Color *radiosity_prog(int n, Poly **p, Color *e, Color *rho)
{
int src, rcv, iter = 0;
Real ff, mts, *a = NEWTARRAY(n, Real);
Color d, *dm = NEWTARRAY(n, Color);
Color ma, *m = NEWTARRAY(n, Color);
initialize(n, m, dm, a, p, e);
while (iter-- < max_iter) {
src = select_shooter(n, dm, a);
if (converged(src, dm))
break;
for (rcv = 0; rcv < n; rcv++) {
if (rcv == src || (ff = formfactor(src, rcv, n, p, a)) < REL_EPS)
continue;
d = c_scale(ff, c_mult(rho[rcv], dm[src]));
m[rcv] = c_add(m[rcv], d);
dm[rcv] = c_add(dm[rcv], d);
}
dm[src] = c_make(0,0,0);
}
ma = ambient_rad(n, dm, a);
for (rcv = 0; rcv < n; rcv++)
m[rcv] = c_add(m[rcv], ma);
efree(a), efree(dm);
return m;
}
static void make_filter(T *result_b, T *result_a, const complex4c *alpha, const complex4c *beta, int K, T sigma)
{
const double denom = sigma * M_SQRT2PI;
complex4c b[DERICHE_MAX_K], a[DERICHE_MAX_K + 1];
int k, j;
b[0] = alpha[0]; /* Initialize b/a = alpha[0] / (1 + beta[0] z^-1) */
a[0] = make_complex(1, 0);
a[1] = beta[0];
for (k = 1; k < K; ++k)
{ /* Add kth term, b/a += alpha[k] / (1 + beta[k] z^-1) */
b[k] = c_mul(beta[k], b[k-1]);
for (j = k - 1; j > 0; --j)
b[j] = c_add(b[j], c_mul(beta[k], b[j - 1]));
for (j = 0; j <= k; ++j)
b[j] = c_add(b[j], c_mul(alpha[k], a[j]));
a[k + 1] = c_mul(beta[k], a[k]);
for (j = k; j > 0; --j)
a[j] = c_add(a[j], c_mul(beta[k], a[j - 1]));
}
for (k = 0; k < K; ++k)
{
result_b[k] = (T)(b[k].real / denom);
result_a[k + 1] = (T)a[k + 1].real;
}
return;
}
Color ray_shade(int level, Real w, Ray v, RContext *rc, Object *ol)
{
Inode *i = ray_intersect(ol, v);
if (i != NULL) { Light *l; Real wf;
Material *m = i->m;
Vector3 p = ray_point(v, i->t);
Cone recv = cone_make(p, i->n, PIOVER2);
Color c = c_mult(m->c, c_scale(m->ka, ambient(rc)));
rc->p = p;
for (l = rc->l; l != NULL; l = l->next)
if ((*l->transport)(l, recv, rc) && (wf = shadow(l, p, ol)) > RAY_WF_MIN)
c = c_add(c, c_mult(m->c,
c_scale(wf * m->kd * v3_dot(l->outdir,i->n), l->outcol)));
if (level++ < MAX_RAY_LEVEL) {
if ((wf = w * m->ks) > RAY_WF_MIN) {
Ray r = ray_make(p, reflect_dir(v.d, i->n));
c = c_add(c, c_mult(m->s,
c_scale(m->ks, ray_shade(level, wf, r, rc, ol))));
}
if ((wf = w * m->kt) > RAY_WF_MIN) {
Ray t = ray_make(p, refract_dir(v.d, i->n, (i->enter)? 1/m->ir: m->ir));
if (v3_sqrnorm(t.d) > 0) {
c = c_add(c, c_mult(m->s,
c_scale(m->kt, ray_shade(level, wf, t, rc, ol))));
}
}
}
inode_free(i);
return c;
} else {
return BG_COLOR;
}
}
Color shiny_surface(RContext *rc)
{
Color ce = environment_map(rc->m->tinfo, reflect_dir(rc->v, rc->n));
return c_add(c_scale(rc->m->ka, ambient(rc)),
c_scale(rc->m->ks, c_add(ce, specular(rc))));
}
Color textured_plastic(RContext *rc)
{
Color c, ct = texture_map(rc->m->tinfo, rc->t);
c = c_add(c_mult(ct, c_add(c_scale(rc->m->ka, ambient(rc)),
c_scale(rc->m->kd, diffuse(rc)))),
c_mult(rc->m->s, c_scale(rc->m->ks, specular(rc))));
return c;
}
void polymult (header *hd)
{ header *st=hd,*hd1,*result;
int c,c1,c2,i,r,j,k;
double *m1,*m2,*mr,x;
complex *mc1,*mc2,*mcr,xc,hc;
interval *mi1,*mi2,*mir,xi,hi;
hd1=next_param(st);
equal_params_2(&hd,&hd1); if (error) return;
getmatrix(hd,&r,&c1,&m1); if (r!=1) wrong_arg();
getmatrix(hd1,&r,&c2,&m2); if (r!=1) wrong_arg();
if ((LONG)c1+c2-1>INT_MAX) wrong_arg();
c=c1+c2-1;
if (iscomplex(hd))
{ mc1=(complex *)m1; mc2=(complex *)m2;
result=new_cmatrix(1,c,""); if (error) return;
mcr=(complex *)matrixof(result);
c_copy(xc,*mc1); mc1++;
for (i=0; i<c2; i++) c_mult(xc,mc2[i],mcr[i]);
for (j=1; j<c1; j++)
{ c_copy(xc,*mc1); mc1++;
for (k=j,i=0; i<c2-1; i++,k++)
{ c_mult(xc,mc2[i],hc);
c_add(hc,mcr[k],mcr[k]);
}
c_mult(xc,mc2[i],mcr[k]);
}
}
else if (isinterval(hd))
{ mi1=(interval *)m1; mi2=(interval *)m2;
result=new_imatrix(1,c,""); if (error) return;
mir=(interval *)matrixof(result);
i_copy(xi,*mi1); mi1++;
for (i=0; i<c2; i++) i_mult(xi,mi2[i],mir[i]);
for (j=1; j<c1; j++)
{ i_copy(xi,*mi1); mi1++;
for (k=j,i=0; i<c2-1; i++,k++)
{ i_mult(xi,mi2[i],hi);
c_add(hi,mir[k],mir[k]);
}
c_mult(xi,mi2[i],mir[k]);
}
}
else if (isreal(hd))
{ result=new_matrix(1,c,""); if (error) return;
mr=matrixof(result);
x=*m1++;
for (i=0; i<c2; i++) mr[i]=x*m2[i];
for (j=1; j<c1; j++)
{ x=*m1++;
for (k=j,i=0; i<c2-1; i++,k++) mr[k]+=x*m2[i];
mr[k]=x*m2[i];
}
}
else wrong_arg();
moveresult(st,result);
}
struct c_complex Lentz_Dn(struct c_complex z,long n)
/*:10*/
#line 126 "./mie.w"
{
struct c_complex alpha_j1,alpha_j2,zinv,aj;
struct c_complex alpha,result,ratio,runratio;
/*12:*/
#line 156 "./mie.w"
zinv= c_sdiv(2.0,z);
alpha= c_smul(n+0.5,zinv);
aj= c_smul(-n-1.5,zinv);
alpha_j1= c_add(aj,c_inv(alpha));
alpha_j2= aj;
ratio= c_div(alpha_j1,alpha_j2);
runratio= c_mul(alpha,ratio);
/*:12*/
#line 131 "./mie.w"
do
/*13:*/
#line 179 "./mie.w"
{
aj.re= zinv.re-aj.re;
aj.im= zinv.im-aj.im;
alpha_j1= c_add(c_inv(alpha_j1),aj);
alpha_j2= c_add(c_inv(alpha_j2),aj);
ratio= c_div(alpha_j1,alpha_j2);
zinv.re*= -1;
zinv.im*= -1;
runratio= c_mul(ratio,runratio);
}
/*:13*/
#line 134 "./mie.w"
while(fabs(c_abs(ratio)-1.0)> 1e-12);
result= c_add(c_sdiv((double)-n,z),runratio);
return result;
}
void calc_aver(FILE *fp,int nframes,int npair,t_pair pair[],t_sij *spec,
real maxdist)
{
int i,j,m;
real nf_1,fac,md_6;
complex c1,c2,dc2;
md_6 = pow(maxdist,-6.0);
fac = 4*M_PI/5;
nf_1 = 1.0/nframes;
for(i=0; (i<npair); i++) {
c2.re = 0;
c2.im = 0;
fprintf(fp,"%5d %5d",pair[i].ai,pair[i].aj);
for(m=0; (m<5); m++) {
c1.re = spec[i].Ylm[m].re*nf_1;
c1.im = spec[i].Ylm[m].im*nf_1;
dc2 = c_sqr(c1);
c2 = c_add(dc2,c2);
if (c1.im > 0)
fprintf(fp," %8.3f+i%8.3f",c1.re,c1.im);
else
fprintf(fp," %8.3f-i%8.3f",c1.re,-c1.im);
}
fprintf(fp,"\n");
spec[i].rij_3 *= nf_1;
spec[i].rij_6 *= nf_1;
spec[i].y2.re = fac*c2.re;
spec[i].y2.im = fac*c2.im;
spec[i].bNOE = (spec[i].rij_6 > md_6);
}
}
// 1D Fourier Transform: fft_1d_helper(x, N, stride, y)
// if N > 1 then
// 1. take the FFT of the even elements in x, and store it in the first half of y
// x' = x
// stride' = 2 * stride
// y' = y
// 2. take the FFT of the odd elements in x, and store it in the second half of y
// x' = x + stride
// stride' = 2 * stride
// y' = y + N/2
// 3. for each frequency step k in 0 ... N/2
// i. compute the component at Wkn = exp(-j*2*pi/N * k)
// y[k] = y[k] + Wkn * y[k + N/2]
// ii. compute the component at Wkn = exp(-j*2*pi/N * (k + N/2))
// y[k + N/2] = y[k] + Wkn * y[k + N/2]
// otherwise just return the element at x[0]
void fft_1d_helper(complex* x, int N, int stride, complex* y,
complex* Wkn, int Wkn_stride){
if(N > 1) {
fft_1d_helper(x, N/2, 2*stride, y, Wkn, 2*Wkn_stride);
fft_1d_helper(x + stride, N/2, 2*stride, y + N/2, Wkn, 2*Wkn_stride);
for(int k=0; k<N/2; k++) {
complex Wkn1 = Wkn[k*Wkn_stride];
complex Wkn2 = Wkn[(k + N/2)*Wkn_stride];
complex yk1 = c_add(y[k], c_mult(Wkn1, y[k + N/2]));
complex yk2 = c_add(y[k], c_mult(Wkn2, y[k + N/2]));
y[k] = yk1;
y[k+N/2] = yk2;
}
} else {
y[0] = x[0];
}
}
static Color ambient_rad(int n, Color *dm, Real *a)
{
int i;
Real aa = 0;
Color ma = c_make(0,0,0);
for (i = 0; i < n; i++) {
ma = c_add(ma, c_scale(a[i], dm[i]));
aa += a[i];
}
return c_scale(1.0/aa, ma);
}
static struct complex *pol_value(
struct polynom *p,
struct complex *z, /* pointer to argument */
struct complex *v /* pointer to function value */
)
{
int i;
v->x=p->a[p->n].x;
v->y=p->a[p->n].y;
for (i=p->n-1; i>=0; --i)
c_add(v,&(p->a[i]),c_mult(v,z,v));
return(v);
}
/**
* \brief Compute the variance of the impulse response
* \param poles0 unscaled pole locations
* \param q rescaling parameter
* \param K number of poles
* \return variance achieved by poles = poles0^(1/q)
* \ingroup vyv_gaussian
*/
static double variance(const complex4c *poles0, int K, double q)
{
complex4c sum = { 0, 0 };
int k;
for (k = 0; k < K; ++k)
{
complex4c z = c_real_pow(poles0[k], 1 / q), denom = z;
denom.real -= 1;
/* Compute sum += z / (z - 1)^2. */
sum = c_add(sum, c_div(z, c_mul(denom, denom)));
}
return 2 * sum.real;
}
//SUMMARY
// Adds the cache entry 'ent' to the end
//NOTES
// Must be locked over the cache - can be unlocked immediately following the call
// No other synchronization needed, since old entries will automatically be displaced once cache is full
void cache_putEntry(struct cache_entry* ent) {
assert( ent->filedata != NULL );
fprintf(stderr, "cache_putEntry: Adding entry for filename %s\n", ent->filename);
assert( cache_size <= max_cache_size );
if (cache_size == max_cache_size) {
fprintf(stderr, "cache_putEntry: Cache full, shifting (%d, %d)\n", cache_size, max_cache_size);
destroyCacheEntry( c_shift() ); // This will also free the file data stored in the entry
}
c_add(ent);
}
static struct polynom *pol_mult(struct polynom *p,struct polynom *p1,struct polynom *p2)
{
int i,j;
struct complex tulo;
p->n=p1->n+p2->n;
if (p->n>MAXN-1) { pol_dim_overflow(); return(p); }
for (i=0; i<=p->n; ++i)
p->a[i].x=p->a[i].y=0.0;
for (i=0; i<=p1->n; ++i)
for (j=0; j<=p2->n; ++j)
c_add(&(p->a[i+j]),&(p->a[i+j]),
c_mult(&tulo,&(p1->a[i]),&(p2->a[j])));
return(p);
}
void Dn_down(struct c_complex z,long nstop,struct c_complex*D)
/*:19*/
#line 247 "./mie.w"
{
long k;
struct c_complex zinv,k_over_z;
D[nstop-1]= Lentz_Dn(z,nstop);
zinv= c_inv(z);
for(k= nstop-1;k>=1;k--){
k_over_z= c_smul((double)k,zinv);
D[k-1]= c_sub(k_over_z,c_inv(c_add(D[k],k_over_z)));
}
}
/**
* \brief Derivative of variance with respect to q
* \param poles0 unscaled pole locations
* \param q rescaling parameter
* \param K number of poles
* \return derivative of variance with respect to q
* \ingroup vyv_gaussian
*
* This function is used by compute_q() in solving for q.
*/
static double dq_variance(const complex4c *poles0, int K, double q)
{
complex4c sum = { 0, 0 };
int k;
for (k = 0; k < K; ++k)
{
complex4c z = c_real_pow(poles0[k], 1 / q), w = z, denom = z;
w.real += 1;
denom.real -= 1;
/* Compute sum += z log(z) (z + 1) / (z - 1)^3 */
sum = c_add(sum, c_div(c_mul(c_mul(z, c_log(z)), w),
c_real_pow(denom, 3)));
}
return (2 / q) * sum.real;
}
void polyadd (header *hd)
{ header *st=hd,*hd1,*result;
int c,c1,c2,i,r;
double *m1,*m2,*mr;
complex *mc1,*mc2,*mcr;
interval *mi1,*mi2,*mir;
hd1=next_param(st);
equal_params_2(&hd,&hd1); if (error) return;
getmatrix(hd,&r,&c1,&m1); if (r!=1) wrong_arg();
getmatrix(hd1,&r,&c2,&m2); if (r!=1) wrong_arg();
c=max(c1,c2);
if (iscomplex(hd)) /* complex values */
{ mc1=(complex *)m1; mc2=(complex *)m2;
result=new_cmatrix(1,c,""); if (error) return;
mcr=(complex *)matrixof(result);
for (i=0; i<c; i++)
{ if (i>=c1) { c_copy(*mcr,*mc2); mcr++; mc2++; }
else if (i>=c2) { c_copy(*mcr,*mc1); mcr++; mc1++; }
else { c_add(*mc1,*mc2,*mcr); mc1++; mc2++; mcr++; }
}
}
else if (isinterval(hd))
{ mi1=(interval *)m1; mi2=(interval *)m2;
result=new_imatrix(1,c,""); if (error) return;
mir=(interval *)matrixof(result);
for (i=0; i<c; i++)
{ if (i>=c1) { i_copy(*mir,*mi2); mir++; mi2++; }
else if (i>=c2) { i_copy(*mir,*mi1); mir++; mi1++; }
else { i_add(*mi1,*mi2,*mir); mi1++; mi2++; mir++; }
}
}
else if (isreal(hd))
{ result=new_matrix(1,c,""); if (error) return;
mr=matrixof(result);
for (i=0; i<c; i++)
{ if (i>=c1) { *mr++ = *m2++; }
else if (i>=c2) { *mr++ = *m1++; }
else { *mr++ = *m1++ + *m2++; }
}
}
else wrong_arg();
moveresult(st,result);
}
void rfft (long m0, long p0, long q0, long n)
/***** rfft
make a fft on x[m],x[m+q0],...,x[m+(p0-1)*q0] (p points).
one has xi_p0 = xi_n^n = zz[n] ; i.e., p0*n=nn.
