From: Nicole C. Tedesco (phlorgian@hotmail.com)
Date: Wed Dec 30 2009 - 13:16:33 MST
Actually, the best thing about Amit's associative neural networks is that it
is NOT DEPENDENT UPON QUANTUM COMPUTING in order to generate a quantum-like
result in terms of dynamics (not of time)! This mirrors my own academic
studies in the 1980s and that of many others: cellular automata networks
(e.g., neural networks) are really good at simulating single and
multi-particle quantum systems (see also, Wolfram). No surprise, there.
Does D-Wave move the "toy computer" simulation (thanks Jack, those words
have stuck in my head for over 10 years now) into the realm of real quantum
manipulation? This is where D-Wave can even fool its builders: A SIMULATED
QUANTUM COMPUTING ENVIRONMENT MIGHT VERY WELL LOOK LIKE A REAL ONE.
The D-Wave computing system is hybrid, like a GPU, in that the array must be
set up, allowed to compute, results read and integrated into a separately
running classical system. The overall order-of timing of the system is
reduced from the theoretical maximum of a total quantum computing system
(geometric but not quite quadratic) and left somewhere above the current
state of the art of a classical multiprocessing. The precise theoretical
timing for a true quantum/classical hybrid system is difficult to determine
since the expected variance in the computing times would be quite high.
This leaves experimenters with a really difficult task of determining if
D-Wave is actually computing in the quantum realm or merely simulating the
expected spectrum of results. Remember, cellular automata networks are
really good a simulating quantum systems from the perspective of value
expectation.
- Nicole
-----Original Message-----
From: owner-sl4@sl4.org [mailto:owner-sl4@sl4.org] On Behalf Of Matt Mahoney
Sent: Wednesday, December 30, 2009 7:29 AM
To: sl4@sl4.org
Subject: Re: [sl4] D-Wave
The real test is whether D-Wave can break RSA. All forms of public key
encryption including RSA, Diffie-Hellman, Rabin, and their ECC analogues are
vulnerable to Shor's algorithm on a quantum computer. However it requires 2
qubits per key bit, and so far they are only claiming 28 qubits. The largest
number factored by Shor's algorithm so far is 15, in which a 7 qubit
computer found the factors 3 x 5.
Practical quantum computing with thousands of qubits would be a setback for
distributed AI systems that depend on digital signatures for reliable sender
authentication (like my proposal at http://mattmahoney.net/agi2.html ).
-- Matt Mahoney, matmahoney@yahoo.com
----- Original Message ----
From: Nicole C. Tedesco <phlorgian@hotmail.com>
To: sl4@sl4.org
Sent: Wed, December 30, 2009 9:36:02 AM
Subject: RE: [sl4] D-Wave
In the Google case D-Wave is probably an ultra-cold system that enjoys some
advantage from being ultra-cold and perhaps even takes advantage of Ising
glass dynamics in a way described originally by Daniel J. Amit ("Modeling
Brain Function: The World of Attractor Neural Networks"). That book, I
remember, was in my opinion one of the first to describe a coherent method
of taking advantage of quantum effects to obtain a real world computing
advantage (in this case, in the case of pattern recognition).
- Nicole
-----Original Message-----
From: owner-sl4@sl4.org [mailto:owner-sl4@sl4.org] On Behalf Of John K Clark
Sent: Tuesday, December 29, 2009 10:05 PM
To: sl4 sl4
Subject: [sl4] D-Wave
I was very skeptical of D-waves claim to have made a working Quantum
Computer, it's still probably Bullshit but I'm no longer quite as
certain. It seems that Google has taken an interest in D-Wave through
their head of image recognition, Hartmut Neven. Neven started a image
recognition company called Neven Vision and about 3 years ago Google
bought the company. Either Neven and Google are not as smart as I
thought they were or D Wave is not a total scam after all.
http://www.theregister.co.uk/2009/12/15/google_quantum_computing_research/
John K Clark
-- John K Clark johnkclark@fastmail.fm
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:05 MDT