ëIf you wanted a random number, historically you could do worse than to pick a sequence from the string of digits in pi. But Purdue University scientists now say other sources might be better.
Physicists including Purdue's Ephraim Fischbach have completed a study comparing the "randomness" in pi to that produced by 30 software random number generators and one chaos-generating physical machine. After conducting several tests, they have found that while sequences of digits from pi are indeed an acceptable source of randomness ââ¬â often an important factor in data encryption and in solving certain physics problems ââ¬â pi's digit string does not always produce randomness as effectively as manufactured generators do.û
- Pi, the naturally random number, may not be the best source of randomness,
and may or may not be cost-efficient, but it's better quality than many intentional artificial attempts.
-
Purdue,
via EurekAlert
;
Earlier work via MAA index of Science News
and LBL.GOV,
2
Maybe I should just FTFL or some such slashdot acronym.
Re:How can something be more random ...
jmm on 2005-04-27T20:41:17
Gradations of randomness are quite meaningful. The degree of randomness is how accurately you can guess what value is coming next. A two-headed coin is non-random, since you know it always comes up heads. A regular coin potentially provides one bit of randomness, since it can come up either heads or tails, and you don't know which. But, it the coin is lopsided it could have a greater tendency to come up on one side than the other, say 50.1% heads and 49.9% tails. That is no longer absolutely random, but it is still quite close. Make the coin curved so that one side is concave and the other is convex and the probabilities could get even more imbalanced.With a random number generator, you can test its output for similar imbalances. Do you get a reasonably even distribution of values? Is there any inconsistancy in the way one value follows another? (Perhaps whenever the value 65535 comes up, the next value is always even.) Are there any longer term trends? Does the sequence of values match the decimal expansion of pi? (Any software based random number generator has the potential to be guessed exactly. If there is hardware involved, it might be predictable too, but there is at least the potential for it to be unpredictable.)
this posting was cited in a scholarly journal. http://interstat.statjournals.net/YEAR/2006/articles/0601001.pdf