*****/
{ long p,q,m,l;
long mh,ml;
int found=0;
complex sum,h;
if (p0==1) return;
if (test_key()==escape) { error=301; return; }
if (p0%2==0) { p=p0/2; q=2; }
else
{ q=3;
while (q*q<=p0)
{ if (p0%q==0)
{ found=1; break; }
q+=2;
}
if (found) p=p0/q;
else { q=p0; p=1; }
}
if (p>1) for (m=0; m<q; m++)
rfft((m0+m*q0)%nn,p,q*q0,nn/p);
mh=m0;
for (l=0; l<p0; l++)
{ ml=l%p;
c_copy(sum,ff[(m0+ml*q*q0)%nn]);
for (m=1; m<q; m++)
{ c_mult(ff[(m0+(m+ml*q)*q0)%nn],zz[(n*l*m)%nn],h);
c_add(sum,h,sum);
}
sum[0]/=q; sum[1]/=q;
c_copy(fh[mh],sum);
mh+=q0; if (mh>=nn) mh-=nn;
}
for (l=0; l<p0; l++)
{ c_copy(ff[m0],fh[m0]);
m0+=q0;
}
}
// resample_1d(complex* x, int N, int stride, float* n)
// 1. create a temporary buffer of size N
// 2. for step i in 0 ... N
// 3. if n[i] is outside of range 0 ... N, then x_interp = 0
// 4. get the lower element x_low = x[floor(n[i]) * stride]
// 5. get the higher element x_high = x[ceil(n[i]) * stride]
// 6. compute the interpolation mixing ratio, ratio = n[i] - floor(n[i])
// 7. compute the interpolated value, x_interp = (1 - ratio) * x_low + ratio * x_high
// 8. store the interpolated value into the buffer
// 9. copy the values from the temporary buffer back into x
void resample_1d(complex* x, int N, int stride, float* n) {
int i;
complex buffer[N];
for(i=0; i<N; i++){
complex x_interp;
if(n[i] < 0 || n[i] > N - 1){
x_interp.real = 0;
x_interp.imag = 0;
} else {
complex x_low = x[(int)floor(n[i]) * stride];
complex x_high = x[(int)ceil(n[i]) * stride];
float ratio = n[i] - floor(n[i]);
x_interp = c_add(c_scalar_mult(x_low, 1 - ratio), c_scalar_mult(x_high, ratio));
}
buffer[i] = x_interp;
}
for(i=0; i<N; i++)
x[i*stride] = buffer[i];
}
void polyzeros (header *hd)
{ header *st=hd,*result;
int i,j,r,c;
double *m,*mr,x;
complex *mc,*mcr,xc,hc;
hd=getvalue(hd); if (error) return;
if (hd->type==s_real || hd->type==s_matrix)
{ getmatrix(hd,&r,&c,&m);
if (r!=1) wrong_arg();
result=new_matrix(1,c+1,""); if (error) return;
mr=matrixof(result);
mr[0]=-m[0]; mr[1]=1.0;
for (i=1; i<c; i++)
{ x=-m[i]; mr[i+1]=1.0;
for (j=i; j>=1; j--) mr[j]=mr[j-1]+x*mr[j];
mr[0]*=x;
}
}
else if (hd->type==s_complex || hd->type==s_cmatrix)
{ getmatrix(hd,&r,&c,&m); mc=(complex *)m;
if (r!=1) wrong_arg();
result=new_cmatrix(1,c+1,""); if (error) return;
mcr=(complex *)matrixof(result);
mcr[0][0]=-mc[0][0]; mcr[0][1]=-mc[0][1];
mcr[1][0]=1.0; mcr[1][1]=0.0;
for (i=1; i<c; i++)
{ xc[0]=-mc[i][0]; xc[1]=-mc[i][1];
mcr[i+1][0]=1.0; mcr[i+1][1]=0.0;
for (j=i; j>=1; j--)
{ c_mult(xc,mcr[j],hc);
c_add(hc,mcr[j-1],mcr[j]);
}
c_mult(xc,mcr[0],mcr[0]);
}
}
else wrong_arg();
moveresult(st,result);
}
static struct polynom *pol_add(struct polynom *p,struct polynom *p1,struct polynom *p2)
{
int i;
// RS REM struct complex tulo;
p->n=(p1->n>p2->n)? (p1->n):(p2->n);
for (i=0; i<=p->n; ++i)
{
if (i<=p1->n)
{
if (i<=p2->n)
c_add(&(p->a[i]),&(p1->a[i]),&(p2->a[i]));
else
{ p->a[i].x=p1->a[i].x; p->a[i].y=p1->a[i].y; }
}
else
{ p->a[i].x=p2->a[i].x; p->a[i].y=p2->a[i].y; }
}
i=p->n;
while (c_zero(&(p->a[i])) && i>0) --i;
p->n=i;
return(p);
}
/////////////////////////////////////////////////////////////////////////////////////////////
//int main(int argc, char *argv[ ])
//{
int main(void)
{
sortingindex = 0;
if((fpp = fopen(INPUT_FILE, "r")) == NULL)
{
printf("Cannot open 'parameter_file'.\n");
}
for(j = 0; j < 11; j++)
{
if(fscanf(fpp, "%lf", &tmp) != EOF)
{
my_array[j] = tmp;
} else {
printf("Not enough data in 'input_parameter'!");
}
}
fclose(fpp);
In_n= (int) my_array[0];
In_vect_n= (int) my_array[1];
Out_n= (int) my_array[2];
Mf_n= (int) my_array[3];
training_data_n= (int) my_array[4];
checking_data_n= (int) my_array[5];
epoch_n= (int) my_array[6];
step_size=my_array[7];
increase_rate=my_array[8];
decrease_rate=my_array[9];
threshold = my_array[10];
Rule_n = (int)pow((double)Mf_n, (double)In_n); //number of rules
Node_n = In_n + In_n*Mf_n + 3*Rule_n + In_n*Rule_n + Out_n;
/* allocate matrices and memories */
int trnnumcheck[training_data_n + 1];
int trnnumchecku[training_data_n + 1];
for(i=0; i<training_data_n +1; i++)
{
trnnumcheck[i]=0;
trnnumchecku[i]=0;
}
diff =(double **)create_matrix(Out_n, training_data_n, sizeof(double));
double chkvar[checking_data_n];
double cdavg[checking_data_n];
double chkvar_un[checking_data_n];
double cdavg_un[checking_data_n];
target = calloc(Out_n, sizeof(double));
de_out = calloc(Out_n, sizeof(double));
node_p = (NODE_T **)create_array(Node_n, sizeof(NODE_T *));
config = (int **)create_matrix(Node_n, Node_n, sizeof(int));
training_data_matrix = (double **)create_matrix(training_data_n, In_n*In_vect_n + Out_n, sizeof(double));
if(checking_data_n > 0)
{
checking_data_matrix = (double **)create_matrix(checking_data_n, In_n*In_vect_n +Out_n, sizeof(double));
checking_data_matrix_un = (double **)create_matrix(checking_data_n, Out_n, sizeof(double));
chk_output = (double **)create_matrix(checking_data_n, Out_n, sizeof(double));
}
layer_1_to_4_output = (COMPLEX_T **)create_matrix(training_data_n, In_n*Mf_n + 3*Rule_n, sizeof(COMPLEX_T));
trn_rmse_error = calloc(epoch_n, sizeof(double));
trnNMSE = calloc(epoch_n, sizeof(double));
chk_rmse_error = calloc(epoch_n, sizeof(double));
kalman_parameter = (double **)create_matrix(Out_n ,(In_n*In_vect_n + 1)*Rule_n, sizeof(double));
kalman_data = (double **)create_matrix(Out_n ,(In_n*In_vect_n + 1)*Rule_n, sizeof(double));
step_size_array = calloc(epoch_n, sizeof(double));
ancfis_output = (double **)create_matrix(training_data_n , Out_n, sizeof(double));
trn_error =calloc(Out_n +1, sizeof(double));
chk_error_n = calloc(Out_n +1, sizeof(double));// changing size for adding new error measures
chk_error_un = calloc(Out_n +1, sizeof(double));// changing size for adding new error measures
trn_datapair_error = calloc(training_data_n, sizeof(double));
trn_datapair_error_sorted = (double **)create_matrix(2,training_data_n, sizeof(double));
NMSE = calloc(Out_n, sizeof(double));
NDEI = calloc(Out_n, sizeof(double));
unNMSE = calloc(Out_n, sizeof(double));
unNDEI = calloc(Out_n, sizeof(double));
//Build Matrix of 0 nd 1 to show the connected nodes
gen_config(In_n, Mf_n,Out_n, config);//gen_config.c
//With the above matrix, build ANCFIS connected nodes
build_ancfis(config); //datastru.c
//Find total number of nodes in layer 1 and 5
parameter_n = set_parameter_mode(); //datastru.c
parameter_array = calloc(parameter_n, sizeof(double));
initpara(TRAIN_DATA_FILE, training_data_n, In_n, In_vect_n+1, Mf_n); // initpara.c
// after this step, the parameters (they are present in layer 1 and layer 5 only) are assigned a random initial value
// using some basic algebra and these value are then stored in "para.ini"
get_parameter(node_p,Node_n,INIT_PARA_FILE); //input.c
// after this step, the initial random values of the parametrs are read from "oara.ini" and assigned to the appropriate nodes in the node structure by accessing their para list.
//Get training and testing data
get_data(TRAIN_DATA_FILE, training_data_n, training_data_matrix); //input.c
// after this step, the training input data is read from the "data.trn" fle and stroed in the training data matrix.
get_data(CHECK_DATA_FILE, checking_data_n, checking_data_matrix); //input.c
// after the above step, the checking data is read from the "data.chk" file and then stored in the checking data matrix.
for(i=0; i< Out_n; i++)
{
for(j=0; j<training_data_n; j++)
{
trnavg = trnavg + training_data_matrix[j][(i+1)*In_vect_n +i];
}
}
trnavg = trnavg /(Out_n * training_data_n);
for(i=0; i< Out_n; i++)
{
for(j=0; j<training_data_n; j++)
{
temp = training_data_matrix[j][(i+1)*In_vect_n +i]- trnavg;
temp = temp*temp;
trnvariance = trnvariance + temp;
}
}
trnvariance = trnvariance /((Out_n * training_data_n)-1);
temp = 0.0;
for(i=0; i< Out_n; i++)
{
for(j=0; j< checking_data_n; j++)
{
chkavg = chkavg + checking_data_matrix[j][(i+1)*In_vect_n +i];
}
}
chkavg = chkavg /(Out_n * checking_data_n);
for(i=0; i< Out_n; i++)
{
for(j=0; j<checking_data_n; j++)
{
temp = checking_data_matrix[j][(i+1)*In_vect_n +i]- chkavg;
temp = temp*temp;
chkvariance = chkvariance + temp;
}
}
chkvariance = chkvariance /((Out_n * checking_data_n)-1);
printf("epochs \t trn error \t tst error\n");
printf("------ \t --------- \t ---------\n");
//printf("not entering the epoch loop and the i loop yoyoyo\n");
/**************
for(ep_n = 0; ep_n < epoch_n; ep_n++)
{
//step_size_pointer= &step_size;
//printf("epoch numbernumber %d \n", ep_n);
//step_size_array[ep_n] = step_size_pointer;
step_size_array[ep_n] = step_size;
// after the above step, the updated stepsize at the end of the last loop is stored in the step_size_array.
// this will keep happening every time we start en epoch and hence at the end of the loop, step_size_array will
// have a list of all the updated step sizes. Since this is a offline version, step sizes are updated only
// at the end of an epoch.
for(m = 0; m < Out_n; m++)
{
//printf("m loop number %d \n", m);
for(j = 0; j < training_data_n; j++)
{
//printf("j loop number %d \n", j);
//copy the input vector(s) to input node(s)
put_input_data(node_p,j, training_data_matrix); //input.c
// after this(above) step, the input data is transferred frm the training data matrix to the "node" structure.
//printf("testing \n");
//printf("reeeetesting \n");
target[m] = training_data_matrix[j][(m+1)*In_vect_n+m]; // ***
// this step assigns the value of the "m"th output of "j" th trainig data pair to target.
//printf("testing \n");
//forward pass, get node outputs from layer 1 to layer 4
calculate_output(In_n, In_n + In_n*Mf_n + 3*Rule_n - 1, j); //forward.c
// after this step, output of nodes in layer 1 to 4 is calculated. Please note that when this happens for the first
// time, i.e. when ep_n=0, our network parametrs are already initialized. thus, it is possible to get the
// output of each node using the function definitios proposed in forward.c. After first epoch, our parametrs get
// updated and this output is then calculated using teh new parameters. The essential point to note here is that
// we can always calculate the output of each node since we have already initialized our parameters.
//printf("testing \n");
//put outputs of layer 1 to 4 into layer_1_to_4_output
for(k = 0; k < Mf_n*In_n + 3*Rule_n; k++)
{
//printf("testing \n");
layer_1_to_4_output[j][k] = *node_p[k + In_n]->value;
}
// the above loop simply puts the values of nodes from layer 1 to layer 4 in the layer_1_to_4_output matrix.
//identify layer 5 params using LSE (Kalman filter)
//printf("testing \n");
get_kalman_data(kalman_data, target); //kalman.c
// this function call finds out the values of O4iXnl .. these are basically the coefficients
// of the kalman parametrs for a given training data pair
//puts them in kalman_data matrix.
// this kalman_data matrix has In_n number of rows and number of columns equal to number of parametrs that are
// responsible for determining each output... as stated above, the outputs are actually the coefficients of the
// parameters.
//printf("testing \n");
//calculate Kalman parameters
kalman(ep_n, j+(m*training_data_n), m, kalman_data, kalman_parameter,target); //kalman.c
// this function call evaluates kalman parametrs for a given output, for a given epoch.. that is it takes the epoch
// number from us, takes the info about how many times has kalman been invoked before, also takes in the
// output number(row number) for whihc the parametrs are to be found out... it also takes kalman_data and reads
// from it to estimate the kalman parameters... it also takes target .. and stores the output in the mth row of
// kalman_parameter.
