spinor random_spinor(void)
{
spinor s;
s.s0=random_su3_vector();
s.s1=random_su3_vector();
s.s2=random_su3_vector();
s.s3=random_su3_vector();
_vector_mul(s.s0,0.5,s.s0);
_vector_mul(s.s1,0.5,s.s1);
_vector_mul(s.s2,0.5,s.s2);
_vector_mul(s.s3,0.5,s.s3);
return(s);
}
su3 random_su3(void)
{
double norm,fact;
complex z;
su3_vector z1,z2,z3;
su3 u;
/*
z1=random_su3_vector();
*/
z1=unif_su3_vector();
for (;;)
{
/*
z2=random_su3_vector();
*/
z2=unif_su3_vector();
z.re=_vector_prod_re(z1,z2);
z.im=_vector_prod_im(z1,z2);
_vector_project(z2,z,z1);
norm=_vector_prod_re(z2,z2);
norm=sqrt(norm);
if (1.0!=(1.0+norm))
break;
}
fact=1.0/norm;
_vector_mul(z2,fact,z2);
z3.c0.re= (z1.c1.re*z2.c2.re-z1.c1.im*z2.c2.im)
-(z1.c2.re*z2.c1.re-z1.c2.im*z2.c1.im);
z3.c0.im=-(z1.c1.re*z2.c2.im+z1.c1.im*z2.c2.re)
+(z1.c2.re*z2.c1.im+z1.c2.im*z2.c1.re);
z3.c1.re= (z1.c2.re*z2.c0.re-z1.c2.im*z2.c0.im)
-(z1.c0.re*z2.c2.re-z1.c0.im*z2.c2.im);
z3.c1.im=-(z1.c2.re*z2.c0.im+z1.c2.im*z2.c0.re)
+(z1.c0.re*z2.c2.im+z1.c0.im*z2.c2.re);
z3.c2.re= (z1.c0.re*z2.c1.re-z1.c0.im*z2.c1.im)
-(z1.c1.re*z2.c0.re-z1.c1.im*z2.c0.im);
z3.c2.im=-(z1.c0.re*z2.c1.im+z1.c0.im*z2.c1.re)
+(z1.c1.re*z2.c0.im+z1.c1.im*z2.c0.re);
u.c00=z1.c0;
u.c01=z1.c1;
u.c02=z1.c2;
u.c10=z2.c0;
u.c11=z2.c1;
u.c12=z2.c2;
u.c20=z3.c0;
u.c21=z3.c1;
u.c22=z3.c2;
return(u);
}
extern void jacobi_source(spinor_dble *sd, srcdef src, int idirac, int icolor) {
int i = 0, ix = 0, iy = 0, n = 0, mu = 0, my_rank, ip1 = 0, ip2 = 0, iw = 0;
su3_dble *up, *um;
spinor_dble *chi, *psi, *phi, **wsd0, **wsd1, **wsd2, *p, *(*p0)[NSPIN];
const spinor_dble sd0 = { { { 0.0 } } };
double zw[18] = { 0 }, d;
double norm = 1. / (1. + 6. * src.kappa);
MPI_Status status1;
MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
p0 = amalloc(3 * sizeof(*p0), 3);
p = amalloc(3 * NSPIN * sizeof(spinor_dble), ALIGN);
for (i = 0; i < 3; i++) {
for (ix = 0; ix < NSPIN; ix++) {
p0[i][ix] = p;
*p = sd0;
p += 1;
}
}
wsd0 = &(p0[0][0]);
wsd1 = &(p0[1][0]);
wsd2 = &(p0[2][0]);
message("sourceposition = (%d, %d, %d, %d)\n", src.pos[0], src.pos[1],
src.pos[2], src.pos[3]);
for (n = 0; n < src.n; n++) {
if (n > 0) {
for (ix = 0; ix < VOLUME; ix++)
*wsd0[ix] = *wsd2[ix];
}
for (ix = 0; ix < VOLUME; ix++) {
if (coords[ix].t == src.pos[0]) {
phi = wsd1[ix];
if (n == 0)
*wsd2[ix] = sd[ix];
psi = wsd2[ix];
for (mu = 1; mu < 4; mu++) {
iy = idn[ix][mu];
if ((iy >= VOLUME) && (iy < (VOLUME + BNDRY))) {
ip1 = npr[2 * mu];
ip2 = npr[2 * mu + 1];
iw = map[iy - VOLUME];
if (n == 0)
*wsd0[iw] = sd[iw];
*phi = *wsd0[iw];
chi = wsd1[iw];
um = pud_sm1[iw][mu];
MPI_Sendrecv((double*) (um), 18, MPI_DOUBLE, ip2, 37,
zw, 18, MPI_DOUBLE, ip1, 37, MPI_COMM_WORLD,
&status1);
um = (su3_dble*) zw;
MPI_Sendrecv((double*) (phi), 24, MPI_DOUBLE, ip2, 37,
(double*) (chi), 24, MPI_DOUBLE, ip1, 37,
MPI_COMM_WORLD, &status1);
} else {
um = pud_sm1[iy][mu];
if (n == 0)
*wsd0[iy] = sd[iy];
chi = wsd0[iy];
}
_su3_inverse_multiply((*phi).c1, (*um), (*chi).c1);
_su3_inverse_multiply((*phi).c2, (*um), (*chi).c2);
_su3_inverse_multiply((*phi).c3, (*um), (*chi).c3);
_su3_inverse_multiply((*phi).c4, (*um), (*chi).c4);
_vector_add_assign((*psi).c1, (*phi).c1);
