tools.c

C-package of "Long Short-Term Memory" for Protein classification

/*
 * Some helper functions
 * 
 * $Id: tools.c 1269 2007-05-09 07:49:08Z mhe $
 *
*/

#include <stdio.h>
#include <stdlib.h>

#include "lstm.h"
#include "tools.h"


int seprand(int k) {
  long l;
  int f;
  l = random();
  f = l % k;
  return(f);
}


/* shuffle a dataset of sequences */
void shuffle (int *a, int num_prot) {
 
 int i,pick;
 int *dd,*ss,b[500000];
 
 
 for (i = 0; i < num_prot; i++) {
    b[i] = i;
 }
 
 for (i = 0; i < num_prot; i++) {
    
    int rand = seprand(num_prot-i); 
    pick = b[rand];
    a[i] = pick;
    
    //
    // "shift" routine here eliminates the "picked" base from b:
    // dd starts at the picked position: ss is one beyond that:
    //
    for( dd = b+rand , ss = dd + 1 ; *dd ; *dd++=*ss++ );
    
 }
}