//printf("testing \n");
}
// let me tell u what the abopve loop is doing.. after observing closely, it is easy to see that in the above loop,
// for a given output node, one by one, all the training data are taken inside the ANCFIS structure, outputs
// are calculated from one to 4, then a recursive kalman filetr is used to identify the kalman
// parametrs corresponding to the output node.. these kalman parameters are updated after every tarining data pair
// and finally at the end of all the training data, we have an estimate for the kalman parametrs corresponding to // the output node.
}
// thus, at the of the above loop, the kalman parametrs for all the output nodes are evaluated...
// now, we are ready to actually calculate the outputs.. plase remember that, all this while, the actual
// values of the parametrs of nodes in layer 1 and layer 5 are the ones that were randomly initialized.
for(j = 0; j < training_data_n; j++)
{
//printf("testing 1\n");
put_input_data(node_p,j, training_data_matrix); //input.c
//printf("testing 2 \n");
for(k = 0; k < Mf_n*In_n + 3*Rule_n; k++)
{
*node_p[k + In_n]->value = layer_1_to_4_output[j][k];
}
// u must be able to see that in the above two loops, each time, whatever output we got for a given training
// datta pair, it was safely stored in layer_1_to_4 array...and each time, the value on the actual nodes in the
// structure got changed.. due to new incoming training data pair..this was periodic with period trainingdata_n..
// that is for each output node, we got the same results for a given training dat aapir.. that is the node values
// were independent of m. Now, for a given traing data pair, we are getting those value back in the actual node
// node structure from that laye blh blah matrix..
//printf("testing 3\n");
put_kalman_parameter(Out_n,kalman_parameter); //kalman.c
// using this function call, we are placing the setimated value of the layer 5 parametrs in the node structure
// by accessing each node and its parameter list.
// Do forward pass for L5 and L6
calculate_output(In_n + In_n*Mf_n + 3*Rule_n, Node_n, j); //forward.c
// for a given value of the training data pair, this function calculates the output of layer 5 and layer 6 nodes
// and places them in the node structure.
calculate_root(training_data_matrix,diff,j,node_p); //debug.c
// this function call calculates the square of the erro between the predicted value of an output node and the
// corresponding actual value in the training data matrix..(actual output) and stores it in diff.
// this call performs the above action for a given training data pair and for all output nodes.
// Do backward pass and calculate all error rate for each node
calculate_de_do(j,Out_n,target,ancfis_output); //backward.c
// calculates de_do for each node fora given training data pair
update_de_dp(j); //de_dp.c
// updates de_do for each node....
}
// thus at the end of this loop, estimated outputs for all the training data are calculated..also back propogatin
// is done and de_dp for all the nodes is updated.
//printf("testing 1\n");
calculate_trn_err(diff,trn_error,trn_datapair_error,training_data_n); //debug.c
//printf("testing 2 \n");
//training_error_measure(target,ancfis_output,diff, training_data_n, trn_error,out_n); //trn_err.c
trn_rmse_error[ep_n] = trn_error[Out_n];
printf("%3d \t %.11f \n", ep_n+1, trn_error[Out_n]);
//Find RMSE of testing error
/************************************* if(checking_data_n != 0)
{
printf("testing 3 \n");
epoch_checking_error(checking_data_matrix, checking_data_n, chk_error, training_data_n, chk_output, ep_n); //chk_err.c writes to tracking.txt
printf("testing 4 \n");
chk_rmse_error[ep_n] = chk_error[Out_n];
for (i=0; i<Out_n; i++)
//printf("%3d \t %.11f \t %.11f\n", ep_n+1, trn_error[i], chk_error[i]);
printf("%3d \t %.11f \n", ep_n+1, trn_error[i]);
//printf("%.11f\t %.11f\n", trn_error[Out_n],chk_error[Out_n]);
printf("%.11f\t %.11f\n", trn_error[Out_n]);
write_result(ep_n+1,Out_n,trn_error,chk_error); //debug.c writes to result.txt
}
else
{
for (i=0; i<Out_n; i++)
printf("%4d \t %.11f\n", ep_n+1, trn_error[i]);
}
***************************/
/**
//Find minimum training error and its epoch-number
if(trn_rmse_error[ep_n] < min_trn_RMSE) {
min_trn_RMSE_epoch = ep_n +1;
min_trn_RMSE = trn_rmse_error[ep_n];
record_parameter(parameter_array);
}
if(ep_n < epoch_n-1)
{
//update parameters in 1st layer (Using VNCSA)
update_parameter(1, step_size); //new_para.c
//update stepsize
update_step_size(trn_rmse_error, ep_n, &step_size, decrease_rate, increase_rate); //stepsize.c
}
}
***/
////////////////////////////////////////////////////////////
fppp = (FILE *)open_file("status.txt", "w");
fpppp = (FILE *)open_file("trn.txt", "w");
ep_n=0;
do
{
//step_size_pointer= &step_size;
printf("epoch numbernumber %d \n", ep_n+1);
//step_size_array[ep_n] = step_size_pointer;
step_size_array[ep_n] = step_size;
// after the above step, the updated stepsize at the end of the last loop is stored in the step_size_array.
// this will keep happening every time we start en epoch and hence at the end of the loop, step_size_array will
// have a list of all the updated step sizes. Since this is a offline version, step sizes are updated only
// at the end of an epoch.
for(m = 0; m < Out_n; m++)
{
//printf("m loop number %d \n", m);
for(j = 0; j < training_data_n; j++)
{
//printf("j loop number %d \n", j);
//copy the input vector(s) to input node(s)
put_input_data(node_p,j, training_data_matrix); //input.c
// after this(above) step, the input data is transferred frm the training data matrix to the "node" structure.
//printf("testing \n");
//printf("reeeetesting \n");
target[m] = training_data_matrix[j][(m+1)*In_vect_n+m]; // ***
// this step assigns the value of the "m"th output of "j" th trainig data pair to target.
//printf("testing \n");
//forward pass, get node outputs from layer 1 to layer 4
calculate_output(In_n, In_n + In_n*Mf_n + 3*Rule_n, j); //forward.c
// after this step, output of nodes in layer 1 to 4 is calculated. Please note that when this happens for the first
// time, i.e. when ep_n=0, our network parametrs are already initialized. thus, it is possible to get the
// output of each node using the function definitios proposed in forward.c. After first epoch, our parametrs get
// updated and this output is then calculated using teh new parameters. The essential point to note here is that
// we can always calculate the output of each node since we have already initialized our parameters.
//printf("testing \n");
//put outputs of layer 1 to 4 into layer_1_to_4_output
for(k = 0; k < Mf_n*In_n + 3*Rule_n; k++)
{
//printf("testing \n");
layer_1_to_4_output[j][k] = *node_p[k + In_n]->value;
//fprintf(fppp, "%lf \t %lf \t \n", (layer_1_to_4_output[j][k]).real, (layer_1_to_4_output[j][k]).imag);
}
// the above loop simply puts the values of nodes from layer 1 to layer 4 in the layer_1_to_4_output matrix.
//identify layer 5 params using LSE (Kalman filter)
//printf("testing \n");
get_kalman_data(kalman_data, target); //kalman.c
// this function call finds out the values of O4iXnl .. these are basically the coefficients
// of the kalman parametrs for a given training data pair
//puts them in kalman_data matrix.
// this kalman_data matrix has In_n number of rows and number of columns equal to number of parametrs that are
// responsible for determining each output... as stated above, the outputs are actually the coefficients of the
// parameters.
//printf("testing \n");
//calculate Kalman parameters
kalman(ep_n, j+(m*training_data_n), m, kalman_data, kalman_parameter,target); //kalman.c
// this function call evaluates kalman parametrs for a given output, for a given epoch.. that is it takes the epoch
// number from us, takes the info about how many times has kalman been invoked before, also takes in the
// output number(row number) for whihc the parametrs are to be found out... it also takes kalman_data and reads
// from it to estimate the kalman parameters... it also takes target .. and stores the output in the mth row of
// kalman_parameter.
//printf("testing \n");
}
// let me tell u what the abopve loop is doing.. after observing closely, it is easy to see that in the above loop,
// for a given output node, one by one, all the training data are taken inside the ANCFIS structure, outputs
// are calculated from one to 4, then a recursive kalman filetr is used to identify the kalman
// parametrs corresponding to the output node.. these kalman parameters are updated after every tarining data pair
// and finally at the end of all the training data, we have an estimate for the kalman parametrs corresponding to // the output node.
}
// thus, at the of the above loop, the kalman parametrs for all the output nodes are evaluated...
// now, we are ready to actually calculate the outputs.. plase remember that, all this while, the actual
// values of the parametrs of nodes in layer 1 and layer 5 are the ones that were randomly initialized.
for(j = 0; j < training_data_n; j++)
{
//printf("testing 1\n");
put_input_data(node_p,j, training_data_matrix); //input.c
//printf("testing 2 \n");
for(k = 0; k < Mf_n*In_n + 3*Rule_n; k++)
{
*node_p[k + In_n]->value = layer_1_to_4_output[j][k];
/*if(ep_n==1)
{
fprintf(fppp, "%d.\t %lf \t + \t i%lf \n", k, (layer_1_to_4_output[j][k]).real,(layer_1_to_4_output[j][k]).imag);
}*/
}
// u must be able to see that in the above two loops, each time, whatever output we got for a given training
// datta pair, it was safely stored in layer_1_to_4 array...and each time, the value on the actual nodes in the
// structure got changed.. due to new incoming training data pair..this was periodic with period trainingdata_n..
// that is for each output node, we got the same results for a given training dat aapir.. that is the node values
// were independent of m. Now, for a given traing data pair, we are getting those value back in the actual node
// node structure from that laye blh blah matrix..
//printf("testing 3\n");
put_kalman_parameter(Out_n,kalman_parameter); //kalman.c
//printf("hihahahha \n");
// using this function call, we are placing the setimated value of the layer 5 parametrs in the node structure
// by accessing each node and its parameter list.
// Do forward pass for L5 and L6
calculate_output(In_n + In_n*Mf_n + 3*Rule_n, Node_n, j); //forward.c
// for a given value of the training data pair, this function calculates the output of layer 5 and layer 6 nodes
// and places them in the node structure.
//printf("hihahahha no 2 \n");
calculate_root(training_data_matrix,diff,j,node_p); //debug.c
// this function call calculates the square of the erro between the predicted value of an output node and the
// corresponding actual value in the training data matrix..(actual output) and stores it in diff.
// this call performs the above action for a given training data pair and for all output nodes.
// Do backward pass and calculate all error rate for each node
calculate_de_do(j,Out_n,target,ancfis_output); //backward.c
//printf("hihahahha no 3 \n");
// calculates de_do for each node fora given training data pair
update_de_dp(j); //de_dp.c
// updates de_do for each node....
}
// thus at the end of this loop, estimated outputs for all the training data are calculated..also back propogatin
// is done and de_dp for all the nodes is updated.
//printf("testing 1\n");
calculate_trn_err(diff,trn_error, trn_datapair_error, training_data_n); //debug.c
//printf("testing 2 \n");
//training_error_measure(target,ancfis_output,diff, training_data_n, trn_error,out_n); //trn_err.c
trn_rmse_error[ep_n] = trn_error[Out_n];
trnNMSE[ep_n] = trn_rmse_error[ep_n]*trn_rmse_error[ep_n]/trnvariance;
fprintf(fppp, "epoch number is %d \t trn RMSE is %.11f \t trn NMSE is %lf \t \n", ep_n + 1, trn_rmse_error[ep_n], trnNMSE[ep_n]);
//fprintf(fpppp, "\n");
fprintf(fpppp, "epoch number is %d \t trn RMSE is %.11f \t trn NMSE is %lf \t \n", ep_n + 1, trn_rmse_error[ep_n], trnNMSE[ep_n]);
printf("trn RMSE is \t %lf \n", trn_rmse_error[ep_n]);
printf("trn NMSE is \t %lf \n", trnNMSE[ep_n]);
for(i=0; i<training_data_n; i++)
{
trn_datapair_error_sorted[0][i]=trn_datapair_error[i];
trn_datapair_error_sorted[1][i]= i+1;
}
for(j=1; j<training_data_n; j++)
{
for(i=0; i<training_data_n-j; i++)
{
if(trn_datapair_error_sorted[0][i]>trn_datapair_error_sorted[0][i+1])
{
sorting=trn_datapair_error_sorted[0][i+1];
trn_datapair_error_sorted[0][i+1]=trn_datapair_error_sorted[0][i];
