From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rutgers!jvnc.net!darwin.sura.net!cs.ucf.edu!news Tue Jun  9 10:06:25 EDT 1992
Article 6039 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rutgers!jvnc.net!darwin.sura.net!cs.ucf.edu!news
>From: clarke@acme.ucf.edu (Thomas Clarke)
Newsgroups: comp.ai.philosophy
Subject: Quantum computing from sci.physics
Message-ID: <1992Jun2.134530.19168@cs.ucf.edu>
Date: 2 Jun 92 13:45:30 GMT
References: <73218@ut-emx.uucp>
Sender: news@cs.ucf.edu (News system)
Organization: University of Central Florida
Lines: 100

Saw this in sci.physics, somewhat relevant

In article <73218@ut-emx.uucp> johncobb@ut-emx.cc.utexas.edu (John W. Cobb)  
writes:
 In article <1992May29.194627.20577@hellgate.utah.edu>,
 tolman%asylum.utah.edu@cs.utah.edu (Kenneth Tolman) writes:
 |>
 |>  Does quantum computing make sense?  Is not the essence of computing 
 |>deterministic transitions from state to state, which is at the far end
 |>of the court opposite quantum mechanics?  I realize that a fellow named
 |>David Deutsch has been writing about it (I am not sure where, but I 
 |>remember his name).  Am I misinterpreting something about quantum
 |>computing, does it somehow overcome the superposition of states which 
 |>leads to indeterminancy?
 |>  I had always assumed that computation needed to take place in a somewhat
 |>macroscopic domain in which QM effects "cancelled" out.  That is, super
 |>tiny computation would be too subject to noise from its own support
 |>mechanism.
 |>
> A few years ago I had the privilage to sit in on a seminar by Deutsch
> (if my memory serves me right it was he). The idea is that with a normal
> computer you start off in a state and then you let your general Turing
> machine process the state (i.e. run your program) until it halts. You
> make sure your program halts suring verification, etc. A physicist would
> say that you have a dynamical system and the Turing machine formulation can
> be recast as a Hamiltonian time-evolution. (I know some may object to the
> precision of this statement i.e. not all turing machines are equivalent to
> Hamiltonian systems , etc. well just back off for now.)
> 
> So Deutsch's idea is that instead of starting with a classical system
> in an exact initial condition, start with a quantum system prepared
> to be in a superposition of different initial states. Then your Turing
> machine/evolution operator acts to transform your input into output.
> when you are done you will have a superposition of states that are the
> results of the computation on your starting states. For instance, suppose
> you are doing a production run of a large scale simulation (my field of
> work). Usually you do several runs each with differeing input parameters.
> Deutsch might suggest that instead I prepare my input deck to be a
> superposition of all parameter sets that I want answers for. Then when I
> am done, the portion of the wavefunction that corresponds to each input
> will be the answer to that run. That is I can do several runs simultaneously
> in the same calculation. The jest of it is that instead of running 
> the program
> once for each answer as in classical computing one uses quantumn 
> superposition to get many answers for one run.
> 
> Okay, so what's the catch? Well, when you are done, you don't get an
> answer spit out of the computer on a card like in the old Spencer Tracy
> movies. Your answer consists of a quantum state vector that is not in
> a pure eigenstate. So you have to measure it. So to extract your answer or
> answers you have to construct and use a measuring device. If you use a naive
> device such as what was the answer to my first question, then you will (or
> perhaps may depending on your device) get your answer, but in the process,
> you will destroy the information about the solution to the other problems. So
> you're no better off than a classical computer. But wait, there's more. If
> you construct your measuring device differently, you might find out other
> usefule information. Remember the state vector collapses into an 
> eigenfunction of the measuring device. So instead of asking what 
> was the answer to each of
> your computations, you might ask something more general about all of the 
> computations. Then you might be able to extract more information. Now I'm
> a little murky here, but my recollection of Deutsch's remarks were that
> you could construct just such a measuring system for many hypothetical
> problems and you could get more information, however, you also had a 
> probability to measure a an iegentstate of this apparatus that would
> yield no information about your question. Thus now instead of getting
> one answer for each run of the computer, you get a 50% chance of getting
> 2 answers and a 50% chance of getting no information. 
> 
> In a gross sense you have gained nothing, but in another sense you may
> find this useful. For instance, suppose I caould predict the stock market,
> but the program takes 2 days to compute the next day's closing prices. If
> I use a classical computer, I make no money, I'm always a day late. If I use
> a quantum computer then half of the time, I will be able to know the closing
> price beforehand.
> 
> Now all that I have been talking about above was motivated by theoretical
> considerations about Turing machines and quantum evolution operators, etc.
> How would any physical device utilize these ideas? Well I'm out on a limb 
> even further here, but one speculation is to use two superconducting loops
> with weak link josephson junctions that are connected together 
> (linked SQUIDS) Now you don't know if the electron is in one or the 
> other of the loops until
> you actually look. So this might form the basic logic element of a 
> quantum computer much as gates do for present chips.
> 
> It's been over 5 years since I heard Deutsch talk about these things, so
> others out there may have heard more recent versions and might comment on
> how these ideas have matured. I just replied because I didn't see anyone
> else pick up the ball on this cute idea.
> 
> 
> John W. Cobb
> jwc@fusion.ph.utexas.edu

--
Thomas Clarke
Institute for Simulation and Training, University of Central FL
12424 Research Parkway, Suite 300, Orlando, FL 32826
(407)658-5030, FAX: (407)658-5059, clarke@acme.ucf.edu


