Analogue Quantum Computer: I Call Bullshit!

Just stumbled on reports of a new company called InfinityQ that makes an analogue version of a quantum computer.

Seems like a REALLY good idea to me - and they've been in the news since April/May - so why haven't I heard of them?

Looking carefully at the reports about what they claim, my nose started twitching: thankfully, I happen to know a bit about the Travelling Salesman Problem (TSP). It has been solved exactly to a size of 85,900 points (link). So when they claimed that their computer had solved it to 128 points (link), and that this was supposedly superior to other methods (same article), I wondered what they were talking about. On looking deeper, I found that they were comparing not with integer programming techniques, but with a purpose-built piece of hardware (link).

So now I know why I hadn't heard of a company building what sounds like a brilliant idea - analogue quantum computing. It's crap.

Looking into it still more, what they're ACTUALLY doing seems to be closely analogous to simulated annealing. Lo and behold - the Wiki article about simulated annealing says it can solve the Travelling Salesman problem exactly to...

... drum roll...

... 125 points (link)!  

A serious point, though: this is just a suspicion, and I could well be wrong (but if I'm right, remember that you heard it from me first! ), but... could it be that what real quantum computers are doing is actually a form of simulated annealing? If so, if might be more limited than many people are hoping for right now.

Damn and blast! I'm not the first person to have said this - link.

When a quantum particle changes value, how much time does it take for its entangled partner to also change? An Israeli-German team have been studying this question - link. As you'd expect, the process is not instantaneous - it does take a finite length of time.

On a related issue, it seems as though quantum tunnelling enables quantum particles to travel faster than the speed of light (link), but I don't think that any of today's quantum computers are actually using quantum tunnelling right now (please correct me if I am wrong!).

I am not an expert in quantum computing, but I am under the impression that if the technology can be used to build computers that can solve practical problems many orders of magnitude faster than normal computers, then we're still at the very early stages of that process. In contrast, when the first electronic computers were built in the 1940s, they were immediately able to do useful work more quickly than could be done before.

Nice FT article on quantum computing - link.

I am still not convinced, though. IMO, the more closely you look at the technology, the worse it gets. It seems to me that the source of all the current optimism is those who have a vested interest (companies that make them, or people in companies like Google and IBM whose jobs depend on these companies continuing to fund research).

The original electronic computers from the 1940s were slow, heavy, expensive, unreliable, and had MASSIVE running costs. Take ENIAC for example:

* 500 calculations per second
* storage for 20 10-digit variables (yes - that few!)
* consumed 160 kw (174,000 watts according to another source)
* wasn't running half the time (I think reliability was improved a lot before its end of life)
* cost $500,000 (over $7 million in today's money)
* had 5 million solder joints
* was big enough to park a school bus inside it

However - it had one thing going for it that today's quantum computers don't have: it was better than anything that came before it, and it was immediately able to do useful work (printing ballistics tables).