[halfoflessthan50p]

Hi.

Im doing a project about Chaos Theory for my physics undergrad course and need a little help with some programming.

I've found a website (here) which has some code for analysing a dataset that would be really useful for me but I don't know how to get it to run.

It looks like this:

Spoiler:

/*******************************************************************

minfo.c -- Eric R. Weeks -- started 2/28/97

does the mutual information algorithm discussed by Fraser & Swinney
(Phys Rev A 33 (1986) p1134-1140)

v01:  2-28-97: taken from shell.c (7/1/96)
			quicksort routine taken from sane.c (original version)
v02:  2-28-97: revised sorting for s[] (different than q[])
			sped up math
v03:  2-28-97: add in tau loop
	 3-01-97: fix for variable number of input; add -b option
v04:  3-01-97: take out chi2 tests for substructure
v05:  3-01-97: realize that with chi2 tests taken out, number()
			function doesn't need to be called very often.  remove
			a[] and b[][] arrays!  Much faster now.

This program is public domain, although please leave my name and
email address attached.

email: weeks@physics.emory.edu
web: http://www.physics.emory.edu/~weeks/

explanation of how to use this program:
    http://www.physics.emory.edu/~weeks/software/minfo.html

 *******************************************************************/
#include <stdio.h>
#include <math.h>
#include <stdlib.h>
#define PI 3.14159265358979323846264338328
#define EE 2.71828182845904523536
#define MAXNUM 100000
#define KMAX 25

int s[MAXNUM],q[MAXNUM];
int numpts,n0,mmax;
float pop[MAXNUM];
int pindex[MAXNUM];
int pow2[KMAX];
int bin;
int globalivalue[4];
float logtwo;

int number(int karray2[],int m2);
float findisq();
float ffunct(int kmarray[],int m);
void saneqsort(int p,int r);
int qpartition_neurons(int p,int r);

main(argc,argv)
int argc;
char **argv;
{
	int c;
	double x,y;
	FILE *fp;
	extern int optind;
	extern char *optarg;
	float array[MAXNUM];
	int taumax,i,j,tau;
	float info;

	bin = -1;
	taumax = 20;
	while ((c = getopt(argc, argv, "ht:b:")) != EOF)
	switch (c)  {
case 'h': 
fprintf(stderr,"Usage: %s [options] [<] [file]\n\n",argv[0]);
fprintf(stderr," -h   : this help message\n");
fprintf(stderr," -t # : set taumax, max tau to find mut. info [%d]\n",taumax);
fprintf(stderr," -b # : bin data into 2^# bins [default: all possible]\n");
			exit(1);
			break;
	case 't': taumax = atoi(optarg);
			break;
	case 'b': bin = atoi(optarg);
			break;
        }

	argc -= (optind-1) ; argv += (optind-1) ;
	fp = (argc > 1) ? fopen(*++argv, "r") : stdin;

	numpts = 0;
	while ((fscanf(fp,"%lf",&x)) == 1)  {
		array[numpts] = x;
		++numpts;
	}	/*----- done reading in data -----*/

	logtwo = 1.0/log(2.0);
	n0=1;
	while ((n0+taumax)<=numpts)  n0 *= 2;
	n0 /= 2;

	/* n0 = numpts - taumax; */
	j = n0;
	for (i=0;i<KMAX;i++)  {
		pow2[i] = j;
		j /= 2;
	}
	mmax = ((int)(log(((float)n0))*logtwo+0.1));
	fprintf(stderr,"n0: %d mmax: %d\n",n0,mmax);


	for (i=0;i<n0;i++)  {
		pop[i] = array[i];
		pindex[i] = i;
	}
	saneqsort(0,n0-1);
	/* for (i=0;i<numpts-taumax;i++)  s[pindex[i]] = i; */
	/* WARNING!!!!!! Note that this definition is somewhat opposite
	 * of what 'makes sense' but will make number() routine faster */
	for (i=0;i<n0;i++)  s[i] = pindex[i]; 

	for (tau=0;tau<=taumax;tau++)  {
		/* now do tau offset for q[] array */
		for (i=tau;i<tau+n0;i++)  {
			pop[i-tau] = array[i];
			pindex[i-tau] = i-tau;
		}
		saneqsort(0,n0-1);
		for (i=0;i<n0;i++)  q[pindex[i]] = i;
		/* for (i=0;i<n0;i++)  fprintf(stderr,"%d ",q[i]); */
		/* fprintf(stderr,"\n"); */