_vector_add_assign((*psi).c2, (*phi).c2);
_vector_add_assign((*psi).c3, (*phi).c3);
_vector_add_assign((*psi).c4, (*phi).c4);
iy = iup[ix][mu];
if ((iy >= VOLUME) && (iy < (VOLUME + BNDRY))) {
ip1 = npr[2 * mu];
ip2 = npr[2 * mu + 1];
iw = map[iy - VOLUME];
if (n == 0)
*wsd0[iw] = sd[iw];
*phi = *wsd0[iw];
chi = wsd1[iw];
MPI_Sendrecv((double*) (phi), 24, MPI_DOUBLE, ip1, 37,
(double*) (chi), 24, MPI_DOUBLE, ip2, 37,
MPI_COMM_WORLD, &status1);
} else {
if (n == 0)
*wsd0[iy] = sd[iy];
chi = wsd0[iy];
}
up = pud_sm1[ix][mu];
_su3_multiply((*phi).c1, (*up), (*chi).c1);
_su3_multiply((*phi).c2, (*up), (*chi).c2);
_su3_multiply((*phi).c3, (*up), (*chi).c3);
_su3_multiply((*phi).c4, (*up), (*chi).c4);
_vector_add_assign((*psi).c1, (*phi).c1);
_vector_add_assign((*psi).c2, (*phi).c2);
_vector_add_assign((*psi).c3, (*phi).c3);
_vector_add_assign((*psi).c4, (*phi).c4);
}
_vector_mul((*psi).c1, src.kappa, (*psi).c1);
_vector_mul((*psi).c2, src.kappa, (*psi).c2);
_vector_mul((*psi).c3, src.kappa, (*psi).c3);
_vector_mul((*psi).c4, src.kappa, (*psi).c4);
if (n == 0)
*wsd1[ix] = sd[ix];
else
*wsd1[ix] = *wsd0[ix];
phi = wsd1[ix];
_vector_add_assign((*psi).c1, (*phi).c1);
_vector_add_assign((*psi).c2, (*phi).c2);
_vector_add_assign((*psi).c3, (*phi).c3);
_vector_add_assign((*psi).c4, (*phi).c4);
_spinor_mul(*(psi),norm, (*psi));
}
}
if(n == (src.n - 1)) {
for(ix = 0; ix < VOLUME; ix++) {
sd[ix] = *wsd2[ix];
_spinor_mul(sd[ix],1./(sqrt((double)(src.n*VOLUME*NPROC))),sd[ix]);
}
d = norm_square_dble(VOLUME, 1, sd);
message("norm of the jacobi-smeared vector %.2e\n", sqrt(d));
}
}
SourceRadius(src, sd);
/* SourceMom4(src, sd);
*/ afree(p);
afree(p0[0][0]);
afree(p0[0]);
afree(p0);
p0 = NULL;
}
void spinor_fft_transpose_xp_t(spinor *fieldout,spinor* fieldin,int dim0,int dim1,int forward,double mulp){
int LXYZ=LX*LY*LZ;
int xyz,tp,xp,t;
spinor *spin1,*spin2;
if(forward == TRUE){
for(tp=0;tp<dim0;tp++){
for(xp=0;xp<dim1;xp++){
for(t=0;t<T;t++){
spin1=fieldin + LXYZ * ((tp * dim1 + xp ) * T + t );
/** ^ ^ ^ ^
* \ \ / |
* \ \ / /
* \ \ / /
* \ X /
* \ / \ /
* \ / \ /
* \/ X
* /\ / \
* / \ / \
* / \/ \
* / /\ \
* / / \ \
* / / \ \
* / / \ |
* V V V V */
spin2=fieldout + LXYZ * ((tp * T + t ) * dim1 + xp );
for(xyz=0;xyz<LXYZ;xyz++){
memcpy(spin2+xyz,spin1+xyz,sizeof(spinor));
_vector_mul((spin2+xyz)->s0,mulp,(spin2+xyz)->s0);
_vector_mul((spin2+xyz)->s1,mulp,(spin2+xyz)->s1);
_vector_mul((spin2+xyz)->s2,mulp,(spin2+xyz)->s2);
_vector_mul((spin2+xyz)->s3,mulp,(spin2+xyz)->s3);
}
/* optionally multiply with mulp */
}
}
}
} else {
for(tp=0;tp<dim0;tp++){
for(xp=0;xp<dim1;xp++){
for(t=0;t<T;t++){
spin1=fieldin + LXYZ * ((tp * T + t ) * dim1 + xp );
/** ^ ^ ^ ^
* \ \ / |
* \ \ / /
* \ \ / /
* \ X /
* \ / \ /
* \ / \ /
* \/ X
* /\ / \
* / \ / \
* / \/ \
* / /\ \
* / / \ \
* / / \ \
* / / \ |
* V V V V */
spin2=fieldout + LXYZ * ((tp * dim1 + xp ) * T + t );
for(xyz=0;xyz<LXYZ;xyz++){
memcpy(spin2+xyz,spin1+xyz,sizeof(spinor));
_vector_mul((spin2+xyz)->s0,mulp,(spin2+xyz)->s0);
_vector_mul((spin2+xyz)->s1,mulp,(spin2+xyz)->s1);
_vector_mul((spin2+xyz)->s2,mulp,(spin2+xyz)->s2);
_vector_mul((spin2+xyz)->s3,mulp,(spin2+xyz)->s3);
}
/* optionally multiply with mulp */
}
}
}
}
}
