Re: [Haskell-cafe] [Haskell] Postdoc ad: quantum-computing programming languages
Hello. Le 31/03/2016 22:04, Scott Pakin a écrit :
My institution just bought a D-Wave 2X adiabatic quantum computer. The problem is, no one really has a grasp on how to *program* an adiabatic quantum computer. It's a totally different beast from the gate-model quantum computers that most people imply when they talk about quantum computing. I find all this a bit disturbing... Los Alamos buys an expensive device that nobody knows how to use??
Moreover, in circumstances where the doubts about the real performance of the D-Wave computer stii persist? Several physicists refuse to call this contraption a "quantum computer". The statements about their "qubits" in their public materials are not always serious, there is practically nothing about a genuine state superposition, no educated physicist will buy such pseudo-definition as "having simultaneously the values 0 and 1" (being the result of two currents flowing in opposite directions ; what about phase?). Their "white paper" about the map colouring shows a model which is more similar to a Hopfield (or similar) neural network, rather than a quantum computing device. The optimization is a natural application domain of such networks, but where are some more universal examples? Surely, there are quantum elements in it: superconducting niobium rings, Josephson junctions, etc. But, actually, even a plain transistor is a quantum device as well, and nobody dares to call it a "qubit". Their native code seems to be extremely far from quantum theory, as we know it. = But, if the device works, has some affinities with neural stuff and with Monte-Carlo techniques (annealing), perhaps a good playground for testing it would be a Go player? Jerzy Karczmarczuk
Research interest is in the way such a machine might work, rather than the
possible (unlikely) affront to ontology.
Demanding a priori documentation is a bit disingenuous as the land is
undiscovered.
Cheers,
Darren
On Mar 31, 2016 19:24, "Jerzy Karczmarczuk"
Hello.
Le 31/03/2016 22:04, Scott Pakin a écrit :
My institution just bought a D-Wave 2X adiabatic quantum computer. The problem is, no one really has a grasp on how to *program* an adiabatic quantum computer. It's a totally different beast from the gate-model quantum computers that most people imply when they talk about quantum computing.
I find all this a bit disturbing... Los Alamos buys an expensive device that nobody knows how to use??
Moreover, in circumstances where the doubts about the real performance of the D-Wave computer stii persist?
Several physicists refuse to call this contraption a "quantum computer". The statements about their "qubits" in their public materials are not always serious, there is practically nothing about a genuine state superposition, no educated physicist will buy such pseudo-definition as "having simultaneously the values 0 and 1" (being the result of two currents flowing in opposite directions ; what about phase?).
Their "white paper" about the map colouring shows a model which is more similar to a Hopfield (or similar) neural network, rather than a quantum computing device. The optimization is a natural application domain of such networks, but where are some more universal examples?
Surely, there are quantum elements in it: superconducting niobium rings, Josephson junctions, etc. But, actually, even a plain transistor is a quantum device as well, and nobody dares to call it a "qubit". Their native code seems to be extremely far from quantum theory, as we know it.
= But, if the device works, has some affinities with neural stuff and with Monte-Carlo techniques (annealing), perhaps a good playground for testing it would be a Go player?
Jerzy Karczmarczuk _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
Darren Grant decided to become the lawyer for D-Wave...:
Research interest is in the way such a machine might work, rather than the possible (unlikely) affront to ontology.
Demanding a priori documentation is a bit disingenuous as the land is undiscovered.
Dear Darren, 1. I did not speak about an "affront to ontology" (the phrase which I don't understand). I said something about their abusive terminology, according to some dozens of texts I read. Several competent people raised doubts about the nature of this device, if you are one of them (there is one D.G., physicist in Alberta), point me to *serious* texts which defend the D-W, you should find them, sincerely. I am NOT accusing them of cheating with physics, or selling rubbish, come on, but Vazirani criticised them, Aaronson as well, the Science article in 2014 also. 2. "...the way such a machine might work" is a technological issue, not JUST a "research interest". They SELL it. "A priori documentation"? "Land is undiscovered"?? What are you talking about? The story has 10 years. Google tested some optimization stuff with the machine in 2009. Some Ising model simulations took place in 2013, and *apparently* the speedup obtained had nothing to do with the "standard hype" of quantum computation. The whole issue of quantum entanglement in their device, is one big mess. 3. I will comment privately your public use of the word "disingenuous", when I learn what can you SAY about the domain of quantum computations. My competence is rather weak, but I taught it a bit. Regards. Jerzy Karczmarczuk
participants (2)
-
Darren Grant -
Jerzy Karczmarczuk