eq_fv(g_spinor_field[0]+_GSI(ix), g_spinor_field[2]+_GSI(ix));
}
spinor_scalar_product_re(&norm2, g_spinor_field[2], g_spinor_field[2], VOLUME*L5);
spinor_scalar_product_re(&norm, g_spinor_field[0], g_spinor_field[0], VOLUME*L5);
if(g_cart_id==0) fprintf(stdout, "\n# [invert_dw_quda] absolut residuum squared: %e; relative residuum %e\n", norm, sqrt(norm/norm2) );
}
if(dummy_flag>0) {
/***********************************************
* create 4-dim. propagator
***********************************************/
if(convert_sign == 0) {
spinor_5d_to_4d(g_spinor_field[1], g_spinor_field[1]);
} else if(convert_sign == -1 || convert_sign == +1) {
spinor_5d_to_4d_sign(g_spinor_field[1], g_spinor_field[1], convert_sign);
}
/***********************************************
* write the solution
***********************************************/
sprintf(filename, "%s.inverted", source_filename_write);
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] writing propagator to file %s\n", filename);
check_error(write_propagator(g_spinor_field[1], filename, 0, g_propagator_precision), "write_propagator", NULL, 22);
//sprintf(filename, "prop.ascii.4d.%.2d.%.2d.%.2d", isc, g_nproc, g_cart_id);
//ofs = fopen(filename, "w");
//printf_spinor_field(g_spinor_field[1], ofs);
//fclose(ofs);
}
if(check_residuum) memcpy(g_spinor_field[2], smearing_spinor_field[0], 24*VOLUME*L5*sizeof(double));
} // of omp single
} // of omp parallel region
if(dummy_flag > 0) strcpy(source_filename_write, source_filename);
dummy_flag++;
} // of loop on momenta
} // of isc
#if 0
// last inversion
{
memcpy(g_spinor_field[0], smearing_spinor_field[1], 24*VOLUME*L5*sizeof(double));
if(g_cart_id==0) fprintf(stdout, "# [] proc%.2d starting last inversion\n", g_cart_id);
/***********************************************
* perform the inversion
***********************************************/
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] starting inversion\n");
xchange_field_5d(g_spinor_field[0]);
memset(g_spinor_field[1], 0, (VOLUME+RAND)*L5*24*sizeof(double));
ratime = CLOCK;
#ifdef MPI
if(inv_param.inv_type == QUDA_BICGSTAB_INVERTER || inv_param.inv_type == QUDA_GCR_INVERTER) {
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] calling invertQuda\n");
invertQuda(g_spinor_field[1], g_spinor_field[0], &inv_param);
} else if(inv_param.inv_type == QUDA_CG_INVERTER) {
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] calling testCG\n");
testCG(g_spinor_field[1], g_spinor_field[0], &inv_param);
} else {
if(g_cart_id==0) fprintf(stderr, "# [invert_dw_quda] unrecognized inverter\n");
}
#else
invertQuda(g_spinor_field[1], g_spinor_field[0], &inv_param);
#endif
retime = CLOCK;
if(g_cart_id==0) {
fprintf(stdout, "# [invert_dw_quda] QUDA time: %e seconds\n", inv_param.secs);
fprintf(stdout, "# [invert_dw_quda] QUDA Gflops: %e\n", inv_param.gflops/inv_param.secs);
fprintf(stdout, "# [invert_dw_quda] wall time: %e seconds\n", retime-ratime);
fprintf(stdout, "# [invert_dw_quda] Device memory used:\n\tSpinor: %f GiB\n\tGauge: %f GiB\n",
inv_param.spinorGiB, gauge_param.gaugeGiB);
}
omp_set_num_threads(g_num_threads);
#pragma omp parallel private(threadid,_2_kappa,is,ix,iy,iix) shared(VOLUME,L5,g_kappa,g_spinor_field,g_num_threads)
{
threadid = omp_get_thread_num();
if(inv_param.mass_normalization == QUDA_KAPPA_NORMALIZATION) {
_2_kappa = 2. * g_kappa5d;
for(ix=threadid; ix<VOLUME*L5;ix+=g_num_threads) {
_fv_ti_eq_re(g_spinor_field[1]+_GSI(ix), _2_kappa );
}
}
#pragma omp barrier
// reorder, multiply with g2
for(is=0;is<L5;is++) {
for(ix=threadid; ix<VOLUME; ix+=g_num_threads) {
iy = lexic2eot_5d(is, ix);
iix = is*VOLUME + ix;
_fv_eq_fv(g_spinor_field[0]+_GSI(iix), g_spinor_field[1]+_GSI(iy));
}}
#pragma omp barrier
if(rotate_gamma_basis) {
for(ix=threadid; ix<VOLUME*L5; ix+=g_num_threads) {
_fv_eq_gamma_ti_fv(g_spinor_field[1]+_GSI(ix), 2, g_spinor_field[0]+_GSI(ix));
}
} else {
for(ix=threadid; ix<VOLUME*L5;ix+=g_num_threads) {
_fv_eq_fv(g_spinor_field[1]+_GSI(ix), g_spinor_field[0]+_GSI(ix));
}
}
} // end of parallel region
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] inversion done in %e seconds\n", retime-ratime);
#ifdef MPI
xchange_field_5d(g_spinor_field[1]);
#endif
/***********************************************
* check residuum
***********************************************/
if(check_residuum && dummy_flag>0) {
// apply the Wilson Dirac operator in the gamma-basis defined in cvc_linalg,
// which uses the tmLQCD conventions (same as in contractions)
// without explicit boundary conditions
#ifdef MPI
xchange_field_5d(g_spinor_field[2]);
#endif
memset(g_spinor_field[0], 0, 24*(VOLUME+RAND)*L5*sizeof(double));
//sprintf(filename, "%s.inverted.ascii.%.2d", source_filename, g_cart_id);
//ofs = fopen(filename, "w");
//printf_spinor_field_5d(g_spinor_field[1], ofs);
//fclose(ofs);
Q_DW_Wilson_phi(g_spinor_field[0], g_spinor_field[1]);
for(ix=0;ix<VOLUME*L5;ix++) {
_fv_mi_eq_fv(g_spinor_field[0]+_GSI(ix), g_spinor_field[2]+_GSI(ix));
}
spinor_scalar_product_re(&norm, g_spinor_field[0], g_spinor_field[0], VOLUME*L5);
spinor_scalar_product_re(&norm2, g_spinor_field[2], g_spinor_field[2], VOLUME*L5);
if(g_cart_id==0) fprintf(stdout, "\n# [invert_dw_quda] absolut residuum squared: %e; relative residuum %e\n", norm, sqrt(norm/norm2) );
}
/***********************************************
* create 4-dim. propagator
***********************************************/
if(convert_sign == 0) {
spinor_5d_to_4d(g_spinor_field[1], g_spinor_field[1]);
} else if(convert_sign == -1 || convert_sign == +1) {
spinor_5d_to_4d_sign(g_spinor_field[1], g_spinor_field[1], convert_sign);
}
/***********************************************
* write the solution
***********************************************/
sprintf(filename, "%s.inverted", source_filename_write);
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] writing propagator to file %s\n", filename);
check_error(write_propagator(g_spinor_field[1], filename, 0, g_propagator_precision), "write_propagator", NULL, 22);
//sprintf(filename, "prop.ascii.4d.%.2d.%.2d.%.2d", isc, g_nproc, g_cart_id);
//ofs = fopen(filename, "w");
//printf_spinor_field(g_spinor_field[1], ofs);
//fclose(ofs);
} // of last inversion
#endif // of if 0
/***********************************************
* free the allocated memory, finalize
***********************************************/
#ifdef HAVE_QUDA
// finalize the QUDA library
if(g_cart_id==0) fprintf(stdout, "# [invert_dw_quda] finalizing quda\n");
#ifdef MPI
freeGaugeQuda();
#endif
endQuda();
#endif
if(g_gauge_field != NULL) free(g_gauge_field);
if(gauge_field_smeared != NULL) free(gauge_field_smeared);
if(no_fields>0) {
if(g_spinor_field!=NULL) {
for(i=0; i<no_fields; i++) if(g_spinor_field[i]!=NULL) free(g_spinor_field[i]);
free(g_spinor_field);
}
}
free_geometry();
if(g_source_momentum_set && full_orbit) {
finalize_q_orbits(&qlatt_id, &qlatt_count, &qlatt_list, &qlatt_rep);
if(qlatt_map != NULL) {
free(qlatt_map[0]);
free(qlatt_map);
}
}
if(source_momentum != NULL) free(source_momentum);
if(lck != NULL) free(lck);
#ifdef MPI
#ifdef HAVE_QUDA
endCommsQuda();
#else
MPI_Finalize();
#endif
#endif
if(g_cart_id==0) {
g_the_time = time(NULL);
fprintf(stdout, "\n# [invert_dw_quda] %s# [invert_dw_quda] end of run\n", ctime(&g_the_time));
fprintf(stderr, "\n# [invert_dw_quda] %s# [invert_dw_quda] end of run\n", ctime(&g_the_time));
}
return(0);
}
int main(int argc, char **argv) {
int c, mu, nu, status;
int i, j, ncon=-1, ir, is, ic, id;
int filename_set = 0;
int x0, x1, x2, x3, ix, iix;
int y0, y1, y2, y3, iy, iiy;
int start_valuet=0, start_valuex=0, start_valuey=0;
int num_threads=1, threadid, nthreads;
int seed, seed_set=0;
double diff1, diff2;
/* double *chi=NULL, *psi=NULL; */
double plaq=0., pl_ts, pl_xs, pl_global;
double *gauge_field_smeared = NULL;
double s[18], t[18], u[18], pl_loc;
double spinor1[24], spinor2[24];
double *pl_gather=NULL;
double dtmp;
complex prod, w, w2;
int verbose = 0;
char filename[200];
char file1[200];
char file2[200];
FILE *ofs=NULL;
double norm, norm2;
fermion_propagator_type *prop=NULL, prop2=NULL, seq_prop=NULL, seq_prop2=NULL, prop_aux=NULL, prop_aux2=NULL;
int idx, eoflag, shift;
float *buffer = NULL;
unsigned int VOL3;
size_t items, bytes;
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "h?vf:g:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'g':
strcpy(file1, optarg);
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
if(g_cart_id==0) fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
/* initialize T etc. */
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] T_global = %3d\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] LX_global = %3d\n"\
"# [%2d] LX = %3d\n"\
"# [%2d] LXstart = %3d\n"\
"# [%2d] LY_global = %3d\n"\
"# [%2d] LY = %3d\n"\
"# [%2d] LYstart = %3d\n",\
g_cart_id, g_cart_id, T_global, g_cart_id, T, g_cart_id, Tstart,
g_cart_id, LX_global, g_cart_id, LX, g_cart_id, LXstart,
g_cart_id, LY_global, g_cart_id, LY, g_cart_id, LYstart);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(101);
}
geometry();
if(init_geometry_5d() != 0) {
fprintf(stderr, "ERROR from init_geometry_5d\n");
exit(102);
}
geometry_5d();
VOL3 = LX*LY*LZ;
/* read the gauge field */
alloc_gauge_field(&g_gauge_field, VOLUMEPLUSRAND);
if(g_cart_id==0) fprintf(stdout, "# gauge field file name %s\n", file1);
// status = read_nersc_gauge_field_3x3(g_gauge_field, filename, &plaq);
// status = read_ildg_nersc_gauge_field(g_gauge_field, filename);
status = read_lime_gauge_field_doubleprec(file1);
// status = read_nersc_gauge_field(g_gauge_field, filename, &plaq);
// status = 0;
if(status != 0) {
fprintf(stderr, "[apply_Dtm] Error, could not read gauge field\n");
EXIT(11);
}
#ifdef MPI
xchange_gauge();
#endif
// measure the plaquette
if(g_cart_id==0) fprintf(stdout, "# read plaquette value 1st field: %25.16e\n", plaq);
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "# measured plaquette value 1st field: %25.16e\n", plaq);
sprintf(filename, "%s.dbl", file1);
if(g_cart_id==0) fprintf(stdout, "# [] writing gauge field in double precision to file %s\n", filename);
status = write_lime_gauge_field(filename, plaq, Nconf, 64);
if(status != 0) {
fprintf(stderr, "[apply_Dtm] Error, could not write gauge field\n");
EXIT(12);
}
/***********************************************
* free the allocated memory, finalize
***********************************************/
free(g_gauge_field);
free_geometry();
g_the_time = time(NULL);
fprintf(stdout, "# [] %s# [] end fo run\n", ctime(&g_the_time));
fflush(stdout);
fprintf(stderr, "# [] %s# [] end fo run\n", ctime(&g_the_time));
fflush(stderr);
#ifdef MPI
MPI_Finalize();
#endif
return(0);
}
int main(int argc, char **argv) {
int c, i, mu, nu;
int count = 0;
int filename_set = 0;
int dims[4] = {0,0,0,0};
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, iix;
int sid, status, gid;
double *disc = (double*)NULL;
double *work = (double*)NULL;
double q[4], fnorm;
int verbose = 0;
int do_gt = 0;
char filename[100], contype[200];
double ratime, retime;
double plaq;
double spinor1[24], spinor2[24], U_[18];
double *gauge_trafo=(double*)NULL;
complex w, w1, *cp1, *cp2, *cp3;
FILE *ofs;
#ifdef MPI
// MPI_Init(&argc, &argv);
fprintf(stderr, "[jc_ud_x] Error, only non-mpi version implemented\n");
exit(1);
#endif
while ((c = getopt(argc, argv, "h?f:")) != -1) {
switch (c) {
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
fprintf(stdout, "\n**************************************************\n");
fprintf(stdout, "* jc_ud_x\n");
fprintf(stdout, "**************************************************\n\n");
/*********************************
* initialize MPI parameters
*********************************/
// mpi_init(argc, argv);
/* initialize fftw */
dims[0]=T_global; dims[1]=LX; dims[2]=LY; dims[3]=LZ;
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
/*************************************************
* allocate mem for gauge field and spinor fields
*************************************************/
alloc_gauge_field(&g_gauge_field, VOLUMEPLUSRAND);
no_fields = 2;
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields; i++) alloc_spinor_field(&g_spinor_field[i], VOLUMEPLUSRAND);
/****************************************
* allocate memory for the contractions
****************************************/
disc = (double*)calloc( 8*VOLUME, sizeof(double));
if( disc == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc\n");
exit(3);
}
/***********************************************
* start loop on gauge id.s
***********************************************/
for(gid=g_gaugeid; gid<=g_gaugeid2; gid++) {
for(ix=0; ix<8*VOLUME; ix++) disc[ix] = 0.;
sprintf(filename, "%s.%.4d", gaugefilename_prefix, gid);
