3 ms·
People have been generating this kind of graph of years, often with more meaningful techniques. For example, you can get another dimension of information by loo
by bdr 18y ago
People have been generating this kind of graph of years, often with more meaningful techniques. For example, you can get another dimension of information by looking at the difference between one number and the last one generated, and plotting it on the z- or color axis.
- pmjordan 18y agoHowever, even that isn't enough to measure true randomness. What this article doesn't mention is that you quickly run into trouble trying to visualise non-random behaviour in RNGs. At the higher end, the analysis isn't quite so easy.
- bdr 18y agoYeah, you need actual math to get anywhere. The pictures are mostly for prettiness and visual proof of non-randomness. But even with math you aren't measuring "true randomness". The best you can do is to confirm statistical properties that random information would also have. It's impossible to tell in general whether a string is "true randomness" or, say, a sequence of bits from pi.
- antirez 18y agoFFT and Chi-square tests, simple bits/bytes distributions are all more interesting tests probably but actually a good generator will pass all this tests while being predictable. The only way to check if a PRNG is of cryptographically use is to analyze the way and sources used to generate the output.