On Feb 16, 2006, at 9:29 AM, Jed Rothwell wrote:
Horace Heffner and improved method of using a thermal amplifier to
produce a random sequence of numbers. Amplifies have been used for
this purpose for a long time I believe. I saw a reference to one in
the 1950s.
They sure have. The problems are (a) speed and (b) sensing the
signal such that the probability of a 0 or 1 is 0.500..., the very
problems addressed in:
http://www.cryptography.com/resources/whitepapers/VIA_rng.pdf.
Anyway, Heffner wrote:
In other words, both the clock timer and the measured interval
must be random and independent. In the case where the time
interval between radioactive disintegration events is used, the
timer flip-flop state needs to be driven by a nonuniform clock,
say by filtered noise from a high gain amplifier.
If you have a method of detecting radioactive disintegration, why
would you use the thermal amplifier technique in the first place?
Speed and cost. Particle detectors are (by comparison) expensive and
have a comparatively long latency between counts. They might also
affected by half-life depending on the isotope used.
Also, wouldn't the radioactive technique be perfectly random?
The counting itself is slow. More importantly, the triggering or
clocking mechanism is going to be biased (by circuit hysteresis) so
as not to obtain an exactly 50/50 split. Even worse, the amount of
the bias will change with temperature and possibly other influences,
like line noise.
I suppose that nowadays with cheap americium smoke detectors
available, it would not be difficult or dangerous to make a
radioactive disintegration random number generator.
- Jed
It would be slow as molasses compared with what is possible, more
expensive, and not achieve the entropy of the new method suggested in
the update at the bottom of the posting. The issues you bring up are
based on the first part of the post, which is just a repeat of vortex
posts from April 2000.
Horace Heffner