if(g_cart_id==0) fprintf(stdout, "# reading gauge field from file %s\n", filename);
read_lime_gauge_field_doubleprec(filename);
xchange_gauge();
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "# measured plaquette value: %25.16e\n", plaq);
/***********************************************
* start loop on source id.s
***********************************************/
for(sid=g_sourceid; sid<=g_sourceid2; sid+=g_sourceid_step) {
/* reset disc to zero */
for(ix=0; ix<8*VOLUME; ix++) disc[ix] = 0.;
/* read the new propagator to g_spinor_field[0] */
ratime = (double)clock() / CLOCKS_PER_SEC;
if(format==0) {
sprintf(filename, "%s.%.4d.%.2d.inverted", filename_prefix, gid, sid);
if(read_lime_spinor(g_spinor_field[0], filename, 0) != 0) break;
}
else if(format==1) {
sprintf(filename, "%s.%.4d.%.5d.inverted", filename_prefix, gid, sid);
if(read_cmi(g_spinor_field[0], filename) != 0) break;
}
xchange_field(g_spinor_field[0]);
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id==0) fprintf(stdout, "# time to read prop.: %e seconds\n", retime-ratime);
ratime = (double)clock() / CLOCKS_PER_SEC;
/* apply D_W once, save in g_spinor_field[1] */
Hopping(g_spinor_field[1], g_spinor_field[0]);
for(ix=0; ix<VOLUME; ix++) {
_fv_pl_eq_fv(g_spinor_field[1]+_GSI(ix), g_spinor_field[0]+_GSI(ix));
_fv_ti_eq_re(g_spinor_field[1]+_GSI(ix), 1./(2.*g_kappa));
}
xchange_field(g_spinor_field[1]);
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id==0) fprintf(stdout, "# time to apply D_W: %e seconds\n", retime-ratime);
ratime = (double)clock() / CLOCKS_PER_SEC;
/* calculate real and imaginary part */
for(mu=0; mu<4; mu++) {
for(ix=0; ix<VOLUME; ix++) {
_cm_eq_cm_ti_co(U_, g_gauge_field+_GGI(ix,mu), &(co_phase_up[mu]));
_fv_eq_gamma_ti_fv(spinor1, 5, g_spinor_field[0]+_GSI(g_iup[ix][mu]));
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_pl_eq_fv(spinor2, spinor1);
_fv_eq_cm_ti_fv(spinor1, U_, spinor2);
_co_eq_fv_dag_ti_fv(&w, g_spinor_field[0]+_GSI(ix), spinor1);
disc[_GWI(mu,ix,VOLUME) ] = g_mu * w.im;
_fv_eq_gamma_ti_fv(spinor1, mu, g_spinor_field[1]+_GSI(g_iup[ix][mu]));
_fv_pl_eq_fv(spinor1, g_spinor_field[1]+_GSI(g_iup[ix][mu]));
_fv_eq_cm_ti_fv(spinor2, U_, spinor1);
_co_eq_fv_dag_ti_fv(&w, g_spinor_field[0]+_GSI(ix), spinor2);
disc[_GWI(mu,ix,VOLUME)+1] = w.im / 3.;
}
}
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id==0) fprintf(stdout, "# time to calculate contractions: %e seconds\n", retime-ratime);
/************************************************
* save results
************************************************/
if(g_cart_id == 0) fprintf(stdout, "# save results for gauge id %d and sid %d\n", gid, sid);
/* save the result in position space */
fnorm = 1. / g_prop_normsqr;
if(g_cart_id==0) fprintf(stdout, "X-fnorm = %e\n", fnorm);
for(mu=0; mu<4; mu++) {
for(ix=0; ix<VOLUME; ix++) {
disc[_GWI(mu,ix,VOLUME) ] *= fnorm;
disc[_GWI(mu,ix,VOLUME)+1] *= fnorm;
}
}
sprintf(filename, "jc_ud_x.%.4d.%.4d", gid, sid);
sprintf(contype, "jc-u_and_d-X");
write_lime_contraction(disc, filename, 64, 4, contype, gid, sid);
//sprintf(filename, "jc_ud_x.%.4d.%.4d.ascii", gid, sid);
//write_contraction (disc, NULL, filename, 4, 2, 0);
} /* of loop on sid */
} /* of loop on gid */
/***********************************************
* free the allocated memory, finalize
***********************************************/
free(g_gauge_field);
for(i=0; i<no_fields; i++) free(g_spinor_field[i]);
free(g_spinor_field);
free_geometry();
free(disc);
return(0);
}
int main(int argc, char **argv) {
int c, mu, nu, status, gid;
int filename_set = 0;
int source_location, have_source_flag = 0;
int x0, x1, x2, x3, ix, iix;
int sx0, sx1, sx2, sx3;
int tsize = 0;
double *conn = NULL;
double *conn2 = NULL;
double *conn3 = NULL;
int verbose = 0;
char filename[200];
double ratime, retime;
FILE *ofs;
double q[4], wre, wim, dtmp;
int check_WI = 0, write_ascii=0;
unsigned int VOL3=0;
while ((c = getopt(argc, argv, "AWh?vf:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'W':
check_WI = 1;
fprintf(stdout, "# [get_corr_v5] check Ward Identity\n");
break;
case 'A':
write_ascii = 1;
fprintf(stdout, "# [get_corr_v5] write Pi_mn in ASCII format\n");
break;
case 'h':
case '?':
default:
usage();
break;
}
}
g_the_time = time(NULL);
// set the default values
set_default_input_values();
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# [get_corr_v5] reading input parameters from file %s\n", filename);
read_input_parser(filename);
// some checks on the input data
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
fprintf(stdout, "# [get_corr_v5] T=%d, LX=%d, LY=%d, LZ=%d\n", T_global, LX, LY, LZ);
if(g_proc_id==0) fprintf(stderr, "[get_corr_v5] Error, T and L's must be set\n");
usage();
}
// initialize MPI parameters
mpi_init(argc, argv);
T = T_global;
Tstart = 0;
fprintf(stdout, "# [%2d] fftw parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart);
if(init_geometry() != 0) {
fprintf(stderr, "[get_corr_v5] Error from init_geometry\n");
EXIT(1);
}
geometry();
VOL3 = LX*LY*LZ;
/****************************************
* allocate memory for the contractions *
****************************************/
conn = (double*)calloc(32 * VOLUME, sizeof(double));
if( (conn==NULL) ) {
fprintf(stderr, "[get_corr_v5] Error, could not allocate memory for contr. fields\n");
EXIT(2);
}
conn2= (double*)calloc(2 * T, sizeof(double));
if( (conn2==NULL) ) {
fprintf(stderr, "[get_corr_v5] Error, could not allocate memory for corr.\n");
EXIT(3);
}
conn3= (double*)calloc(2 * T, sizeof(double));
if( (conn3==NULL) ) {
fprintf(stderr, "[get_corr_v5] Error, could not allocate memory for corr.\n");
EXIT(3);
}
/********************************
* determine source coordinates *
********************************/
/*
have_source_flag = (int)(g_source_location/(LX*LY*LZ)>=Tstart && g_source_location/(LX*LY*LZ)<(Tstart+T));
if(have_source_flag==1) fprintf(stdout, "# [get_corr_v5] process %2d has source location\n", g_cart_id);
sx0 = g_source_location/(LX*LY*LZ)-Tstart;
sx1 = (g_source_location%(LX*LY*LZ)) / (LY*LZ);
sx2 = (g_source_location%(LY*LZ)) / LZ;
sx3 = (g_source_location%LZ);
if(have_source_flag==1) {
fprintf(stdout, "# [get_corr_v5] local source coordinates: (%3d,%3d,%3d,%3d)\n", sx0, sx1, sx2, sx3);
source_location = g_ipt[sx0][sx1][sx2][sx3];
}
have_source_flag = 0;
*/
for(gid=g_gaugeid; gid<=g_gaugeid2; gid+=g_gauge_step) {
memset(conn, 0, 32*VOLUME*sizeof(double));
/***********************
* read contractions *
***********************/
ratime = CLOCK;
sprintf(filename, "%s.%.4d", filename_prefix, gid);
if(format==2 || format==3) {
status = read_contraction(conn, NULL, filename, 16);
} else if( format==0) {
status = read_lime_contraction(conn, filename, 16, 0);
}
if(status != 0) {
// fprintf(stderr, "[get_corr_v5] Error from read_contractions, status was %d\n", status);
// EXIT(5);
fprintf(stderr, "[get_corr_v5] Warning, could not read contractions for gid %d, status was %d\n", gid, status);
continue;
}
retime = CLOCK;
fprintf(stdout, "# [get_corr_v5] time to read contractions %e seconds\n", retime-ratime);
// TEST Pi_mm
if(write_ascii) {
sprintf(filename, "pimm_test.%.4d", gid);
ofs = fopen(filename, "w");
if(ofs == NULL) exit(33);
fprintf(ofs, "# Pi_mm\n# %s", ctime(&g_the_time));
for(x0=0; x0<T; x0++) {
for(x1=0; x1<LX; x1++) {
for(x2=0; x2<LY; x2++) {
for(x3=0; x3<LZ; x3++) {
fprintf(ofs, "# t=%3d x=%3d y=%3d z=%3d\n", x0, x1, x2, x3);
ix = g_ipt[x0][x1][x2][x3];
for(nu=0;nu<4;nu++) {
wre = conn[_GWI(5*nu,ix,VOLUME)];
wim = conn[_GWI(5*nu,ix,VOLUME)+1];
fprintf(ofs, "%3d%16.7e%16.7e\n", nu, wre, wim);
}
}}}}
fclose(ofs);
} // of if write_ascii
// TEST Ward Identity
if(check_WI) {
fprintf(stdout, "# [get_corr_v5] Ward identity\n");
sprintf(filename, "WI.%.4d", gid);
ofs = fopen(filename, "w");
if(ofs == NULL) exit(32);
for(x0=0; x0<T; x0++) {
q[0] = 2. * sin(M_PI * (double)x0 / (double)T);
for(x1=0; x1<LX; x1++) {
q[1] = 2. * sin(M_PI * (double)x1 / (double)LX);
for(x2=0; x2<LY; x2++) {
q[2] = 2. * sin(M_PI * (double)x2 / (double)LY);
for(x3=0; x3<LZ; x3++) {
q[3] = 2. * sin(M_PI * (double)x3 / (double)LZ);
ix = g_ipt[x0][x1][x2][x3];
for(nu=0;nu<4;nu++) {
wre = q[0] * conn[_GWI(4*0+nu,ix,VOLUME)] + q[1] * conn[_GWI(4*1+nu,ix,VOLUME)] \
+ q[2] * conn[_GWI(4*2+nu,ix,VOLUME)] + q[3] * conn[_GWI(4*3+nu,ix,VOLUME)];
wim = q[0] * conn[_GWI(4*0+nu,ix,VOLUME)+1] + q[1] * conn[_GWI(4*1+nu,ix,VOLUME)+1] \
+ q[2] * conn[_GWI(4*2+nu,ix,VOLUME)+1] + q[3] * conn[_GWI(4*3+nu,ix,VOLUME)+1];
fprintf(ofs, "\t%3d%3d%3d%3d%3d%16.7e%16.7e\n", nu, x0, x1, x2, x3, wre, wim);
}
}}}}
fclose(ofs);
}
/***********************
* fill the correlator *
***********************/
ratime = CLOCK;
memset(conn2, 0, 2*T*sizeof(double));
// (1) V0V0
for(x0=0; x0<T; x0++) {
for(ix=0; ix<VOL3; ix++) {
iix = _GWI(0,x0*VOL3+ix,VOLUME);
conn2[2*x0 ] += conn[iix ];
conn2[2*x0+1] += conn[iix+1];
}
}
// (2) VKVK
memset(conn3, 0, 2*T*sizeof(double));
for(x0=0; x0<T; x0++) {
for(ix=0; ix<VOL3; ix++) {
iix = x0 * VOL3 + ix;
conn3[2*x0 ] += conn[_GWI(5,iix,VOLUME) ] + conn[_GWI(10,iix,VOLUME) ] + conn[_GWI(15,iix,VOLUME) ];
conn3[2*x0+1] += conn[_GWI(5,iix,VOLUME)+1] + conn[_GWI(10,iix,VOLUME)+1] + conn[_GWI(15,iix,VOLUME)+1];
}
}
// normalization
dtmp = 1. / (double)VOL3;
for(x0=0; x0<2*T; x0++) { conn2[x0] *= dtmp; }
for(x0=0; x0<2*T; x0++) { conn3[x0] *= dtmp; }
retime = CLOCK;
fprintf(stdout, "# [get_corr_v5] time to fill correlator %e seconds\n", retime-ratime);
// TEST
/*
fprintf(stdout, "# [get_corr_v5] V0V0 correlator\n");
for(x0=0; x0<T; x0++) {
fprintf(stdout, "\t%3d%25.16e%25.16e\n",x0, conn2[2*x0], conn2[2*x0+1]);
}
fprintf(stdout, "# [get_corr_v5] VKVK correlator\n");
for(x0=0; x0<T; x0++) {
fprintf(stdout, "\t%3d%25.16e%25.16e\n",x0, conn3[2*x0], conn3[2*x0+1]);
}
*/
/*****************************************
* write to file
*****************************************/
ratime = CLOCK;
sprintf(filename, "p00_corr.%.4d", gid);
if( (ofs=fopen(filename, "w")) == NULL ) {
fprintf(stderr, "[get_corr_v5] Error, could not open file %s for writing\n", filename);
EXIT(6);
}
x0 = 0;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*x0], 0., gid);
for(x0=1; x0<T/2; x0++) {
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*x0], conn2[2*(T-x0)], gid);
}
x0 = T / 2;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn2[2*x0], 0., gid);
fclose(ofs);
sprintf(filename, "pkk_corr.%.4d", gid);
if( (ofs=fopen(filename, "w")) == NULL ) {
fprintf(stderr, "[get_corr_v5] Error, could not open file %s for writing\n", filename);
EXIT(7);
}
x0 = 0;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn3[2*x0], 0., gid);
for(x0=1; x0<T/2; x0++) {
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn3[2*x0], conn3[2*(T-x0)], gid);
}
x0 = T / 2;
fprintf(ofs, "%3d%3d%3d%25.16e%25.16e%6d\n", 5, 1, x0, conn3[2*x0], 0., gid);
fclose(ofs);
retime = CLOCK;
fprintf(stdout, "# [get_corr_v5] time to write correlator %e seconds\n", retime-ratime);
} // of loop on gid
/***************************************
* free the allocated memory, finalize *
***************************************/
free_geometry();
if(conn != NULL) free(conn);
if(conn2 != NULL) free(conn2);
if(conn3 != NULL) free(conn3);
fprintf(stdout, "# [get_corr_v5] %s# [get_corr_v5] end of run\n", ctime(&g_the_time));
fflush(stdout);
fprintf(stderr, "[get_corr_v5] %s[get_corr_v5] end of run\n", ctime(&g_the_time));
fflush(stderr);
return(0);
}
int main(int argc, char **argv) {
int c, i, mu, nu;
int count = 0;
int filename_set = 0;
int dims[4] = {0,0,0,0};
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, iix;
int dxm[4], dxn[4], ixpm, ixpn;
int sid;
double *disc = (double*)NULL;
double *work = (double*)NULL;
double q[4], fnorm;
int verbose = 0;
int do_gt = 0;
char filename[100];
double ratime, retime;
double plaq, _2kappamu, hpe3_coeff, onepmutilde2, mutilde2;
double spinor1[24], spinor2[24], U_[18], U1_[18], U2_[18];
double *gauge_trafo=(double*)NULL;
complex w, w1, w2, *cp1, *cp2, *cp3;
FILE *ofs;
fftw_complex *in=(fftw_complex*)NULL;
#ifdef MPI
fftwnd_mpi_plan plan_p, plan_m;
int *status;
#else
fftwnd_plan plan_p, plan_m;
#endif
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "h?vgf:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'g':
do_gt = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
set_default_input_values();
if(filename_set==0) strcpy(filename, "cvc.input");
/* read the input file */
read_input(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
#ifdef MPI
if((status = (int*)calloc(g_nproc, sizeof(int))) == (int*)NULL) {
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
exit(7);
}
#endif
/* initialize fftw */
dims[0]=T_global; dims[1]=LX; dims[2]=LY; dims[3]=LZ;
#ifdef MPI