trn_datapair_error_sorted[0][i]=sorting;
sortingindex = sorting=trn_datapair_error_sorted[1][i+1];
trn_datapair_error_sorted[1][i+1]=trn_datapair_error_sorted[1][i];
trn_datapair_error_sorted[1][i]=sortingindex;
}
}
}
for(j=0; j<training_data_n; j++)
{
fprintf(fppp, "\n");
fprintf(fppp, "training data pair sorted number \t %d \n", j+1);
fprintf(fppp, "training data pair original number \t %d \n", (int)(trn_datapair_error_sorted[1][j]));
fprintf(fppp, "training data pair sorted error in RMSE is \t %lf \n",trn_datapair_error_sorted[0][j]);
fprintf(fpppp, "%d \t", (int)(trn_datapair_error_sorted[1][j]));
complexsum = complex(0.0, 0.0);
fprintf(fppp,"Normalized layer 3 outputs are as follows \n");
for(k= In_n*Mf_n + Rule_n; k< In_n*Mf_n + 2*Rule_n; k++)
{
fprintf(fppp, "%d.\t %lf + i%lf \t %lf < %lf \n", k, (layer_1_to_4_output[j][k]).real,(layer_1_to_4_output[j][k]).imag, c_abs(layer_1_to_4_output[j][k]), c_phase(layer_1_to_4_output[j][k])*180/PI);
complexsum = c_add(complexsum, layer_1_to_4_output[j][k]);
}
fprintf(fppp, "Sum of the outputs of layer 3 is \t %lf+i%lf \t %lf<%lf \n", complexsum.real, complexsum.imag, c_abs(complexsum), c_phase(complexsum)*180/PI);
complexsum = complex(0.0, 0.0);
fprintf(fppp,"dot producted layer 4 outputs are as follows \n");
for(k=In_n*Mf_n + 2*Rule_n; k< In_n*Mf_n + 3*Rule_n; k++)
{
fprintf(fppp, "%d.\t %lf + i%lf \t %lf < %lf \n", k, (layer_1_to_4_output[j][k]).real,(layer_1_to_4_output[j][k]).imag, c_abs(layer_1_to_4_output[j][k]), c_phase(layer_1_to_4_output[j][k])*180/PI);
complexsum = c_add(complexsum, layer_1_to_4_output[j][k]);
}
fprintf(fppp, "sum of the outputs of layer 4 is \t %lf +i%lf \t %lf<%lf \n", complexsum.real, complexsum.imag, c_abs(complexsum), c_phase(complexsum)*180/PI);
if(j> training_data_n -6 )
{
trnnumcheck[(int)(trn_datapair_error_sorted[1][j])]= trnnumcheck[(int)(trn_datapair_error_sorted[1][j])] +1;
}
if(j<5 )
{
trnnumchecku[(int)(trn_datapair_error_sorted[1][j])]= trnnumchecku[(int)(trn_datapair_error_sorted[1][j])] +1;
}
}
fprintf(fpppp, "\n");
//Find RMSE of testing error
/********************************************************************************
if(checking_data_n != 0)
{
printf("testing 3 \n");
epoch_checking_error(checking_data_matrix, checking_data_n, chk_error, training_data_n, chk_output, ep_n); //chk_err.c writes to tracking.txt
printf("testing 4 \n");
chk_rmse_error[ep_n] = chk_error[Out_n];
for (i=0; i<Out_n; i++)
printf("%3d \t %.11f \t %.11f\n", ep_n+1, trn_error[i], chk_error[i]);
printf("%.11f\t %.11f\n", trn_error[Out_n],chk_error[Out_n]);
write_result(ep_n+1,Out_n,trn_error,chk_error); //debug.c writes to result.txt
}
else
{
for (i=0; i<Out_n; i++)
printf("%4d \t %.11f\n", ep_n+1, trn_error[i]);
}
**************************************************************************************/
// check whether the current training RMSE is less than the threhold and store its epch number and parametrs
if(trn_rmse_error[ep_n] < min_trn_RMSE)
{
min_trn_RMSE_epoch = ep_n +1;
min_trn_RMSE = trn_rmse_error[ep_n];
min_trnNMSE = trnNMSE[ep_n];
record_parameter(parameter_array);
}
if(ep_n < epoch_n-1)
{
//update parameters in 1st layer (Using VNCSA)
update_parameter(1, step_size); //new_para.c
//update stepsize
update_step_size(trn_rmse_error, ep_n, &step_size, decrease_rate, increase_rate); //stepsize.c
}
ep_n++;
} while((trnNMSE[ep_n -1]>= threshold) && (ep_n <= epoch_n -1));
for(i=1; i<=training_data_n; i++)
{
fprintf(fpppp, "%d \t %d \n", i, trnnumcheck[i]);
}
for(i=1; i<=training_data_n; i++)
{
fprintf(fpppp, "%d \t %d \n", i, trnnumchecku[i]);
}
if(trnNMSE[ep_n -1]< threshold)
{
fprintf(fppp, "\n");
fprintf(fppp, "We have gone below the threshold value \n");
fprintf(fppp, "the epoch number in which this happened is %d \n", min_trn_RMSE_epoch);
}
else
{
fprintf(fppp, "\n");
fprintf(fppp, "We exhausted the available epochs and threshold was not broken :( \n");
fprintf(fppp, "the epoch number which yielded minimum training RMSE is %d \n", min_trn_RMSE_epoch);
}
fclose(fppp);
fclose(fpppp);
double *minmaxc;
minmaxc= (double *)calloc(2*In_n, sizeof(double));
if((fpp = fopen("minmax.txt", "r")) == NULL)
{
printf("Cannot open 'parameter_file'.\n");
}
for(j = 0; j < 2*In_n; j++)
{
if(fscanf(fpp, "%lf", &tmp) != EOF)
{
minmaxc[j] = tmp;
} else {
printf("Not enough data in 'input_parameter'!");
}
}
fclose(fpp);
//////////////////////////////////////////////////////////////
restore_parameter(parameter_array); //output.c
write_parameter(FINA_PARA_FILE); //output.c
write_array(trnNMSE, epoch_n, TRAIN_ERR_FILE); //lib.c
if (checking_data_n != 0)
{
//printf("testing 3 \n");
epoch_checking_error(checking_data_matrix, checking_data_n, chk_error_n, chk_error_un, training_data_n, chk_output, ep_n -1, minmaxc); //chk_err.c writes to tracking.txt
//printf("testing 4 \n");
//chk_rmse_error[ep_n] = chk_error[Out_n];
min_chk_RMSE_n = chk_error_n[Out_n];
printf(" initial checking RMSE is %lf \n ", min_chk_RMSE_n);
min_chk_RMSE_un = chk_error_un[Out_n];
//for (i=0; i<Out_n; i++)
//printf("%3d \t %.11f \t %.11f\n", ep_n+1, trn_error[i], chk_error[i]);
//printf("%3d \t %.11f \n", ep_n+1, trn_error[i]);
//printf("%.11f\t %.11f\n", trn_error[Out_n],chk_error[Out_n]);
//printf("%.11f\t \n", trn_error[Out_n]);
//write_result(min_trn_RMSE_epoch ,Out_n,trn_rmse_error,chk_error); //debug.c writes to result.txt about the epoch number at which the stopping was done and the corresponding training RMSE and checking RMSE
}
//write_array(chk_rmse_error, epoch_n, CHECK_ERR_FILE); //lib.c
//}
write_array(step_size_array, epoch_n, STEP_SIZE_FILE); //lib.c
/**************************************************************************
min_chk_RMSE = chk_rmse_error[epoch_n -1];
min_chk_RMSE_epoch = epoch_n -1;
for(j=0; j< epoch_n; j++)
{
if(chk_rmse_error[j]< min_chk_RMSE)
{
min_chk_RMSE = chk_rmse_error[j];
min_chk_RMSE_epoch = j;
}
}
*************************************************************************/
/**************************************************************
double minmaxc[2*In_n];
if((fpp = fopen("minmax.txt", "r")) == NULL)
{
printf("Cannot open 'parameter_file'.\n");
}
for(j = 0; j < 2*In_n; j++)
{
if(fscanf(fpp, "%lf", &tmp) != EOF)
{
minmaxc[j] = tmp;
} else {
printf("Not enough data in 'input_parameter'!");
}
}
fclose(fpp);
***************************************************************************/
for(k=0; k< Out_n; k++)
{
for(j=0; j< checking_data_n; j++)
{
checking_data_matrix_un[j][k]= (checking_data_matrix[j][(k+1)*In_vect_n +k])* (minmaxc[(2*k) +1] - minmaxc[2*k]) + minmaxc[2*k];
}
}
// the following code calculates the cdavg_un and checking datat average bothe un normalized
for(k=0; k< Out_n; k++)
{
for(j=0; j< checking_data_n; j++)
{
checking_data_average_un = checking_data_average_un + checking_data_matrix_un[j][k];
}
cdavg_un[k]=checking_data_average_un/checking_data_n;
checking_data_average_un_temp=checking_data_average_un_temp+checking_data_average_un;
checking_data_average_un=0;
}
checking_data_average_un = checking_data_average_un_temp/(Out_n*checking_data_n);
printf("%lf is the checking datat average non normalized\n", checking_data_average_un);
// the following code calcuates the chkvar_un un normalized
for(k=0; k< Out_n; k++)
{
for(j=0; j< checking_data_n; j++)
{
temp= temp + (checking_data_matrix_un[j][k] - cdavg_un[k])*(checking_data_matrix_un[j][k] - cdavg_un[k]);
}
chkvar_un[k]=temp/(checking_data_n-1);
//checking_variance_un = checking_variance_un + temp;
temp=0;
}
temp =0.0;
// the following code cacluates the un normalized checking varinace
for(j=0; j< checking_data_n; j++)
{
for(k=0; k< Out_n; k++)
{
temp = checking_data_matrix_un[j][k] - checking_data_average_un;
temp = temp*temp;
checking_variance_un = checking_variance_un + temp;
}
}
checking_variance_un = checking_variance_un/((Out_n*checking_data_n)-1);
printf("%lf is the checking variance non normalized \n", checking_variance_un);
temp =0.0;
checking_data_average_n=0.0;
checking_data_average_n_temp=0.0;
// the following code calculates the cdavg and checking data average both normalized
for(k=0; k< Out_n; k++)
{
for(j=0; j< checking_data_n; j++)
{
checking_data_average_n = checking_data_average_n + checking_data_matrix[j][(k+1)*In_vect_n +k];
}
cdavg[k]=checking_data_average_n/checking_data_n;
checking_data_average_n_temp=checking_data_average_n_temp+checking_data_average_n;
checking_data_average_n=0;
}
checking_data_average_n = checking_data_average_n_temp/(Out_n*checking_data_n);
printf("%lf is the checking datat average normalized\n", checking_data_average_n);
temp =0.0;
checking_variance_n =0.0;
// the following code cacluates the normalized checking varinace
for(j=0; j< checking_data_n; j++)
{
for(k=0; k< Out_n; k++)
{
temp = checking_data_matrix[j][(k+1)*In_vect_n +k] - checking_data_average_n;
temp = temp*temp;
checking_variance_n = checking_variance_n + temp;
}
}
checking_variance_n = checking_variance_n/((Out_n*checking_data_n)-1);
temp = 0.0;
printf("%lf is the checking variance normalized \n", checking_variance_n);
// the following code calcuatres the normalized chkvar[k]
temp=0.0;
for(k=0; k< Out_n; k++)
{
for(j=0; j< checking_data_n; j++)
{
temp= temp + (checking_data_matrix[j][(k+1)*In_vect_n +k] - cdavg[k])*(checking_data_matrix[j][(k+1)*In_vect_n +k] - cdavg[k]);
}
chkvar[k]=temp/(checking_data_n-1);
//checking_variance_n = checking_variance_n + temp;
temp=0;
}
NMSE_un = min_chk_RMSE_un * min_chk_RMSE_un / checking_variance_un;
NMSE_n = min_chk_RMSE_n * min_chk_RMSE_n / checking_variance_n;
NMSE_n2 = min_chk_RMSE_n * min_chk_RMSE_n / chkvariance;
NDEI_un = sqrt(NMSE_un);
NDEI_n = sqrt(NMSE_n);
for(k=0;k<Out_n;k++)
{
NMSE[k]=chk_error_n[k]*chk_error_n[k]/chkvar[k];
NDEI[k]=sqrt(NMSE[k]);
unNMSE[k]=chk_error_un[k]*chk_error_un[k]/chkvar_un[k];
unNDEI[k]=sqrt(unNMSE[k]);
}
write_result(min_trn_RMSE_epoch ,Out_n,trn_rmse_error,chk_error_un, chk_error_n, NDEI_un, NMSE_un, NDEI_n, NMSE_n, NMSE, NDEI, unNMSE, unNDEI); //debug.c writes to result.txt about the epoch number at which the stopping was done and the corresponding training RMSE and checking RMSE
printf("Minimum training RMSE is \t %f \t \n",min_trn_RMSE);
printf("Minimum training RMSE epoch is \t %d \n",min_trn_RMSE_epoch);
printf("Minimum training NMSE is \t %f \t \n",min_trnNMSE);
//printf("Minimum training RMSE epoch is \t %d \n",min_trnNMSE_epoch);
//printf("Minimum training RMSE is \t %f \t \n",min_trn_RMSE);
//printf("Minimum training RMSE epoch is \t %d \n",min_trn_RMSE_epoch);
printf("%f \t is the checking RMSE non normalized\n",min_chk_RMSE_un);
printf("%f \t is the checking RMSE normalized\n",min_chk_RMSE_n);
//printf("%f \t is the checking RMSE normalized22222222 \n",min_chk_RMSE_n2);
printf(" checking NMSE non normlized is %f \t NDEI non normalized is %f \n",NMSE_un, NDEI_un);
printf("checking NMSE normalized is %f \t NDEI normalized is %f \n",NMSE_n, NDEI_n);
printf("checking NMSE2 normalized is %f \n",NMSE_n2);
printf("traning data variance is %f \n",trnvariance);
return(0);
/*! \brief
* <pre>
* Purpose
* =======
* ilu_cdrop_row() - Drop some small rows from the previous
* supernode (L-part only).
* </pre>
*/
int ilu_cdrop_row(
superlu_options_t *options, /* options */
int first, /* index of the first column in the supernode */
int last, /* index of the last column in the supernode */
double drop_tol, /* dropping parameter */
int quota, /* maximum nonzero entries allowed */
int *nnzLj, /* in/out number of nonzeros in L(:, 1:last) */
double *fill_tol, /* in/out - on exit, fill_tol=-num_zero_pivots,
* does not change if options->ILU_MILU != SMILU1 */
GlobalLU_t *Glu, /* modified */
float swork[], /* working space
* the length of swork[] should be no less than
* the number of rows in the supernode */
float swork2[], /* working space with the same size as swork[],
* used only by the second dropping rule */
int lastc /* if lastc == 0, there is nothing after the
* working supernode [first:last];
* if lastc == 1, there is one more column after
* the working supernode. */ )
{
register int i, j, k, m1;
register int nzlc; /* number of nonzeros in column last+1 */
register int xlusup_first, xlsub_first;
int m, n; /* m x n is the size of the supernode */
int r = 0; /* number of dropped rows */
register float *temp;
register complex *lusup = (complex *) Glu->lusup;
register int *lsub = Glu->lsub;
register int *xlsub = Glu->xlsub;
register int *xlusup = Glu->xlusup;
register float d_max = 0.0, d_min = 1.0;
