I am not an expert, but my assessment is that it's pure marketing. They talk about needing millions of qubits to do useful computations. This is flat out false. Thousands will already solve interesting problems that no classical computer can currently solve, e.g. factoring RSA moduli. They also say that many qubits operating together will be required to build a practical quantum computer (true), and that their technology allows this (false). The problems involved in maintaining coherence and correcting errors, at reasonable operating temperatures do not boil down to how to fit multiple qubits on a silicon wafer. This is like saying a new process for manufacturing nuts and bolts will allow the construction of Boeing 747's. The paper may contain more sensible statements, but as usual, the press release is probably nonsense.

You're probably confusing physical qubits (inside the actual physical device) and logical qubits (on which the platonic ideal calculation takes place). If we had 1000 logical qubits, we could indeed do useful things, but real-world implementations require huge amounts of error correction, which is much tricker on quantum computers than classical ones. Asymptotically the ratio of physical to logical qubits can approach 1 (just polylog overhead), but for early calculations the ratio will literally be more than a factor of a thousand. (This is true for all concrete proposals to date; people have speculated that there might be something useful to do with 1000 physical qubits for certain problems if the error could be small enough to be ignored and not corrected, but no one has actually shown explicitly how to do this, or produced the components with the quality that would be necessary.)

Yes, I was certainly confused. When IBM announced they are building a 7x7 = 49 qubit machine, I naturally assumed they meant logical qubits, not physical ones. Using surface codes, they can probably get one logical bit. But what confuses me are past records where they used Shor's algorithm to factor small integers on a quantum computer. How is this possible if we haven't even built a single logical qubit yet? It's surely necessary to have ~2N logical qubits to factor an integer of N bits.

I have an idea that might help though: if logical qubits are so difficult to build and require thousands of times more physical qubits, and physical qubits are also difficult to build, perhaps we can simulate physical qubits with thousands of times more "ultra-physical" qubits.

Two excellent questions.

Regarding claims of factoring small numbers: First note that error correction becomes exponentially more important for longer computations. Second, see this

https://arxiv.org/abs/1301.7007

Regarding building a tower of ever less noisy qubits: it turns out there is something called a threshold theorem which requires a certain minimal level of noise before concatenating error correction schemes makes things better rather than worse.

https://en.m.wikipedia.org/wiki/Quantum_threshold_theorem

Reaching this threshold for physical qubits in the laboratory is perhaps the primary goal of experimental quantum computing.

That first article you cite certainly puts the whole thing in perspective. I was tempted to put an article on Arxiv announcing a new architecture for billions of "super-physical" qubits, but you've convinced me that not only would the real experts see right through this, but that it might bring me down to the level of some of the qubit measuring contests currently on display from the likes of IBM, Google and Microsoft. Thanks for the genuinely interesting answers, especially to the not completely serious question!

When I took a course on quantum computing in 2007 with a standford prof, he said that things would get interesting after we get to 128 qubits :)

I guess someone will eventually introduce some kind of terminology like, "qubit equivalent". But I think that if you could build an error free 128 qubit machine, this would be interesting. But the main reason I see for that is if you can build one with 128 qubits, almost certainly someone with a lot more money than you, with a black budget, can build one much larger. The basic thrust of the article, and the paper, seems to be that if you can engineer a large enough quantum computer than doesn't really work, you can simulate one that doesn't exist. It's truly great if they are right, but I can't find a single statement about this paper on the entire web by an expert that is enthused about this announcement. The paper is short enough that there should be some comment by now.

Someone has already done that:

https://www.research.ibm.com/ibm-q/resources/quantum-volume....

That's a very specific number, do you know what the relevance of 128 is?

No I don't think it was specific. I believe he was just guesstimating the exponential power required to do anything more interesting than basic math. We were all rather young students at the time so I don't think he had a specific application in mind. It was all rather new back then anyways, hard to imagine all the applications.

He is probably referencing the 128 bit security threshold most crypto systems seek to achieve. However this doesn’t really line up with the number of qubits needed to break said crypto, which is typically much larger (thousands of qubits, plus error correction to survive millions of operations).

Edit: less than a doubling. Eg from 128 to 192 or from 256 to 384. I meant to say a doubling is sufficient.

> Thousands will already solve interesting problems that no classical computer can currently solve, e.g. factoring RSA moduli.

With hundreds you can already speed-up chemistry simulations, what will have a much larger impact than breaking RSA.

> With hundreds you can already speed-up chemistry simulations

You can? How?

You need an exponential increase in performance (power of N) to simulate an increasing number of molecules, while for quantum computers you may need only a polynomial increase in performance (power of 2).

