/* Program do generacji polaczen synaptycznych
na wyjsciu ma produkowac plik z polaczeniami z zapisana para neuronow waga i
delay. Nastenamepny program (lub druga czesc tego samego, moze wykorzystac to
do generacji struktury neuronow i synaps we wlasciwym formacie.
05/29/2003 P. Kudela, now multi-class supported
07/27/2005 S. Anderson
*/
#define RISC
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/types.h>
#include <time.h>
#include <math.h>
#include "lnet.h"
#define randomm(num) (int)((double)num*(double)random()/((double)(RAND_MAX+1.0)))
#define randomize() srandom((unsigned)time(NULL))
int lrodz,histoint; /* number of neuron classes */
int *NN; /* NN[lrodz] total number of neurons within each class */
int *NOF; /* NOF[lrodz] table of index offsets */
int *NB;
int **N; /* N[lrodz][3] size of subnetwork for each class */
int **k; /* k[lrodz][3] scaling factor to align class networks */
int ***o; /* o[lrodz][lrodz][3] neighborhood array */
int **s; /* s[lrodz][lrodz] synapsy na we/wy z otoczenie */
int **w,**sdw; /* w[lrodz][lrodz] sdw[][] synaptic weights and dispersions */
int **d,**sdd; /* d[lrodz][lrodz] sdd[][] synaptic delays and dispersions */
int *tab; /* temporary neighborhood array tab[maxot]*/
int get_kind(int nr)
{
int kind=0;
while(NOF[++kind]<=nr);
return kind-1;
}
struct SYN_TMP
{
NEURON_INDEX neuron_pre;
NEURON_INDEX neuron_post;
NEURON_INDEX preclass_num;
WEIGHT weight;
DELAY delay;
};
int sort_delay(const void *, const void *);
int sort_neu_num(const void *, const void *);
void set_neu(struct NEURON *);
unsigned seed_save();
FILE *cfg;
int histo_f,out_f,vout_f,noise_f,iext_f,link_f,stim_f; /* flags for simulator */
char flags[100];
void set_flags()
{
/*signed*/ char z;
/*unsigned*/ char buf[100],*l;
l=flags;
histo_f=out_f=vout_f=noise_f=iext_f=link_f=stim_f=0; /* flags for simulator */
fscanf(cfg,"%s",flags);
do
{
sscanf(l,"%c,",&z);
l+=2;
switch(z)
{
case 'N': noise_f = 1;
break;
case 'H': histo_f = 1;
break;
case 'V': vout_f = 1;
break;
case 'O': out_f = 1;
break;
case 'I': iext_f = 1;
break;
case 'L': link_f = 1;
break;
case 'S': stim_f = 1;
iext_f = 1;
case '$': break;
default :
fprintf(stderr,"Unknown option %c in cfg",z);
}
}while (z != '$');
strcpy(buf,"Options:");
if (noise_f) strcat(buf,"NOISE,");
if (iext_f) strcat(buf,"IEXT,");
if (histo_f) strcat(buf,"HISTO,");
if (out_f) strcat(buf,"OUT,");
if (vout_f) strcat(buf,"VOUT,");
if (link_f) strcat(buf,"LINK,");
if (stim_f) strcat(buf,"STIM,");
puts(buf);
}
int main(int argc, char *argv[])
{
int i,j,l,m,n,jp;
int nr,ir,ix,iy,iz; /* index for neurons */
