I wrote a random graph generation program that generates a /single/ random graph. Typically, a run of the program requires about 50 random integers. As you could have predicted I then had a need to generate a stream of random graphs.
So taking the lazy way I wrapped my graph gen program in a perl script and passed it a seed generated randomly by the perl script. FWIW my perl code to generate the seed was
my ($runCt) = 40000;
srand();
:
my ($seed) = int(rand($runCt*$runCt));
Via the perl script I ran my graph gen program 40,000 times; all 40,000 seeds were unique.
However, of all the graphs generated only 256 of them were unique. Is 256 a coincidence? My suspicion is that RNGs aren't designed to be very "resistant" to different seeds. Can somebody confirm my suspicion or provide a better reason for why I am getting so few "random streams"?
Obviously I could refactor the graph gen program so that with a single seed it generates all 40,000 graphs but I'd be curious to know what is happening here.
Míle buíochais,