plan_p = fftwnd_mpi_create_plan(g_cart_grid, 4, dims, FFTW_BACKWARD, FFTW_MEASURE);
plan_m = fftwnd_mpi_create_plan(g_cart_grid, 4, dims, FFTW_FORWARD, FFTW_MEASURE);
fftwnd_mpi_local_sizes(plan_p, &T, &Tstart, &l_LX_at, &l_LXstart_at, &FFTW_LOC_VOLUME);
#else
plan_p = fftwnd_create_plan(4, dims, FFTW_BACKWARD, FFTW_MEASURE | FFTW_IN_PLACE);
plan_m = fftwnd_create_plan(4, dims, FFTW_FORWARD, FFTW_MEASURE | FFTW_IN_PLACE);
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
#endif
fprintf(stdout, "# [%2d] fftw parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
#ifdef MPI
if(T==0) {
fprintf(stderr, "[%2d] local T is zero; exit\n", g_cart_id);
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
exit(2);
}
#endif
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(1);
}
geometry();
/* read the gauge field */
alloc_gauge_field(&g_gauge_field, VOLUMEPLUSRAND);
sprintf(filename, "%s.%.4d", gaugefilename_prefix, Nconf);
if(g_cart_id==0) fprintf(stdout, "reading gauge field from file %s\n", filename);
read_lime_gauge_field_doubleprec(filename);
xchange_gauge();
/* measure the plaquette */
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "measured plaquette value: %25.16e\n", plaq);
if(do_gt==1) {
/***********************************
* initialize gauge transformation
***********************************/
init_gauge_trafo(&gauge_trafo, 1.);
apply_gt_gauge(gauge_trafo);
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "measured plaquette value after gauge trafo: %25.16e\n", plaq);
}
/****************************************
* allocate memory for the spinor fields
****************************************/
no_fields = 3;
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields; i++) alloc_spinor_field(&g_spinor_field[i], VOLUMEPLUSRAND);
/****************************************
* allocate memory for the contractions
****************************************/
disc = (double*)calloc( 8*VOLUME, sizeof(double));
if( disc == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(3);
}
for(ix=0; ix<8*VOLUME; ix++) disc[ix] = 0.;
work = (double*)calloc(48*VOLUME, sizeof(double));
if( work == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for work\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(3);
}
/****************************************
* prepare Fourier transformation arrays
****************************************/
in = (fftw_complex*)malloc(FFTW_LOC_VOLUME*sizeof(fftw_complex));
if(in==(fftw_complex*)NULL) {
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(4);
}
/***********************************************
* start loop on source id.s
***********************************************/
for(sid=g_sourceid; sid<=g_sourceid2; sid++) {
/********************************
* read the first propagator
********************************/
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(format==0) {
sprintf(filename, "%s.%.4d.%.2d.inverted", filename_prefix, Nconf, sid);
if(read_lime_spinor(g_spinor_field[2], filename, 0) != 0) break;
}
else if(format==1) {
sprintf(filename, "%s.%.4d.%.5d.inverted", filename_prefix, Nconf, sid);
if(read_cmi(g_spinor_field[2], filename) != 0) break;
}
xchange_field(g_spinor_field[2]);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
fprintf(stdout, "time to read prop.: %e seconds\n", retime-ratime);
if(do_gt==1) {
/******************************************
* gauge transform the propagators for sid
******************************************/
for(ix=0; ix<VOLUME; ix++) {
_fv_eq_cm_ti_fv(spinor1, gauge_trafo+18*ix, g_spinor_field[2]+_GSI(ix));
_fv_eq_fv(g_spinor_field[2]+_GSI(ix), spinor1);
}
xchange_field(g_spinor_field[2]);
}
/************************************************
* calculate the source: apply Q_phi_tbc
************************************************/
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
Q_phi_tbc(g_spinor_field[0], g_spinor_field[2]);
xchange_field(g_spinor_field[0]);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "time to calculate source: %e seconds\n", retime-ratime);
/************************************************
* HPE: apply BH5
************************************************/
BH5(g_spinor_field[1], g_spinor_field[2]);
for(ix=0; ix<8*VOLUME; ix++) {disc[ix] = 0.;}
/* add new contractions to (existing) disc */
# ifdef MPI
ratime = MPI_Wtime();
# else
ratime = (double)clock() / CLOCKS_PER_SEC;
# endif
for(mu=0; mu<4; mu++) { /* loop on Lorentz index of the current */
iix = _GWI(mu,0,VOLUME);
for(ix=0; ix<VOLUME; ix++) { /* loop on lattice sites */
_cm_eq_cm_ti_co(U_, &g_gauge_field[_GGI(ix, mu)], &co_phase_up[mu]);
/* first contribution */
_fv_eq_cm_ti_fv(spinor1, U_, &g_spinor_field[1][_GSI(g_iup[ix][mu])]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_mi_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[0][_GSI(ix)], spinor2);
disc[iix ] -= 0.5 * w.re;
disc[iix+1] -= 0.5 * w.im;
/* second contribution */
_fv_eq_cm_dag_ti_fv(spinor1, U_, &g_spinor_field[1][_GSI(ix)]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_pl_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[0][_GSI(g_iup[ix][mu])], spinor2);
disc[iix ] -= 0.5 * w.re;
disc[iix+1] -= 0.5 * w.im;
iix += 2;
} /* of ix */
} /* of mu */
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "[%2d] time to contract cvc: %e seconds\n", g_cart_id, retime-ratime);
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
/* Fourier transform data, copy to work */
for(mu=0; mu<4; mu++) {
memcpy((void*)in, (void*)(disc+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_p, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_p, in, NULL);
#endif
memcpy((void*)(work+_GWI(mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
}
/********************************
* read the second propagator
********************************/
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(format==0) {
sprintf(filename, "%s.%.4d.%.2d.inverted", filename_prefix, Nconf, sid+g_resume);
if(read_lime_spinor(g_spinor_field[2], filename, 0) != 0) break;
}
else if(format==1) {
sprintf(filename, "%s.%.4d.%.5d.inverted", filename_prefix, Nconf, sid+g_resume);
if(read_cmi(g_spinor_field[2], filename) != 0) break;
}
xchange_field(g_spinor_field[2]);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
fprintf(stdout, "time to read prop.: %e seconds\n", retime-ratime);
if(do_gt==1) {
/******************************************
* gauge transform the propagators for sid
******************************************/
for(ix=0; ix<VOLUME; ix++) {
_fv_eq_cm_ti_fv(spinor1, gauge_trafo+18*ix, g_spinor_field[2]+_GSI(ix));
_fv_eq_fv(g_spinor_field[2]+_GSI(ix), spinor1);
}
xchange_field(g_spinor_field[2]);
}
/************************************************
* calculate the source: apply Q_phi_tbc
************************************************/
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
Q_phi_tbc(g_spinor_field[0], g_spinor_field[2]);
xchange_field(g_spinor_field[0]);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "time to calculate source: %e seconds\n", retime-ratime);
/************************************************
* HPE: apply BH5
************************************************/
BH5(g_spinor_field[1], g_spinor_field[2]);
for(ix=0; ix<8*VOLUME; ix++) disc[ix] = 0.;
/* add new contractions to (existing) disc */
# ifdef MPI
ratime = MPI_Wtime();
# else
ratime = (double)clock() / CLOCKS_PER_SEC;
# endif
for(mu=0; mu<4; mu++) { /* loop on Lorentz index of the current */
iix = _GWI(mu,0,VOLUME);
for(ix=0; ix<VOLUME; ix++) { /* loop on lattice sites */
_cm_eq_cm_ti_co(U_, &g_gauge_field[_GGI(ix, mu)], &co_phase_up[mu]);
/* first contribution */
_fv_eq_cm_ti_fv(spinor1, U_, &g_spinor_field[1][_GSI(g_iup[ix][mu])]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_mi_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[0][_GSI(ix)], spinor2);
disc[iix ] -= 0.5 * w.re;
disc[iix+1] -= 0.5 * w.im;
/* second contribution */
_fv_eq_cm_dag_ti_fv(spinor1, U_, &g_spinor_field[1][_GSI(ix)]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_pl_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[0][_GSI(g_iup[ix][mu])], spinor2);
disc[iix ] -= 0.5 * w.re;
disc[iix+1] -= 0.5 * w.im;
iix += 2;
} /* of ix */
} /* of mu */
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "[%2d] time to contract cvc: %e seconds\n", g_cart_id, retime-ratime);
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
/* Fourier transform data, copy to work */
for(mu=0; mu<4; mu++) {
memcpy((void*)in, (void*)(disc+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_m, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_m, in, NULL);
#endif
memcpy((void*)(work+_GWI(4+mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
}
fnorm = 1. / ((double)(T_global*LX*LY*LZ));
fprintf(stdout, "fnorm = %e\n", fnorm);
for(mu=0; mu<4; mu++) {
for(nu=0; nu<4; nu++) {
cp1 = (complex*)(work+_GWI(mu,0,VOLUME));
cp2 = (complex*)(work+_GWI(4+nu,0,VOLUME));
cp3 = (complex*)(work+_GWI(8+4*mu+nu,0,VOLUME));
for(x0=0; x0<T; x0++) {
q[0] = (double)(x0+Tstart) / (double)T_global;
for(x1=0; x1<LX; x1++) {
q[1] = (double)(x1) / (double)LX;
for(x2=0; x2<LY; x2++) {
q[2] = (double)(x2) / (double)LY;
for(x3=0; x3<LZ; x3++) {
q[3] = (double)(x3) / (double)LZ;
ix = g_ipt[x0][x1][x2][x3];
w.re = cos( M_PI * (q[mu]-q[nu]) );
w.im = sin( M_PI * (q[mu]-q[nu]) );
_co_eq_co_ti_co(&w1, cp1, cp2);
_co_eq_co_ti_co(cp3, &w1, &w);
_co_ti_eq_re(cp3, fnorm);
cp1++; cp2++; cp3++;
}
}
}
}
}
}
/* save the result in momentum space */
sprintf(filename, "cvc_hpe5_ft.%.4d.%.2d", Nconf, sid);
write_contraction(work+_GWI(8,0,VOLUME), NULL, filename, 16, 0, 0);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "time to save cvc results: %e seconds\n", retime-ratime);
} /* of loop on sid */
/***********************************************
* free the allocated memory, finalize
***********************************************/
free(g_gauge_field);
for(i=0; i<no_fields; i++) free(g_spinor_field[i]);
free(g_spinor_field);
free_geometry();
fftw_free(in);
free(disc);
free(work);
#ifdef MPI
fftwnd_mpi_destroy_plan(plan_p);
fftwnd_mpi_destroy_plan(plan_m);
free(status);
MPI_Finalize();
#else
fftwnd_destroy_plan(plan_p);
fftwnd_destroy_plan(plan_m);
#endif
return(0);
}
int main(int argc, char **argv) {
int c, i, mu;
int count = 0;
int filename_set = 0;
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, it, iy;
int sid1, sid2, status, gid;
int Thp1, Lhp1, nmom, shift[4], shift2[4], nperm;
double *disc1=NULL, *disc2=NULL;
double *work = NULL;
double r2, fnorm;
char filename[100];
double ratime, retime;
complex w;
int *mom_tab=NULL, *mom_members=NULL, *mom_perm=NULL;
FILE *ofs;
int perm_tab_3[6][3];
perm_tab_3[0][0] = 0;
perm_tab_3[0][1] = 1;
perm_tab_3[0][2] = 2;
perm_tab_3[1][0] = 1;
perm_tab_3[1][1] = 2;
perm_tab_3[1][2] = 0;
perm_tab_3[2][0] = 2;
perm_tab_3[2][1] = 0;
perm_tab_3[2][2] = 1;
perm_tab_3[3][0] = 0;
perm_tab_3[3][1] = 2;
perm_tab_3[3][2] = 1;
perm_tab_3[4][0] = 1;
perm_tab_3[4][1] = 0;
perm_tab_3[4][2] = 2;
perm_tab_3[5][0] = 2;
perm_tab_3[5][1] = 1;
perm_tab_3[5][2] = 0;
while ((c = getopt(argc, argv, "h?f:")) != -1) {
switch (c) {
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
fprintf(stdout, "\n**************************************************\n");
fprintf(stdout, "* jc_ud_tr\n");
fprintf(stdout, "**************************************************\n\n");
/* initialize */
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
fprintf(stdout, "# [%2d] parameters:\n"\
"# T = %3d\n"\
"# Tstart = %3d\n"\
"# l_LX_at = %3d\n"\
"# l_LXstart_at = %3d\n"\
"# FFTW_LOC_VOLUME = %3d\n",
g_cart_id, T, Tstart, l_LX_at, l_LXstart_at, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
Thp1 = T /2 + 1;
Lhp1 = LX/2 + 1;
nmom = 3*Lhp1 - 2;
/****************************************
* initialize the momenta
****************************************/
mom_tab = (int*)calloc(3*nmom, sizeof(int));
if( mom_tab==NULL) {
fprintf(stderr, "could not allocate memory for mom_tab\n");
exit(4);
}
mom_tab[0] = 0; mom_tab[1] = 0; mom_tab[2] = 0;
count=3;
for(x1=1; x1<Lhp1; x1++) {
mom_tab[count ] = x1;
mom_tab[count+1] = 0;
mom_tab[count+2] = 0;
mom_tab[count+3] = x1;
mom_tab[count+4] = 1;
mom_tab[count+5] = 0;
mom_tab[count+6] = x1;
mom_tab[count+7] = 1;
mom_tab[count+8] = 1;
count+=9;
}
mom_members = (int*)calloc(Thp1*nmom, sizeof(int));
mom_perm = (int*)calloc(nmom, sizeof(int));
mom_perm[0] = 1;
mom_perm[1] = 3;
mom_perm[2] = 3;
mom_perm[3] = 1;
for (i=2; i<Lhp1; i++) {
mom_perm[3*i-2] = 3;
mom_perm[3*i-1] = 6;
mom_perm[3*i ] = 3;
}
for (i=0; i<nmom; i++)
fprintf(stdout, "# %d\t(%d, %d, %d)\t%d\n", i, mom_tab[3*i], mom_tab[3*i+1], mom_tab[3*i+2], mom_perm[i]);
/****************************************
* allocate memory for the contractions
