September 14, 2006

Superconducting quantum computer details

The key finding that Dwave systems is disclosing is that quantum error correction (and fault tolerance) can be achieved PASSIVELY instead of actively. They have also released two papers describing their work on quantum computers

Here is an Aug 11, 2006 paper (pdf) about their sign and magnitude tuneable superconducting couplers

The second pdf describes the underlying way their qunatum computing system will work. It is thermally assisted adiabatic quantum computing. The paper describes simulations up to 20 qubits.

Further reading:
Previous survey of quantum computer progress


Jonathan Pugh said...

Hey Brian. I was checking out one of your previous posts on DWave and saw that you said:

"Based on funding and progress I am thinking 32-40 qubits when they release and 100-200 qubits within a year or so after."

So are these guys releasing their product next year with 32-40 qubits? I haven't researched these guys a lot so please forgive me if I am a little ignorant.

Also, I understand that some of the applications of a QC would be in encryption and any field that has to do quantum simulations such as physics, biology, chemistry, nanotechnology. So my question is, what do you think the most useful near tear-term applications will be for QC. How about long term?

bw said...

Hi Jonathan

My statement about 32-40 qubits and then more afterwards was speculation on my part.

How did I arrive at that ?
From a 2002 article interviewing the CTO of the company Mr Rose. He said at the time he was targeting something like 27 qubits. Which would have the power of a then Cray computer...if one converts the size of the quantum search space with flops. I upgraded it to 32-40 qubits in order to get it up the petaflop power we are having today.

Dwave has been secretive, but the target of a 2007 release has been mentioned. They also are now putting information at the which I site. That is written by the same CTO. On the blog they are talking about pre-releasing some service even later this year. Based on the fact that they are emerging from complete stealth mode into actually sharing info...suggests that they think they are close to launching something.

the 100-200 qubits number ..where did that come from ?
I read the superconducting quantum computing papers for the type that they are working on. One of those indicated that the number of theoretical qubits was based on how they use up energy/heat and that would start pushing up to critical temperatures. Superconductors need to work below critical temperatures. That article speculated about hundreds and maybe thousands. the thousands sounded like they needed better superconductors and maybe some more efficient designs. therefore, I read that to think that progressing to 100-200 qubits sounded achievable..particularly if they get a successful 20-40 qubits from the 12 million or so in funding that they have. If they did that I would see the US gov't giving them 100-200 million to push it as big and as fact as they could.

In terms of applications, they have a specialized system. they are targeting a particular set of problems. I understand those to be optimization problems. Things like laying out more efficient networks etc... Fedex, shipping and other transportation companies would pay to get their networks more efficiently laid out. So better logistics. It seems that the Dwave QC will not at first be suited to the encryption problem. But they will make bank improving the solutions of logistic problems.

I think they will be able to do simulations of other quantum systems. that will be good for science including nano.

I expect a flood of QC after the first practical one. Guys in Japan are also doing work on superconducting QC. There are other types (trapped ion, electron bubbles etc...)

A commercial operation may be less focused on the decryption because the gov't would probably restrict sales.

Jonathan Pugh said...

Hi Brian. Thanks for posting back on my comment. Always appreciate you insight.

So yesterday Geordie commented on my question on release date & number of qubits. Check out comment 7 on his blog, it seems to agree with what you were saying:


Jonathan - September 21, 2006
Can you give an estimate of:
1) release date
2) qubits utilized on release date

Geordie - September 23, 2006

1) first public demo of the technology some time in Q4/2006

2) the demo will either use a 16- or a 64-qubit chip - which we use will be determined after completing some tests in October.

The first demo has as its objective a demonstration of what the machine is designed to do; we’re going to run a couple of applications on it. It’s not supposed to be faster than conventional methods, but the approach is eminently scalable; in a relatively short period of time (at least on the problem type it’s designed for) nothing else will be able to compete with it.