int maxot, tmpot;
int xyz[3]={0,0,0};
int no,nl;
ulong ns,is;
int *neu_pre;
int Nneu;
int torus_flag;
int syn_n;
char label[]="xyz",name[100],fname[100],buff[10],strdummy[10];
FILE *inputfile,*outputfile,*par,*synfile,*neufile,*opar,*nodefile;
struct SYN_TMP *syn_tmp;
struct SYNAPSE syn;
struct NEURON *neu;
struct PARAMETERS p;
int nproc,nsub,iproc,isub,vsub,nproc2,nsub2;
SPIKE_INTERVAL delays[MAX_KIND][MAX_KIND];
int otoczenie(int,int,int*,int);
int otoczenie2(int,int,int*,int);
int losuj(int*,int);
int dummy,gumba;
double testrandnoise;
extern int zapisz(int,int,int,int,FILE *);
extern int sort_neu_num(const void *, const void *);
int **flag;
int **flagv;
if(argc<5)
{fprintf(stderr,"USAGE:\n\tmknodes <inp_name> <out_fname> <sqrt(subnets/node)> <sqrt(nodes)> [<seed>]\n");
return 1;
}
if(argc>5)
srandom((unsigned)atoi(argv[5]));
else
printf("seed %u\n",seed_save()); /* seed for srandom */
strcpy(name,argv[1]);
inputfile = fopen(strcat(name,".sub"),"r");
if(inputfile==NULL){fprintf(stderr, "Subnetwork definition file not present %s\n",name);return -1;}
strcpy(name,argv[1]); /*input file name*/
strcpy(fname,argv[2]); /*output file name*/
nsub2=atoi(argv[3]); /*sqrt subnets per node*/
nproc2=atoi(argv[4]); /*sqrt nodes*/
nsub=nsub2*nsub2; /* number of subnets per node */
nproc=nproc2*nproc2; /* number of nodes */
fscanf(inputfile,"%d",&torus_flag);
fscanf(inputfile,"%d",&lrodz);
cfg=fopen(strcat(name,".cfg"),"r");
if(cfg==NULL)
{fprintf(stderr,"Can't open configuration file %s\n",name);return 2;}
set_flags();
fscanf(cfg,"%d",&dummy);
fscanf(cfg,"%d",&dummy);
fscanf(cfg,"%s",&strdummy);
fscanf(cfg,"%d",&dummy);
fscanf(cfg,"%s",&strdummy);
fscanf(cfg,"%d",&dummy);
fscanf(cfg,"%d,%d",&dummy,&histoint);
/* printf("dummy=%d,histoint=%d\n",dummy,histoint); */
fclose(cfg);
flag=(int**)calloc(nsub,sizeof(int*));
for(i=0;i<nsub;i++)
flag[i]=(int*)calloc(nproc,sizeof(int)); /* Subnet# X Node# flag array */
flagv=(int**)calloc(nsub,sizeof(int*)); /* Subnet# X Node# flag array */
for(i=0;i<nsub;i++)
flagv[i]=(int*)calloc(nproc,sizeof(int));
/* table for rodzajow neurons */
N=(int**)malloc(lrodz*sizeof(int*)); /* table allocation for size of networks for each class */
for(i=0;i<lrodz;i++)
N[i]=(int*)malloc(3*sizeof(int));
k=(int**)malloc(lrodz*3*sizeof(int)); /* table allocation for network scaling factor for each class */
for(i=0;i<lrodz;i++)
k[i]=(int*)malloc(3*sizeof(int));
o=(int ***)malloc(lrodz*sizeof(int **)); /* allocating neighborhood tables */
for(i=0;i<lrodz;i++)
o[i]=(int **)malloc(lrodz*sizeof(int *));
