[vox-tech] random number in C

[Peter Jay Salzman]

begin Bill Broadley <bill@math.ucdavis.edu> 
> On Tue, Apr 02, 2002 at 10:05:34AM -0800, Peter Jay Salzman wrote:
> > i wrote some monte carlo code which obtains a random generator seed
> > through a read to /dev/random.
> > 
> > on a lark, i did a run:
> > 
> > 	T: 4.0e+00  beta: 2.500e-01  trials: 1000000
> > 	N: 10  M: 10  seed: 208006379
> 
> What code did you use to read from /dev/random(provide source)?  
> What exact call did you call with rand() (provide source)?  
> How exactly did you initialize the random number generator (provide source)?
> 
> You did pass the seed to srand() and not rand() right?
 
here's the code:

   int main(int argc, char *argv[])
   {
		unsigned int seed;
      ...
      seed = SeedRandomGenerator();
      ...
      printf("\nT: %.1Le  beta: %.3Le  trials: %lu  N: %u  M: %u  seed: %u\n",
         T, beta, trials, N, M, seed);
   }

and here is my seed function:

   unsigned int SeedRandomGenerator(void)
   {
   	unsigned int seed;
   	FILE *fp;
   
   	if( (fp = fopen("/dev/random", "r")) == NULL)
   		die("Can't open /dev/random for reading.");
   	fread(&seed, sizeof(unsigned int), 1, fp);
   	fclose(fp);
   	srand(seed);
   	return seed;
   }


ok, so i run the program.  here's what i get:

   p@satan% ./ising2-jfunc 4.0
   T: 4.0e+00  beta: 2.500e-01  trials: 1000  N: 10  M: 10  seed: 3497451914
   Average Energy: -1.273893
   Average magnetization: 0.425220
   Average |magnetization|: 0.531100
   Average magnetic susceptibility: 34.746600



so now i comment out the call to SeedRandomGenerator() and replace it by

   int main(int argc, char *argv[])
   {
   	...
      /* seed = SeedRandomGenerator(); */
      seed = 3497451914;
      ...
      printf("\nT: %.1Le  beta: %.3Le  trials: %lu  N: %u  M: %u  seed: %u\n",
         T, beta, trials, N, M, seed);
   }

and here's what i get:

   T: 4.0e+00  beta: 2.500e-01  trials: 1000  N: 10  M: 10  seed: 3497451914
   Average Energy: -1.326747
   Average magnetization: 0.495320
   Average |magnetization|: 0.495320
   Average magnetic susceptibility: 28.645040

the seed is supposedly the same, but i get different results.  from
here, when i re-run the program, i keep getting:

   T: 4.0e+00  beta: 2.500e-01  trials: 1000  N: 10  M: 10  seed: 3497451914
   Average Energy: -1.326747
   Average magnetization: 0.495320
   Average |magnetization|: 0.495320
   Average magnetic susceptibility: 28.645040

which is what i would expect to happen.

i take this to mean that seed doesn't have the same value between the
run with SeedRandomGenerator() and the run with SeedRandomGenerator()
commented out.

but darned if i know.  they look the same.   :(


> > the seed for these two runs is the same.  i would've expected that calls
> > to rand() should've generated the same set of random numbers.  yet the
> > different runs of the monte carlo code yield different results, which is
> > fine, except i was expecting it to be exactly the same since the seed is
> > the same.
> 
> A reasonable expectation.
> 
> If you port to any other platform you should use random and srandom
> which provides for higher quality implemtnations on most platforms
> (besides linux).   It also allows you to set a more complex seed.

i'll keep this in mind.  thanks for the tip.

i can't port this code to other platforms as-is since i have code that
catches SIGFPE, which is linux specific.  i'm thinking of writing a
solaris port so i can run my jobs on isun.


just kidding.   :)

pete