[107929] in Cypherpunks
Re: CDR: A digital way to filter... (fwd)
daemon@ATHENA.MIT.EDU (Bill Stewart)
Fri Jan 29 00:38:16 1999
Date: Thu, 28 Jan 1999 21:23:16 -0800
To: Jim Choate <ravage@einstein.ssz.com>, cypherpunks@einstein.ssz.com
From: Bill Stewart <bill.stewart@pobox.com>
In-Reply-To: <199901251424.IAA06675@einstein.ssz.com>
Reply-To: Bill Stewart <bill.stewart@pobox.com>
Jim Choate wrote:
>Which is why I said it would need to be done in optical way back at the
>beginning of the entire track. I believe you asked me how I'd build it in
>hardware and somehow that's extrapolated into me stating
[....]
>no limit on the number of effective photons in the universe at any point,
You were far less than obvious about making the point that you planned
on building this thing out of unanchored photons rather than matter;
by the time I saw you talking about counters and clock frequencies,
I assumed you meant counters made out of _things_.
If you can find a way to attach more than one counter to an atom
or electron and both set and retrieve its state, I'd be surprised.
(retrieving the superposition of more than one at a time may be possible,
if you're only trying to detect all-0 vs. mixed vs. all-1;
setting multiple counter spaces simultaneously seems like the hard part,
and detecting which of many non-equal counters are which also seems hard.)
But even if you can fit a few quadrillion counters per atom,
you won't fit in this universe if you want it to finish any time soon.
If you can pull off the job strictly by herding photons,
maybe you've got some chance, but I doubt it.
>Opticaly we need a representation of large n that is ameniable to
>being built as a tuned trap. Then take a tunable source (ie tunable dye
>laser) and tune it for each of the known primes less than n. If a standing
>wave forms then n is a multiple of i and n isn't prime.
....
>But all in all I agree, a QM methodology is the most productive. So what is
>needed is a QM counter of modulo n. Pump it n times and have it output a
>zero if any of the sub-counts are zero, indicating a non-Prime n.
You're still proposing architectures that are dependent on time,
if not necessarily on space. As long as you pump it at least i
clock ticks per test prime i, you've lost, because you've only got
2**61 seconds to finish the job (if the universe is closed),
which means you've got to achieve 2**450 clock ticks per second
to test even one moderately interesting prime factor,
which means you need about 2**185 clock ticks per second per atom,
if you're still messing around with atoms.
For 1024-bit RSA moduli, there are about 2**500 potential prime factors;
I'd leave you the choice of generating them on the fly
or calculating them all first, but as discussed in a previous message,
it's difficult to find storage for more than about 2**270 things.
This means you've still lost even if you only need to pump once per prime.
>I happen to be a supporter of David Bohm's Many Worlds hypothesis for about
>the last 30 years.
I'm not convinced, though it's at least attractive, but this instantiation
of me in this universe never did much quantum beyond Schr"odinger equations.
Basically, you need to use a *lot* of universes to get an answer.
There was one suggestion that, if Many Worlds is true,
you can pick an arbitrary solution, blow up the universe if it's wrong,
and know that any remaining universes have the right answer.
>Opticaly we need a representation of large n that is ameniable to
>being built as a tuned trap. Then take a tunable source (ie tunable dye
>laser) and tune it for each of the known primes less than n. If a standing
>wave forms then n is a multiple of i and n isn't prime.
If you can tune your source to a precision of 2**-512,
you still need to try it ~2**500 times, so you still lose.
>One could also take two large n and determine if they shared any commen
>factors. Set the tank circuit to one n and then pump it with the other n.
>If any standing waves form then there is a shared factor.
>I will give you credit for the nice strawman in which you exchange a
>discussion of technical merits into one of economics.
Economics? We're talking about whether using up all the atoms
in the universe will do the job -- we left economics behind long ago,
somewhere around deciding whether a single planet is enough.
This is simple counting. If you wanted to talk quantum physics,
I'd suggest exploring whether Heisenberg's principle affects
the constraints here - Planck's Constant is about 2**-100,
so it still shouldn't be a dealbreaker.
>Tim>No, Bill was (more politely than I can be) reminding you that tradeoffs
>No, Bill was euphamistic. Polite means you don't insult people in the
>first place. Both of you could learn some manners and a lesson on when
>to keep your mouth shut.
Huh? I was going out of my way to address the substance of what
you were saying and question the ideas, rather than insulting you.
Should I be over in Room 12A? *
You are getting close to detweiling here, though....
* http://www.univnorthco.edu/philosophy/python.html
Thanks!
Bill
Bill Stewart, bill.stewart@pobox.com
PGP Fingerprint D454 E202 CBC8 40BF 3C85 B884 0ABE 4639