int drop_rule = options->ILU_DropRule;
milu_t milu = options->ILU_MILU;
norm_t nrm = options->ILU_Norm;
complex zero = {0.0, 0.0};
complex one = {1.0, 0.0};
complex none = {-1.0, 0.0};
int i_1 = 1;
int inc_diag; /* inc_diag = m + 1 */
int nzp = 0; /* number of zero pivots */
float alpha = pow((double)(Glu->n), -1.0 / options->ILU_MILU_Dim);
xlusup_first = xlusup[first];
xlsub_first = xlsub[first];
m = xlusup[first + 1] - xlusup_first;
n = last - first + 1;
m1 = m - 1;
inc_diag = m + 1;
nzlc = lastc ? (xlusup[last + 2] - xlusup[last + 1]) : 0;
temp = swork - n;
/* Quick return if nothing to do. */
if (m == 0 || m == n || drop_rule == NODROP)
{
*nnzLj += m * n;
return 0;
}
/* basic dropping: ILU(tau) */
for (i = n; i <= m1; )
{
/* the average abs value of ith row */
switch (nrm)
{
case ONE_NORM:
temp[i] = scasum_(&n, &lusup[xlusup_first + i], &m) / (double)n;
break;
case TWO_NORM:
temp[i] = scnrm2_(&n, &lusup[xlusup_first + i], &m)
/ sqrt((double)n);
break;
case INF_NORM:
default:
k = icamax_(&n, &lusup[xlusup_first + i], &m) - 1;
temp[i] = c_abs1(&lusup[xlusup_first + i + m * k]);
break;
}
/* drop small entries due to drop_tol */
if (drop_rule & DROP_BASIC && temp[i] < drop_tol)
{
r++;
/* drop the current row and move the last undropped row here */
if (r > 1) /* add to last row */
{
/* accumulate the sum (for MILU) */
switch (milu)
{
case SMILU_1:
case SMILU_2:
caxpy_(&n, &one, &lusup[xlusup_first + i], &m,
&lusup[xlusup_first + m - 1], &m);
break;
case SMILU_3:
for (j = 0; j < n; j++)
lusup[xlusup_first + (m - 1) + j * m].r +=
c_abs1(&lusup[xlusup_first + i + j * m]);
break;
case SILU:
default:
break;
}
ccopy_(&n, &lusup[xlusup_first + m1], &m,
&lusup[xlusup_first + i], &m);
} /* if (r > 1) */
else /* move to last row */
{
cswap_(&n, &lusup[xlusup_first + m1], &m,
&lusup[xlusup_first + i], &m);
if (milu == SMILU_3)
for (j = 0; j < n; j++) {
lusup[xlusup_first + m1 + j * m].r =
c_abs1(&lusup[xlusup_first + m1 + j * m]);
lusup[xlusup_first + m1 + j * m].i = 0.0;
}
}
lsub[xlsub_first + i] = lsub[xlsub_first + m1];
m1--;
continue;
} /* if dropping */
else
{
if (temp[i] > d_max) d_max = temp[i];
if (temp[i] < d_min) d_min = temp[i];
}
i++;
} /* for */
/* Secondary dropping: drop more rows according to the quota. */
quota = ceil((double)quota / (double)n);
if (drop_rule & DROP_SECONDARY && m - r > quota)
{
register double tol = d_max;
/* Calculate the second dropping tolerance */
if (quota > n)
{
if (drop_rule & DROP_INTERP) /* by interpolation */
{
d_max = 1.0 / d_max; d_min = 1.0 / d_min;
tol = 1.0 / (d_max + (d_min - d_max) * quota / (m - n - r));
}
else /* by quick select */
{
int len = m1 - n + 1;
scopy_(&len, swork, &i_1, swork2, &i_1);
tol = sqselect(len, swork2, quota - n);
#if 0
register int *itemp = iwork - n;
A = temp;
for (i = n; i <= m1; i++) itemp[i] = i;
qsort(iwork, m1 - n + 1, sizeof(int), _compare_);
tol = temp[itemp[quota]];
#endif
}
}
for (i = n; i <= m1; )
{
if (temp[i] <= tol)
{
register int j;
r++;
/* drop the current row and move the last undropped row here */
if (r > 1) /* add to last row */
{
/* accumulate the sum (for MILU) */
switch (milu)
{
case SMILU_1:
case SMILU_2:
caxpy_(&n, &one, &lusup[xlusup_first + i], &m,
&lusup[xlusup_first + m - 1], &m);
break;
case SMILU_3:
for (j = 0; j < n; j++)
lusup[xlusup_first + (m - 1) + j * m].r +=
c_abs1(&lusup[xlusup_first + i + j * m]);
break;
case SILU:
default:
break;
}
ccopy_(&n, &lusup[xlusup_first + m1], &m,
&lusup[xlusup_first + i], &m);
} /* if (r > 1) */
else /* move to last row */
{
cswap_(&n, &lusup[xlusup_first + m1], &m,
&lusup[xlusup_first + i], &m);
if (milu == SMILU_3)
for (j = 0; j < n; j++) {
lusup[xlusup_first + m1 + j * m].r =
c_abs1(&lusup[xlusup_first + m1 + j * m]);
lusup[xlusup_first + m1 + j * m].i = 0.0;
}
}
lsub[xlsub_first + i] = lsub[xlsub_first + m1];
m1--;
temp[i] = temp[m1];
continue;
}
i++;
} /* for */
} /* if secondary dropping */
for (i = n; i < m; i++) temp[i] = 0.0;
if (r == 0)
{
*nnzLj += m * n;
return 0;
}
/* add dropped entries to the diagnal */
if (milu != SILU)
{
register int j;
complex t;
float omega;
for (j = 0; j < n; j++)
{
t = lusup[xlusup_first + (m - 1) + j * m];
if (t.r == 0.0 && t.i == 0.0) continue;
omega = SUPERLU_MIN(2.0 * (1.0 - alpha) / c_abs1(&t), 1.0);
cs_mult(&t, &t, omega);
switch (milu)
{
case SMILU_1:
if ( !(c_eq(&t, &none)) ) {
c_add(&t, &t, &one);
cc_mult(&lusup[xlusup_first + j * inc_diag],
&lusup[xlusup_first + j * inc_diag],
&t);
}
else
{
cs_mult(
&lusup[xlusup_first + j * inc_diag],
&lusup[xlusup_first + j * inc_diag],
*fill_tol);
#ifdef DEBUG
printf("[1] ZERO PIVOT: FILL col %d.\n", first + j);
fflush(stdout);
#endif
nzp++;
}
break;
case SMILU_2:
cs_mult(&lusup[xlusup_first + j * inc_diag],
&lusup[xlusup_first + j * inc_diag],
1.0 + c_abs1(&t));
break;
case SMILU_3:
c_add(&t, &t, &one);
cc_mult(&lusup[xlusup_first + j * inc_diag],
&lusup[xlusup_first + j * inc_diag],
&t);
break;
case SILU:
default:
break;
}
}
if (nzp > 0) *fill_tol = -nzp;
}
/* Remove dropped entries from the memory and fix the pointers. */
m1 = m - r;
for (j = 1; j < n; j++)
{
register int tmp1, tmp2;
tmp1 = xlusup_first + j * m1;
tmp2 = xlusup_first + j * m;
for (i = 0; i < m1; i++)
lusup[i + tmp1] = lusup[i + tmp2];
}
for (i = 0; i < nzlc; i++)
lusup[xlusup_first + i + n * m1] = lusup[xlusup_first + i + n * m];
for (i = 0; i < nzlc; i++)
lsub[xlsub[last + 1] - r + i] = lsub[xlsub[last + 1] + i];
for (i = first + 1; i <= last + 1; i++)
{
xlusup[i] -= r * (i - first);
xlsub[i] -= r;
}
if (lastc)
{
xlusup[last + 2] -= r * n;
xlsub[last + 2] -= r;
}
*nnzLj += (m - r) * n;
return r;
}
void schurFactorization(long n, complex **A, complex **T, complex **U)
{
/* Schur factorization: A = U*T*U', T = upper triangular, U = unitary */
long i,j,iter,maxIter;
double tol, diff1,diff2;
complex T11, T12, T21, T22;
complex sigma1, sigma2, sigma;
complex z, z1, z2;
complex **P, **Q, **R;
/* Allocate auxiliary matrices */
P = (complex **)Mem(MEM_ALLOC,n, sizeof(complex *));
Q = (complex **)Mem(MEM_ALLOC,n, sizeof(complex *));
R = (complex **)Mem(MEM_ALLOC,n, sizeof(complex *));
for (i=0; i<n; i++){
P[i] = (complex *)Mem(MEM_ALLOC,n, sizeof(complex));
Q[i] = (complex *)Mem(MEM_ALLOC,n, sizeof(complex));
R[i] = (complex *)Mem(MEM_ALLOC,n, sizeof(complex));
}
/* ------------------------------------------------------------*/
/* Parameters for iteration */
maxIter = 500;
tol = 1E-30;
/* ------------------------------------------------------------*/
/* Init U = eye(n) (identity matrix) */
for (i=0; i<n; i++){
U[i][i].re = 1.0;
U[i][i].im = 0.0;
}
/* ------------------------------------------------------------*/
/* Reduce A to Hessenberg form */
hessFactorization(n,A,P,T);
/* ------------------------------------------------------------*/
/* Compute Schur factorization of Hessenberg matrix T */
for (j=n-1; j>0; j--){ /* Main loop */
for (iter=0; iter<maxIter; iter++){ /* Iteration loop */
sigma.re = T[j][j].re;
sigma.im = T[j][j].im;
/* -- Use Wilkinson shift -- */
/* submatrix considered in the shift */
T11 = T[j-1][j-1];
T12 = T[j-1][j];
T21 = T[j][j-1];
T22 = T[j][j];
/* Compute eigenvalues of submatrix */
z.re = 0.0;
z.im = 0.0;
z2.re = 0.0;
z2.im = 0.0;
/* z = T11*T11 + T22*T22 - 2*T11*T22 + 4*T12*T21 */
z1 = c_mul(T11,T11);
z = c_add(z ,z1);
z2 = c_add(z2,z1);
z1 = c_mul(T22,T22);
z = c_add(z ,z1);
z2 = c_add(z2,z1);
z1 = c_mul(T11,T22);
z1.re = -2.0 * z1.re;
z1.im = -2.0 * z1.im;
z = c_add(z,z1);
z1 = c_mul(T12,T21);
z1.re = 4.0 * z1.re;
z1.im = 4.0 * z1.im;
z = c_add(z,z1);
/* Square root*/
z = c_sqrt(z);
/* Eigenvalues */
sigma1 = c_add(z2,z);
sigma2 = c_sub(z2,z);
/* printf("sigma1 = %e %e\n", sigma1.re, sigma1.im); */
/* printf("sigma2 = %e %e\n", sigma2.re, sigma2.im); */
/* Select eigenvalue for shift*/
diff1 = c_norm( c_sub(T[j][j], sigma1) );
diff2 = c_norm( c_sub(T[j][j], sigma2) );
if (diff1 < diff2){
sigma.re = sigma1.re;
sigma.im = sigma1.im;
}else{
sigma.re = sigma2.re;
sigma.im = sigma2.im;
}
/* --- QR step with Wilkinson shift --- */
/* Shift: T(1:j,1:j) = T(1:j,1:j) - sigma * eye(j) */
for (i=0; i<j+1; i++){
CheckValue(FUNCTION_NAME, "T[i][i].re","", T[i][i].re, -INFTY, INFTY);
CheckValue(FUNCTION_NAME, "T[i][i].im","", T[i][i].im, -INFTY, INFTY);
T[i][i].re = T[i][i].re - sigma.re;
T[i][i].im = T[i][i].im - sigma.im;
}
/* Compute QR factorization of shifted Hessenberg matrix */
for (i=0; i<n; i++){
memset(Q[i], 0, n*sizeof(complex));
memset(R[i], 0, n*sizeof(complex));
}
QRfactorization(n,T,Q,R);
/* T = T_new = R * Q */
for (i=0; i<n; i++){
memset(T[i], 0, n*sizeof(complex));
}
matProduct(n, n, n, R, Q, T);
/* T(1:j,1:j) = T(1:j,1:j) + sigma * eye(j) */
for (i=0; i<j+1; i++){
T[i][i].re = T[i][i].re + sigma.re;
T[i][i].im = T[i][i].im + sigma.im;
}
/* R = U_new = U * Q */
for (i=0; i<n; i++){
memset(R[i], 0, n*sizeof(complex));
}
matProduct(n,n,n,U,Q,R);
/* U = R */
for (i=0; i<n; i++){
memcpy(U[i],R[i], n*sizeof(complex));
}
/* Check convergence */
if (c_norm( T[j][j-1] ) <= tol * (c_norm(T[j-1][j-1]) + c_norm(T[j][j]))){
T[j][j-1].re = 0.0;
T[j][j-1].im = 0.0;
break;
}
} /* end of iter loop */
} /* end of main loop */
/* -------------------------------------------------------------*/
/* U = P*U */
for (i=0; i<n; i++){
memset(U[i], 0, n*sizeof(complex));
}
matProduct(n,n,n,P,R,U);
/* -------------------------------------------------------------*/
/* Free auxiliary variables */
for (i=0; i<n; i++){
Mem(MEM_FREE,P[i]);
Mem(MEM_FREE,Q[i]);
Mem(MEM_FREE,R[i]);
}
Mem(MEM_FREE,P);
Mem(MEM_FREE,Q);
Mem(MEM_FREE,R);
/* Return */
return;
}
void QRfactorization(long n, complex **A, complex **Q, complex **R)
{
/* QR factorization based on Householder transformations */
long i,j,k,m;
double nrm;
complex z,z1,z2;
complex *vj;
/* Init. Q = eye(n) (identity matrix) */
for (i=0; i<n; i++){
Q[i][i].re = 1.0;
Q[i][i].im = 0.0;
}
/* Init. R = A */
for(j=0; j<n; j++){
for (i=0; i<n; i++){
R[i][j].re = A[i][j].re;
R[i][j].im = A[i][j].im;
CheckValue(FUNCTION_NAME, "A[i][j].re","", A[i][j].re, -INFTY, INFTY);
CheckValue(FUNCTION_NAME, "A[i][j].im","", A[i][j].im, -INFTY, INFTY);
}
}
/* Allocate auxiliary variables*/
vj = (complex *)Mem(MEM_ALLOC, n, sizeof(complex));
/* printf("begin calc, n = %d\n", n); */
/* ------------------------------------------------------------*/
for (j=0; j<n; j++){ /* Main loop */
/* R(j:end, j) */
for (i=j; i<n; i++){
vj[i-j].re = R[i][j].re;
vj[i-j].im = R[i][j].im;
}
nrm = vectorNorm(n-j, vj);
/* v(1) = v(1) + sign(R(j,j)) * norm(v) */
vj[0].re = vj[0].re + R[j][j].re / c_norm(R[j][j]) * nrm;
vj[0].im = vj[0].im + R[j][j].im / c_norm(R[j][j]) * nrm;
/* Update norm */
nrm = vectorNorm(n-j, vj);
/* v = v./norm(v) */
for (i=0; i<n-j; i++){
vj[i].re = vj[i].re / nrm;
vj[i].im = vj[i].im / nrm;
}
/* Update */
/* R(j:end, :) = R(j:end,:) - 2 * vj * vj' * R(j:end,:), : */
/* Q(:,j:end) = Q(:,j:end) - 2 * Q(:,j:end) * vj * vj^T */
for (k=0; k<n; k++){
/* (v * v' * A)_ik = v_i SUM_m Conj(v_m) A_mk */
z.re = 0.0;
z.im = 0.0;
for (m=j; m<n; m++){
z1 = c_con(vj[m-j]);
z1 = c_mul(z1, R[m][k]);
z = c_add(z,z1);
}
for (i=j; i<n; i++){
z2 = c_mul(vj[i-j],z);
/* Update R(i,k) */
R[i][k].re = R[i][k].re - 2.0 * z2.re;
R[i][k].im = R[i][k].im - 2.0 * z2.im;
CheckValue(FUNCTION_NAME, "R[i][k].re","", R[i][k].re, -INFTY, INFTY);
CheckValue(FUNCTION_NAME, "R[i][k].im","", R[i][k].im, -INFTY, INFTY);
}
/* (A * v * v^')_ki = v_i * SUM_m Conj(v_m) A_km */
z.re = 0.0;
z.im = 0.0;
for (m=j; m<n; m++){
z1 = vj[m-j];
z1 = c_mul(z1, Q[k][m]);
z = c_add(z,z1);
}
for (i=j; i<n; i++){
z1 = c_con(vj[i-j]);
z2 = c_mul(z1,z);
/* Update Q(k,i)*/
Q[k][i].re = Q[k][i].re - 2.0 * z2.re;
Q[k][i].im = Q[k][i].im - 2.0 * z2.im;
CheckValue(FUNCTION_NAME, "Q[k][i].re","", Q[k][i].re, -INFTY, INFTY);
CheckValue(FUNCTION_NAME, "Q[k][i].im","", Q[k][i].im, -INFTY, INFTY);
}
}
} /* End of main loop (j) */
/* -------------------------------------------------------------*/
/* Free auxiliary variables */
Mem(MEM_FREE, vj);
return;
}
int
ilu_ccopy_to_ucol(
int jcol, /* in */
int nseg, /* in */
int *segrep, /* in */
int *repfnz, /* in */
int *perm_r, /* in */
complex *dense, /* modified - reset to zero on return */
int drop_rule,/* in */
milu_t milu, /* in */
double drop_tol, /* in */
int quota, /* maximum nonzero entries allowed */
complex *sum, /* out - the sum of dropped entries */
int *nnzUj, /* in - out */
GlobalLU_t *Glu, /* modified */
float *work /* working space with minimum size n,
* used by the second dropping rule */
)
{
/*
* Gather from SPA dense[*] to global ucol[*].
