Reply to post:

Big Blue's quantum rainmaker jumps to room-temp diamond quantum accelerator company

Michael Wojcik Silver badge

I thought magical infrastructure was the essential requirement.

Perhaps not. Scott Aaronson did an ACM TechTalk on this just last week, and the USTC group just released a new result claiming 60 qubits using the optical approach with a linear cross-entropy benchmark (LXEB, the current gold standard for establishing that there's an actual QC process happening) that's just high enough to be convincing.

This betters USTC's result from last year by a handful of qubits, and is seven more than Google's result, though Google's used a different approach (random quantum circuit evaluation in a hybrid chip) and Google had an LXEB that was quite a bit better.

Of course, Google's system also has to be cooled to 0.01K, so you're probably not going to be running it at home, unless you have a spare dilution fridge.

I'm more or less a QC skeptic – more than Aaronson, who is well known for not suffering QC hype gladly. (He repeatedly punctured claims about the D-wave machine, for example.)

But my reservations are primarily around scaling (particularly the glaring lack of practical Quantum Error Correction in any published results), manufacture, and applicability (the algorithms in BQP are limited). I used to be dubious about the feasibility of even non-practical QC systems, based on arguments (from physicists) about parameter explosion and other fundamental problems. Now Google and USTC have arguably showed quantum supremacy.1 So it looks like "fundamentally impossible" is pretty much out the window. But "not useful in practice" is still very present.

Now, as someone else posted, nitrogen-vacancy systems do have the very nice property of long decoherence times at room temperature. But making use of them has so far been difficult (AIUI), and I am very reluctant to believe that even if diamond semiconductors solve those problems, the manufacturing issues will be addressed in a few years. That seems wildly optimistic.

Personally, I think the chances of this startup getting anything working are small, and having something commercially available in five years ... well, I'd be keeping the CV up to date.

1Aka quantum advantage: That is, a result which is not feasible to calculate on any conventional computer. The point of the LXEB is to demonstrate, in effect, how many of the output set are correct; since we know the problem domain and its computational complexity, we can easily calculate how fast an arbitrarily large and fast (within physical constraints) conventional computer could get a comparable result. With 50-60 qubits it doesn't take long to get to a problem size where that becomes infeasible within whatever runtime you want to choose (a day, a year, the lifetime of the universe...).

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2022