Testing Random Number Generator

Hello All,

I have code that generates 32 bytes random numbers taking input as 32 bit value. i.e 2^32 possible inputs.

Now I passed 1-2^32 numbers to generate the random number 2^32 random numbers. I believe random numbers should not be duplicated.

How do I test to make sure that it is truely random and are not duplicated? Any idea.

Currently, I directed the random number result to txt file and developed a perl script that will go tro' each random number and compares with all other results.

Wondering is there any better method as the above method is very time consuimg.

Thanks,

http://www.google.co.uk/search?q=tests+for+randomness
Quick test, generate lots of data and see how well it zips. If it compresses at all, then it isn't random.

if you have a limited sized number, eventually it will come
up with the same number, and in a large enough
section there will be a pattern, the only way is
to seed it with different numbers at each run time.

and that assuming that each run time is short enough
to not reach the pattern in the run.

there something else you could try, but i dont know if it legal
code because i havent tested it so i wont mention it here,
at least not till after i try it.

P.S.
it comming up with the same number doesnt make it
any less random, beable to predict the next number
make it less random.

Can I use WinZip to compress?

Yes the seed is dynamic in nature.

Random numbers do repeat:

Random number = all numbers over the given range are equally possible at every "draw" of a new number.

[edit]
PS: do not use your random number generator for cryptography.[/edit]

Originally Posted by jim mcnamara

[edit]
PS: do not use your random number generator for cryptography.[/edit]

unless of course you are allowing the user to generate themselves a key
in which case it could be useful, as long as you display the key
for them of course.

> Can I use WinZip to compress?
Yes, any commonly available compress program will work.
Save the binary data, not the text version.