for(i=0;i<lrodz;i++)
for(j=0;j<lrodz;j++)
o[i][j]=(int *)malloc(3*sizeof(int));
s=(int **)malloc(lrodz*sizeof(int*)); /* synaptic table allocation */
for(i=0;i<lrodz;i++)
s[i]=(int*)malloc(lrodz*sizeof(int));
w=(int **)malloc(lrodz*sizeof(int*)); /* allocation for synaptic weight storage */
for(i=0;i<lrodz;i++)
w[i]=(int*)malloc(lrodz*sizeof(int));
sdw=(int **)malloc(lrodz*sizeof(int*)); /* allocation for synaptic weight dispersion */
for(i=0;i<lrodz;i++)
sdw[i]=(int*)malloc(lrodz*sizeof(int));
d=(int **)malloc(lrodz*sizeof(int*)); /* allocation for synaptic delay */
for(i=0;i<lrodz;i++)
d[i]=(int*)malloc(lrodz*sizeof(int));
sdd=(int **)malloc(lrodz*sizeof(int*)); /* allocation for synaptic delay dispersion */
for(i=0;i<lrodz;i++)
sdd[i]=(int*)malloc(lrodz*sizeof(int));
/* input parameters from files to fill arrays */
strcpy(buff,fname); /* copying output file name to 'buff'*/
nodefile = fopen(strcat(buff,".node"),"wt"); /* file used by jhist program for linking histograms */
fprintf(nodefile,"%d\n",lrodz);
for(i=0;i<lrodz;i++)
{
fscanf(inputfile,"%d,",&nr);
fprintf(nodefile,"%d\n",nr);
N[i][0]=N[i][1]=nr; /* square mapping */
N[i][2]=1;
k[i][0]=k[i][1]=k[i][2]=1;
}
fprintf(nodefile,"%d\n",nsub2);
fprintf(nodefile,"%d\n",nproc2);
fclose(nodefile);
fscanf(inputfile,"%d",&nr); /* reading summation check on total number of neurons per subnetwork */
/* setting neighborhood and other input parameters */
for(i=0;i<lrodz;i++)
for(j=0;j<lrodz;j++)
{
fscanf(inputfile,"%d,%d,%d",&o[j][i][0],&o[j][i][1],&o[j][i][2]); /* connection range for given neuron */
fscanf(inputfile,"%d",&s[j][i]); /* number of links from given neuron */
fscanf(inputfile,"%d",&w[j][i]); /* synaptic weights for given neuron */
fscanf(inputfile,"%d",&sdw[j][i]);/* weight dispersion */
fscanf(inputfile,"%d",&d[j][i]); /* synaptic delay */
fscanf(inputfile,"%d",&sdd[j][i]);/* delay dispersion */
}
fclose(inputfile);
/* Setting up noise flags in array */
iproc=(nproc2/2)*(nproc2+1); /* works for even case */
if(nproc%2)
isub=0;
else
isub=(nsub2/2)*(nsub2+1);
/* Old Central Noise Source Routine */
//if(isub=0) /* central noise source for oddXodd node array including case of one processor */
// {
// isub=(nsub2/2)*(nsub2+1);
// iproc = (nproc-1)/2;
// flag[isub-nsub2-1][iproc]=flag[isub-nsub2][iproc]=flag[isub-1][iproc]=flag[isub][iproc]=4; // 4 neurons marked with noise flag
// printf("4 neurons with NOISE in subnets %d,%d,%d,%d in node %d\n",isub-nsub2-1,isub-nsub2,isub-1,isub,iproc);
// }
// else /*sets central four neurons with noise flags in evenXeven case*/