		/* assume at this time that s[], q[] contain integers from
		 * 0 to 2^N-1.   q[] is based on the time lagged data from
		 * array[].  s[] is just array[].
		 *
		 * example: s[] = 0 4 5 3 6 1 2 7
		 *          q[] = 7 4 5 3 6 2 0 1
		 *
		 *     o.......      so the first partition is s: 0-3 | 4-7
		 *     ......o.                                q: 0-3 | 4-7
		 *     .....o..
		 *     ....o...      with 3 in LL, 1 in UL, 3 in UR, 1 in LR
		 *     ...o....      quadrants.
		 *     .o......
		 *     .......o      second partition is s: 01 | 23 | 45 | 67
		 *     ..o.....                          q: 01 | 23 | 45 | 67
		 *                   with distribution: 1001
		 *				                    0020
		 *								1100
		 *								0101
		 */

		/* now find I(S,Q) according to formula (19) */
		info=findisq();
		printf("%d %f\n",tau,info);
		fprintf(stderr,"%d %f\n",tau,info);
	}

	exit(0);
}	/* END OF MAIN */



float findisq()
{
	float info;
	int kmarray[KMAX];
	float x,y;

	kmarray[0] = 0;
	x = ((float)n0);
	y = ffunct(kmarray,0);
	info = (1.0/x)*y - log(x)*logtwo;

	return info;
}

float ffunct(int kmarray[],int m)
{
	/* THIS FUNCTION CAN CALL ITSELF RECURSIVELY */
	float value;
	int n,j;
	int temparray[KMAX];

	for (j=0;j<=m;j++)  temparray[j] = kmarray[j];

	n = number(temparray,m);
	
	value = ((float)n);
	if (n<=1)  {
		value = 0.0;
	} else if (n==2)  {
		value = 4.0;
	} else if (m==bin)  {
		/* no substructure */
		value = value*log(value)*logtwo;
	} else {
		/* assume substructure exists */
		value = value*2.0;
		for (j=0;j<=3;j++)  {
			temparray[m+1] = j;
			value += ffunct(temparray,m+1);
		}
	}

	return value;
}


int number(int karray2[],int m2)
{
	/* THIS FUNCTION IS NOT RECURSIVE */
	int ivalue;
	int los,his,loq,hiq;
	register int i,j;

	if (m2>0)  {
		los = 0;loq = 0;
		his = n0; hiq = n0;
		for (i=1;i<=m2;i++)  {
			if (karray2[i]%2==0)  his -= pow2[i];
			else                  los += pow2[i];
			if (karray2[i]<2)     hiq -= pow2[i];
			else                  loq += pow2[i];
		}
		ivalue = 0;
		for (i=los;i<his;i++)  {
			j = q[s[i]];
			if ((j>=loq)&&(j<hiq))  ivalue++;
		}
	} else {
		ivalue = n0;
	}

	return ivalue;
}



/* quicksort, taken from sane.c by David E. Moriarty */
/* modified for this purpose */

void saneqsort(p,r)
  int p,r;
{
  int q;
  if (p<r) {
    q = qpartition_neurons(p,r);
    saneqsort(p,q);
    saneqsort(q+1,r);
  }
}

