Go Down

Topic: Good source of random numbers? (Read 4 times) previous topic - next topic

dkl65

I just skimmed through the thread, so I don't know if someone mentioned this already. How about trying http://code.google.com/p/tinkerit/wiki/TrueRandom. They claim that it is very random. It uses analog input A0, and some other stuff.

wanderson


I just skimmed through the thread, so I don't know if someone mentioned this already. How about trying http://code.google.com/p/tinkerit/wiki/TrueRandom. They claim that it is very random. It uses analog input A0, and some other stuff.

They claim it is random; however, it fails even the simple tests

Code: [Select]

Entropy = 6.440816 bits per byte.

Optimum compression would reduce the size
of this 1,002,424 byte file by 19.49%

Chi square distribution for 1,002,424 samples is 4802982.71, and randomly
would exceed this value 0.00% percent of the time.

Arithetic mean value of data bytes is 100.3535 (127.5 = random)
Serial correlation coefficient is 0.042774 (totally uncorrelated = 0.0).
New true random number library available at: http://code.google.com/p/avr-hardware-random-number-generation/

Current version 1.0.1

xolroc

I'm intrigued as to where you got those test results from... whatever you used for it seems like it would be very useful!

Coding Badly


ent can be downloaded here...
http://www.fourmilab.ch/random/

Just feed it a file of binary data.  Try a few JPGs.

wanderson

As CodingBadly mentioned ent is the source of the test results I posted; however, I made a python script that performed those tests with a few options along with some graphs which I find useful; histogram and scatter.

That python script can be found here; http://code.google.com/p/avr-hardware-random-number-generation/source/browse/analyze.py
New true random number library available at: http://code.google.com/p/avr-hardware-random-number-generation/

Current version 1.0.1

Go Up