// {
// flag[nsub-1][iproc-nproc2-1]=flag[nsub-nsub2][iproc-nproc2]=flag[nsub2-1][iproc-1]=flag[0][iproc]=4;
// printf("4 neurons with NOISE in subnet/node %d/%d,%d/%d,%d/%d,%d/%d\n",
// nsub-1,iproc-nproc2-1,nsub-nsub2,iproc-nproc2,nsub2-1,iproc-1,0,iproc);
//for(i=0;i<nsub;i++)
// for(j=0;j<nproc;j++)
// flag[i][j]=4;
// }
/* Setting up vout flags in array for use with SVE program */
if(nsub2%2)
vsub=(nsub2*(nsub2-1))/2+(nsub2-1)/2; /* for odd, provides index of middle subnetwork */
else
vsub=nsub2*nsub2/2-nsub2/2;
flagv[vsub][0]=flagv[vsub][nproc2-1]=flagv[vsub][nproc2*(nproc2-1)]=flagv[vsub][nproc-1]=nr; /* nr is the total number of neurons/subnet */
/*
OFFSET CALCULATIONS TO FILL TABLES
*/
NN=(int*)malloc(lrodz*sizeof(long int)); /* table with numbers of neurons within each class */
for(i=0;i<lrodz;i++)
NN[i]=(N[i][0]?N[i][0]:1)*(N[i][1]?N[i][1]:1)*(N[i][2]?N[i][2]:1);
NOF=(int*)malloc((lrodz+1)*sizeof(int)); /* table with offsets */
for(i=1,NOF[0]=0;i<=lrodz;i++)
NOF[i]=NOF[i-1]+NN[i-1];
NB=(int*)calloc(lrodz,sizeof(int)); /* base no of neurons in subnets */
if (NB==NULL){fprintf(stderr,"Insufficient memory for NB\n");exit(1);}
/* maximal neighborhood calculation */
maxot=0;
if(torus_flag!=2)
for(i=0;i<lrodz;i++)
for(j=0;j<lrodz;j++)
{
tmpot=1;
for(l=0;l<3;l++)
tmpot*=2*o[i][j][l]+1;
if(tmpot > maxot)
maxot=tmpot;
}
else
for(i=0;i<lrodz;i++)
{
if(maxot<NN[i])
maxot=NN[i];
}
tab=(int*)malloc(maxot*sizeof(int));
Nneu=0;
for(i=0;i<lrodz;i++)
Nneu+=NN[i]; /* Calculate the total number of neurons in the subnetwork */
if(Nneu!=nr) {fprintf(stderr,"\nerror: check lines 2 & 3 in input file\n");
return 2;
}else
Nneu*=nsub; /* Total number of neurons of all classes per node */
if(Nneu>MAX_NEURON)
{fprintf(stderr,"No of Neurons per node : %d > %d\n",Nneu,MAX_NEURON);
return 2;
}
//printf("Nneu=%i\n",Nneu);
neu=(struct NEURON *)calloc(Nneu,sizeof(struct NEURON));
if (neu==NULL){fprintf(stderr,"Insufficient memory for neu");exit(1);}
for(iproc=0;iproc<nproc;iproc++) /* nproc is number of nodes, big loop over nodes */
{
ulong nsum=0;
sprintf(name,"%s.syncomp.%d",fname,iproc);
if(!(synfile=fopen(name,"w")))
{fprintf(stderr,"Can't open %s\n",fname);return 2;}
sprintf(name,"%s.neutmp.%d",fname,iproc);
if(!(neufile=fopen(name,"w")))
{fprintf(stderr,"Can't open %s\n",fname);return 2;}
for(i=0;i<lrodz;i++)
NB[i]=NOF[i]*(nsub-1);
ns=0; /* number of synapses, talleyed below when writing synapse file */
sprintf(name,"tmpsyn.%d",iproc);
if(!(outputfile=fopen(name,"wb")))
{fprintf(stderr,"Can't open %s\n",name);return 2;}
