Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

[Phys-L] high-quality random number generator

On Feb 22, 2014, at 5:07 PM, Bernard Cleyet <bernard@cleyet.org> wrote:

Is there a “perfect” RNG?

All the noise generators of which I think have flaws.

Two answers:
1) Nothing on this earth is perfect.
2) You can make an RNG that is so close to perfect that is
will be the least of your worries. For example, you can
have one that produces 159.9999 bits of entropy in a 160 bit
word ... with the entropy evenly distributed over all 160
bit positions. For details: http://www.av8n.com/turbid/

On 02/22/2014 04:13 PM, Chuck Britton wrote:

Radioactive decays and Thermal Fluctuations are sometimes used.

Yes. Turbid uses the thermal fluctuations (Johnson noise) in
an electrical resistor.

You need to understand the physics and calibrate the front-end
source, where the entropy really comes from ... and then you
need to worry about a gazillion implementation details, to make
sure the entropy doesn't get mishandled.

Radioactivity and other "quantum" phenomena are no better than
thermal noise. They are no worse except insofar as the details
are harder to handle, which they usually are.

====

By way of contrast, tossing a coin is not nearly as random as
people think it is.
Persi Diaconis, Susan Holmes, and Richard Montgomery
"Dynamical Bias in the Coin Toss"
SIAM Review, 49(2):211-235 (2007)
http://statweb.stanford.edu/~cgates/PERSI/papers/dyn_coin_07.pdf

Dice have their own problems.

Shuffling cards can be made to work if the guy doing the shuffling
/wants/ it to work:
Dave Bayer and Persi Diaconis
"Trailing the Dovetail Shuffle to its Lair"
http://statweb.stanford.edu/~cgates/PERSI/papers/bayer92.pdf


The "powerball" machines that use air to stir the balls might be
OK in principle, based on chaotic dynamics. However, the devil
is in the details, and I don't actually know how well calibrated
and validated those machines are.

The "turbid" approach has the advantage of very high quality, low
cost, and high throughput.