Major Quantum Computing Strategy Suffers Serious Setbacks
quantamagazine.org
quantamagazine.org
Anyway, topological quantum computing and Majorana fermions are interesting but the research in that area is at a stage where we've been with superconducting systems 30-40 years ago, i.e. people making the first fundamental experiments and trying to assemble the basic building blocks for a qubit. So I don't think anyone seriously bets on topological quantum computers to win the race.
This quantum computing strategy neither was nor is practiced by the vast majority of quantum institutions—commercial or otherwise. It was attempted by a group at Microsoft (and a small collection of other university groups) and was known from the start that it would be a search for essentially fundamentally new observations.
Other quantum computing players, like Rigetti, Google, HRL Laboratories, IBM, Amazon, Honeywell, and others are doing an approach that is nothing like the article, and have demonstrated significant results.
"Quantum qubit substrate found to be apparently insufficient" (Given the given methods and probably available resources)
And then - in an attempt to use terminology from Constructor Theory https://en.m.wikipedia.org/wiki/Constructor_theory :
> In constructor theory, a transformation or change is described as a task. A constructor is a physical entity which is able to carry out a given task repeatedly. A task is only possible if a constructor capable of carrying it out exists, otherwise it is impossible. To work with constructor theory everything is expressed in terms of tasks. The properties of information are then expressed as relationships between possible- and impossible tasks. Counterfactuals are thus fundamental statements and the properties of information may be described by physical laws.[4] If a system has a set of attributes, the set of permutations of these attributes is seen as a set of tasks. A computation medium is a system whose attributes permute to always produce a possible task. The set of permutations, and hence of tasks, is a computation set. If it is possible to copy the attributes in the computation set, the computation medium is also an information medium.
> Information, or a given task, does not rely on a specific constructor. Any suitable constructor will serve. This ability of information to be carried on different physical systems or media is described as interoperability, and arises as the principle that the combination of two information media is also an information medium.[4] Media capable of carrying out quantum computations are called superinformation media, and are characterised by specific properties. Broadly, certain copying tasks on their states are impossible tasks. This is claimed to give rise to all the known differences between quantum and classical information.[4]
"Subsequent attempts to reproduce [Quantized Majorana conductance (topological qubits of arranged electrons) within indium antimonide nanowires] eventually as a (quantum) computation medium for the given tasks failed"
"Quantum computation by Majorana zero-mode (MZM) quasiparticles in indium antimonide nanowires not actually apparently possible"
... "But what about in DDR5?" Which leads us to a more generally interesting: "Rowhammer for qubits", which is already an actual Quantum on Silicon (QoS) thing.
I guess I wouldn't call it major but rather "exiting" strategy. The theoretical result from Kitaev [1], from Microsoft research, was certainly interesting. Having a way not deal with error corrections would be really great.
[1] https://arxiv.org/pdf/cond-mat/0010440.pdf
> Thus an isolated Majorana site (usually called a Majorana fermion) is immune to any kind of error!
And he comes back to find that the plant he used to work at got exported to the country he just got back from. And the guy who put the shrapnel in his ass got his old job, cause he'll work for fifteen cents a day and no bathroom breaks. Meanwhile, he realizes the only reason he was over there in the first place was so we could install a government that would sell us oil at a good price. And, of course, the oil companies used the skirmish over there to scare up domestic oil prices. A cute little ancillary benefit for them, but it ain't helping my buddy at two-fifty a gallon.
And they're takin' their sweet time bringin' the oil back, of course, and maybe even took the liberty of hiring an alcoholic skipper who likes to drink martinis and fuckin' play slalom with the icebergs, and it ain't too long 'til he hits one, spills the oil and kills all the sea life in the North Atlantic. So now my buddy's out of work and he can't afford to drive, so he's got to walk to the fuckin' job interviews, which sucks cause the shrapnel in his ass is givin' him chronic hemorrhoids. And meanwhile he's starvin', cause every time he tries to get a bite to eat, the only blue plate special they're servin' is North Atlantic scrod with Quaker State.
So what did I think? I'm holdin' out for somethin' better. I figure fuck it, while I'm at it why not just shoot my buddy, take his job, give it to his sworn enemy, hike up gas prices, bomb a village, club a baby seal, hit the hash pipe and join the National Guard? I could be elected president."
Good Will Hunting
Some people blew the result out of proportion, making claims that finally quantum computers are provably useful/the best/etc. Google (who got the result) know these claims are not true, but in their typical character, turned a blind eye to the hype that ensued, because who doesn’t want free press?
See section 16.4 of these notes (pdf warning) [1].
1. Do theoretical quantum computers have different properties than theoretical classical computers? (or, QBP = P? )
2. Does some particular physical device show the properties associated with a theoretical quantum computer?
Of course, if 1 is false, than 2 is more or less irrelevant. However, even if 1 true, that still leaves the question of 2 for any particular device.
The Google and Chinese experiments have proven 2 for their own particular devices.
This is an important milestone not just for these particular devices, but also because, prior to this, there was also a question 3: do QM systems actually exhibit the properties of a theoretical quantum computer? After the Google result, this has been almost entirely put to rest: the answer is yes.