//outputfile=fopen("syn.tmp","wb");
//if(outputfile==NULL){fprintf(stderr,"\nError in opening output file");exit(1);}
for(isub=0;isub<nsub;isub++)
{
//for(i=0;i<lrodz;i++)
//printf("\nNN[%d]=%d, NOF[%d]=%d",i,NN[i],i,NOF[i]); /* checks */
//printf("\nmaxot=%d,Nneu=%d\n",maxot,Nneu);
/*
3.Neighborhood calculations
*/
/*randomize();*/
nr=0; /* neighborhood counting */
for(ir=0;ir<lrodz;ir++) /* loop over neuron class */
for (ix=0,xyz[0]=0;ix<N[ir][0];ix++,xyz[0]+=k[ir][0]) /* x-loop */
for(iy=0,xyz[1]=0;iy<N[ir][1];iy++,xyz[1]+=k[ir][1]) /* y-loop */
for(iz=0,xyz[2]=0;iz<N[ir][2];iz++,xyz[2]+=k[ir][2],nr++) /* z-loop */
for(jp=0;jp<lrodz;jp++) /* loop on classes of neurons */
{
/* wygeneruj tablice z numerami neuronow rodzaju jp
w otoczeniu neuronu nr (czyli neuronu rodzaju ir,
w polozeniu ix,iy,iz) np.*/
switch(torus_flag) /* tabulating number of neurons available for connection */
{
case 0:
no=otoczenie(nr,ir,xyz,jp); /* no torus boundary condition */
break;
case 1:
no=otoczenie2(nr,ir,xyz,jp); /* torus boundary condition */
break;
case 2: /* case 2 choosing from whole array */
for(i=NOF[jp],no=0;i<NN[jp]+NOF[jp];i++)
if(i!=nr)
tab[no++]=i;
break;
default:
fprintf(stderr,"wrong option for torus flag %d\n",torus_flag);
exit(0);
}
if(!no)continue;
l=(no<s[ir][jp])?no:s[ir][jp];
/* random connection formation */
for(i=0;i<l;i++)
{
/* random connection formation */
nl=losuj(tab,no); /*printf("%d,%d,%d,%d\n",ir,jp, nr,nl);*/
/* filling of tables for delay/synaptic weights */
ns+=zapisz(nl,nr,ir,jp,outputfile); /*writing output to file 0.neu, 0.syn */
/* printf("\npostindex=%d, postclass=%d, preindex=%d, preclass=%d",nr,ir+1,nl,jp+1); */ /* print check variables */
} /* end of drawing */
} /* jp */
for(i=0;i<lrodz;i++)
NB[i]+=NN[i];
} // end of isub
fclose(outputfile);
sprintf(name,"tmpsyn.%d",iproc);
if(!(outputfile=fopen(name,"r")))
{fprintf(stderr,"Can't open %s\n",name);return 2;}
//outputfile=fopen("syn.tmp","r");
syn_tmp=(struct SYN_TMP *)calloc(ns,sizeof(struct SYN_TMP)); /* tablica wszystkch syn_tmp */
if (syn_tmp==NULL){fprintf(stderr,"Can't allocate temporary synapse memory space\n");exit(1);}
fread(syn_tmp,sizeof(struct SYN_TMP),(size_t)ns,outputfile); /* czytaj syn_tmp z syn.tmp */
fclose(outputfile);
//printf("\nns=%d iproc=%d\n",ns,iproc);/* print check variables */
//unlink(name);
//unlink("syn.tmp"); /* syn.tmp nie jest dalej potrzebny */
qsort( (void *)syn_tmp,(size_t)ns,sizeof(struct SYN_TMP),sort_neu_num ); /* sorting synapses by presynaptic neuron index for simulator use
sort_neu_num defined below*/