****************************************/
disc1 = (double*)calloc(8*VOLUME, sizeof(double));
disc2 = (double*)calloc(8*VOLUME, sizeof(double));
if( disc1==NULL || disc2==NULL) {
fprintf(stderr, "could not allocate memory for disc\n");
exit(3);
}
work = (double*)calloc(8*Thp1*nmom, sizeof(double));
if( work == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for work\n");
exit(3);
}
/***********************************************
* start loop on gauge id.s
***********************************************/
for(gid=g_gaugeid; gid<=g_gaugeid2; gid++) {
for(ix=0; ix<8*Thp1*nmom; ix++) work[ix] = 0.;
for(ix=0; ix<8*VOLUME; ix++) disc2[ix] = 0.;
for (i=0; i<Thp1*nmom; i++) mom_members[i] = 0;
/***********************************************
* start loop on source id.s
***********************************************/
ratime = (double)clock() / CLOCKS_PER_SEC;
for(sid1=g_sourceid; sid1<=g_sourceid2; sid1+=g_sourceid_step) {
sprintf(filename, "jc_ud_x.%.4d.%.4d", gid, sid1);
if(read_lime_contraction(disc1, filename, 4, 0) != 0) break;
for(ix=0; ix<8*VOLUME; ix++) disc2[ix] += disc1[ix];
count=0;
for (it=0; it<Thp1; it++) {
shift[0] = it; shift2[0] = it;
for (i=0; i<nmom; i++) {
shift[1] = mom_tab[3*i ];
shift[2] = mom_tab[3*i+1];
shift[3] = mom_tab[3*i+2];
for (mu=0; mu<mom_perm[i]; mu++) {
// fprintf(stdout, "# mom=%d,\tperm=%d\n", i, mom_perm[i]);
shift2[1] = shift[perm_tab_3[mu][0]+1];
shift2[2] = shift[perm_tab_3[mu][1]+1];
shift2[3] = shift[perm_tab_3[mu][2]+1];
for(x0=shift2[0]; x0<T; x0++) {
for(x1=shift2[1]; x1<LX; x1++) {
for(x2=shift2[2]; x2<LY; x2++) {
for(x3=shift2[3]; x3<LZ; x3++) {
ix = g_ipt[x0][x1][x2][x3];
iy = g_ipt[x0-shift2[0]][x1-shift2[1]][x2-shift2[2]][x3-shift2[3]];
// fprintf(stdout, "shift2=(%d,%d,%d,%d); x=(%d,%d,%d,%d); ix=%d, iy=%d\n",
// shift2[0], shift2[1],shift2[2],shift2[3], x0, x1, x2, x3, ix, iy);
_co_eq_co_ti_co(&w, (complex*)(disc1+_GWI(0,ix,VOLUME)), (complex*)(disc1+_GWI(0,iy,VOLUME)));
work[2*( count) ] -= w.re;
work[2*( count)+1] -= w.im;
_co_eq_co_ti_co(&w, (complex*)(disc1+_GWI(1,ix,VOLUME)), (complex*)(disc1+_GWI(1,iy,VOLUME)));
work[2*( Thp1*nmom+count) ] -= w.re;
work[2*( Thp1*nmom+count)+1] -= w.im;
_co_eq_co_ti_co(&w, (complex*)(disc1+_GWI(2,ix,VOLUME)), (complex*)(disc1+_GWI(2,iy,VOLUME)));
work[2*(2*Thp1*nmom+count) ] -= w.re;
work[2*(2*Thp1*nmom+count)+1] -= w.im;
_co_eq_co_ti_co(&w, (complex*)(disc1+_GWI(3,ix,VOLUME)), (complex*)(disc1+_GWI(3,iy,VOLUME)));
work[2*(3*Thp1*nmom+count) ] -= w.re;
work[2*(3*Thp1*nmom+count)+1] -= w.im;
}}}}
}
count++;
}
} /* of it=0,...,T/2 */
} /* of loop on sid1 */
count=0;
for (it=0; it<Thp1; it++) {
shift[0] = it; shift2[0] = it;
for (i=0; i<nmom; i++) {
shift[1] = mom_tab[3*i ];
shift[2] = mom_tab[3*i+1];
shift[3] = mom_tab[3*i+2];
for (mu=0; mu<mom_perm[i]; mu++) {
// fprintf(stdout, "# mom=%d,\tperm=%d\n", i, mom_perm[i]);
shift2[1] = shift[perm_tab_3[mu][0]+1];
shift2[2] = shift[perm_tab_3[mu][1]+1];
shift2[3] = shift[perm_tab_3[mu][2]+1];
for(x0=shift2[0]; x0<T; x0++) {
for(x1=shift2[1]; x1<LX; x1++) {
for(x2=shift2[2]; x2<LY; x2++) {
for(x3=shift2[3]; x3<LZ; x3++) {
ix = g_ipt[x0][x1][x2][x3];
iy = g_ipt[x0-shift2[0]][x1-shift2[1]][x2-shift2[2]][x3-shift2[3]];
// fprintf(stdout, "shift2=(%d,%d,%d,%d); x=(%d,%d,%d,%d); ix=%d, iy=%d\n",
// shift2[0], shift2[1],shift2[2],shift2[3], x0, x1, x2, x3, ix, iy);
_co_eq_co_ti_co(&w, (complex*)(disc2+_GWI(0,ix,VOLUME)), (complex*)(disc2+_GWI(0,iy,VOLUME)));
work[2*( count) ] += w.re;
work[2*( count)+1] += w.im;
_co_eq_co_ti_co(&w, (complex*)(disc2+_GWI(1,ix,VOLUME)), (complex*)(disc2+_GWI(1,iy,VOLUME)));
work[2*( Thp1*nmom+count) ] += w.re;
work[2*( Thp1*nmom+count)+1] += w.im;
_co_eq_co_ti_co(&w, (complex*)(disc2+_GWI(2,ix,VOLUME)), (complex*)(disc2+_GWI(2,iy,VOLUME)));
work[2*(2*Thp1*nmom+count) ] += w.re;
work[2*(2*Thp1*nmom+count)+1] += w.im;
_co_eq_co_ti_co(&w, (complex*)(disc2+_GWI(3,ix,VOLUME)), (complex*)(disc2+_GWI(3,iy,VOLUME)));
work[2*(3*Thp1*nmom+count) ] += w.re;
work[2*(3*Thp1*nmom+count)+1] += w.im;
mom_members[count]++;
}}}}
}
count++;
}
} /* of it=0,...,T/2 */
/* normalization */
count=0;
for (it=0; it<Thp1; it++) {
for (i=0; i<nmom; i++) {
fprintf(stdout, "%d\t%d\t%d\n", it, i, mom_members[count]);
count++;
}
}
for (mu=0; mu<4; mu++) {
count=0;
for (it=0; it<Thp1; it++) {
for(i=0; i<nmom; i++) {
fnorm = 1. / ( (double)mom_members[count]
* (double)(g_sourceid2-g_sourceid+1) * (double)(g_sourceid2-g_sourceid) );
// fprintf(stdout, "# fnorm(%d,%2d) = %25.16e\n", mu, count, fnorm);
work[2*(mu*Thp1*nmom+count) ] *= fnorm;
work[2*(mu*Thp1*nmom+count)+1] *= fnorm;
count++;
}
}
}
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id == 0) fprintf(stdout, "# time for building correl.: %e seconds\n", retime-ratime);
/************************************************
* save results
************************************************/
sprintf(filename, "jc_ud_tr.%4d", gid);
ofs = fopen(filename, "w");
if (ofs==NULL) {
fprintf(stderr, "Error, could not open file %s for writing\n", filename);
}
for(mu=0; mu<4; mu++) {
count=0;
for (it=0; it<Thp1; it++) {
for (i=0; i<nmom; i++) {
r2 = sqrt( mom_tab[3*i]*mom_tab[3*i] + mom_tab[3*i+1]*mom_tab[3*i+1]
+ mom_tab[3*i+2]*mom_tab[3*i+2] );
fprintf(ofs, "%3d%3d%3d%3d%16.7e%25.16e%25.16e\n", it,
mom_tab[3*i], mom_tab[3*i+1],mom_tab[3*i+2], r2,
work[2*(mu*Thp1*nmom+count)], work[2*(mu*Thp1*nmom+count)+1]);
count++;
}
}
}
fclose(ofs);
} /* of loop on gid */
/***********************************************
* free the allocated memory, finalize
***********************************************/
free_geometry();
free(disc1);
free(disc2);
free(work);
free(mom_tab);
free(mom_perm);
free(mom_members);
return(0);
}
int main(int argc, char **argv) {
int c, i, mu, nu;
int count = 0;
int filename_set = 0;
int dims[4] = {0,0,0,0};
int l_LX_at, l_LXstart_at;
int x0, x1, x2, x3, ix, iix;
int dxm[4], dxn[4], ixpm, ixpn;
int sid;
double *disc = (double*)NULL;
double *disc2 = (double*)NULL;
double *work = (double*)NULL;
double q[4], fnorm;
int verbose = 0;
int do_gt = 0;
char filename[100], contype[200];
double ratime, retime;
double plaq, _2kappamu, hpe3_coeff, onepmutilde2, mutilde2;
double spinor1[24], spinor2[24], U_[18], U1_[18], U2_[18];
double *gauge_trafo=(double*)NULL;
complex w, w1, w2, *cp1, *cp2, *cp3;
FILE *ofs;
fftw_complex *in=(fftw_complex*)NULL;
#ifdef MPI
fftwnd_mpi_plan plan_p, plan_m;
#else
fftwnd_plan plan_p, plan_m;
#endif
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "h?vgf:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'g':
do_gt = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# Reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
/* initialize fftw */
dims[0]=T_global; dims[1]=LX; dims[2]=LY; dims[3]=LZ;
#ifdef MPI
plan_p = fftwnd_mpi_create_plan(g_cart_grid, 4, dims, FFTW_BACKWARD, FFTW_MEASURE);
plan_m = fftwnd_mpi_create_plan(g_cart_grid, 4, dims, FFTW_FORWARD, FFTW_MEASURE);
fftwnd_mpi_local_sizes(plan_p, &T, &Tstart, &l_LX_at, &l_LXstart_at, &FFTW_LOC_VOLUME);
#else
plan_p = fftwnd_create_plan(4, dims, FFTW_BACKWARD, FFTW_MEASURE | FFTW_IN_PLACE);
plan_m = fftwnd_create_plan(4, dims, FFTW_FORWARD, FFTW_MEASURE | FFTW_IN_PLACE);
T = T_global;
Tstart = 0;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
#endif
fprintf(stdout, "# [%2d] fftw parameters:\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, T, g_cart_id, Tstart, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
#ifdef MPI
if(T==0) {
fprintf(stderr, "[%2d] local T is zero; exit\n", g_cart_id);
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
exit(2);
}
#endif
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(1);
}
geometry();
/* read the gauge field */
alloc_gauge_field(&g_gauge_field, VOLUMEPLUSRAND);
sprintf(filename, "%s.%.4d", gaugefilename_prefix, Nconf);
if(g_cart_id==0) fprintf(stdout, "reading gauge field from file %s\n", filename);
read_lime_gauge_field_doubleprec(filename);
#ifdef MPI
xchange_gauge();
#endif
/* measure the plaquette */
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "measured plaquette value: %25.16e\n", plaq);
if(do_gt==1) {
/***********************************
* initialize gauge transformation
***********************************/
init_gauge_trafo(&gauge_trafo, 1.);
apply_gt_gauge(gauge_trafo);
plaquette(&plaq);
if(g_cart_id==0) fprintf(stdout, "measured plaquette value after gauge trafo: %25.16e\n", plaq);
}
/****************************************
* allocate memory for the spinor fields
****************************************/
no_fields = 3;
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields; i++) alloc_spinor_field(&g_spinor_field[i], VOLUMEPLUSRAND);
/****************************************
* allocate memory for the contractions
****************************************/
disc = (double*)calloc( 8*VOLUME, sizeof(double));
if( disc == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(3);
}
for(ix=0; ix<8*VOLUME; ix++) disc[ix] = 0.;
disc2 = (double*)calloc( 8*VOLUME, sizeof(double));
if( disc2 == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for disc2\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(3);
}
for(ix=0; ix<8*VOLUME; ix++) disc2[ix] = 0.;
work = (double*)calloc(48*VOLUME, sizeof(double));
if( work == (double*)NULL ) {
fprintf(stderr, "could not allocate memory for work\n");
# ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
# endif
exit(3);
}
/****************************************
* prepare Fourier transformation arrays
****************************************/
in = (fftw_complex*)malloc(FFTW_LOC_VOLUME*sizeof(fftw_complex));
if(in==(fftw_complex*)NULL) {
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 1);
MPI_Finalize();
#endif
exit(4);
}
/************************************************
* HPE: calculate coeff. of 3rd order term
************************************************/
_2kappamu = 2. * g_kappa * g_mu;
onepmutilde2 = 1. + _2kappamu * _2kappamu;
mutilde2 = _2kappamu * _2kappamu;
hpe3_coeff = 16. * g_kappa*g_kappa*g_kappa*g_kappa * (1. + 6. * mutilde2 + mutilde2*mutilde2) / onepmutilde2 / onepmutilde2 / onepmutilde2 / onepmutilde2;
/*
hpe3_coeff = 8. * g_kappa*g_kappa*g_kappa * \
(1. + 6.*_2kappamu*_2kappamu + _2kappamu*_2kappamu*_2kappamu*_2kappamu) / (1. + _2kappamu*_2kappamu) / (1. + _2kappamu*_2kappamu) / (1. + _2kappamu*_2kappamu) / (1. + _2kappamu*_2kappamu);
*/
fprintf(stdout, "hpe3_coeff = %25.16e\n", hpe3_coeff);
/************************************************
* HPE: calculate the plaquette terms
************************************************/
for(ix=0; ix<VOLUME; ix++) {
for(mu=0; mu<4; mu++) {
for(i=1; i<4; i++) {
nu = (mu+i)%4;
_cm_eq_cm_ti_cm(U1_, g_gauge_field+_GGI(ix,mu), g_gauge_field+_GGI(g_iup[ix][mu],nu) );
_cm_eq_cm_ti_cm(U2_, g_gauge_field+_GGI(ix,nu), g_gauge_field+_GGI(g_iup[ix][nu],mu) );
_cm_eq_cm_ti_cm_dag(U_, U1_, U2_);
_co_eq_tr_cm(&w1, U_);
iix = g_idn[ix][nu];
_cm_eq_cm_ti_cm(U1_, g_gauge_field+_GGI(iix,mu), g_gauge_field+_GGI(g_iup[iix][mu],nu) );
_cm_eq_cm_ti_cm(U2_, g_gauge_field+_GGI(iix,nu), g_gauge_field+_GGI(g_iup[iix][nu],mu) );
_cm_eq_cm_ti_cm_dag(U_, U1_, U2_);
_co_eq_tr_cm(&w2, U_);
disc2[_GWI(mu,ix,VOLUME)+1] += hpe3_coeff * (w1.im - w2.im);
/*
_cm_eq_cm_ti_cm(U1_, g_gauge_field+_GGI(g_idn[ix][nu],nu), g_gauge_field+_GGI(ix,mu) );
_cm_eq_cm_ti_cm(U2_, g_gauge_field+_GGI(g_idn[ix][nu],mu), g_gauge_field+_GGI(g_iup[g_idn[ix][nu]][mu], nu) );
_cm_eq_cm_ti_cm_dag(U_, U1_, U2_);
_co_eq_tr_cm(&w2, U_);
disc2[_GWI(mu,ix,VOLUME)+1] += hpe3_coeff * (w1.im + w2.im);
*/
/* fprintf(stdout, "mu=%1d, ix=%5d, nu=%1d, w1=%25.16e +i %25.16e; w2=%25.16e +i %25.16e\n",
mu, ix, nu, w1.re, w1.im, w2.re, w2.im); */
} /* of nu */
/****************************************
* - in case lattice size equals 4
* calculate additional loop term
* - _NOTE_ the possible minus sign from
* the fermionic boundary conditions
****************************************/
if(dims[mu]==4) {
wilson_loop(&w, ix, mu, dims[mu]);
fnorm = -64. * g_kappa*g_kappa*g_kappa*g_kappa / onepmutilde2 / onepmutilde2 / onepmutilde2 / onepmutilde2;
disc2[_GWI(mu,ix,VOLUME)+1] += fnorm * w.im;
/* fprintf(stdout, "loop contribution: ix=%5d, mu=%2d, fnorm=%25.16e, w=%25.16e\n", ix, mu, fnorm, w.im); */
}
/*
fprintf(stdout, "-------------------------------------------\n");
fprintf(stdout, "disc2[ix=%d,mu=%d] = %25.16e +i %25.16e\n", ix, mu, disc2[_GWI(mu,ix,VOLUME)], disc2[_GWI(mu,ix,VOLUME)+1]);
fprintf(stdout, "-------------------------------------------\n");
*/
}
}
/*
sprintf(filename, "avc_disc_hpe5_3rd.%.4d", Nconf);
ofs = fopen(filename, "w");
for(ix=0; ix<VOLUME; ix++) {
for(mu=0; mu<4; mu++) {
fprintf(ofs, "%6d%3d%25.16e\t%25.16e\n", ix, mu, disc[_GWI(mu,ix,VOLUME)], disc[_GWI(mu,ix,VOLUME)+1]);
}
}
fclose(ofs);
