09 February, 2007

Adiabatic Quantum Computing

Well someone is giving it a go.

It will be significant if they succeed but I am going to give this one long odds.

Here is some background on adiabatic quantum computing if you are interested in some pretty detailed reading.

The key hurdle I found that casts some doubt on the attempt in the tech world article is that increasing complexities require exponentially increasing time. The test is being done with 16 qubits which is below the previous experiments of 20 qubits mentioned in the detailed paper so this potential problem won't show up even if the rest of the test is successful. My warning to any VC's or Angel's out there is to check about this before sinking any money. Don't just assume it is just going to scale up based on the initial test.

That said this could have some pretty significant effects on other areas (such as cryptography) if the test is successful. Depending on the speed and architecture individual scaling may not be necessary for it to have value. I do have some trouble with thinking it will make much of a difference though since 64K calculations isn't that much.

Not to be confused with adiabatic methods in standard computing (Here)

Update:
Here is the blog link to the the announcement and some good details as well. I am slightly more optimistic at this point because they are taking a somewhat different approach than I originally thought. Since it is a blog perhaps we can take advantage of the medium and ask him to explain some of the differences.

More Update:

I like his Blogroll - UFC, Dawkin's and Kurzweil - Can't be that bad of a guy. :)

Update 3:

I got some comments back from the blogger. I did read the paper but he was right about it only being a theoretical assesment. That would make their attempt the first real atempt. There are two major questions that they need to answer before this has any chance of being a truely significant breakthoguh. How long does it take for them to perform the 64K calculations (actually 1 calcualation that asseses 64K prosibilities) and if they add additional qubits how does that affect the time of the calculations? Is it a linear impact or exponential?

3 comments:

Anonymous said...

"The test is being done with 16 qubits which is below the previous experiments of 20 qubits mentioned in the detailed paper so this potential problem won't show up even if the rest of the test is successful."

Looks like you didn't read the paper you referenced -- the 20 qubits were THEORETICALLY investigated, not experimentally.

James said...

Ah thanks

Jim C said...

and I did read it but it certainly isn't the easiest read in the world. Hence me asking questions about it.