/** partition function for saneqsort.
     **/

int qpartition_neurons(p,r)
  int p,r;
{
  register int i,j;
  float x,temp;
  int tempi;

  x = pop[p];
  i = p - 1;
  j = r + 1;
  while(1) {
    do{
      --j;
    }while(pop[j] < x);
    do{
      ++i;
    }while(pop[i] > x);
    if (i < j) {
	 /* here's where the in-place swap takes place */
	 /* fortunately pindex[] stores the *original* location */
      temp = pop[i]; pop[i] = pop[j]; pop[j] = temp;
	 tempi = pindex[i]; pindex[i] = pindex[j]; pindex[j] = tempi;
    }
    else
      return j;
  }
}

I dont know much about programming (although I have used C++, can do headers, loops, basic stuff.) the website suggests the code will work on a Unix C compiler but I don't understand what this means. And also it was written quite a few years ago so i dont know if it will be compatible with modern programs i could use to compile and run it.

I can't waste too much time trying to get this to work but if anyone knows of a program I can download for free that would compile and run the code without me having to rewrite any of it that would help me out an enormous amount.

thanks. rep for anyone who manage to help

[Choad]

well it compiled on my comp fine. i assume you use windows... i have no idea about compiling stuff on windows. maybe it would be simplest to set up a cygwin environment (a unix-like environment used from within windows)

http://www.cygwin.com/

[SunderX]

By a 'UNIX Compiler' they probably mean something along the lines of gcc. (GNU C Compiler or something)

As far as I know, gcc is packaged on almost any Linux machine, including ones you might find at your lab.

Basically, log in, bring up the command line (right clicking the desktop usually brings up a menu to do this) and type something along the lines of: gcc <file>, where <file> is a text file with the code in.

It will compile the code to a file called a.out, which you can execute by typing ./a.out

If I remember correctly, Cygwin (the link above) has a Windows port of gcc, and should also work if you don't want to go to the hassle of finding a Linux box.

Also, any C compiler should work in theory provided the code they wrote is standards-compliant. gcc is pretty flexible either way.

[halfoflessthan50p]

okay ive downloaded cygwin and the gcc package.

i get the error

gcc: minfo5: no such file or directory
gcc: no input files

I guess this means the file is in the wrong format or location?

Ive tried putting copies of the code as .txt and .c files into the directories c://cygwin, c://cygwin/home and c://cygwin/home/administrator?

Sorry if im being stupid here, ive never done this kind of thing before.

[INTit]

Get choad to send you the binary :P

[laser]

Originally Posted by halfoflessthan50p

okay ive downloaded cygwin and the gcc package.

i get the error

gcc: minfo5: no such file or directory
gcc: no input files

I guess this means the file is in the wrong format or location?

Ive tried putting copies of the code as .txt and .c files into the directories c://cygwin, c://cygwin/home and c://cygwin/home/administrator?

Sorry if im being stupid here, ive never done this kind of thing before.

Sounds like you forgot the ".c". You probably want to do something like:

gcc -o minfo5.exe minfo5.c

[Choad]

Originally Posted by INTit

Get choad to send you the binary :P

cygwin doesn't help u run native linux binaries in windows it just allows you to compile linux stuff for windows if you have the source

[Kyle_S-C]

Originally Posted by halfoflessthan50p

Hi.

Im doing a project about Chaos Theory for my physics undergrad course and need a little help with some programming.

I've found a website (here) which has some code for analysing a dataset that would be really useful for me but I don't know how to get it to run.

It looks like this:

Spoiler:

/*******************************************************************

minfo.c -- Eric R. Weeks -- started 2/28/97

does the mutual information algorithm discussed by Fraser & Swinney
(Phys Rev A 33 (1986) p1134-1140)

v01:  2-28-97: taken from shell.c (7/1/96)
			quicksort routine taken from sane.c (original version)
v02:  2-28-97: revised sorting for s[] (different than q[])
			sped up math
v03:  2-28-97: add in tau loop
	 3-01-97: fix for variable number of input; add -b option
v04:  3-01-97: take out chi2 tests for substructure
v05:  3-01-97: realize that with chi2 tests taken out, number()
			function doesn't need to be called very often.  remove
			a[] and b[][] arrays!  Much faster now.