*/
int ksub, krep, ksupno;
int i, k, kfnz, segsze;
int fsupc, isub, irow;
int jsupno, nextu;
int new_next, mem_error;
int *xsup, *supno;
int *lsub, *xlsub;
complex *ucol;
int *usub, *xusub;
int nzumax;
int m; /* number of entries in the nonzero U-segments */
register float d_max = 0.0, d_min = 1.0 / dlamch_("Safe minimum");
register double tmp;
complex zero = {0.0, 0.0};
int i_1 = 1;
xsup = Glu->xsup;
supno = Glu->supno;
lsub = Glu->lsub;
xlsub = Glu->xlsub;
ucol = Glu->ucol;
usub = Glu->usub;
xusub = Glu->xusub;
nzumax = Glu->nzumax;
*sum = zero;
if (drop_rule == NODROP) {
drop_tol = -1.0, quota = Glu->n;
}
jsupno = supno[jcol];
nextu = xusub[jcol];
k = nseg - 1;
for (ksub = 0; ksub < nseg; ksub++) {
krep = segrep[k--];
ksupno = supno[krep];
if ( ksupno != jsupno ) { /* Should go into ucol[] */
kfnz = repfnz[krep];
if ( kfnz != EMPTY ) { /* Nonzero U-segment */
fsupc = xsup[ksupno];
isub = xlsub[fsupc] + kfnz - fsupc;
segsze = krep - kfnz + 1;
new_next = nextu + segsze;
while ( new_next > nzumax ) {
if ((mem_error = cLUMemXpand(jcol, nextu, UCOL, &nzumax,
Glu)) != 0)
return (mem_error);
ucol = Glu->ucol;
if ((mem_error = cLUMemXpand(jcol, nextu, USUB, &nzumax,
Glu)) != 0)
return (mem_error);
usub = Glu->usub;
lsub = Glu->lsub;
}
for (i = 0; i < segsze; i++) {
irow = lsub[isub++];
tmp = c_abs1(&dense[irow]);
/* first dropping rule */
if (quota > 0 && tmp >= drop_tol) {
if (tmp > d_max) d_max = tmp;
if (tmp < d_min) d_min = tmp;
usub[nextu] = perm_r[irow];
ucol[nextu] = dense[irow];
nextu++;
} else {
switch (milu) {
case SMILU_1:
case SMILU_2:
c_add(sum, sum, &dense[irow]);
break;
case SMILU_3:
/* *sum += fabs(dense[irow]);*/
sum->r += tmp;
break;
case SILU:
default:
break;
}
#ifdef DEBUG
num_drop_U++;
#endif
}
dense[irow] = zero;
}
}
}
} /* for each segment... */
xusub[jcol + 1] = nextu; /* Close U[*,jcol] */
m = xusub[jcol + 1] - xusub[jcol];
/* second dropping rule */
if (drop_rule & DROP_SECONDARY && m > quota) {
register double tol = d_max;
register int m0 = xusub[jcol] + m - 1;
if (quota > 0) {
if (drop_rule & DROP_INTERP) {
d_max = 1.0 / d_max; d_min = 1.0 / d_min;
tol = 1.0 / (d_max + (d_min - d_max) * quota / m);
} else {
i_1 = xusub[jcol];
for (i = 0; i < m; ++i, ++i_1) work[i] = c_abs1(&ucol[i_1]);
tol = sqselect(m, work, quota);
#if 0
A = &ucol[xusub[jcol]];
for (i = 0; i < m; i++) work[i] = i;
qsort(work, m, sizeof(int), _compare_);
tol = fabs(usub[xusub[jcol] + work[quota]]);
#endif
}
}
for (i = xusub[jcol]; i <= m0; ) {
if (c_abs1(&ucol[i]) <= tol) {
switch (milu) {
case SMILU_1:
case SMILU_2:
c_add(sum, sum, &ucol[i]);
break;
case SMILU_3:
sum->r += tmp;
break;
case SILU:
default:
break;
}
ucol[i] = ucol[m0];
usub[i] = usub[m0];
m0--;
m--;
#ifdef DEBUG
num_drop_U++;
#endif
xusub[jcol + 1]--;
continue;
}
i++;
}
}
if (milu == SMILU_2) {
sum->r = c_abs1(sum); sum->i = 0.0;
}
if (milu == SMILU_3) sum->i = 0.0;
*nnzUj += m;
return 0;
}
/*! \brief Performs one of the matrix-vector operations y := alpha*A*x + beta*y, or y := alpha*A'*x + beta*y
*
* <pre>
* Purpose
* =======
*
* sp_cgemv() performs one of the matrix-vector operations
* y := alpha*A*x + beta*y, or y := alpha*A'*x + beta*y,
* where alpha and beta are scalars, x and y are vectors and A is a
* sparse A->nrow by A->ncol matrix.
*
* Parameters
* ==========
*
* TRANS - (input) char*
* On entry, TRANS specifies the operation to be performed as
* follows:
* TRANS = 'N' or 'n' y := alpha*A*x + beta*y.
* TRANS = 'T' or 't' y := alpha*A'*x + beta*y.
* TRANS = 'C' or 'c' y := alpha*A^H*x + beta*y.
*
* ALPHA - (input) complex
* On entry, ALPHA specifies the scalar alpha.
*
* A - (input) SuperMatrix*
* Before entry, the leading m by n part of the array A must
* contain the matrix of coefficients.
*
* X - (input) complex*, array of DIMENSION at least
* ( 1 + ( n - 1 )*abs( INCX ) ) when TRANS = 'N' or 'n'
* and at least
* ( 1 + ( m - 1 )*abs( INCX ) ) otherwise.
* Before entry, the incremented array X must contain the
* vector x.
*
* INCX - (input) int
* On entry, INCX specifies the increment for the elements of
* X. INCX must not be zero.
*
* BETA - (input) complex
* On entry, BETA specifies the scalar beta. When BETA is
* supplied as zero then Y need not be set on input.
*
* Y - (output) complex*, array of DIMENSION at least
* ( 1 + ( m - 1 )*abs( INCY ) ) when TRANS = 'N' or 'n'
* and at least
* ( 1 + ( n - 1 )*abs( INCY ) ) otherwise.
* Before entry with BETA non-zero, the incremented array Y
* must contain the vector y. On exit, Y is overwritten by the
* updated vector y.
*
* INCY - (input) int
* On entry, INCY specifies the increment for the elements of
* Y. INCY must not be zero.
*
* ==== Sparse Level 2 Blas routine.
* </pre>
*/
int
sp_cgemv(char *trans, complex alpha, SuperMatrix *A, complex *x,
int incx, complex beta, complex *y, int incy)
{
/* Local variables */
NCformat *Astore;
complex *Aval;
int info;
complex temp, temp1;
int lenx, leny, i, j, irow;
int iy, jx, jy, kx, ky;
int notran;
complex comp_zero = {0.0, 0.0};
complex comp_one = {1.0, 0.0};
notran = ( strncmp(trans, "N", 1)==0 || strncmp(trans, "n", 1)==0 );
Astore = A->Store;
Aval = Astore->nzval;
/* Test the input parameters */
info = 0;
if ( !notran && strncmp(trans, "T", 1)!=0 && strncmp(trans, "C", 1)!=0)
info = 1;
else if ( A->nrow < 0 || A->ncol < 0 ) info = 3;
else if (incx == 0) info = 5;
else if (incy == 0) info = 8;
if (info != 0) {
input_error("sp_cgemv ", &info);
return 0;
}
/* Quick return if possible. */
if (A->nrow == 0 || A->ncol == 0 ||
c_eq(&alpha, &comp_zero) &&
c_eq(&beta, &comp_one))
return 0;
/* Set LENX and LENY, the lengths of the vectors x and y, and set
up the start points in X and Y. */
if ( notran ) {
lenx = A->ncol;
leny = A->nrow;
} else {
lenx = A->nrow;
leny = A->ncol;
}
if (incx > 0) kx = 0;
else kx = - (lenx - 1) * incx;
if (incy > 0) ky = 0;
else ky = - (leny - 1) * incy;
/* Start the operations. In this version the elements of A are
accessed sequentially with one pass through A. */
/* First form y := beta*y. */
if ( !c_eq(&beta, &comp_one) ) {
if (incy == 1) {
if ( c_eq(&beta, &comp_zero) )
for (i = 0; i < leny; ++i) y[i] = comp_zero;
else
for (i = 0; i < leny; ++i)
cc_mult(&y[i], &beta, &y[i]);
} else {
iy = ky;
if ( c_eq(&beta, &comp_zero) )
for (i = 0; i < leny; ++i) {
y[iy] = comp_zero;
iy += incy;
}
else
for (i = 0; i < leny; ++i) {
cc_mult(&y[iy], &beta, &y[iy]);
iy += incy;
}
}
}
if ( c_eq(&alpha, &comp_zero) ) return 0;
if ( notran ) {
/* Form y := alpha*A*x + y. */
jx = kx;
if (incy == 1) {
for (j = 0; j < A->ncol; ++j) {
if ( !c_eq(&x[jx], &comp_zero) ) {
cc_mult(&temp, &alpha, &x[jx]);
for (i = Astore->colptr[j]; i < Astore->colptr[j+1]; ++i) {
irow = Astore->rowind[i];
cc_mult(&temp1, &temp, &Aval[i]);
c_add(&y[irow], &y[irow], &temp1);
}
}
jx += incx;
}
} else {
ABORT("Not implemented.");
}
} else if (strncmp(trans, "T", 1) == 0 || strncmp(trans, "t", 1) == 0) {
/* Form y := alpha*A'*x + y. */
jy = ky;
if (incx == 1) {
for (j = 0; j < A->ncol; ++j) {
temp = comp_zero;
for (i = Astore->colptr[j]; i < Astore->colptr[j+1]; ++i) {
irow = Astore->rowind[i];
cc_mult(&temp1, &Aval[i], &x[irow]);
c_add(&temp, &temp, &temp1);
}
cc_mult(&temp1, &alpha, &temp);
c_add(&y[jy], &y[jy], &temp1);
jy += incy;
}
} else {
ABORT("Not implemented.");
}
} else { /* trans == 'C' or 'c' */
/* Form y := alpha * conj(A) * x + y. */
complex temp2;
jy = ky;
if (incx == 1) {
for (j = 0; j < A->ncol; ++j) {
temp = comp_zero;
for (i = Astore->colptr[j]; i < Astore->colptr[j+1]; ++i) {
irow = Astore->rowind[i];
temp2.r = Aval[i].r;
temp2.i = -Aval[i].i; /* conjugation */
cc_mult(&temp1, &temp2, &x[irow]);
c_add(&temp, &temp, &temp1);
}
cc_mult(&temp1, &alpha, &temp);
c_add(&y[jy], &y[jy], &temp1);
jy += incy;
}
} else {
ABORT("Not implemented.");
}
}
return 0;
} /* sp_cgemv */
int
pcgstrf_column_bmod(
const int pnum, /* process number */
const int jcol, /* current column in the panel */
const int fpanelc,/* first column in the panel */
const int nseg, /* number of s-nodes to update jcol */
int *segrep,/* in */
int *repfnz,/* in */
complex *dense, /* modified */
complex *tempv, /* working array */
pxgstrf_shared_t *pxgstrf_shared, /* modified */
Gstat_t *Gstat /* modified */
)
{
/*
* -- SuperLU MT routine (version 2.0) --
* Lawrence Berkeley National Lab, Univ. of California Berkeley,
* and Xerox Palo Alto Research Center.
* September 10, 2007
*
* Purpose:
* ========
* Performs numeric block updates (sup-col) in topological order.
* It features: col-col, 2cols-col, 3cols-col, and sup-col updates.
* Special processing on the supernodal portion of L\U[*,j].
*
* Return value:
* =============
* 0 - successful return
* > 0 - number of bytes allocated when run out of space
*
*/
#if ( MACH==CRAY_PVP )
_fcd ftcs1 = _cptofcd("L", strlen("L")),
ftcs2 = _cptofcd("N", strlen("N")),
ftcs3 = _cptofcd("U", strlen("U"));
#endif
#ifdef USE_VENDOR_BLAS
int incx = 1, incy = 1;
complex alpha, beta;
#endif
GlobalLU_t *Glu = pxgstrf_shared->Glu; /* modified */
/* krep = representative of current k-th supernode
* fsupc = first supernodal column
* nsupc = no of columns in supernode
* nsupr = no of rows in supernode (used as leading dimension)
* luptr = location of supernodal LU-block in storage
* kfnz = first nonz in the k-th supernodal segment
* no_zeros = no of leading zeros in a supernodal U-segment
*/
complex ukj, ukj1, ukj2;
register int lptr, kfnz, isub, irow, i, no_zeros;
register int luptr, luptr1, luptr2;
int fsupc, nsupc, nsupr, segsze;
int nrow; /* No of rows in the matrix of matrix-vector */
int jsupno, k, ksub, krep, krep_ind, ksupno;
int ufirst, nextlu;
int fst_col; /* First column within small LU update */
int d_fsupc; /* Distance between the first column of the current
panel and the first column of the current snode.*/
int *xsup, *supno;
int *lsub, *xlsub, *xlsub_end;
complex *lusup;
int *xlusup, *xlusup_end;
complex *tempv1;
int mem_error;
register float flopcnt;
complex zero = {0.0, 0.0};
complex one = {1.0, 0.0};
complex none = {-1.0, 0.0};
complex comp_temp, comp_temp1;
xsup = Glu->xsup;
supno = Glu->supno;
lsub = Glu->lsub;
xlsub = Glu->xlsub;
xlsub_end = Glu->xlsub_end;
lusup = Glu->lusup;
xlusup = Glu->xlusup;
xlusup_end = Glu->xlusup_end;
jsupno = supno[jcol];
/*
* For each nonz supernode segment of U[*,j] in topological order
*/
k = nseg - 1;
for (ksub = 0; ksub < nseg; ksub++) {
krep = segrep[k];
k--;
ksupno = supno[krep];
#if ( DEBUGlvel>=2 )
if (jcol==BADCOL)
printf("(%d) pcgstrf_column_bmod[1]: %d, nseg %d, krep %d, jsupno %d, ksupno %d\n",
pnum, jcol, nseg, krep, jsupno, ksupno);
#endif
if ( jsupno != ksupno ) { /* Outside the rectangular supernode */
fsupc = xsup[ksupno];
fst_col = SUPERLU_MAX ( fsupc, fpanelc );
/* Distance from the current supernode to the current panel;
d_fsupc=0 if fsupc >= fpanelc. */
d_fsupc = fst_col - fsupc;
luptr = xlusup[fst_col] + d_fsupc;
lptr = xlsub[fsupc] + d_fsupc;
kfnz = repfnz[krep];
kfnz = SUPERLU_MAX ( kfnz, fpanelc );
segsze = krep - kfnz + 1;
nsupc = krep - fst_col + 1;
nsupr = xlsub_end[fsupc] - xlsub[fsupc]; /* Leading dimension */
nrow = nsupr - d_fsupc - nsupc;
krep_ind = lptr + nsupc - 1;
flopcnt = segsze * (segsze - 1) + 2 * nrow * segsze;//sj
Gstat->procstat[pnum].fcops += flopcnt;
#if ( DEBUGlevel>=2 )
if (jcol==BADCOL)
printf("(%d) pcgstrf_column_bmod[2]: %d, krep %d, kfnz %d, segsze %d, d_fsupc %d,\
fsupc %d, nsupr %d, nsupc %d\n",
pnum, jcol, krep, kfnz, segsze, d_fsupc, fsupc, nsupr, nsupc);
#endif
/*
* Case 1: Update U-segment of size 1 -- col-col update
*/
if ( segsze == 1 ) {
ukj = dense[lsub[krep_ind]];
luptr += nsupr*(nsupc-1) + nsupc;
for (i = lptr + nsupc; i < xlsub_end[fsupc]; ++i) {
irow = lsub[i];
cc_mult(&comp_temp, &ukj, &lusup[luptr]);
c_sub(&dense[irow], &dense[irow], &comp_temp);
luptr++;
}
} else if ( segsze <= 3 ) {
ukj = dense[lsub[krep_ind]];
luptr += nsupr*(nsupc-1) + nsupc-1;
ukj1 = dense[lsub[krep_ind - 1]];
luptr1 = luptr - nsupr;
if ( segsze == 2 ) { /* Case 2: 2cols-col update */
cc_mult(&comp_temp, &ukj1, &lusup[luptr1]);
c_sub(&ukj, &ukj, &comp_temp);