neu_pre=calloc(Nneu,sizeof(int)); /* table of neurons with number of synapses (presynaptic neurons)*/
is=0;syn_n=0;
for(i=0;i<Nneu;i++)
{
neu_pre[i]=0;
while(i==(syn_tmp+syn_n)->neuron_pre){
syn_n++;
neu_pre[i]++; }
qsort((void *)(syn_tmp+is),(size_t)neu_pre[i],sizeof(struct SYN_TMP), sort_delay ); /* sort by delay */
is=syn_n;
}
for(is=0;is<ns;is++)
{
// int kind=get_kind((syn_tmp+is)->neuron_post);
syn.neuron=(syn_tmp+is)->neuron_post;
syn.weight=(syn_tmp+is)->weight;
syn.delay=(syn_tmp+is)->delay;
//printf("=%d,iproc=%d,i=%d,j=%d\n",isub,iproc,i,j);
syn.preclass_num=(syn_tmp+is)->preclass_num;
fwrite(&syn,sizeof(struct SYNAPSE),1,synfile);
}
nsum+=ns;
free(syn_tmp);
/* teraz neurony */
for(i=0;i<Nneu;i++){
(neu+i)->syn_num=neu_pre[i]; /* preparing to write neuron files, structure NEURONS */
set_neu(neu+i);} /* setting initial voltage and other values */
for(i=0;i<lrodz;i++)
for(j=0;j<NN[i]*nsub;j++)
(neu+NOF[i]*nsub+j)->param=i; /* assigning type of neuron */
/* noise flags */
//for(isub=0;isub<nsub;isub++)
// { l=isub*NN[0];
// if(flag[isub][iproc]){
// for(i=0;i<flag[isub][iproc];i++)
// {
//j=randomm(NN[0])+l;
//j=l+N[0][0]/2*(N[0][0]+1)+i;
// j=l+i;
//j=isub*N[0][0]*N[0][0]+i;
// printf("Noise flags ** isub=%d,iproc=%d,i=%d,j=%d\n",isub,iproc,i,j);
// (neu+j)->flags|=NOISE; /* only 1st kind with NOISE */
// }
// } /* end if */
// }
/* noise out flags */
gumba=0;
for(isub=0;isub<nsub;isub++)
{
l=0;
for(n=0;n<lrodz;n++)
{
for(i=0;i<NN[n];i++)
{
j=l+isub*NN[n]+i;
//printf("Noise flags ** isub=%d,iproc=%d,i=%d,j=%d\n",isub,iproc,i,j);
testrandnoise=(double)random()/((double)RAND_MAX+1);
// printf("testrandnoise=%f\n",testrandnoise);
if( (testrandnoise<0.01) && ( (n==0) || (n==2) ) ){
(neu+j)->flags|=NOISE;
}
}
l+=nsub*NN[n];
}
}
/* vout flags */
for(isub=0;isub<nsub;isub++)
{
//l=isub*NN[0];
//m=nsub*NN[0]+isub*NN[1];
if(flagv[isub][iproc])
{
l=0;
for(n=0;n<lrodz;n++)
{
for(i=0;i<NN[n];i++)
{
j=l+isub*NN[n]+i;
printf("Voltage out flags ** vsub=%d,iproc=%d,i=%d,j=%d,class=%d\n",isub,iproc,i,j,n);
(neu+j)->flags|=VOUT; /* only 1st kind VOUT */
//(neu+j)->flags|=IEXT;
}
l+=nsub*NN[n];
}
}
/*
for(i=0;i<flagv[isub][iproc];i++)
{
j=l+i;
printf("vsub=%d,iproc=%d,i=%d,j=%d\n",isub,iproc,i,j);
(neu+j)->flags|=VOUT; // only 1st kind VOUT
}
for(i=0;i<sqrt(flagv[isub][iproc]);i++) // inhibitorty
{
j=m+i;
printf("vsub=%d,iproc=%d,i=%d,j=%d\n",isub,iproc,i,j);
(neu+j)->flags|=VOUT; // 2nd kind VOUT
}
*/
}
fwrite(neu,sizeof(struct NEURON),Nneu,neufile);
fclose(synfile);
fclose(neufile);
sprintf(name,"%s.cfg.%d",fname,iproc);
