[RC5] Quantum Computing
AZilinskas at SolutionsIQ.com
Fri Apr 16 15:30:05 EDT 1999
Kind of interesting. Not quite sure if it could be adapted
to be used in d.net or not. Maybe all the different
processors available on d.net would simulate the
quantum bits they talk about.
I have no way gotten near to digesting this stuff yet
(the online radio talk that will be available via RealAudio
in a week or so after 4/16 ). They mentioned you
are getting close to understanding this when you get dizzy.
Maybe if I spin my deskchair around twenty times really fast...
This is a repost in that the original form came out in MIME.
Since then I did some dizzy thinking. I may be close or
really off in left field.
Lets see, quantum physics stuff has the state of things are
in probabilities. The Hiesenburg Uncertainy Principle said
that you can measure where a particle is, or its speed/direction
but can't know both at once. The act of measuring fixes
the state, you know a flipped coin is 50% heads 50% tails,
but looking at the coin, it is in its head or tails state.
Well this means something with quantum states could be in any
of its states, where it actually is described by a probability.
Now if you put constraints on the particle, most likely bound in a
molecule or a crystal, the probabilities are adjusted. Do it the
right way, you can form logic-like functions (if the electron of
an hydrogen atom is given a certain spin AND certain energy,
it will have a 90% chance of being in their particular orbit)
Now you just have to set up the situation just right and in
taking a measurement, you get a solution state. One of the
web links showed that they are using NMR (Nucleo Magneto
Resonance) to set stuff like calcium flouride (toothpaste)
spinning in just the right way and the radio wave it emits
gives them an answer.
The closest thing I can think of are Laplace curves. A soap
bubble film over a wire frame will take on the lowest energy
surface on its own. I have heard of an MIT architect that designed
some very swoopy organic structures this way. They also turned
out to be quite stable as the shape never took on a stressful
situation (the soap film would pop or flow to a better configuration).
Now what could this means to d.net ?
Well, given enough work, there may someday be a dollop of
Crest toothpaste that could crack RC5 in the time it takes
to set it up with proper constraints and read its state.
Another possibility which makes me think of speculative
out of order instruction fetching on new processors. If there
is a problem like a local minimum search, d.net could distribute
out all the possible local states out to be solved at the same time.
Better solutions would be saved and a finer search distributed
based upon that.
If I understand the RC5 problem, we cannot tell if we are getting
close, we must plod on all possible keys until one fits. But for
problems like chess moves, we could search all promising moves
and the response moves they generate. The problem is coordinating
this tree search. Hmmm, BigBlue can do chess in that way, has anyone
done the equivalent for the Chinese game Go ?
azilinskas at solutionsiq.com
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest
More information about the rc5