dense[lsub[krep_ind]] = ukj;
for (i = lptr + nsupc; i < xlsub_end[fsupc]; ++i) {
irow = lsub[i];
luptr++;
luptr1++;
cc_mult(&comp_temp, &ukj, &lusup[luptr]);
cc_mult(&comp_temp1, &ukj1, &lusup[luptr1]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
c_sub(&dense[irow], &dense[irow], &comp_temp);
}
} else { /* Case 3: 3cols-col update */
ukj2 = dense[lsub[krep_ind - 2]];
luptr2 = luptr1 - nsupr;
cc_mult(&comp_temp, &ukj2, &lusup[luptr2-1]);
c_sub(&ukj1, &ukj1, &comp_temp);
cc_mult(&comp_temp, &ukj1, &lusup[luptr1]);
cc_mult(&comp_temp1, &ukj2, &lusup[luptr2]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
c_sub(&ukj, &ukj, &comp_temp);
dense[lsub[krep_ind]] = ukj;
dense[lsub[krep_ind-1]] = ukj1;
for (i = lptr + nsupc; i < xlsub_end[fsupc]; ++i) {
irow = lsub[i];
luptr++;
luptr1++;
luptr2++;
cc_mult(&comp_temp, &ukj, &lusup[luptr]);
cc_mult(&comp_temp1, &ukj1, &lusup[luptr1]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
cc_mult(&comp_temp1, &ukj2, &lusup[luptr2]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
c_sub(&dense[irow], &dense[irow], &comp_temp);
}
}
} else {
/*
* Case: sup-col update
* Perform a triangular solve and block update,
* then scatter the result of sup-col update to dense
*/
no_zeros = kfnz - fst_col;
/* Copy U[*,j] segment from dense[*] to tempv[*] */
isub = lptr + no_zeros;
for (i = 0; i < segsze; i++) {
irow = lsub[isub];
tempv[i] = dense[irow];
++isub;
}
/* Dense triangular solve -- start effective triangle */
luptr += nsupr * no_zeros + no_zeros;
#ifdef USE_VENDOR_BLAS
#if ( MACH==CRAY_PVP )
CTRSV( ftcs1, ftcs2, ftcs3, &segsze, &lusup[luptr],
&nsupr, tempv, &incx );
#else
ctrsv_( "L", "N", "U", &segsze, &lusup[luptr],
&nsupr, tempv, &incx );
#endif
luptr += segsze; /* Dense matrix-vector */
tempv1 = &tempv[segsze];
alpha = one;
beta = zero;
#if ( MACH==CRAY_PVP )
CGEMV( ftcs2, &nrow, &segsze, &alpha, &lusup[luptr],
&nsupr, tempv, &incx, &beta, tempv1, &incy );
#else
cgemv_( "N", &nrow, &segsze, &alpha, &lusup[luptr],
&nsupr, tempv, &incx, &beta, tempv1, &incy );
#endif
#else
clsolve ( nsupr, segsze, &lusup[luptr], tempv );
luptr += segsze; /* Dense matrix-vector */
tempv1 = &tempv[segsze];
cmatvec (nsupr, nrow , segsze, &lusup[luptr], tempv, tempv1);
#endif
/* Scatter tempv[] into SPA dense[*] */
isub = lptr + no_zeros;
for (i = 0; i < segsze; i++) {
irow = lsub[isub];
dense[irow] = tempv[i]; /* Scatter */
tempv[i] = zero;
isub++;
}
/* Scatter tempv1[] into SPA dense[*] */
for (i = 0; i < nrow; i++) {
irow = lsub[isub];
c_sub(&dense[irow], &dense[irow], &tempv1[i]);
tempv1[i] = zero;
++isub;
}
} /* else segsze >= 4 */
} /* if jsupno ... */
} /* for each segment... */
/* ------------------------------------------
Process the supernodal portion of L\U[*,j]
------------------------------------------ */
fsupc = SUPER_FSUPC (jsupno);
nsupr = xlsub_end[fsupc] - xlsub[fsupc];
if ( (mem_error = Glu_alloc(pnum, jcol, nsupr, LUSUP, &nextlu,
pxgstrf_shared)) )
return mem_error;
xlusup[jcol] = nextlu;
lusup = Glu->lusup;
/* Gather the nonzeros from SPA dense[*,j] into L\U[*,j] */
for (isub = xlsub[fsupc]; isub < xlsub_end[fsupc]; ++isub) {
irow = lsub[isub];
lusup[nextlu] = dense[irow];
dense[irow] = zero;
#ifdef DEBUG
if (jcol == -1)
printf("(%d) pcgstrf_column_bmod[lusup] jcol %d, irow %d, lusup %.10e\n",
pnum, jcol, irow, lusup[nextlu]);
#endif
++nextlu;
}
xlusup_end[jcol] = nextlu; /* close L\U[*,jcol] */
#if ( DEBUGlevel>=2 )
if (jcol == -1) {
nrow = xlusup_end[jcol] - xlusup[jcol];
print_double_vec("before sup-col update", nrow, &lsub[xlsub[fsupc]],
&lusup[xlusup[jcol]]);
}
#endif
/*
* For more updates within the panel (also within the current supernode),
* should start from the first column of the panel, or the first column
* of the supernode, whichever is bigger. There are 2 cases:
* (1) fsupc < fpanelc, then fst_col := fpanelc
* (2) fsupc >= fpanelc, then fst_col := fsupc
*/
fst_col = SUPERLU_MAX ( fsupc, fpanelc );
if ( fst_col < jcol ) {
/* distance between the current supernode and the current panel;
d_fsupc=0 if fsupc >= fpanelc. */
d_fsupc = fst_col - fsupc;
lptr = xlsub[fsupc] + d_fsupc;
luptr = xlusup[fst_col] + d_fsupc;
nsupr = xlsub_end[fsupc] - xlsub[fsupc]; /* Leading dimension */
nsupc = jcol - fst_col; /* Excluding jcol */
nrow = nsupr - d_fsupc - nsupc;
/* points to the beginning of jcol in supernode L\U[*,jsupno] */
ufirst = xlusup[jcol] + d_fsupc;
#if ( DEBUGlevel>=2 )
if (jcol==BADCOL)
printf("(%d) pcgstrf_column_bmod[3] jcol %d, fsupc %d, nsupr %d, nsupc %d, nrow %d\n",
pnum, jcol, fsupc, nsupr, nsupc, nrow);
#endif
flopcnt = nsupc * (nsupc - 1) + 2 * nrow * nsupc; //sj
Gstat->procstat[pnum].fcops += flopcnt;
/* ops[TRSV] += nsupc * (nsupc - 1);
ops[GEMV] += 2 * nrow * nsupc; */
#ifdef USE_VENDOR_BLAS
alpha = none; beta = one; /* y := beta*y + alpha*A*x */
#if ( MACH==CRAY_PVP )
CTRSV( ftcs1, ftcs2, ftcs3, &nsupc, &lusup[luptr],
&nsupr, &lusup[ufirst], &incx );
CGEMV( ftcs2, &nrow, &nsupc, &alpha, &lusup[luptr+nsupc], &nsupr,
&lusup[ufirst], &incx, &beta, &lusup[ufirst+nsupc], &incy );
#else
ctrsv_( "L", "N", "U", &nsupc, &lusup[luptr],
&nsupr, &lusup[ufirst], &incx );
cgemv_( "N", &nrow, &nsupc, &alpha, &lusup[luptr+nsupc], &nsupr,
&lusup[ufirst], &incx, &beta, &lusup[ufirst+nsupc], &incy );
#endif
#else
clsolve ( nsupr, nsupc, &lusup[luptr], &lusup[ufirst] );
cmatvec ( nsupr, nrow, nsupc, &lusup[luptr+nsupc],
&lusup[ufirst], tempv );
/* Copy updates from tempv[*] into lusup[*] */
isub = ufirst + nsupc;
for (i = 0; i < nrow; i++) {
c_sub(&lusup[isub], &lusup[isub], &tempv[i]);
tempv[i] = zero;
++isub;
}
#endif
} /* if fst_col < jcol ... */
return 0;
}
void
cpanel_bmod (
const int m, /* in - number of rows in the matrix */
const int w, /* in */
const int jcol, /* in */
const int nseg, /* in */
complex *dense, /* out, of size n by w */
complex *tempv, /* working array */
int *segrep, /* in */
int *repfnz, /* in, of size n by w */
GlobalLU_t *Glu, /* modified */
SuperLUStat_t *stat /* output */
)
{
#ifdef USE_VENDOR_BLAS
#ifdef _CRAY
_fcd ftcs1 = _cptofcd("L", strlen("L")),
ftcs2 = _cptofcd("N", strlen("N")),
ftcs3 = _cptofcd("U", strlen("U"));
#endif
int incx = 1, incy = 1;
complex alpha, beta;
#endif
register int k, ksub;
int fsupc, nsupc, nsupr, nrow;
int krep, krep_ind;
complex ukj, ukj1, ukj2;
int luptr, luptr1, luptr2;
int segsze;
int block_nrow; /* no of rows in a block row */
register int lptr; /* Points to the row subscripts of a supernode */
int kfnz, irow, no_zeros;
register int isub, isub1, i;
register int jj; /* Index through each column in the panel */
int *xsup, *supno;
int *lsub, *xlsub;
complex *lusup;
int *xlusup;
int *repfnz_col; /* repfnz[] for a column in the panel */
complex *dense_col; /* dense[] for a column in the panel */
complex *tempv1; /* Used in 1-D update */
complex *TriTmp, *MatvecTmp; /* used in 2-D update */
complex zero = {0.0, 0.0};
complex one = {1.0, 0.0};
complex comp_temp, comp_temp1;
register int ldaTmp;
register int r_ind, r_hi;
static int first = 1, maxsuper, rowblk, colblk;
flops_t *ops = stat->ops;
xsup = Glu->xsup;
supno = Glu->supno;
lsub = Glu->lsub;
xlsub = Glu->xlsub;
lusup = Glu->lusup;
xlusup = Glu->xlusup;
if ( first ) {
maxsuper = SUPERLU_MAX( sp_ienv(3), sp_ienv(7) );
rowblk = sp_ienv(4);
colblk = sp_ienv(5);
first = 0;
}
ldaTmp = maxsuper + rowblk;
/*
* For each nonz supernode segment of U[*,j] in topological order
*/
k = nseg - 1;
for (ksub = 0; ksub < nseg; ksub++) { /* for each updating supernode */
/* krep = representative of current k-th supernode
* fsupc = first supernodal column
* nsupc = no of columns in a supernode
* nsupr = no of rows in a supernode
*/
krep = segrep[k--];
fsupc = xsup[supno[krep]];
nsupc = krep - fsupc + 1;
nsupr = xlsub[fsupc+1] - xlsub[fsupc];
nrow = nsupr - nsupc;
lptr = xlsub[fsupc];
krep_ind = lptr + nsupc - 1;
repfnz_col = repfnz;
dense_col = dense;
if ( nsupc >= colblk && nrow > rowblk ) { /* 2-D block update */
TriTmp = tempv;
/* Sequence through each column in panel -- triangular solves */
for (jj = jcol; jj < jcol + w; jj++,
repfnz_col += m, dense_col += m, TriTmp += ldaTmp ) {
kfnz = repfnz_col[krep];
if ( kfnz == EMPTY ) continue; /* Skip any zero segment */
segsze = krep - kfnz + 1;
luptr = xlusup[fsupc];
ops[TRSV] += 4 * segsze * (segsze - 1);
ops[GEMV] += 8 * nrow * segsze;
/* Case 1: Update U-segment of size 1 -- col-col update */
if ( segsze == 1 ) {
ukj = dense_col[lsub[krep_ind]];
luptr += nsupr*(nsupc-1) + nsupc;
for (i = lptr + nsupc; i < xlsub[fsupc+1]; i++) {
irow = lsub[i];
cc_mult(&comp_temp, &ukj, &lusup[luptr]);
c_sub(&dense_col[irow], &dense_col[irow], &comp_temp);
++luptr;
}
} else if ( segsze <= 3 ) {
ukj = dense_col[lsub[krep_ind]];
ukj1 = dense_col[lsub[krep_ind - 1]];
luptr += nsupr*(nsupc-1) + nsupc-1;
luptr1 = luptr - nsupr;
if ( segsze == 2 ) {
cc_mult(&comp_temp, &ukj1, &lusup[luptr1]);
c_sub(&ukj, &ukj, &comp_temp);
dense_col[lsub[krep_ind]] = ukj;
for (i = lptr + nsupc; i < xlsub[fsupc+1]; ++i) {
irow = lsub[i];
luptr++; luptr1++;
cc_mult(&comp_temp, &ukj, &lusup[luptr]);
cc_mult(&comp_temp1, &ukj1, &lusup[luptr1]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
c_sub(&dense_col[irow], &dense_col[irow], &comp_temp);
}
} else {
ukj2 = dense_col[lsub[krep_ind - 2]];
luptr2 = luptr1 - nsupr;
cc_mult(&comp_temp, &ukj2, &lusup[luptr2-1]);
c_sub(&ukj1, &ukj1, &comp_temp);
cc_mult(&comp_temp, &ukj1, &lusup[luptr1]);
cc_mult(&comp_temp1, &ukj2, &lusup[luptr2]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
c_sub(&ukj, &ukj, &comp_temp);
dense_col[lsub[krep_ind]] = ukj;
dense_col[lsub[krep_ind-1]] = ukj1;
for (i = lptr + nsupc; i < xlsub[fsupc+1]; ++i) {
irow = lsub[i];
luptr++; luptr1++; luptr2++;
cc_mult(&comp_temp, &ukj, &lusup[luptr]);
cc_mult(&comp_temp1, &ukj1, &lusup[luptr1]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
cc_mult(&comp_temp1, &ukj2, &lusup[luptr2]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
c_sub(&dense_col[irow], &dense_col[irow], &comp_temp);
}
}
} else { /* segsze >= 4 */
/* Copy U[*,j] segment from dense[*] to TriTmp[*], which
holds the result of triangular solves. */
no_zeros = kfnz - fsupc;
isub = lptr + no_zeros;
for (i = 0; i < segsze; ++i) {
irow = lsub[isub];
TriTmp[i] = dense_col[irow]; /* Gather */
++isub;
}
/* start effective triangle */
luptr += nsupr * no_zeros + no_zeros;
#ifdef USE_VENDOR_BLAS
#ifdef _CRAY
CTRSV( ftcs1, ftcs2, ftcs3, &segsze, &lusup[luptr],
&nsupr, TriTmp, &incx );
#else
ctrsv_( "L", "N", "U", &segsze, &lusup[luptr],
&nsupr, TriTmp, &incx );
#endif
#else
clsolve ( nsupr, segsze, &lusup[luptr], TriTmp );
#endif
} /* else ... */
} /* for jj ... end tri-solves */
/* Block row updates; push all the way into dense[*] block */
for ( r_ind = 0; r_ind < nrow; r_ind += rowblk ) {
r_hi = SUPERLU_MIN(nrow, r_ind + rowblk);
block_nrow = SUPERLU_MIN(rowblk, r_hi - r_ind);
luptr = xlusup[fsupc] + nsupc + r_ind;
isub1 = lptr + nsupc + r_ind;
repfnz_col = repfnz;
TriTmp = tempv;
dense_col = dense;
/* Sequence through each column in panel -- matrix-vector */
for (jj = jcol; jj < jcol + w; jj++,
repfnz_col += m, dense_col += m, TriTmp += ldaTmp) {
kfnz = repfnz_col[krep];
if ( kfnz == EMPTY ) continue; /* Skip any zero segment */
segsze = krep - kfnz + 1;
if ( segsze <= 3 ) continue; /* skip unrolled cases */
/* Perform a block update, and scatter the result of
matrix-vector to dense[]. */
no_zeros = kfnz - fsupc;
luptr1 = luptr + nsupr * no_zeros;
MatvecTmp = &TriTmp[maxsuper];
#ifdef USE_VENDOR_BLAS
alpha = one;
beta = zero;
#ifdef _CRAY
CGEMV(ftcs2, &block_nrow, &segsze, &alpha, &lusup[luptr1],
&nsupr, TriTmp, &incx, &beta, MatvecTmp, &incy);
#else
cgemv_("N", &block_nrow, &segsze, &alpha, &lusup[luptr1],
&nsupr, TriTmp, &incx, &beta, MatvecTmp, &incy);
#endif
#else
cmatvec(nsupr, block_nrow, segsze, &lusup[luptr1],
TriTmp, MatvecTmp);
#endif
/* Scatter MatvecTmp[*] into SPA dense[*] temporarily
* such that MatvecTmp[*] can be re-used for the
* the next blok row update. dense[] will be copied into
* global store after the whole panel has been finished.
