Greg? ðŸ™‚

]]>Would you consider maybe taking a look at the Wikipedia entry on D-wave? The article on D-wave demonstrates such an utter lack of skepticism that I feel it constitutes a violation of NPOV. The whole thing reads like a slightly toned-down press release, and the only part that suggests there exists anywhere active skepticism the D-wave works is the following sentence, offered without citation:

D-Wave themselves admit that they “are not sure” if the device is actually doing quantum computations, instead stating that it simply may be using quantum mechanics to do essentially classical computation.

*What?*

I do not think the current content needs to go away, but I think the wikipedia entry would benefit from the addition of a “criticism” section by someone familiar enough with the standing criticisms to state them fairly. Meanwhile the article links to “adiabatic quantum computation”, but no article by that name exists on wikipedia and there are standing requests on the Talk page for someone to explain exactly what this means. Basically the wikipedia coverage of this issue is badly in need of the kind of informed perspective that this blog discussion comprises…

]]>According to CBC News Herb Martin (D-Wave CEO) stated that *the company has not published its findings because they are seeking patents on their technology*.

At the same time he announced that *if there’s any scientists out there who want to take a look at what we’re doing, they would be our guest.*

# 3 | Jonathan Vos Post | April 11, 2007 11:37 AM

New York Times, April 1, 2008:

Roast Beef Sandwich Computer: DNA-based Quantum Computer In 2nd Round Venture Capital

After dazzling NASA with a demonstration that solved the Dark Energy-Dark Matter-Dark Entropy equations to 22 decimal points accuracy, the start-up Earlofsandwich.com demonstrated their prototype Roast Beef Sandwich Computer capabilities further.

It successfully applied hybrid DNA-Quantum computing technology to dazzling speeds in a set of benchmark tests:

(1) The 5-dimensional Sudoku Puzzle;

(2) computing human brain tomography based on horseradish peroxidase-stained neuron photographs;

(3) automated theorem proving of Geometry’s Ham Sandwich Theorem;

(4) Retrodiction of the “Pick a Number Win a Book” conundrum;

(5) Using Twistor Theory to determine the plotline of Bon Dylan’s “Tangled up in Blue”;

(6) Predicting the NCAA College Baketball rankings based on Gatorade extrapolation metrics;

(7) Using the Reagan Institute Trickle-down Econometric model to prove that there is a way out of Iraq, but it is not Godel-decidable;

(8) Spoofing the Google page-index algorithm to boost Scienceblogs to the #1 ranking in the world.

“This one doesn’t look like hype,” said Scott Aaronson, a theoretical computer scientist at the Institute for Quantum Computing at the University of Waterloo in Canada. “Previous Quantum Computer demos were suspect, based on the Penrose-Conway-Turing Loop Quantum Gravity surreal-number matrices.”

“Let’s see them top this in Bangalore” said a grizzled Silicon Valley Venture Capitalist.

]]>thanks!

by the way, i thought you’ve been working on the subject of quantum adiabatic error correction? or am i confusing something?..

]]>Quite a few. A start is quant-ph/0108048. A more recent paper is quant-ph/0512170. However, this being said, I can pretty confidently tell you that as of this moment, no one knows how to make adiabatic quantum computation fault-tolerant (in either its original form for building quantum algorithms or in its newer incarnation as a universal quantum computer.)

]]>Well actually the statement is that X+Z+ZZ probably isn’t universal for adiabatic quantum computation. Certainly if you have Hamiltonians with X,Z, and ZZ terms which you can turn on and off to produce quantum gates then this is universal (X and Z generate SU(2) and you only need any nontrivial interaction between two qubits to get universality once you have single qubit gates.) But in the adiabatic model you don’t get to apply these gates directly. And as of yet there is no proof of universality of these interactions with adiabatic quantum computation. Indeed there is even some evidence that quantum computations with these interactions is not universal because it turns out that Hamiltonians with these interactions are (or can always be easily transformed into) what are called a stoquastic Hamiltonians. There is evidence that stoquastic Hamiltonian’s are easier to simulate (and by easier I don’t mean easy!) Some of this evidence is in quant-ph/0611021.

]]>scott, thank you very much for your blog! ðŸ™‚

i’ve some questions on quantum adiabatic computation…

**quantum simulation**

geordie rose stated on his blog about the orion processor that *‘the reason why we donâ€™t think we can run quantum simulation […] in general is that X+Z+ZZ isnâ€™t universal for BQP’*

could someone please make some comments on this statement? what are the computational complexity classes for quantum simulation of quantum systems and classical simulation of quantum systems?

**Ising model**

*an ability to tune the rate of the adiabatic process â€” something which appears to researchers to be extremely hard if not impossible for the Ising problem*

are there some papers on the subject of quantum adiabatic computation and Ising model?

**the robustness**

the robustness of quantum adiabatic computation seems to be a much discussed topic… geordie rose stated on his blog that *‘AQC approach is naturally shielded from errors in a way that the gate model isnâ€™t’*.

are there some peer-reviewed results on quantum adiabatic error correction?

**degenerated ground state**

geordie rose stated on his blog that *‘the two theoretically degenerate solutions are each obtained roughly 50% of the time is evidence that the control we have over the machine language parameters is pretty good’*.

so if the energy minimum of the final hamiltonian is twofold degenerated, the quantum adiabatic protocol drives the initial ground state into a quantum state which is a superposition of two corresponding basis states.

|psi_f> = 1/sqrt(2) ( |psi_min_1> + |psi_min_2> )

the projective measurement results each of this states |psi_min_i> with the probability prob_i=1/2. is this correct?

**thanks!**

Just as a clarifying remark, with regard to the devices of the above citation, if the gates are made noisy enough to allow classical simulation of the device, then the device’s performance is so crippled by that noise that the device cannot even simulate a classical computation—at least that’s what the article states.

Hopefully, there is a lot of room to improve this result!

If there is a more recent and powerful result, a citation would be very interesting and useful to me, and I suspect, to many.

]]>