for(ix=0; ix<8*VOLUME; ix++) disc[ix] = 0.;
*/
/*
for(x0=0; x0<T; x0++) {
for(x1=0; x1<LX; x1++) {
for(x2=0; x2<LY; x2++) {
for(x3=0; x3<LZ; x3++) {
ix = g_ipt[x0][x1][x2][x3];
for(mu=0; mu<4; mu++) {
dxm[0]=0; dxm[1]=0; dxm[2]=0; dxm[3]=0; dxm[mu]=1;
for(i=1; i<4; i++) {
nu = (mu+i)%4;
dxn[0]=0; dxn[1]=0; dxn[2]=0; dxn[3]=0; dxn[nu]=1;
ixpm = g_ipt[(x0+dxm[0]+T)%T][(x1+dxm[1]+LX)%LX][(x2+dxm[2]+LY)%LY][(x3+dxm[3]+LZ)%LZ];
ixpn = g_ipt[(x0+dxn[0]+T)%T][(x1+dxn[1]+LX)%LX][(x2+dxn[2]+LY)%LY][(x3+dxn[3]+LZ)%LZ];
_cm_eq_cm_ti_cm(U1_, g_gauge_field + 72*ix+18*mu, g_gauge_field + 72*ixpm+18*nu );
_cm_eq_cm_ti_cm(U2_, g_gauge_field + 72*ix+18*nu, g_gauge_field + 72*ixpn+18*mu );
_cm_eq_cm_ti_cm_dag(U_, U1_, U2_);
_co_eq_tr_cm(&w1, U_);
ixpm = g_ipt[(x0+dxm[0]-dxn[0]+T)%T][(x1+dxm[1]-dxn[1]+LX)%LX][(x2+dxm[2]-dxn[2]+LY)%LY][(x3+dxm[3]-dxn[3]+LZ)%LZ];
ixpn = g_ipt[(x0-dxn[0]+T)%T][(x1-dxn[1]+LX)%LX][(x2-dxn[2]+LY)%LY][(x3-dxn[3]+LZ)%LZ];
_cm_eq_cm_ti_cm(U1_, g_gauge_field + 72*ixpn+18*nu, g_gauge_field + 72*ix+18*mu);
_cm_eq_cm_ti_cm(U2_, g_gauge_field + 72*ixpn+18*mu, g_gauge_field + 72*ixpm+18*nu);
_cm_eq_cm_ti_cm_dag(U_, U1_, U2_);
_co_eq_tr_cm(&w2, U_);
disc2[_GWI(mu,ix,VOLUME)+1] += hpe3_coeff * (w1.im + w2.im);
fprintf(stdout, "mu=%1d, ix=%5d, nu=%1d, w1=%25.16e; w2=%25.16e\n", mu, ix, nu, w1.im, w2.im);
}
fprintf(stdout, "-------------------------------------------\n");
fprintf(stdout, "disc2[ix=%d,mu=%d] = %25.16e +i %25.16e\n", ix, mu, disc2[_GWI(mu,ix,VOLUME)], disc2[_GWI(mu,ix,VOLUME)+1]);
fprintf(stdout, "-------------------------------------------\n");
}
}
}
}
}
*/
/***********************************************
* start loop on source id.s
***********************************************/
for(sid=g_sourceid; sid<=g_sourceid2; sid+=g_sourceid_step) {
/* read the new propagator */
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(format==0) {
sprintf(filename, "%s.%.4d.%.2d.inverted", filename_prefix, Nconf, sid);
if(read_lime_spinor(g_spinor_field[2], filename, 0) != 0) break;
}
else if(format==1) {
sprintf(filename, "%s.%.4d.%.5d.inverted", filename_prefix, Nconf, sid);
if(read_cmi(g_spinor_field[2], filename) != 0) break;
}
xchange_field(g_spinor_field[2]);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
fprintf(stdout, "time to read prop.: %e seconds\n", retime-ratime);
if(do_gt==1) {
/******************************************
* gauge transform the propagators for sid
******************************************/
for(ix=0; ix<VOLUME; ix++) {
_fv_eq_cm_ti_fv(spinor1, gauge_trafo+18*ix, g_spinor_field[2]+_GSI(ix));
_fv_eq_fv(g_spinor_field[2]+_GSI(ix), spinor1);
}
xchange_field(g_spinor_field[2]);
}
count++;
/************************************************
* calculate the source: apply Q_phi_tbc
************************************************/
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
Q_phi_tbc(g_spinor_field[0], g_spinor_field[2]);
xchange_field(g_spinor_field[0]);
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "time to calculate source: %e seconds\n", retime-ratime);
/************************************************
* HPE: apply BH5
************************************************/
BH5(g_spinor_field[1], g_spinor_field[2]);
/* add new contractions to (existing) disc */
# ifdef MPI
ratime = MPI_Wtime();
# else
ratime = (double)clock() / CLOCKS_PER_SEC;
# endif
for(mu=0; mu<4; mu++) { /* loop on Lorentz index of the current */
iix = _GWI(mu,0,VOLUME);
for(ix=0; ix<VOLUME; ix++) { /* loop on lattice sites */
_cm_eq_cm_ti_co(U_, &g_gauge_field[_GGI(ix, mu)], &co_phase_up[mu]);
/* first contribution */
_fv_eq_cm_ti_fv(spinor1, U_, &g_spinor_field[1][_GSI(g_iup[ix][mu])]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_mi_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[0][_GSI(ix)], spinor2);
disc[iix ] -= 0.5 * w.re;
disc[iix+1] -= 0.5 * w.im;
/* second contribution */
_fv_eq_cm_dag_ti_fv(spinor1, U_, &g_spinor_field[1][_GSI(ix)]);
_fv_eq_gamma_ti_fv(spinor2, mu, spinor1);
_fv_pl_eq_fv(spinor2, spinor1);
_co_eq_fv_dag_ti_fv(&w, &g_spinor_field[0][_GSI(g_iup[ix][mu])], spinor2);
disc[iix ] -= 0.5 * w.re;
disc[iix+1] -= 0.5 * w.im;
iix += 2;
} /* of ix */
} /* of mu */
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time to contract cvc: %e seconds\n", retime-ratime);
/************************************************
* save results for count = multiple of Nsave
************************************************/
if(count%Nsave == 0) {
if(g_cart_id == 0) fprintf(stdout, "save results for count = %d\n", count);
fnorm = 1. / ( (double)count * g_prop_normsqr );
if(g_cart_id==0) fprintf(stdout, "# X-fnorm = %e\n", fnorm);
for(mu=0; mu<4; mu++) {
for(ix=0; ix<VOLUME; ix++) {
work[_GWI(mu,ix,VOLUME) ] = disc[_GWI(mu,ix,VOLUME) ] * fnorm + disc2[_GWI(mu,ix,VOLUME) ];
work[_GWI(mu,ix,VOLUME)+1] = disc[_GWI(mu,ix,VOLUME)+1] * fnorm + disc2[_GWI(mu,ix,VOLUME)+1];
}
}
/* save the result in position space */
sprintf(filename, "cvc_hpe5_X.%.4d.%.4d", Nconf, count);
sprintf(contype, "cvc-disc-all-hpe-05-X");
write_lime_contraction(work, filename, 64, 4, contype, Nconf, count);
/*
sprintf(filename, "cvc_hpe5_Xascii.%.4d.%.4d", Nconf, count);
write_contraction(work, NULL, filename, 4, 2, 0);
*/
#ifdef MPI
ratime = MPI_Wtime();
#else
ratime = (double)clock() / CLOCKS_PER_SEC;
#endif
/* Fourier transform data, copy to work */
for(mu=0; mu<4; mu++) {
memcpy((void*)in, (void*)(work+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_m, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_m, in, NULL);
#endif
memcpy((void*)(work+_GWI(4+mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
memcpy((void*)in, (void*)(work+_GWI(mu,0,VOLUME)), 2*VOLUME*sizeof(double));
#ifdef MPI
fftwnd_mpi(plan_p, 1, in, NULL, FFTW_NORMAL_ORDER);
#else
fftwnd_one(plan_p, in, NULL);
#endif
memcpy((void*)(work+_GWI(mu,0,VOLUME)), (void*)in, 2*VOLUME*sizeof(double));
} /* of mu =0 ,..., 3*/
fnorm = 1. / (double)(T_global*LX*LY*LZ);
if(g_cart_id==0) fprintf(stdout, "# P-fnorm = %e\n", fnorm);
for(mu=0; mu<4; mu++) {
for(nu=0; nu<4; nu++) {
cp1 = (complex*)(work+_GWI(mu,0,VOLUME));
cp2 = (complex*)(work+_GWI(4+nu,0,VOLUME));
cp3 = (complex*)(work+_GWI(8+4*mu+nu,0,VOLUME));
for(x0=0; x0<T; x0++) {
q[0] = (double)(x0+Tstart) / (double)T_global;
for(x1=0; x1<LX; x1++) {
q[1] = (double)(x1) / (double)LX;
for(x2=0; x2<LY; x2++) {
q[2] = (double)(x2) / (double)LY;
for(x3=0; x3<LZ; x3++) {
q[3] = (double)(x3) / (double)LZ;
ix = g_ipt[x0][x1][x2][x3];
w.re = cos( M_PI * (q[mu]-q[nu]) );
w.im = sin( M_PI * (q[mu]-q[nu]) );
_co_eq_co_ti_co(&w1, cp1, cp2);
_co_eq_co_ti_co(cp3, &w1, &w);
_co_ti_eq_re(cp3, fnorm);
cp1++; cp2++; cp3++;
}
}
}
}
}
}
/* save the result in momentum space */
sprintf(filename, "cvc_hpe5_P.%.4d.%.4d", Nconf, count);
sprintf(contype, "cvc-disc-all-hpe-05-P");
write_lime_contraction(work+_GWI(8,0,VOLUME), filename, 64, 16, contype, Nconf, count);
/*
sprintf(filename, "cvc_hpe5_Pascii.%.4d.%.4d", Nconf, count);
write_contraction(work+_GWI(8,0,VOLUME), NULL, filename, 16, 2, 0);
*/
#ifdef MPI
retime = MPI_Wtime();
#else
retime = (double)clock() / CLOCKS_PER_SEC;
#endif
if(g_cart_id==0) fprintf(stdout, "# time to save cvc results: %e seconds\n", retime-ratime);
} /* of count % Nsave == 0 */
} /* of loop on sid */
/***********************************************
* free the allocated memory, finalize
***********************************************/
free(g_gauge_field);
for(i=0; i<no_fields; i++) free(g_spinor_field[i]);
free(g_spinor_field);
free_geometry();
fftw_free(in);
free(disc);
free(work);
#ifdef MPI
fftwnd_mpi_destroy_plan(plan_p);
fftwnd_mpi_destroy_plan(plan_m);
MPI_Finalize();
#else
fftwnd_destroy_plan(plan_p);
fftwnd_destroy_plan(plan_m);
#endif
return(0);
}
int main(int argc, char **argv) {
int K=32, nfc=2, timeslice, status;
int c, i, j, k, ll, t, id, mu, icol;
int count, position=-1, position_set=0;
size_t shift;
long unsigned int VOL3;
int filename_set = 0;
int x0, x1, x2, x3, ix, idx;
int n_c=1, n_s=4;
int *xgindex1=NULL, *xgindex2=NULL, *xisimag=NULL;
int write_ascii=0;
int prop_single_file=0;
double *xvsign=NULL;
void *cconn=NULL;
void *buffer = NULL;
int sigmalight=0, sigmaheavy=0;
double correlator_norm = 1.;
void * vptr;
int source_coords[4];
int verbose = 0;
size_t prec = 64, bytes;
char filename[200];
double ratime, retime;
void *chi=NULL, *psi=NULL;
FILE *ofs=NULL, *ofs2=NULL;
double c_conf_gamma_sign[] = {1., 1., 1., -1., -1., -1., -1., 1., 1., 1., -1., -1., 1., 1., 1., 1.};
double n_conf_gamma_sign[] = {1., 1., 1., -1., -1., -1., -1., 1., 1., 1., 1., 1., -1., -1., 1., 1.};
double *conf_gamma_sign=NULL;
void *spinor_field=NULL;
DML_Checksum *checksum=NULL;
/**************************************************************************************************
* charged stuff
*
* (pseudo-)scalar:
* g5 - g5, g5 - g0g5, g0g5 - g5, g0g5 - g0g5,
* g0 - g0, g5 - g0, g0 - g5, g0g5 - g0,
* g0 - g0g5, 1 - 1, 1 - g5, g5 - 1,
* 1 - g0g5, g0g5 - 1, 1 - g0, g0 - 1
*
* (pseudo-)vector:
* gig0 - gig0, gi - gi, gig5 - gig5, gig0 - gi,
* gi - gig0, gig0 - gig5, gig5 - gig0, gi - gig5,
* gig5 - gi, gig0g5 - gig0g5, gig0 - gig0g5, gig0g5 - gig0,
* gi - gig0g5, gig0g5 - gi, gig5 - gig0g5, gig0g5 - gig5
**************************************************************************************************/
int gindex1[] = {5, 5, 6, 6, 0, 5, 0, 6, 0, 4, 4, 5, 4, 6, 4, 0,
10, 11, 12, 1, 2, 3, 7, 8, 9, 10, 11, 12, 1, 2, 3, 10, 11, 12, 7, 8, 9, 1, 2, 3, 7, 8, 9,
13, 14, 15, 10, 11, 12, 15, 14, 13, 1, 2, 3, 15, 14, 13, 7, 8, 9, 15, 14, 13};
int gindex2[] = {5, 6, 5, 6, 0, 0, 5, 0, 6, 4, 5, 4, 6, 4, 0, 4,
10, 11, 12, 1, 2, 3, 7, 8, 9, 1, 2, 3, 10, 11, 12, 7, 8, 9, 10, 11, 12, 7, 8, 9, 1, 2, 3,
13, 14, 15, 15, 14, 13, 10, 11, 12, 15, 14, 13, 1, 2, 3, 15, 14, 13, 7, 8, 9};
/* due to twisting we have several correlators that are purely imaginary */
int isimag[] = {0, 0, 0, 0,
0, 1, 1, 1,
1, 0, 1, 1,
1, 1, 0, 0,
0, 0, 0, 0,
0, 1, 1, 1,
1, 0, 1, 1,
1, 1, 0, 0};
double vsign[] = {1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., 1., 1., 1., -1., 1., 1., -1., 1.,
1., -1., 1., 1., -1., 1., 1., -1., 1., 1., -1., 1.};
/**************************************************************************************************
* neutral stuff
*
* (pseudo-)scalar:
* g5 - g5, g5 - g0g5, g0g5 - g5, g0g5 - g0g5,
* 1 - 1, g5 - 1, 1 - g5, g0g5 - 1,
* 1 - g0g5, g0 - g0, g0 - g5, g5 - g0,
* g0 - g0g5, g0g5 - g0, g0 - 1, 1 - g0
*
* (pseudo-)vector:
* gig0 - gig0, gi - gi, gig0g5 - gig0g5, gig0 - gi,
* gi - gig0, gig0 - gig0g5, gig0g5 - gig0, gi - gig0g5,
* gig0g5 - gi gig5 - gig5, gig5 - gi, gi - gig5,
* gig5 - gig0, gig0 - gig5, gig5 - gig0g5, gig0g5 - gig5
**************************************************************************************************/
int ngindex1[] = {5, 5, 6, 6, 4, 5, 4, 6, 4, 0, 0, 5, 0, 6, 0, 4,
10, 11, 12, 1, 2, 3, 13, 14, 15, 10, 11, 12, 1, 2, 3, 10, 11, 12, 15, 14, 13, 1, 2, 3, 15, 14, 13,
7, 8, 9, 7, 8, 9, 1, 2, 3, 7, 8, 9, 10, 11, 12, 7, 8, 9, 15, 14, 13};
int ngindex2[] = {5, 6, 5, 6, 4, 4, 5, 4, 6, 0, 5, 0, 6, 0, 4, 0,
10, 11, 12, 1, 2, 3, 13, 14, 15, 1, 2, 3, 10, 11, 12, 15, 14, 13, 10, 11, 12, 15, 14, 13, 1, 2, 3,
7, 8, 9, 1, 2, 3, 7, 8, 9, 10, 11, 12, 7, 8, 9, 15, 14, 13, 7, 8, 9};
int nisimag[] = {0, 0, 0, 0,
0, 1, 1, 1,
1, 0, 1, 1,
1, 1, 0, 0,
0, 0, 0, 0,
0, 1, 1, 1,
1, 0, 1, 1,
1, 1, 0, 0};
double nvsign[] = {1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., -1., 1., 1., -1., 1., 1., -1., 1.,
1., -1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
1., 1., 1., 1., 1., 1., 1., -1., 1., 1., -1., 1. };
/*
double isneg_std[]= {+1., -1., +1., -1., +1., +1., +1., +1., -1., +1., +1., +1., +1., +1., +1., +1.,
-1., +1., -1., -1., +1., +1., +1., -1., +1., -1., +1., +1., +1., +1., +1., +1.};
*/
double isneg_std[]= {+1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1.,
+1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1., +1.};
double isneg[32];
while ((c = getopt(argc, argv, "sah?vf:c:p:n:P:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'c':
n_c = atoi(optarg);
break;
case 'p':
position = atoi(optarg);
position_set = 1;
break;
case 'a':
write_ascii = 1;
fprintf(stdout, "# [] will write in ascii format\n");
break;
case 'n':
nfc = atoi(optarg);
fprintf(stdout, "# [] number of flavor combinations set to %d\n", nfc);
break;
case 's':
prop_single_file = 1;
fprintf(stdout, "# [] will read up and down from same file\n");
break;
case 'P':
prec = (size_t)atoi(optarg);
fprintf(stdout, "# [] set precision to %lu\n", prec);
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* the global time stamp */
g_the_time = time(NULL);
fprintf(stdout, "\n# [ll_conn_x2dep_extract] using global time stamp %s", ctime(&g_the_time));