cfg=fopen(name,"w"); /* zbior *.cfg */
//fprintf(outputfile,"%s\n","ONSET"); /* creating .cfg file */
fprintf(cfg,"%s\n",flags); /* */
fprintf(cfg,"%s\n",fname); /* */
fprintf(cfg,"%d\n",lrodz);
fprintf(cfg,"%s\n",fname); /* */
fprintf(cfg,"%lu\n",nsum); /* */
fprintf(cfg,"%s\n",fname); /* */
fprintf(cfg,"%d\n",Nneu); /* */
fprintf(cfg,"%d,%d\n",Nneu,histoint); /* histogram parameters-length of bin */
fprintf(cfg,"0\n"); /* out_no*/
fprintf(cfg,"0.00003068"); /* the lambda actually used by simulator, was 3.068X10^-4 */
fclose(cfg);
} // end iproc
/* copy par file */
strcpy(name,argv[1]);
par=fopen(strcat(name,".par"),"r");
if(par==NULL)
{fprintf(stderr,"Can't open %s\n",name);return 2;}
for(i=0;i<nproc;i++)
{
int ni=0;
sprintf(name,"%s.par.%d",argv[2],i);
opar=fopen(name,"w");
if(opar==NULL)
{fprintf(stderr,"Can't open %s\n",name);return 2;}
rewind(par);
while(ni<lrodz && fread(&p,sizeof(struct PARAMETERS),1,par)==1 &&
fread(&delays[ni][0],sizeof( SPIKE_INTERVAL ),lrodz,par)==lrodz)
{
fwrite(&p,sizeof(struct PARAMETERS),1,opar);
fwrite(&delays[ni][0],sizeof( SPIKE_INTERVAL ),lrodz,opar);
ni++;
}
fclose(opar);
}
fclose(par);
/* copy stim file */
if(stim_f)
{
strcpy(name,argv[1]);
par=fopen(strcat(name,".stim"),"r");
if(par==NULL)
{fprintf(stderr,"Can't open %s\n",name);return 2;}
for(i=0;i<nproc;i++)
{
rewind(par);
sprintf(name,"%s.stim.%d",argv[2],i);
opar=fopen(name,"w");
if(opar==NULL)
{fprintf(stderr,"Can't open %s\n",name);return 2;}
while(!feof(opar))
{ char buf[100];
if(fscanf(par,"%[^\n]\n",buf)!=1)break;
fprintf(opar,"%s\n",buf);
}
fclose(opar);
}
fclose(par);
}
printf("\n");
return 0;
} /*koniec main()*/
/*
sort_delay - needed in qsort
*/
int sort_delay(const void * a, const void * b )
{
char delay1[1],delay2[1];
struct SYN_TMP *syn1=NULL, *syn2=NULL;
syn1=(struct SYN_TMP *)a;
syn2=(struct SYN_TMP *)b;
*delay1=syn1->delay;
*delay2=syn2->delay;
return( strcmp(delay1,delay2) );
}
/*
sort_neu_num - needed in qsort()
*/
int sort_neu_num(const void * a, const void * b)
{
NEURON_INDEX neu1,neu2;
struct SYN_TMP *syn_tmp1=NULL,*syn_tmp2=NULL;
syn_tmp1=(struct SYN_TMP *)a;
syn_tmp2=(struct SYN_TMP *)b;
neu1=syn_tmp1->neuron_pre;
neu2=syn_tmp2->neuron_pre;
return(neu1-neu2);
}
/*
4. Generation of neighborhood table:
*/
int rozrzut(int sd)
{
int los;
los=randomm(100);
if(los < 50 )
return( -1*randomm(sd) );
else
return( randomm(sd) );
}
int losuj(int *tablica,int index) /* random connections within identified regions of allowed connections */
{
int i,j;
do
{
i=randomm(index);
}while(tablica[i]<0);
j=tablica[i];