This program is public domain, although please leave my name and
email address attached.

email: weeks@physics.emory.edu
web: http://www.physics.emory.edu/~weeks/

explanation of how to use this program:
    http://www.physics.emory.edu/~weeks/software/minfo.html

 *******************************************************************/
#include <stdio.h>
#include <math.h>
#include <stdlib.h>
#define PI 3.14159265358979323846264338328
#define EE 2.71828182845904523536
#define MAXNUM 100000
#define KMAX 25

int s[MAXNUM],q[MAXNUM];
int numpts,n0,mmax;
float pop[MAXNUM];
int pindex[MAXNUM];
int pow2[KMAX];
int bin;
int globalivalue[4];
float logtwo;

int number(int karray2[],int m2);
float findisq();
float ffunct(int kmarray[],int m);
void saneqsort(int p,int r);
int qpartition_neurons(int p,int r);

main(argc,argv)
int argc;
char **argv;
{
	int c;
	double x,y;
	FILE *fp;
	extern int optind;
	extern char *optarg;
	float array[MAXNUM];
	int taumax,i,j,tau;
	float info;

	bin = -1;
	taumax = 20;
	while ((c = getopt(argc, argv, "ht:b:")) != EOF)
	switch (c)  {
case 'h': 
fprintf(stderr,"Usage: %s [options] [<] [file]\n\n",argv[0]);
fprintf(stderr," -h   : this help message\n");
fprintf(stderr," -t # : set taumax, max tau to find mut. info [%d]\n",taumax);
fprintf(stderr," -b # : bin data into 2^# bins [default: all possible]\n");
			exit(1);
			break;
	case 't': taumax = atoi(optarg);
			break;
	case 'b': bin = atoi(optarg);
			break;
        }

	argc -= (optind-1) ; argv += (optind-1) ;
	fp = (argc > 1) ? fopen(*++argv, "r") : stdin;

	numpts = 0;
	while ((fscanf(fp,"%lf",&x)) == 1)  {
		array[numpts] = x;
		++numpts;
	}	/*----- done reading in data -----*/

	logtwo = 1.0/log(2.0);
	n0=1;
	while ((n0+taumax)<=numpts)  n0 *= 2;
	n0 /= 2;

	/* n0 = numpts - taumax; */
	j = n0;
	for (i=0;i<KMAX;i++)  {
		pow2[i] = j;
		j /= 2;
	}
	mmax = ((int)(log(((float)n0))*logtwo+0.1));
	fprintf(stderr,"n0: %d mmax: %d\n",n0,mmax);


	for (i=0;i<n0;i++)  {
		pop[i] = array[i];
		pindex[i] = i;
	}
	saneqsort(0,n0-1);
	/* for (i=0;i<numpts-taumax;i++)  s[pindex[i]] = i; */
	/* WARNING!!!!!! Note that this definition is somewhat opposite
	 * of what 'makes sense' but will make number() routine faster */
	for (i=0;i<n0;i++)  s[i] = pindex[i]; 

	for (tau=0;tau<=taumax;tau++)  {
		/* now do tau offset for q[] array */
		for (i=tau;i<tau+n0;i++)  {
			pop[i-tau] = array[i];
			pindex[i-tau] = i-tau;
		}
		saneqsort(0,n0-1);
		for (i=0;i<n0;i++)  q[pindex[i]] = i;
		/* for (i=0;i<n0;i++)  fprintf(stderr,"%d ",q[i]); */
		/* fprintf(stderr,"\n"); */


		/* assume at this time that s[], q[] contain integers from
		 * 0 to 2^N-1.   q[] is based on the time lagged data from
		 * array[].  s[] is just array[].
		 *
		 * example: s[] = 0 4 5 3 6 1 2 7
		 *          q[] = 7 4 5 3 6 2 0 1
		 *
		 *     o.......      so the first partition is s: 0-3 | 4-7
		 *     ......o.                                q: 0-3 | 4-7
		 *     .....o..
		 *     ....o...      with 3 in LL, 1 in UL, 3 in UR, 1 in LR
		 *     ...o....      quadrants.
		 *     .o......
		 *     .......o      second partition is s: 01 | 23 | 45 | 67
		 *     ..o.....                          q: 01 | 23 | 45 | 67
		 *                   with distribution: 1001
		 *				                    0020
		 *								1100
		 *								0101
		 */