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if( Nlong > 0 ) {
if(g_proc_id==0) fprintf(stdout, "Fuzzing not available in this version.\n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
VOL3 = LX*LY*LZ;
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] T_global = %3d\n"\
"# [%2d] T = %3d\n"\
"# [%2d] Tstart = %3d\n"\
"# [%2d] LX_global = %3d\n"\
"# [%2d] LX = %3d\n"\
"# [%2d] LXstart = %3d\n"\
"# [%2d] LY_global = %3d\n"\
"# [%2d] LY = %3d\n"\
"# [%2d] LYstart = %3d\n",
g_cart_id, g_cart_id, T_global, g_cart_id, T, g_cart_id, Tstart,
g_cart_id, LX_global, g_cart_id, LX, g_cart_id, LXstart,
g_cart_id, LY_global, g_cart_id, LY, g_cart_id, LYstart);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
#ifdef MPI
MPI_Abort(MPI_COMM_WORLD, 2);
MPI_Finalize();
#endif
exit(1);
}
// switch to double precision of single precision was set
if(prec == 32) {
fprintf(stderr, "[] Warning: switching to double precision\n");
prec = 64;
}
geometry();
fprintf(stdout, "# ll_conn wout MPI\n");
fprintf(stdout, "# number of colours = %d\n", n_c);
if(position_set == 0) {
position = g_propagator_position;
if(g_cart_id == 0) fprintf(stdout, "# using input file value for prop pos %d\n", position);
} else {
if(g_cart_id == 0) fprintf(stdout, "# using command line arg value for prop pos %d\n", position);
}
/*********************************************
* set the isneg field
*********************************************/
for(i = 0; i < K; i++) isneg[i] = isneg_std[i];
/*********************************************************
* allocate memory for the spinor fields
*********************************************************/
no_fields = n_s;
if(nfc>1) {
no_fields *= 2;
}
if(prec==64) {
spinor_field = calloc(no_fields, sizeof(double*));
} else {
spinor_field = calloc(no_fields, sizeof(float*));
}
if(g_cart_id==0) fprintf(stdout, "# no. of spinor fields is %d\n", no_fields);
if(prec==64) {
for(i=0; i<no_fields-1; i++) {
((double**)spinor_field)[i] = (double*)malloc(24*VOL3*sizeof(double));
if( ((double**)spinor_field)[i] == NULL) {
fprintf(stderr, "Error, could not alloc spinor field %d\n", i);
exit(12);
}
}
((double**)spinor_field)[i] = (double*)malloc(24*VOL3*sizeof(double));
if( ((double**)spinor_field)[i] == NULL) {
fprintf(stderr, "Error, could not alloc spinor field %d\n", i);
exit(12);
}
} else {
for(i=0; i<no_fields-1; i++) {
((float**)spinor_field)[i] = (float*)malloc(24*VOL3*sizeof(float));
if(((float**)spinor_field)[i] == NULL) {
fprintf(stderr, "Error, could not alloc spinor field %d\n", i);
exit(14);
}
}
((float**)spinor_field)[i] = (float*)malloc(24*VOL3*sizeof(float));
if( ((float**)spinor_field)[i] == NULL) {
fprintf(stderr, "Error, could not alloc spinor field %d\n", i);
exit(14);
}
}
checksum = (DML_Checksum*)malloc(2*n_c*n_s*sizeof(DML_Checksum));
if(checksum == NULL) {
fprintf(stderr, "[] Error, could not alloc checksumßn");
exit(75);
}
/*********************************************************
* allocate memory for the contractions
*********************************************************/
bytes = (prec==64) ? sizeof(double) : sizeof(float);
cconn = calloc(2*nfc*K*VOL3, bytes);
if( cconn==NULL ) {
fprintf(stderr, "could not allocate memory for cconn\n");
exit(3);
}
buffer = calloc(2*nfc*K*LZ, bytes);
if( buffer==NULL) {
fprintf(stderr, "could not allocate memory for buffers\n");
exit(4);
}
/******************************************************************
* calculate source coordinates
******************************************************************/
source_coords[0] = g_source_location / (LX_global*LY_global*LZ);
source_coords[1] = ( g_source_location % (LX_global*LY_global*LZ) ) / (LY_global*LZ);
source_coords[2] = ( g_source_location % (LY_global*LZ) ) / LZ;
source_coords[3] = g_source_location % LZ;
if(g_cart_id==0) fprintf(stdout, "# source coords = %3d%3d%3d%3d\n", source_coords[0], source_coords[1],
source_coords[2], source_coords[3]);
/******************************************************************
* final normalization of the correlators
******************************************************************/
/* correlator_norm = 1. / ( 2. * g_kappa * g_kappa * (double)(LX_global*LY_global*LZ) );*/
correlator_norm = 1.;
if(g_cart_id==0) fprintf(stdout, "# correlator_norm = %12.5e\n", correlator_norm);
/******************************************************************
******************************************************************
** **
** local - local **
** **
******************************************************************
******************************************************************/
if(g_cart_id==0) fprintf(stdout, "# Starting LL\n");
for(timeslice=0;timeslice<T_global;timeslice++) {
if(prec==64) {
for(idx=0; idx<2*nfc*K*VOL3; idx++) ((double*)cconn)[idx] = 0.;
} else {
for(idx=0; idx<2*nfc*K*VOL3; idx++) ((float*)cconn)[idx] = 0.;
}
for(icol=0;icol<n_c;icol++) {
ratime = (double)clock() / CLOCKS_PER_SEC;
for(i=0; i<n_s; i++) {
if(prec==64) {
sprintf(filename, "%s.%.4d.00.%.2d.inverted", filename_prefix, Nconf, n_c*i+icol);
status = read_lime_spinor_timeslice( (double*)(((double**)spinor_field)[i]), timeslice, filename, position, checksum+n_c*i+icol);
} else {
fprintf(stderr, "[] Error, no single precision timeslice-wise reading yet\n");
exit(72);
}
if (status != 0) {
fprintf(stderr, "[] Error, could not read propagator from file %s\n", filename);
exit(73);
}
if(nfc>1) {
if(prop_single_file) {
if(prec==64) {
status = read_lime_spinor_timeslice( (double*)(((double**)spinor_field)[i+n_s]), timeslice, filename, 1-position, checksum+n_c*i+icol+n_s*n_c);
} else {
fprintf(stderr, "[] Error, no single precision timeslice-wise reading yet\n");
exit(72);
}
} else {
if(prec==64) {
sprintf(filename, "%s.%.4d.00.%.2d.inverted", filename_prefix2, Nconf, n_c*i+icol);
status = read_lime_spinor_timeslice((double*)(((double**)spinor_field)[i+n_s]), timeslice, filename, position, checksum+n_c*i+icol+n_s*n_c);
} else {
fprintf(stderr, "[] Error, no single precision timeslice-wise reading yet\n");
exit(72);
}
}
if (status != 0) {
fprintf(stderr, "[] Error, could not read propagator from file %s\n", filename);
exit(74);
}
} // of if nfc > 1
} // of is
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id==0) fprintf(stdout, "# time for preparing light prop.: %e seconds\n", retime-ratime);
ratime = (double)clock() / CLOCKS_PER_SEC;
count = -1;
for(sigmalight=1; sigmalight>=-1; sigmalight-=2) {
for(sigmaheavy=1; sigmaheavy>=-1; sigmaheavy-=2) {
count++;
if(count>=nfc) continue;
if(prec==64) {
chi = (void*) &( ((double**)spinor_field)[ ( (1-sigmalight)/2 )*n_s ] );
psi = (void*) &( ((double**)spinor_field)[ ( (1-sigmaheavy)/2 )*n_s ] );
} else {
chi = (void*) &( ((float**)spinor_field)[ ( (1-sigmalight)/2 )*n_s ] );
psi = (void*) &( ((float**)spinor_field)[ ( (1-sigmaheavy)/2 )*n_s ] );
}
if(sigmalight == sigmaheavy) {
xgindex1 = gindex1; xgindex2 = gindex2; xisimag=isimag; xvsign=vsign; conf_gamma_sign = c_conf_gamma_sign;
} else {
xgindex1 = ngindex1; xgindex2 = ngindex2; xisimag=nisimag; xvsign=nvsign; conf_gamma_sign = n_conf_gamma_sign;
}
// (pseudo-)scalar sector
for(idx=0; idx<16; idx++) {
//fprintf(stdout, "# sigma(%d, %d): (idx,i) = (%d,%d) ---> (%d,%d)\n",
// sigmalight, sigmaheavy, idx, i, xgindex1[idx], xgindex2[idx]);
if(prec==64) {
vptr = (void*)( ((double*)cconn) + 2*(count*K + idx) );
} else {
vptr = (void*)( ((float*)cconn) + 2*(count*K + idx) );
}
contract_twopoint_xdep_timeslice(vptr, xgindex1[idx], xgindex2[idx], chi, psi, 1, nfc*K, 1.0, prec);
}
// (pseudo-)vector sector
for(idx = 16; idx < 64; idx+=3) {
for(i = 0; i < 3; i++) {
//if(xgindex1[idx+i]==xgindex2[idx+i] && (xgindex2[idx+i]==1 || xgindex2[idx+i]==2 || xgindex2[idx+i]==3) ) {
// fprintf(stdout, "# sigma(%d, %d): (idx,i) = (%d,%d) ---> (%d,%d); factor = %e\n",
// sigmalight, sigmaheavy, idx, i, xgindex1[idx+i], xgindex2[idx+i], conf_gamma_sign[(idx-16)/3]*xvsign[idx-16+i]);
//}
if(prec==64) {
vptr = (void*)( ((double*)cconn) + 2*(count*K + (16+(idx-16)/3)) );
} else {
vptr = (void*)( ((float*)cconn) + 2*(count*K + (16+(idx-16)/3)) );
}
contract_twopoint_xdep_timeslice(vptr, xgindex1[idx+i], xgindex2[idx+i], chi, psi, 1, nfc*K,
conf_gamma_sign[(idx-16)/3]*xvsign[idx-16+i], prec);
}
}
}}
} // of loop on colors
/***************************************************************
* write contractions to file
***************************************************************/
ratime = (double)clock() / CLOCKS_PER_SEC;
sprintf(filename, "correl.%.4d.t%.2dx%.2dy%.2dz%.2d", Nconf, source_coords[0],
source_coords[1], source_coords[2], source_coords[3]);
if(timeslice == 0) {
// fprintf(stdout, "# [] opening file %s for writing\n", filename);
ofs = fopen(filename, "w");
} else {
// fprintf(stdout, "# [] opening file %s for appending\n", filename);
ofs = fopen(filename, "a");
}
if(ofs==NULL) {
fprintf(stderr, "Error, could not open file %s for writing\n", filename);
exit(7);
}
if(write_ascii) {
sprintf(filename, "correl.%.4d.t%.2dx%.2dy%.2dz%.2d.ascii", Nconf, source_coords[0],
source_coords[1], source_coords[2], source_coords[3]);
if(timeslice == 0) {
ofs2 = fopen(filename, "w");
} else {
ofs2 = fopen(filename, "a");
}
}
for(x1=0; x1<LX_global; x1++) {
for(x2=0; x2<LY_global; x2++) {
shift = ( (x1 % LX) * LY + (x2 % LY) ) * LZ;
if(prec==64) {
vptr = (void*)( ((double*)cconn)+shift*2*nfc*K);
bytes = sizeof(double);
} else {
vptr = (void*)( ((float*)cconn)+shift*2*nfc*K);
bytes = sizeof(float);
}
if( fwrite(vptr, bytes, 2*nfc*K*LZ, ofs) != 2*nfc*K*LZ ) {
fprintf(stderr, "Error, could not write proper amount of data\n");
exit(8);
}
if(write_ascii) {
for(x3=0; x3<LZ; x3++) {
count = -1;
for(j=0; j<nfc; j++) {
for(i=0; i<K; i++) {
count++;
if(prec==64) {
fprintf(ofs2, "%3d%3d%3d%3d%3d%3d%6lu%25.16e%25.16e\n",
j, i, timeslice, x1, x2, x3, shift,
((double*)cconn)[(shift+x3)*2*nfc*K+2*count], ((double*)cconn)[(shift+x3)*2*nfc*K+2*count+1]);
} else {
fprintf(ofs2, "%3d%3d%3d%3d%3d%3d%6lu%16.7e%16.7e\n",
j, i, timeslice, x1, x2, x3, shift,
((float*)cconn)[(shift+x3)*2*nfc*K+2*count], ((float*)cconn)[(shift+x3)*2*nfc*K+2*count+1]);
}
}}
}
} // of if write_ascii
}}
if(g_cart_id==0) {
if(ofs != NULL) fclose(ofs);
if(ofs2 != NULL) fclose(ofs2);
}
retime = (double)clock() / CLOCKS_PER_SEC;
if(g_cart_id==0) fprintf(stdout, "# time to write LL contractions: %e seconds\n", retime-ratime);
} // of loop on timeslices
if(g_cart_id==0) fprintf(stdout, "# finished LL contractions\n");
/**************************************************
* free the allocated memory, finalize
**************************************************/
if(no_fields>0) {
if(prec==64) {
for(i=0; i<no_fields; i++) free( ((double**)spinor_field)[i]);
} else {
for(i=0; i<no_fields; i++) free( ((float**)spinor_field)[i]);
}
free(spinor_field);
}
free_geometry();
free(cconn);
free(buffer);
#ifdef MPI
MPI_Finalize();
#endif
fprintf(stdout, "\n# [ll_conn] %s# [ll_conn] end of run\n", ctime(&g_the_time));
fflush(stdout);
fprintf(stderr, "\n# [ll_conn] %s# [ll_conn] end of run\n", ctime(&g_the_time));
fflush(stderr);
return(0);
}
int main(int argc, char **argv) {
const int n_c=3;
const int n_s=4;
const char outfile_prefix[] = "delta_pp_2pt_v4";
int c, i, icomp;
int filename_set = 0;
int append, status;
int l_LX_at, l_LXstart_at;
int ix, it, iix, x1,x2,x3;
int ir, ir2, is;
int VOL3;
int do_gt=0;
int dims[3];
double *connt=NULL;
spinor_propagator_type *connq=NULL;
int verbose = 0;
int sx0, sx1, sx2, sx3;
int write_ascii=0;
int fermion_type = 1; // Wilson fermion type
int pos;
char filename[200], contype[200], gauge_field_filename[200];
double ratime, retime;
//double plaq_m, plaq_r;
double *work=NULL;
fermion_propagator_type *fp1=NULL, *fp2=NULL, *fp3=NULL, *uprop=NULL, *dprop=NULL, *fpaux=NULL;
spinor_propagator_type *sp1=NULL, *sp2=NULL;
double q[3], phase, *gauge_trafo=NULL;
complex w, w1;
size_t items, bytes;
FILE *ofs;
int timeslice;
DML_Checksum ildg_gauge_field_checksum, *spinor_field_checksum=NULL, connq_checksum;
uint32_t nersc_gauge_field_checksum;
int threadid, nthreads;
/*******************************************************************
* Gamma components for the Delta:
* */
const int num_component = 4;
int gamma_component[2][4] = { {0, 1, 2, 3},
{0, 1, 2, 3} };
double gamma_component_sign[4] = {+1.,+1.,-1.,+1.};
/*
*******************************************************************/
fftw_complex *in=NULL;
#ifdef MPI
fftwnd_mpi_plan plan_p;
#else
fftwnd_plan plan_p;
#endif
#ifdef MPI
MPI_Status status;
#endif
#ifdef MPI
MPI_Init(&argc, &argv);
#endif
while ((c = getopt(argc, argv, "ah?vgf:F:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'a':
write_ascii = 1;
fprintf(stdout, "# [] will write in ascii format\n");
break;
case 'F':
if(strcmp(optarg, "Wilson") == 0) {