tablica[i]=-1;
return j;
}
int zapisz(int inr, int inl, int i, int j, FILE * outputfile)
{
struct SYN_TMP syn_tmp;
extern int rozrzut(int);
syn_tmp.neuron_post=inl+NB[i];
syn_tmp.neuron_pre=inr+NB[j];
syn_tmp.preclass_num=j;
syn_tmp.weight=w[i][j]+rozrzut(sdw[i][j]);
/* obliczamy najpierw wspolrzedne i[3] neuronu na siatce jp, ktory jest najblizej x,y,z */
syn_tmp.delay =d[i][j]+rozrzut(sdd[i][j]);
//printf("pre= %d,post= %d,preclass= %d, weight= %d,delay= %d\n",syn_tmp.neuron_pre,syn_tmp.neuron_post,syn_tmp.preclass_num,syn_tmp.weight,syn_tmp.delay);
return(fwrite(&syn_tmp,sizeof(struct SYN_TMP),1,outputfile));
}
int otoczenie(int nr ,int ir,int xyz[3],int jp)
{
int nx,ny,no,i[3],ix,iy,iz,mini[3],maxi[3], *ot;
int j;
/* tab must be declared earlier with sufficient size */
ot = o[ir][jp];
/* The next line calculates the coordinates of neurons from one class to the others */
no=0;
for(j=0;j<3;j++)
i[j]=xyz[j]/k[jp][j];
/* No torus boundary condition */
for(j=0;j<3;j++)
{
mini[j]=i[j]-ot[j];
maxi[j]=i[j]+ot[j]+1;
if(mini[j]<0)mini[j]=0;
if(maxi[j]>N[jp][j])maxi[j]=N[jp][j];
}
for(ix=mini[0];ix<maxi[0];ix++)
{ nx=ix*N[jp][1]*N[jp][2]+NOF[jp];
for(iy=mini[1];iy<maxi[1];iy++)
{ ny=nx+iy*N[jp][2];
for(iz=mini[2];iz<maxi[2];iz++)
{
if((j=ny+iz)!=nr)
tab[no++]=j;
}
}
}
return no;
}
int otoczenie2(int nr, int ir,int xyz[3],int jp)
{
int nx,ny,no,i[3],ix,iy,iz,mini[3],maxi[3], *ot;
int j,ii,NX,NY,NZ;
/* Torus boundary condition */
ot = o[ir][jp];
/* calculating location of neurons used to create connections x,y,z */
no=0;
for(j=0;j<3;j++)
i[j]=xyz[j]/k[jp][j];
/* teraz otoczenie
wersja z zawijaniem na torusie */
for(j=0;j<3;j++)
{
mini[j]=i[j]-ot[j];
maxi[j]=i[j]+ot[j]+1;
}
NX=N[jp][0];
NY=N[jp][1];
NZ=N[jp][2];
for(ix=mini[0];ix<maxi[0];ix++)
{ ii=(ix+NX)%NX;
nx=ii*NZ*NY+NOF[jp];
for(iy=mini[1];iy<maxi[1];iy++)
{ ii=(iy+NY)%NY;
ny=nx+ii*NZ;
for(iz=mini[2];iz<maxi[2];iz++)
{
if((j=ny+(iz+NZ)%NZ)!=nr)
tab[no++]=j;
}
}
}
return no;
}
void set_neu(struct NEURON * neu)
{
int j,kk;
switch(neu->param)
{
default:
neu->flags = HISTO;
neu->polarity = 0x0;
for (kk=0;kk<NEXCCLASS;kk++){
neu->Ve_o[kk]=0;
neu->Ve_d[kk]=0;
}
for (kk=0;kk<NINHCLASS;kk++){
neu->Vi_o[kk]=0;
neu->Vi_d[kk]=0;
}
neu->interval=0;
neu->sum_interval=0;
for(j=0;j<SPIKE_LIST_LENGTH;j++)
neu->spikes[j]=0;
neu->first=EMPTY_LIST;
neu->last=0x1;
neu->V=V_0;
neu->W=W_0;
#ifdef CALCIUM
for(j=1;j<7;j++)
{
neu->C[j]=0.;
neu->U[j]=0.;
}
// neu->C=C_0;
neu->X=X_0;
neu->B=B_0;
#endif
}
}
unsigned seed_save()
{
unsigned seed;
seed = (unsigned)time(NULL);
srandom(seed);
return seed;
}