So it's possible to simulate 25-qubit-level chemistry with a laptop, 50-qubit-level chemistry with a supercomputer, and 100-qubit-level simulation with a supercomputer is impossible and will be impossible for a very long time.

But going from a 50-qubit quantum computer to a 100-qubit one may take just 2 years (or 4, it doesn't really matter), and then another 2 years to get to 200 qubits, and so on. Quantum computers will solve chemistry problems in minutes that would otherwise have taken 100 years to simulate.

So what I'm asking is how does a quantum computer solve quantum chemistry problems? To make it concrete, say I want to calculate the band structure of a crystal using a quantum computer? What do I do? What kind of algorithms can the QC run that solves this problem faster/better than, say, a DFT code?

I don't think it's a given that just because I want to simulate a quantum mechanical system in some way, that a quantum computer is a useful tool. Or it might be, I don't know. I just don't think it's enough justification that there happens to be the word "quantum" in both the problem description and in the name of the computer.

Simulating quantum chemistry was one of the original motivations for quantum computers.

See this paper for an overview https://arxiv.org/pdf/1203.1331.pdf

Thank you, exactly what I was looking for!

ETA 2022 until a 10-qubit silicon quantum computer prototype is made.

I doubt it's just marketing. UNSW is the leading university in researching silicon quantum computers and one of the best in researching quantum computers in general, too.

You cannot do anything interesting with 10 qubits. Anything you can do with 10 qubits will likely be faster on conventional computers. IBM says you need at least 50 before it gets interesting.

I realize that, but I think after a 10-qubit chip is made, it will be easier to scale up.

This tech is aiming to be "mainstream". Google, IBM and others' quantum computers will likely prove useful earlier, but we won't see them be paired with our personal computers. This on the other hand, offers that opportunity, but it will lag behind other types of quantum computers in terms of performance.

I don't think we will get devices cooled by liquid helium at our homes in any short time.

But this makes it possible to distribute them at more datacenters that can be much easier to reach than what Google and IBM are creating.

This is not my field, but IIRC dilution refrigerators capable of sub-K temperatures are nowadays available off the shelf from a number of commercial suppliers. Which is a pretty cool (har har!) advance considering not many decades ago labs had to create their own, at great expense.

So in principle, there might not be any really fundamental issues preventing such things in every home, with mass production economies making them cheaper and more reliable.

The real question IMHO is whether this will be useful in some way compared to just connecting over the internet to some data center running them.

Safety is a fundamental issue. Home appliances don't get the best maintenance available, and supercool and high-pressure containers aren't easy to keep around.

What interesting things happen at 50 qubits?

We can prove that quantum computers are faster than classical computers at solving certain problems (like simulating atom interactions). Although IBM recently moved that goal post, and proved that we can simulate 56-qubit (and probably a little more than that) with classical supercomputers.

https://spectrum.ieee.org/tech-talk/computing/hardware/ibms-...

It's debatable how generic this 56-qubit simulation is. We can already "simulate" millions of qubits, with entanglement, as long as they dance in a particular way...

I figured as much. So whoever develops a 50-qubit quantum computer first (Google?) will have to prove that the same simulation can't be done with a supercomputer. And they'd need to be serious about it and use the latest research in classical simulation, including IBM's 56-qubit research to try and do it on a supercomputer.

Otherwise, they may claim "quantum supremacy" and 3 months later someone else will prove that the same simulation can be done on a classical computer.

You can't do anything interesting with ENIAC, but you had to build those before you built anything else :^)

The people that built the ENIAC did it to solve actual problems.

Well, every new technology needs a proof of concept first.

Uhm, without even RTFA, it’s a .edu.au domain, so an educational institution (in Australia), not a commercial venture, so presumably pretty immune from marketing hype.

EDIT: sorry guys, I assumed the sarcasm would be transparent. Apologies.

I assume that was meant as a joke. Universities regularly put out utter nonsense in press releases (the marketing departments at universities are often ground zero for the most atrocious exaggerations that end up in the press). And I doubt universities in Australia are specially immune from this.

Yes it was meant as a joke but unfortunately my prose was a bit too deadpan, I edited it to add a warning, sorry.

It's from the University 'Newsroom' department, and AFAIK these are pretty notorious for over-hyping University research. It essentially is a marketing department for the University.

Yes, sorry, as I mentioned to somebody else I though my sarcasm would be self-evident, but maybe I just dripped with too much caustic fluid... sorry.

At some level, this is marketing hype. They have NOT "solved" quantum computing in the sense that a layman would expect: room temperature quantum integrated circuit able to give exponential speedup over classical algorithms.

I would say the biggest innovation here is that they have made a silicon based integrated circuit with fault-tolerant qbits (using error-correction codes they invented).