fermion_type = _WILSON_FERMION;
} else if(strcmp(optarg, "tm") == 0) {
fermion_type = _TM_FERMION;
} else {
fprintf(stderr, "[] Error, unrecognized fermion type\n");
exit(145);
}
fprintf(stdout, "# [] will use fermion type %s ---> no. %d\n", optarg, fermion_type);
break;
case 'g':
do_gt = 1;
fprintf(stdout, "# [] will perform gauge transform\n");
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/* set the default values */
if(filename_set==0) strcpy(filename, "cvc.input");
fprintf(stdout, "# reading input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
#ifdef OPENMP
omp_set_num_threads(g_num_threads);
#else
fprintf(stdout, "[delta_pp_2pt_v4] Warning, resetting global thread number to 1\n");
g_num_threads = 1;
#endif
/* initialize MPI parameters */
mpi_init(argc, argv);
#ifdef OPENMP
status = fftw_threads_init();
if(status != 0) {
fprintf(stderr, "\n[] Error from fftw_init_threads; status was %d\n", status);
exit(120);
}
#endif
/******************************************************
*
******************************************************/
VOL3 = LX*LY*LZ;
l_LX_at = LX;
l_LXstart_at = 0;
FFTW_LOC_VOLUME = T*LX*LY*LZ;
fprintf(stdout, "# [%2d] parameters:\n"\
"# [%2d] l_LX_at = %3d\n"\
"# [%2d] l_LXstart_at = %3d\n"\
"# [%2d] FFTW_LOC_VOLUME = %3d\n",
g_cart_id, g_cart_id, l_LX_at,
g_cart_id, l_LXstart_at, g_cart_id, FFTW_LOC_VOLUME);
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
if(N_Jacobi>0) {
// alloc the gauge field
alloc_gauge_field(&g_gauge_field, VOL3);
switch(g_gauge_file_format) {
case 0:
sprintf(gauge_field_filename, "%s.%.4d", gaugefilename_prefix, Nconf);
break;
case 1:
sprintf(gauge_field_filename, "%s.%.5d", gaugefilename_prefix, Nconf);
break;
}
} else {
g_gauge_field = NULL;
}
/*********************************************************************
* gauge transformation
*********************************************************************/
if(do_gt) { init_gauge_trafo(&gauge_trafo, 1.); }
// determine the source location
sx0 = g_source_location/(LX*LY*LZ)-Tstart;
sx1 = (g_source_location%(LX*LY*LZ)) / (LY*LZ);
sx2 = (g_source_location%(LY*LZ)) / LZ;
sx3 = (g_source_location%LZ);
// g_source_time_slice = sx0;
fprintf(stdout, "# [] source location %d = (%d,%d,%d,%d)\n", g_source_location, sx0, sx1, sx2, sx3);
// allocate memory for the spinor fields
g_spinor_field = NULL;
no_fields = n_s*n_c;
// if(fermion_type == _TM_FERMION) {
// no_fields *= 2;
// }
if(N_Jacobi>0) no_fields++;
g_spinor_field = (double**)calloc(no_fields, sizeof(double*));
for(i=0; i<no_fields-1; i++) alloc_spinor_field(&g_spinor_field[i], VOL3);
alloc_spinor_field(&g_spinor_field[no_fields-1], VOL3);
work = g_spinor_field[no_fields-1];
spinor_field_checksum = (DML_Checksum*)malloc(no_fields * sizeof(DML_Checksum) );
if(spinor_field_checksum == NULL ) {
fprintf(stderr, "[] Error, could not alloc checksums for spinor fields\n");
exit(73);
}
// allocate memory for the contractions
items = 4* num_component*T;
bytes = sizeof(double);
connt = (double*)malloc(items*bytes);
if(connt == NULL) {
fprintf(stderr, "\n[] Error, could not alloc connt\n");
exit(2);
}
for(ix=0; ix<items; ix++) connt[ix] = 0.;
items = num_component * (size_t)VOL3;
connq = create_sp_field( items );
if(connq == NULL) {
fprintf(stderr, "\n[] Error, could not alloc connq\n");
exit(2);
}
/******************************************************
* initialize FFTW
******************************************************/
items = 2 * num_component * g_sv_dim * g_sv_dim * VOL3;
bytes = sizeof(double);
in = (fftw_complex*)malloc(num_component*g_sv_dim*g_sv_dim*VOL3*sizeof(fftw_complex));
if(in == NULL) {
fprintf(stderr, "[] Error, could not malloc in for FFTW\n");
exit(155);
}
dims[0]=LX; dims[1]=LY; dims[2]=LZ;
//plan_p = fftwnd_create_plan(3, dims, FFTW_FORWARD, FFTW_MEASURE | FFTW_IN_PLACE);
plan_p = fftwnd_create_plan_specific(3, dims, FFTW_FORWARD, FFTW_MEASURE, in, num_component*g_sv_dim*g_sv_dim, (fftw_complex*)( connq[0][0] ), num_component*g_sv_dim*g_sv_dim);
uprop = (fermion_propagator_type*)malloc(g_num_threads * sizeof(fermion_propagator_type) );
fp1 = (fermion_propagator_type*)malloc(g_num_threads * sizeof(fermion_propagator_type) );
fp2 = (fermion_propagator_type*)malloc(g_num_threads * sizeof(fermion_propagator_type) );
fp3 = (fermion_propagator_type*)malloc(g_num_threads * sizeof(fermion_propagator_type) );
fpaux = (fermion_propagator_type*)malloc(g_num_threads * sizeof(fermion_propagator_type) );
if(uprop==NULL || fp1==NULL || fp2==NULL || fp3==NULL || fpaux==NULL ) {
fprintf(stderr, "[] Error, could not alloc fermion propagator points\n");
exit(57);
}
sp1 = (spinor_propagator_type*)malloc(g_num_threads * sizeof(spinor_propagator_type) );
sp2 = (spinor_propagator_type*)malloc(g_num_threads * sizeof(spinor_propagator_type) );
if(sp1==NULL || sp2==NULL) {
fprintf(stderr, "[] Error, could not alloc spinor propagator points\n");
exit(59);
}
for(i=0;i<g_num_threads;i++) { create_fp(uprop+i); }
for(i=0;i<g_num_threads;i++) { create_fp(fp1+i); }
for(i=0;i<g_num_threads;i++) { create_fp(fp2+i); }
for(i=0;i<g_num_threads;i++) { create_fp(fp3+i); }
for(i=0;i<g_num_threads;i++) { create_fp(fpaux+i); }
for(i=0;i<g_num_threads;i++) { create_sp(sp1+i); }
for(i=0;i<g_num_threads;i++) { create_sp(sp2+i); }
/******************************************************
* loop on timeslices
******************************************************/
for(timeslice=0; timeslice<T; timeslice++) {
append = (int)( timeslice != 0 );
// read timeslice of the gauge field
if( N_Jacobi>0) {
switch(g_gauge_file_format) {
case 0:
status = read_lime_gauge_field_doubleprec_timeslice(g_gauge_field, gauge_field_filename, timeslice, &ildg_gauge_field_checksum);
break;
case 1:
status = read_nersc_gauge_field_timeslice(g_gauge_field, gauge_field_filename, timeslice, &nersc_gauge_field_checksum);
break;
}
if(status != 0) {
fprintf(stderr, "[] Error, could not read gauge field\n");
exit(21);
}
#ifdef OPENMP
status = APE_Smearing_Step_Timeslice_threads(g_gauge_field, N_ape, alpha_ape);
#else
for(i=0; i<N_ape; i++) { status = APE_Smearing_Step_Timeslice(g_gauge_field, alpha_ape); }
#endif
}
// read timeslice of the 12 up-type propagators and smear them
for(is=0;is<n_s*n_c;is++) {
if(do_gt == 0) {
sprintf(filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.%.2d.inverted", filename_prefix, Nconf, sx0, sx1, sx2, sx3, is);
status = read_lime_spinor_timeslice(g_spinor_field[is], timeslice, filename, 0, spinor_field_checksum+is);
if(status != 0) {
fprintf(stderr, "[] Error, could not read propagator from file %s\n", filename);
exit(102);
}
if(N_Jacobi > 0) {
fprintf(stdout, "# [] Jacobi smearing propagator no. %d with paramters N_Jacobi=%d, kappa_Jacobi=%f\n",
is, N_Jacobi, kappa_Jacobi);
#ifdef OPENMP
Jacobi_Smearing_Step_one_Timeslice_threads(g_gauge_field, g_spinor_field[is], work, N_Jacobi, kappa_Jacobi);
#else
for(c=0; c<N_Jacobi; c++) {
Jacobi_Smearing_Step_one_Timeslice(g_gauge_field, g_spinor_field[is], work, kappa_Jacobi);
}
#endif
}
} else { // of if do_gt == 0
// apply gt
apply_gt_prop(gauge_trafo, g_spinor_field[is], is/n_c, is%n_c, 4, filename_prefix, g_source_location);
} // of if do_gt == 0
}
/******************************************************
* contractions
******************************************************/
#ifdef OPENMP
omp_set_num_threads(g_num_threads);
#pragma omp parallel private (ix,icomp,threadid) \
firstprivate (fermion_type,gamma_component,num_component,connq,\
gamma_component_sign,VOL3,g_spinor_field,fp1,fp2,fp3,fpaux,uprop,sp1,sp2)
{
threadid = omp_get_thread_num();
#else
threadid = 0;
#endif
for(ix=threadid; ix<VOL3; ix+=g_num_threads)
{
// assign the propagators
_assign_fp_point_from_field(uprop[threadid], g_spinor_field, ix);
if(fermion_type == _TM_FERMION) {
_fp_eq_rot_ti_fp(fp1[threadid], uprop[threadid], +1, fermion_type, fp2[threadid]);
_fp_eq_fp_ti_rot(uprop[threadid], fp1[threadid], +1, fermion_type, fp2[threadid]);
}
for(icomp=0; icomp<num_component; icomp++) {
_sp_eq_zero( connq[ix*num_component+icomp]);
/******************************************************
* prepare propagators
******************************************************/
// fp1[threadid] = C Gamma_1 x S_u = g0 g2 Gamma_1 S_u
_fp_eq_zero(fp1[threadid]);
_fp_eq_zero(fpaux[threadid]);
_fp_eq_gamma_ti_fp(fp1[threadid], gamma_component[0][icomp], uprop[threadid]);
_fp_eq_gamma_ti_fp(fpaux[threadid], 2, fp1[threadid]);
_fp_eq_gamma_ti_fp(fp1[threadid], 0, fpaux[threadid]);
// fp2[threadid] = C Gamma_1 x S_u x C Gamma_2
_fp_eq_zero(fp2[threadid]);
_fp_eq_zero(fpaux[threadid]);
_fp_eq_fp_ti_gamma(fp2[threadid], 0, fp1[threadid]);
_fp_eq_fp_ti_gamma(fpaux[threadid], 2, fp2[threadid]);
_fp_eq_fp_ti_gamma(fp2[threadid], gamma_component[1][icomp], fpaux[threadid]);
// fp3[threadid] = S_u x C Gamma_2 = S_u g0 g2 Gamma_2
_fp_eq_zero(fp3[threadid]);
_fp_eq_zero(fpaux[threadid]);
_fp_eq_fp_ti_gamma(fp3[threadid], 0, uprop[threadid]);
_fp_eq_fp_ti_gamma(fpaux[threadid], 2, fp3[threadid]);
_fp_eq_fp_ti_gamma(fp3[threadid], gamma_component[1][icomp], fpaux[threadid]);
/******************************************************
* contractions
******************************************************/
// (1)
// reduce
_fp_eq_zero(fpaux[threadid]);
_fp_eq_fp_eps_contract13_fp(fpaux[threadid], fp1[threadid], uprop[threadid]);
// reduce to spin propagator
_sp_eq_zero( sp1[threadid] );
_sp_eq_fp_del_contract23_fp(sp1[threadid], fp3[threadid], fpaux[threadid]);
// (2)
// reduce to spin propagator
_sp_eq_zero( sp2[threadid] );
_sp_eq_fp_del_contract24_fp(sp2[threadid], fp3[threadid], fpaux[threadid]);
// add and assign
_sp_pl_eq_sp(sp1[threadid], sp2[threadid]);
_sp_eq_sp_ti_re(sp2[threadid], sp1[threadid], -gamma_component_sign[icomp]);
_sp_eq_sp( connq[ix*num_component+icomp], sp2[threadid]);
// (3)
// reduce
_fp_eq_zero(fpaux[threadid]);
_fp_eq_fp_eps_contract13_fp(fpaux[threadid], fp2[threadid], uprop[threadid]);
// reduce to spin propagator
_sp_eq_zero( sp1[threadid] );
_sp_eq_fp_del_contract23_fp(sp1[threadid], uprop[threadid], fpaux[threadid]);
// (4)
// reduce
_fp_eq_zero(fpaux[threadid]);
_fp_eq_fp_eps_contract13_fp(fpaux[threadid], fp1[threadid], fp3[threadid]);
// reduce to spin propagator
_sp_eq_zero( sp2[threadid] );
_sp_eq_fp_del_contract24_fp(sp2[threadid], uprop[threadid], fpaux[threadid]);
// add and assign
_sp_pl_eq_sp(sp1[threadid], sp2[threadid]);
_sp_eq_sp_ti_re(sp2[threadid], sp1[threadid], -gamma_component_sign[icomp]);
_sp_pl_eq_sp( connq[ix*num_component+icomp], sp2[threadid]);
// (5)
// reduce
_fp_eq_zero(fpaux[threadid]);
_fp_eq_fp_eps_contract13_fp(fpaux[threadid], fp2[threadid], uprop[threadid]);
// reduce to spin propagator
_sp_eq_zero( sp1[threadid] );
_sp_eq_fp_del_contract34_fp(sp1[threadid], uprop[threadid], fpaux[threadid]);
// (6)
// reduce
_fp_eq_zero(fpaux[threadid]);
_fp_eq_fp_eps_contract13_fp(fpaux[threadid], fp1[threadid], fp3[threadid]);
// reduce to spin propagator
_sp_eq_zero( sp2[threadid] );
_sp_eq_fp_del_contract34_fp(sp2[threadid], uprop[threadid], fpaux[threadid]);
// add and assign
_sp_pl_eq_sp(sp1[threadid], sp2[threadid]);
_sp_eq_sp_ti_re(sp2[threadid], sp1[threadid], -gamma_component_sign[icomp]);
_sp_pl_eq_sp( connq[ix*num_component+icomp], sp2[threadid]);
} // of icomp
} // of ix
#ifdef OPENMP
}
#endif
/***********************************************
* finish calculation of connq
***********************************************/
if(g_propagator_bc_type == 0) {
// multiply with phase factor
fprintf(stdout, "# [] multiplying timeslice %d with boundary phase factor\n", timeslice);
ir = (timeslice - sx0 + T_global) % T_global;
w1.re = cos( 3. * M_PI*(double)ir / (double)T_global );
w1.im = sin( 3. * M_PI*(double)ir / (double)T_global );
for(ix=0;ix<num_component*VOL3;ix++) {
_sp_eq_sp(sp1[0], connq[ix] );
_sp_eq_sp_ti_co( connq[ix], sp1[0], w1);
}
} else if (g_propagator_bc_type == 1) {
// multiply with step function
if(timeslice < sx0) {
fprintf(stdout, "# [] multiplying timeslice %d with boundary step function\n", timeslice);
for(ix=0;ix<num_component*VOL3;ix++) {
_sp_eq_sp(sp1[0], connq[ix] );
_sp_eq_sp_ti_re( connq[ix], sp1[0], -1.);
}
}
}
if(write_ascii) {
sprintf(filename, "%s_x.%.4d.t%.2dx%.2dy%.2dz%.2d.ascii", outfile_prefix, Nconf, sx0, sx1, sx2, sx3);
write_contraction2( connq[0][0], filename, num_component*g_sv_dim*g_sv_dim, VOL3, 1, append);
}
/******************************************************************
* Fourier transform
******************************************************************/
items = 2 * num_component * g_sv_dim * g_sv_dim * VOL3;
bytes = sizeof(double);
memcpy(in, connq[0][0], items * bytes);
ir = num_component * g_sv_dim * g_sv_dim;
#ifdef OPENMP