*/
isub = isub1;
for (i = 0; i < block_nrow; i++) {
irow = lsub[isub];
c_sub(&dense_col[irow], &dense_col[irow],
&MatvecTmp[i]);
MatvecTmp[i] = zero;
++isub;
}
} /* for jj ... */
} /* for each block row ... */
/* Scatter the triangular solves into SPA dense[*] */
repfnz_col = repfnz;
TriTmp = tempv;
dense_col = dense;
for (jj = jcol; jj < jcol + w; jj++,
repfnz_col += m, dense_col += m, TriTmp += ldaTmp) {
kfnz = repfnz_col[krep];
if ( kfnz == EMPTY ) continue; /* Skip any zero segment */
segsze = krep - kfnz + 1;
if ( segsze <= 3 ) continue; /* skip unrolled cases */
no_zeros = kfnz - fsupc;
isub = lptr + no_zeros;
for (i = 0; i < segsze; i++) {
irow = lsub[isub];
dense_col[irow] = TriTmp[i];
TriTmp[i] = zero;
++isub;
}
} /* for jj ... */
} else { /* 1-D block modification */
/* Sequence through each column in the panel */
for (jj = jcol; jj < jcol + w; jj++,
repfnz_col += m, dense_col += m) {
kfnz = repfnz_col[krep];
if ( kfnz == EMPTY ) continue; /* Skip any zero segment */
segsze = krep - kfnz + 1;
luptr = xlusup[fsupc];
ops[TRSV] += 4 * segsze * (segsze - 1);
ops[GEMV] += 8 * nrow * segsze;
/* Case 1: Update U-segment of size 1 -- col-col update */
if ( segsze == 1 ) {
ukj = dense_col[lsub[krep_ind]];
luptr += nsupr*(nsupc-1) + nsupc;
for (i = lptr + nsupc; i < xlsub[fsupc+1]; i++) {
irow = lsub[i];
cc_mult(&comp_temp, &ukj, &lusup[luptr]);
c_sub(&dense_col[irow], &dense_col[irow], &comp_temp);
++luptr;
}
} else if ( segsze <= 3 ) {
ukj = dense_col[lsub[krep_ind]];
luptr += nsupr*(nsupc-1) + nsupc-1;
ukj1 = dense_col[lsub[krep_ind - 1]];
luptr1 = luptr - nsupr;
if ( segsze == 2 ) {
cc_mult(&comp_temp, &ukj1, &lusup[luptr1]);
c_sub(&ukj, &ukj, &comp_temp);
dense_col[lsub[krep_ind]] = ukj;
for (i = lptr + nsupc; i < xlsub[fsupc+1]; ++i) {
irow = lsub[i];
++luptr; ++luptr1;
cc_mult(&comp_temp, &ukj, &lusup[luptr]);
cc_mult(&comp_temp1, &ukj1, &lusup[luptr1]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
c_sub(&dense_col[irow], &dense_col[irow], &comp_temp);
}
} else {
ukj2 = dense_col[lsub[krep_ind - 2]];
luptr2 = luptr1 - nsupr;
cc_mult(&comp_temp, &ukj2, &lusup[luptr2-1]);
c_sub(&ukj1, &ukj1, &comp_temp);
cc_mult(&comp_temp, &ukj1, &lusup[luptr1]);
cc_mult(&comp_temp1, &ukj2, &lusup[luptr2]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
c_sub(&ukj, &ukj, &comp_temp);
dense_col[lsub[krep_ind]] = ukj;
dense_col[lsub[krep_ind-1]] = ukj1;
for (i = lptr + nsupc; i < xlsub[fsupc+1]; ++i) {
irow = lsub[i];
++luptr; ++luptr1; ++luptr2;
cc_mult(&comp_temp, &ukj, &lusup[luptr]);
cc_mult(&comp_temp1, &ukj1, &lusup[luptr1]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
cc_mult(&comp_temp1, &ukj2, &lusup[luptr2]);
c_add(&comp_temp, &comp_temp, &comp_temp1);
c_sub(&dense_col[irow], &dense_col[irow], &comp_temp);
}
}
} else { /* segsze >= 4 */
/*
* Perform a triangular solve and block update,
* then scatter the result of sup-col update to dense[].
*/
no_zeros = kfnz - fsupc;
/* Copy U[*,j] segment from dense[*] to tempv[*]:
* The result of triangular solve is in tempv[*];
* The result of matrix vector update is in dense_col[*]
*/
isub = lptr + no_zeros;
for (i = 0; i < segsze; ++i) {
irow = lsub[isub];
tempv[i] = dense_col[irow]; /* Gather */
++isub;
}
/* start effective triangle */
luptr += nsupr * no_zeros + no_zeros;
#ifdef USE_VENDOR_BLAS
#ifdef _CRAY
CTRSV( ftcs1, ftcs2, ftcs3, &segsze, &lusup[luptr],
&nsupr, tempv, &incx );
#else
ctrsv_( "L", "N", "U", &segsze, &lusup[luptr],
&nsupr, tempv, &incx );
#endif
luptr += segsze; /* Dense matrix-vector */
tempv1 = &tempv[segsze];
alpha = one;
beta = zero;
#ifdef _CRAY
CGEMV( ftcs2, &nrow, &segsze, &alpha, &lusup[luptr],
&nsupr, tempv, &incx, &beta, tempv1, &incy );
#else
cgemv_( "N", &nrow, &segsze, &alpha, &lusup[luptr],
&nsupr, tempv, &incx, &beta, tempv1, &incy );
#endif
#else
clsolve ( nsupr, segsze, &lusup[luptr], tempv );
luptr += segsze; /* Dense matrix-vector */
tempv1 = &tempv[segsze];
cmatvec (nsupr, nrow, segsze, &lusup[luptr], tempv, tempv1);
#endif
/* Scatter tempv[*] into SPA dense[*] temporarily, such
* that tempv[*] can be used for the triangular solve of
* the next column of the panel. They will be copied into
* ucol[*] after the whole panel has been finished.
*/
isub = lptr + no_zeros;
for (i = 0; i < segsze; i++) {
irow = lsub[isub];
dense_col[irow] = tempv[i];
tempv[i] = zero;
isub++;
}
/* Scatter the update from tempv1[*] into SPA dense[*] */
/* Start dense rectangular L */
for (i = 0; i < nrow; i++) {
irow = lsub[isub];
c_sub(&dense_col[irow], &dense_col[irow], &tempv1[i]);
tempv1[i] = zero;
++isub;
}
} /* else segsze>=4 ... */
} /* for each column in the panel... */
} /* else 1-D update ... */
} /* for each updating supernode ... */
}
int
ilu_cpivotL(
const int jcol, /* in */
const double u, /* in - diagonal pivoting threshold */
int *usepr, /* re-use the pivot sequence given by
* perm_r/iperm_r */
int *perm_r, /* may be modified */
int diagind, /* diagonal of Pc*A*Pc' */
int *swap, /* in/out record the row permutation */
int *iswap, /* in/out inverse of swap, it is the same as
perm_r after the factorization */
int *marker, /* in */
int *pivrow, /* in/out, as an input if *usepr!=0 */
double fill_tol, /* in - fill tolerance of current column
* used for a singular column */
milu_t milu, /* in */
complex drop_sum, /* in - computed in ilu_ccopy_to_ucol()
(MILU only) */
GlobalLU_t *Glu, /* modified - global LU data structures */
SuperLUStat_t *stat /* output */
)
{
int n; /* number of columns */
int fsupc; /* first column in the supernode */
int nsupc; /* no of columns in the supernode */
int nsupr; /* no of rows in the supernode */
int lptr; /* points to the starting subscript of the supernode */
register int pivptr;
int old_pivptr, diag, ptr0;
register float pivmax, rtemp;
float thresh;
complex temp;
complex *lu_sup_ptr;
complex *lu_col_ptr;
int *lsub_ptr;
register int isub, icol, k, itemp;
int *lsub, *xlsub;
complex *lusup;
int *xlusup;
flops_t *ops = stat->ops;
int info;
complex one = {1.0, 0.0};
/* Initialize pointers */
n = Glu->n;
lsub = Glu->lsub;
xlsub = Glu->xlsub;
lusup = Glu->lusup;
xlusup = Glu->xlusup;
fsupc = (Glu->xsup)[(Glu->supno)[jcol]];
nsupc = jcol - fsupc; /* excluding jcol; nsupc >= 0 */
lptr = xlsub[fsupc];
nsupr = xlsub[fsupc+1] - lptr;
lu_sup_ptr = &lusup[xlusup[fsupc]]; /* start of the current supernode */
lu_col_ptr = &lusup[xlusup[jcol]]; /* start of jcol in the supernode */
lsub_ptr = &lsub[lptr]; /* start of row indices of the supernode */
/* Determine the largest abs numerical value for partial pivoting;
Also search for user-specified pivot, and diagonal element. */
pivmax = -1.0;
pivptr = nsupc;
diag = EMPTY;
old_pivptr = nsupc;
ptr0 = EMPTY;
for (isub = nsupc; isub < nsupr; ++isub) {
if (marker[lsub_ptr[isub]] > jcol)
continue; /* do not overlap with a later relaxed supernode */
switch (milu) {
case SMILU_1:
c_add(&temp, &lu_col_ptr[isub], &drop_sum);
rtemp = c_abs1(&temp);
break;
case SMILU_2:
case SMILU_3:
/* In this case, drop_sum contains the sum of the abs. value */
rtemp = c_abs1(&lu_col_ptr[isub]);
break;
case SILU:
default:
rtemp = c_abs1(&lu_col_ptr[isub]);
break;
}
if (rtemp > pivmax) { pivmax = rtemp; pivptr = isub; }
if (*usepr && lsub_ptr[isub] == *pivrow) old_pivptr = isub;
if (lsub_ptr[isub] == diagind) diag = isub;
if (ptr0 == EMPTY) ptr0 = isub;
}
if (milu == SMILU_2 || milu == SMILU_3) pivmax += drop_sum.r;
/* Test for singularity */
if (pivmax < 0.0) {
fprintf(stderr, "[0]: jcol=%d, SINGULAR!!!\n", jcol);
fflush(stderr);
exit(1);
}
if ( pivmax == 0.0 ) {
if (diag != EMPTY)
*pivrow = lsub_ptr[pivptr = diag];
else if (ptr0 != EMPTY)
*pivrow = lsub_ptr[pivptr = ptr0];
else {
/* look for the first row which does not
belong to any later supernodes */
for (icol = jcol; icol < n; icol++)
if (marker[swap[icol]] <= jcol) break;
if (icol >= n) {
fprintf(stderr, "[1]: jcol=%d, SINGULAR!!!\n", jcol);
fflush(stderr);
exit(1);
}
*pivrow = swap[icol];
/* pick up the pivot row */
for (isub = nsupc; isub < nsupr; ++isub)
if ( lsub_ptr[isub] == *pivrow ) { pivptr = isub; break; }
}
pivmax = fill_tol;
lu_col_ptr[pivptr].r = pivmax;
lu_col_ptr[pivptr].i = 0.0;
*usepr = 0;
#ifdef DEBUG
printf("[0] ZERO PIVOT: FILL (%d, %d).\n", *pivrow, jcol);
fflush(stdout);
#endif
info =jcol + 1;
} /* if (*pivrow == 0.0) */
else {
thresh = u * pivmax;
/* Choose appropriate pivotal element by our policy. */
if ( *usepr ) {
switch (milu) {
case SMILU_1:
c_add(&temp, &lu_col_ptr[old_pivptr], &drop_sum);
rtemp = c_abs1(&temp);
break;
case SMILU_2:
case SMILU_3:
rtemp = c_abs1(&lu_col_ptr[old_pivptr]) + drop_sum.r;
break;
case SILU:
default:
rtemp = c_abs1(&lu_col_ptr[old_pivptr]);
break;
}
if ( rtemp != 0.0 && rtemp >= thresh ) pivptr = old_pivptr;
else *usepr = 0;
}
if ( *usepr == 0 ) {
/* Use diagonal pivot? */
if ( diag >= 0 ) { /* diagonal exists */
switch (milu) {
case SMILU_1:
c_add(&temp, &lu_col_ptr[diag], &drop_sum);
rtemp = c_abs1(&temp);
break;
case SMILU_2:
case SMILU_3:
rtemp = c_abs1(&lu_col_ptr[diag]) + drop_sum.r;
break;
case SILU:
default:
rtemp = c_abs1(&lu_col_ptr[diag]);
break;
}
if ( rtemp != 0.0 && rtemp >= thresh ) pivptr = diag;
}
*pivrow = lsub_ptr[pivptr];
}
info = 0;
/* Reset the diagonal */
switch (milu) {
case SMILU_1:
c_add(&lu_col_ptr[pivptr], &lu_col_ptr[pivptr], &drop_sum);
break;
case SMILU_2:
case SMILU_3:
temp = c_sgn(&lu_col_ptr[pivptr]);
cc_mult(&temp, &temp, &drop_sum);
c_add(&lu_col_ptr[pivptr], &lu_col_ptr[pivptr], &drop_sum);
break;
case SILU:
default:
break;
}
} /* else */
/* Record pivot row */
perm_r[*pivrow] = jcol;
if (jcol < n - 1) {
register int t1, t2, t;
t1 = iswap[*pivrow]; t2 = jcol;
if (t1 != t2) {
t = swap[t1]; swap[t1] = swap[t2]; swap[t2] = t;
t1 = swap[t1]; t2 = t;
t = iswap[t1]; iswap[t1] = iswap[t2]; iswap[t2] = t;
}
} /* if (jcol < n - 1) */
/* Interchange row subscripts */
if ( pivptr != nsupc ) {
itemp = lsub_ptr[pivptr];
lsub_ptr[pivptr] = lsub_ptr[nsupc];
lsub_ptr[nsupc] = itemp;
/* Interchange numerical values as well, for the whole snode, such
* that L is indexed the same way as A.
*/
for (icol = 0; icol <= nsupc; icol++) {
itemp = pivptr + icol * nsupr;
temp = lu_sup_ptr[itemp];
lu_sup_ptr[itemp] = lu_sup_ptr[nsupc + icol*nsupr];
lu_sup_ptr[nsupc + icol*nsupr] = temp;
}
} /* if */
/* cdiv operation */
ops[FACT] += 10 * (nsupr - nsupc);
c_div(&temp, &one, &lu_col_ptr[nsupc]);
for (k = nsupc+1; k < nsupr; k++)
cc_mult(&lu_col_ptr[k], &lu_col_ptr[k], &temp);
return info;
}