		/* now find I(S,Q) according to formula (19) */
		info=findisq();
		printf("%d %f\n",tau,info);
		fprintf(stderr,"%d %f\n",tau,info);
	}

	exit(0);
}	/* END OF MAIN */



float findisq()
{
	float info;
	int kmarray[KMAX];
	float x,y;

	kmarray[0] = 0;
	x = ((float)n0);
	y = ffunct(kmarray,0);
	info = (1.0/x)*y - log(x)*logtwo;

	return info;
}

float ffunct(int kmarray[],int m)
{
	/* THIS FUNCTION CAN CALL ITSELF RECURSIVELY */
	float value;
	int n,j;
	int temparray[KMAX];

	for (j=0;j<=m;j++)  temparray[j] = kmarray[j];

	n = number(temparray,m);
	
	value = ((float)n);
	if (n<=1)  {
		value = 0.0;
	} else if (n==2)  {
		value = 4.0;
	} else if (m==bin)  {
		/* no substructure */
		value = value*log(value)*logtwo;
	} else {
		/* assume substructure exists */
		value = value*2.0;
		for (j=0;j<=3;j++)  {
			temparray[m+1] = j;
			value += ffunct(temparray,m+1);
		}
	}

	return value;
}


int number(int karray2[],int m2)
{
	/* THIS FUNCTION IS NOT RECURSIVE */
	int ivalue;
	int los,his,loq,hiq;
	register int i,j;

	if (m2>0)  {
		los = 0;loq = 0;
		his = n0; hiq = n0;
		for (i=1;i<=m2;i++)  {
			if (karray2[i]%2==0)  his -= pow2[i];
			else                  los += pow2[i];
			if (karray2[i]<2)     hiq -= pow2[i];
			else                  loq += pow2[i];
		}
		ivalue = 0;
		for (i=los;i<his;i++)  {
			j = q[s[i]];
			if ((j>=loq)&&(j<hiq))  ivalue++;
		}
	} else {
		ivalue = n0;
	}

	return ivalue;
}



/* quicksort, taken from sane.c by David E. Moriarty */
/* modified for this purpose */

void saneqsort(p,r)
  int p,r;
{
  int q;
  if (p<r) {
    q = qpartition_neurons(p,r);
    saneqsort(p,q);
    saneqsort(q+1,r);
  }
}

/** partition function for saneqsort.
     **/

int qpartition_neurons(p,r)
  int p,r;
{
  register int i,j;
  float x,temp;
  int tempi;

  x = pop[p];
  i = p - 1;
  j = r + 1;
  while(1) {
    do{
      --j;
    }while(pop[j] < x);
    do{
      ++i;
    }while(pop[i] > x);
    if (i < j) {
	 /* here's where the in-place swap takes place */
	 /* fortunately pindex[] stores the *original* location */
      temp = pop[i]; pop[i] = pop[j]; pop[j] = temp;
	 tempi = pindex[i]; pindex[i] = pindex[j]; pindex[j] = tempi;
    }
    else
      return j;
  }
}

I dont know much about programming (although I have used C++, can do headers, loops, basic stuff.) the website suggests the code will work on a Unix C compiler but I don't understand what this means. And also it was written quite a few years ago so i dont know if it will be compatible with modern programs i could use to compile and run it.

I can't waste too much time trying to get this to work but if anyone knows of a program I can download for free that would compile and run the code without me having to rewrite any of it that would help me out an enormous amount.

thanks. rep for anyone who manage to help

Do you have remote access to a university computer system (e.g. ours is called the Private Workstation Facility)? They'll probably run Linux with all the compilers necessary, since I imagine your course will almost certainly include some programming. That'll compile the code for you most likely.

You might want to have a look at your university's computing service website.