fftwnd_threads(g_num_threads, plan_p, ir, in, ir, 1, (fftw_complex*)(connq[0][0]), ir, 1);
#else
fftwnd(plan_p, ir, in, ir, 1, (fftw_complex*)(connq[0][0]), ir, 1);
#endif
// add phase factor from the source location
iix = 0;
for(x1=0;x1<LX;x1++) {
q[0] = (double)x1 / (double)LX;
for(x2=0;x2<LY;x2++) {
q[1] = (double)x2 / (double)LY;
for(x3=0;x3<LZ;x3++) {
q[2] = (double)x3 / (double)LZ;
phase = 2. * M_PI * ( q[0]*sx1 + q[1]*sx2 + q[2]*sx3 );
w1.re = cos(phase);
w1.im = sin(phase);
for(icomp=0; icomp<num_component; icomp++) {
_sp_eq_sp(sp1[0], connq[iix] );
_sp_eq_sp_ti_co( connq[iix], sp1[0], w1) ;
iix++;
}
}}} // of x3, x2, x1
// write to file
sprintf(filename, "%s_q.%.4d.t%.2dx%.2dy%.2dz%.2d", outfile_prefix, Nconf, sx0, sx1, sx2, sx3);
sprintf(contype, "2-pt. function, (t,q_1,q_2,q_3)-dependent, source_timeslice = %d", sx0);
write_lime_contraction_timeslice(connq[0][0], filename, 64, num_component*g_sv_dim*g_sv_dim, contype, Nconf, 0, &connq_checksum, timeslice);
if(write_ascii) {
strcat(filename, ".ascii");
write_contraction2(connq[0][0],filename, num_component*g_sv_dim*g_sv_dim, VOL3, 1, append);
}
/***********************************************
* calculate connt
***********************************************/
for(icomp=0;icomp<num_component; icomp++) {
// fwd
_sp_eq_sp(sp1[0], connq[icomp]);
_sp_eq_gamma_ti_sp(sp2[0], 0, sp1[0]);
_sp_pl_eq_sp(sp1[0], sp2[0]);
_co_eq_tr_sp(&w, sp1[0]);
connt[2*(icomp*T + timeslice) ] = w.re * 0.25;
connt[2*(icomp*T + timeslice)+1] = w.im * 0.25;
// bwd
_sp_eq_sp(sp1[0], connq[icomp]);
_sp_eq_gamma_ti_sp(sp2[0], 0, sp1[0]);
_sp_mi_eq_sp(sp1[0], sp2[0]);
_co_eq_tr_sp(&w, sp1[0]);
connt[2*(icomp*T+timeslice + num_component*T) ] = w.re * 0.25;
connt[2*(icomp*T+timeslice + num_component*T)+1] = w.im * 0.25;
}
} // of loop on timeslice
// write connt
sprintf(filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.fw", outfile_prefix, Nconf, sx0, sx1, sx2, sx3);
ofs = fopen(filename, "w");
if(ofs == NULL) {
fprintf(stderr, "[] Error, could not open file %s for writing\n", filename);
exit(3);
}
fprintf(ofs, "#%12.8f%3d%3d%3d%3d%8.4f%6d\n", g_kappa, T_global, LX, LY, LZ, g_mu, Nconf);
for(icomp=0; icomp<num_component; icomp++) {
ir = sx0;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], 0, connt[2*(icomp*T+ir)], 0., Nconf);
for(it=1;it<T/2;it++) {
ir = ( it + sx0 ) % T_global;
ir2 = ( (T_global - it) + sx0 ) % T_global;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], it, connt[2*(icomp*T+ir)], connt[2*(icomp*T+ir2)], Nconf);
}
ir = ( it + sx0 ) % T_global;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], it, connt[2*(icomp*T+ir)], 0., Nconf);
}
fclose(ofs);
sprintf(filename, "%s.%.4d.t%.2dx%.2dy%.2dz%.2d.bw", outfile_prefix, Nconf, sx0, sx1, sx2, sx3);
ofs = fopen(filename, "w");
if(ofs == NULL) {
fprintf(stderr, "[] Error, could not open file %s for writing\n", filename);
exit(3);
}
fprintf(ofs, "#%12.8f%3d%3d%3d%3d%8.4f%6d\n", g_kappa, T_global, LX, LY, LZ, g_mu, Nconf);
for(icomp=0; icomp<num_component; icomp++) {
ir = sx0;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], 0, connt[2*(num_component*T+icomp*T+ir)], 0., Nconf);
for(it=1;it<T/2;it++) {
ir = ( it + sx0 ) % T_global;
ir2 = ( (T_global - it) + sx0 ) % T_global;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], it, connt[2*(num_component*T+icomp*T+ir)], connt[2*(num_component*T+icomp*T+ir2)], Nconf);
}
ir = ( it + sx0 ) % T_global;
fprintf(ofs, "%3d%3d%3d%16.7e%16.7e%6d\n", gamma_component[0][icomp], gamma_component[1][icomp], it, connt[2*(num_component*T+icomp*T+ir)], 0., Nconf);
}
fclose(ofs);
/***********************************************
* free the allocated memory, finalize
***********************************************/
free_geometry();
if(connt!= NULL) free(connt);
if(connq!= NULL) free(connq);
if(gauge_trafo != NULL) free(gauge_trafo);
if(g_spinor_field!=NULL) {
for(i=0; i<no_fields; i++) free(g_spinor_field[i]);
free(g_spinor_field); g_spinor_field=(double**)NULL;
}
if(spinor_field_checksum !=NULL) free(spinor_field_checksum);
if(g_gauge_field != NULL) free(g_gauge_field);
for(i=0;i<g_num_threads;i++) { free_fp(uprop+i); }
for(i=0;i<g_num_threads;i++) { free_fp(fp1+i); }
for(i=0;i<g_num_threads;i++) { free_fp(fp2+i); }
for(i=0;i<g_num_threads;i++) { free_fp(fp3+i); }
for(i=0;i<g_num_threads;i++) { free_fp(fpaux+i); }
for(i=0;i<g_num_threads;i++) { free_sp(sp1+i); }
for(i=0;i<g_num_threads;i++) { free_sp(sp2+i); }
if(uprop!=NULL) free(uprop);
if(fp1!=NULL) free(fp1);
if(fp2!=NULL) free(fp2);
if(fp3!=NULL) free(fp3);
if(fpaux!=NULL) free(fpaux);
if(sp1!=NULL) free(sp1);
if(sp2!=NULL) free(sp2);
free(in);
fftwnd_destroy_plan(plan_p);
g_the_time = time(NULL);
fprintf(stdout, "# [] %s# [] end fo run\n", ctime(&g_the_time));
fflush(stdout);
fprintf(stderr, "# [] %s# [] end fo run\n", ctime(&g_the_time));
fflush(stderr);
#ifdef MPI
MPI_Finalize();
#endif
return(0);
}
int main(int argc, char **argv) {
int c, mu;
int filename_set = 0;
int sl0, sl1, sl2, sl3;
double *disc;
double vp1[8], vp2[8], vp3[8], vp4[8], vp5[8];
char filename[200];
while ((c = getopt(argc, argv, "h?vf:")) != -1) {
switch (c) {
case 'v':
verbose = 1;
break;
case 'f':
strcpy(filename, optarg);
filename_set=1;
break;
case 'h':
case '?':
default:
usage();
break;
}
}
/**************************************
* set the default values, read input
**************************************/
if(filename_set==0) strcpy(filename, "cvc.input.test");
fprintf(stdout, "# Reading test input from file %s\n", filename);
read_input_parser(filename);
/* some checks on the input data */
if((T_global == 0) || (LX==0) || (LY==0) || (LZ==0)) {
if(g_proc_id==0) fprintf(stdout, "T and L's must be set\n");
usage();
}
if(g_kappa == 0.) {
if(g_proc_id==0) fprintf(stdout, "kappa should be > 0.n");
usage();
}
/* initialize MPI parameters */
mpi_init(argc, argv);
T = T_global;
if(init_geometry() != 0) {
fprintf(stderr, "ERROR from init_geometry\n");
exit(1);
}
geometry();
/********************************
* the source locaton
********************************/
sl0 = g_source_location/(LX*LY*LZ);
sl1 = ( g_source_location%(LX*LY*LZ) ) / (LY*LZ);
sl2 = ( g_source_location%(LY*LZ) ) / (LZ);
sl3 = g_source_location%LZ;
fprintf(stdout, "# global sl = (%d, %d, %d, %d)\n", sl0, sl1, sl2, sl3);
if( (disc = (double*)malloc(32*VOLUME*sizeof(double))) == (double*)NULL) {
exit(102);
}
/*******************************************************************
* (1) comparison of results from
* - avc_disc_stochastic
* - avc_disc_hpe and avc_disc_hpe5
* - vp_disc_hpe_loops_red/vp_disc_hpe_stoch to 3rd and 5th order
*******************************************************************/
sprintf(filename, "outcvc_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading avc_disc_stochastic data from file %s\n", filename);
read_contraction(disc, NULL, filename, 4);
for(mu=0; mu<4; mu++) {
vp1[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ] / 60.;
vp1[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1] / 60.;
}
sprintf(filename, "cvc_hpe_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading avc_disc_hpe data from file %s\n", filename);
read_contraction(disc, NULL, filename, 4);
for(mu=0; mu<4; mu++) {
vp2[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ] / 60.;
vp2[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1] / 60.;
}
sprintf(filename, "cvc_hpe5_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading avc_disc_hpe5 data from file %s\n", filename);
read_lime_contraction(disc, filename, 4, 0);
for(mu=0; mu<4; mu++) {
vp3[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ];
vp3[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1];
}
sprintf(filename, "vp_disc_hpe03_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading vp_disc_hpe03 data from file %s\n", filename);
read_lime_contraction(disc, filename, 4, 0);
for(mu=0; mu<4; mu++) {
vp4[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ];
vp4[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1];
}
sprintf(filename, "vp_disc_hpe05_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading vp_disc_hpe05 data from file %s\n", filename);
read_lime_contraction(disc, filename, 4, 0);
for(mu=0; mu<4; mu++) {
vp5[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ];
vp5[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1];
}
for(mu=0; mu<4; mu++) {
fprintf(stdout, "\n#--------------------------------------------\n"\
"# mu = %d\n", mu);
fprintf(stdout, "%30s%30s%30s\n", "method", "real part", "imaginary part");
fprintf(stdout, "%30s%30.16e%30.16e\n", "avc_disc_stochastic", vp1[2*mu], vp1[2*mu+1]);
fprintf(stdout, "%30s%30.16e%30.16e\n", "avc_disc_hpe", vp2[2*mu], vp2[2*mu+1]);
fprintf(stdout, "%30s%30.16e%30.16e\n", "avc_disc_hpe5", vp3[2*mu], vp3[2*mu+1]);
fprintf(stdout, "%30s%30.16e%30.16e\n", "vp_disc_hpe03", vp4[2*mu], vp4[2*mu+1]);
fprintf(stdout, "%30s%30.16e%30.16e\n", "vp_disc_hpe05", vp5[2*mu], vp5[2*mu+1]);
}
fprintf(stdout, "\n#=======================================================\n");
/*******************************************************************
* (2) comparison of results from
* - lvc_disc_stochastic
* - lvc_disc_hpe for 4th and 6th order
*******************************************************************/
sprintf(filename, "outlvc_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading lvc_disc_stochastic data from file %s\n", filename);
read_contraction(disc, NULL, filename, 4);
for(mu=0; mu<4; mu++) {
vp1[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ] / 60.;
vp1[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1] / 60.;
}
sprintf(filename, "lvc_disc_hpe04_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading lvc_disc_hpe04 data from file %s\n", filename);
read_lime_contraction(disc, filename, 4, 0);
for(mu=0; mu<4; mu++) {
vp2[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ];
vp2[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1];
}
sprintf(filename, "lvc_disc_hpe06_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading lvc_disc_hpe06 data from file %s\n", filename);
read_lime_contraction(disc, filename, 4, 0);
for(mu=0; mu<4; mu++) {
vp3[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ];
vp3[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1];
}
for(mu=0; mu<4; mu++) {
fprintf(stdout, "\n#--------------------------------------------\n"\
"# mu = %d\n", mu);
fprintf(stdout, "%30s%30s%30s\n", "method", "real part", "imaginary part");
fprintf(stdout, "%30s%30.16e%30.16e\n", "lvc_disc_stochastic", vp1[2*mu], vp1[2*mu+1]);
fprintf(stdout, "%30s%30.16e%30.16e\n", "lvc_disc_hpe04", vp2[2*mu], vp2[2*mu+1]);
fprintf(stdout, "%30s%30.16e%30.16e\n", "lvc_disc_hpe06", vp3[2*mu], vp3[2*mu+1]);
}
fprintf(stdout, "\n#=======================================================\n");
/*******************************************************************
* (3) comparison of results from
* - vp_disc_hpe_mc1/2 for 3rd and 5th order
*******************************************************************/
sprintf(filename, "vp_disc_hpe-01_mc2_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading vp_disc_hpe-01_mc2 data from file %s\n", filename);
read_lime_contraction(disc, filename, 4, 0);
for(mu=0; mu<4; mu++) {
vp1[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ];
vp1[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1];
}
sprintf(filename, "vp_disc_hpe03_mc2_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading vp_disc_hpe03_mc2 data from file %s\n", filename);
read_lime_contraction(disc, filename, 4, 0);
for(mu=0; mu<4; mu++) {
vp2[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ];
vp2[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1];
}
sprintf(filename, "vp_disc_hpe05_mc2_X.%.4d.%.4d", Nconf, Nsave);
fprintf(stdout, "\n# Reading vp_disc_hpe05_mc2 data from file %s\n", filename);
read_lime_contraction(disc, filename, 4, 0);
for(mu=0; mu<4; mu++) {
vp3[2*mu ] = disc[_GWI(mu,g_source_location,VOLUME) ];
vp3[2*mu+1] = disc[_GWI(mu,g_source_location,VOLUME)+1];
}
for(mu=0; mu<4; mu++) {
fprintf(stdout, "\n#--------------------------------------------\n"\
"# mu = %d\n", mu);
fprintf(stdout, "%30s%30s%30s\n", "method", "real part", "imaginary part");
fprintf(stdout, "%30s%30.16e%30.16e\n", "vp_disc_hpe00_mc", vp1[2*mu], vp1[2*mu+1]);
fprintf(stdout, "%30s%30.16e%30.16e\n", "vp_disc_hpe03_mc", vp2[2*mu], vp2[2*mu+1]);
fprintf(stdout, "%30s%30.16e%30.16e\n", "vp_disc_hpe05_mc", vp3[2*mu], vp3[2*mu+1]);
}
fprintf(stdout, "\n#=======================================================\n");
/***********************************************
* free the allocated memory, finalize
***********************************************/
free_geometry();
free(disc);
#ifdef MPI
MPI_Finalize();
#endif
return(0);
}