Bad News for Quantum Computing: Another advantage gone (Patreon)

[This is a transcript with links to references.]

Sometimes people ask me why I’m now doing science news. It’s so that you can fully appreciate the drama of scientific discovery,  in which one result contradicts a previous one and the next one finds a flaw in this other one. I think the entertainment-value of science is greatly underappreciated.

And quantum computing  is without doubt one of the most dramatic areas at the moment. One the one hand you have people claiming it’s the next industrial revolution  and it’s going to change the world,   on the other hand you have people saying it’s all hype, quantum computers are interesting but useless.   This new paper supports the latter camp. Let’s have a look.

Quantum computers are devices that exploit quantum links known as entanglement, and no that’s not what Einstein meant by spooky action at a distance. Entanglement doesn’t exist in conventional computers, and because quantum computers have these additional ways to encode and handle information, they can perform some calculations much faster.

The relevant part of the previous sentence is that they can perform *some calculations much faster.  It isn’t the case for all calculations. Nevertheless, the calculations that quantum computers can speed up cover several commercially relevant cases such as quantum chemistry,  that plays a role in material design and drug development, as well as logistics  and finance . This is why there is hype around quantum computers and that talk about a new revolution.

One the other hand, there are a lot of practical problems in making quantum computers work, I mean besides the problem that no one understands quantum mechanics, though there’s that. The real issue is that these quantum links are incredibly fragile.  Depending on how you create them they might last microseconds to a few seconds, but that’s about it, and your calculation needs to be done within that time. And I mean, I do calculations for my financial management within a few microseconds, but it’s probably not a good thing.

There are many approaches to quantum computing that differ by how you physically realize the quantum bits, qubits for short.  I have talked about this in several previous episodes.  But they all have this problem in common: The quantum information is very sensitive to disturbances, which brings in errors. These errors can be avoided by introducing redundancies,  but that requires making the computers larger, which brings in more problems.  What this means is that the need for error corrections moves useful quantum computers farther into the future, and no one knows how far. That’s unappealing for investors.

This is why in the past years, some physicists, like for example John Preskill  have pushed the idea that even “noisy” quantum computers  can be good for something. If that was true, it would do a lot against the criticism that quantum computing applications are far away and might remain far away forever.

Last summer, IBM said they’d done it, they had done something useful with a noisy quantum computer and the New York Times celebrated this success. The new paper now puts the IBM achievement into doubt. So, ha, you see the drama.

In last year’s paper, the IBM team explains how they used a noisy quantum processor with 127 qubits, called the Eagle processor, to calculate what’s called the Trotterised time evolution of a 2D transverse-field Ising model.

The Ising model is a model of coupled quantum spins, and the trotterised transverse-field and so on means it’s a special version. This model is interesting… if you find that kind of model interesting. To my best knowledge it doesn’t have any uses.  However, the type of calculation is difficult to do on a conventional computer, which is what makes it an interesting test.

The IBM group claimed that their quantum chip outperforms any calculation that could be done on a conventional computer and that, quote “there is indeed merit to pursuing research towards deriving a practical computational advantage from noise-limited quantum circuits”.

The authors of the new paper now say that IBM was somewhat overoptimistic. They found a way to do the calculation on a conventional computer and they managed to do it faster and better than IBM could on their noisy computer. Basically, they show that conventional computers outperform the quantum computer also on this task. They write “we can perform a classical simulation that is significantly more accurate and precise than the results obtained from the quantum processor”. You can see this nicely in this figure. The dots with the error bar are the results from the quantum computer, and the smooth line is the new calculation.

So, another supposed quantum achievement gone.

That’s happened a few times before that one group claimed they’d done something spectacular on a quantum computer just that later another group did it on a conventional computer. The most prominent example was probably Google’s 2019 claim that they had proved a “quantum advantage”. Back then, the scientists from Google published a paperin Nature claiming they’d done a calculation on a 53 qubit quantum computer which would have taken 10 thousand years on a conventional computer. The 10,000 years claim was swiftly questioned by IBM and indeed it was later done by a Chinese group on a conventional computer in only 5 minutes.

Then again, maybe that just means that 10000 years in silicon valley are worth 5 minutes in China.

This new result matters because it’s once again shown that conventional computers  are so advanced already that they are a very strong competition for quantum computers and that will probably remain so for quite some time to come. This is really bad news for anyone who has invested into quantum computing.