Before this milestone, there were still reasons to believe that quantum computers could not, in principle, beat classical computers, that there is some fundamental limitation of the universe that would actually prevent quantum effects from making a QC work.
After this milestone, the only "hope" for such a fundamental limitation lies in the area of quantum error correction - the possibility that, somehow, you would lose the theoretical speed-up from QC by having to re-run the algorithm enough times to get an accurate enough response. As Scott Aaronson explains, while disappointing in terms of QC, this would also be extremely exciting for quantum mechanics itself and for our understanding of the universe.
So, the Google and Chinese quantum supremacy demonstrations, while useless as actual computations, serve a very important role in showing that actual quantum computers have properties that can't be replicated by classical computers, which was not proven before these experiments were run.
Equivalently, you can say that these experiments have proven that the physical systems being called "quantum computers" have at least some properties of the theoretical mathematical concept of a quantum computer - something that no previous systems had proven conclusively. In this way, this can also be seen as a new test of Quantum Mechanics' predictions, since it proves once again that real physical systems do exhibit some of the complex behaviors predicted by the maths.
The laws of Quantum Mechanics clearly allow quantum computers. So those beliefs must have included something that goes beyond known physics i.e. the assumption of some supernatural effect?
The laws of relativity and nuclear physics “allow” a tablespoon of salt to provide 453 GWh of energy.
It may help to understand that a prime motivation for building quantum computers is that they provide the only known way to perform quantum mechanics calculations of general physical systems at significant scale. In other words, although you can simulate a quantum computer in principle, and calculate the effects of the laws of quantum mechanics in principle, in practice classical (non-quantum) computers hit a computional wall doing either of those things above a certain level of complexity which occurs in the natural world.
That makes it impossible to work backwards from physical observations of sufficiently complex systems to confirm that the underlying process is following the laws of quantum mechanics as we currently formulate them. Basically, we need a quantum computer to do the kinds of calculations needed to relate physics observations of complex quantum systems to their quantum mechanics underpinnings. We just can't do the math otherwise.
Quantum mechanics itself has been verified to extraordinary precision over a wide range of phenomena. We know the physics and mathematics works and is remarkably accurate, for the things we can test. That includes many macroscopic observations; it's not a size problem. We have all sorts of clever ways of calculating different kinds of physics observations from underlying quantum mechanics. But the calculations are all limited in the quantum information complexity they can handle, in practice. In particular, nothing with the high level of coherent but complex quantum state required for a substantial size quantum computer is calculatable without a quantum computer, as far as we know.
So, remarkably, quantum mechanics being verified over a wide range of phenomena isn't enough to tell us if quantum computation at significant scale is possible, because we can't do the calculations that show if the observed universe is following mathematical quantum mechanics closely enough to support one.
Ironically, to measure if the universe supports quantum computation, we have to build a quantum computer just to be able to do the calculations that show us if we can build a quantum computer.
I will just add one thing: basically Quantum Supremacy is yet another test of QM. As long as quantum supremacy could not be demonstrated, there remained a possibility that QM was wrong in this prediction - just like until the Higgs boson was detected, there remained a possibility that the Standard Model was wrong/incomplete.
If you're curious of some proposed possible barriers to quantum supremacy from the time before this experiment was completed, Gil Kalai is a major advocate of the impossibility of achieving QS in practice. This paper [0] details some of this arguments. In very short, his argument is that it is fundamentally impossible to control quantum states to a sufficient degree of precision to realize an actual quantum computer, that errors will always compound too much (he doesn't believe the Google paper will stand the test of time). I am not a believer in his view, but as far as I understand it is (or was, before the Google experiments) not completely absurd; and it is definitely not supernatural.
And fundamental developments in physics or theoretical computer science aren’t concerned with what’s “meh”. It was a logical next step to determining whether these machines do something they’re theoretically supposed to be able to do.
That's also not to say that these roles will not reverse in the future. Perhaps some fundamental problem is found with the existing QC approaches, and someone discovers a fusion approach that is straightforward and just requires a big engineering effort. But right now to me it seems like fusion is closer to a Majorana-based QC (though Majorana is slightly more nebulous since even the theory isn't completely there) than it is to QC in general.
The difference here is that we produce a quantum annealer, which is useful for optimization problems instead of for database searches + factoring. It's already delivering some real-world value for early applications.
While gate model machines are interesting, D-Wave took the tack of implementing the model of QC most likely to lead to actual useful applications within our lifetimes. Gate-model QC does seem to be very useful, but until it reaches millions of physical qubits it's not going to be producing any results beyond pet laboratory projects, and it remains to be seen if that's even physically possible. In contrast, quantum annealing has been able to grow at a good rate both in terms of qubit count, degree of connectivity between qubits, and also in terms of reaching lower noise and better quality results.
We also have hybrid solvers that combine the state of the art in classical algorithms with QPU sampling to get the lowest energy state possible with a much larger graph than we can do in hardware.
I think it's a very interesting field to follow, there are huge investments being made and real progress is happening on a number of fronts. Our competitors are trying to bring live systems to market, too, but it's harder to see them being much more useful than simulators for the foreseeable future.
I think what you wrote makes sense as a way of maximising the chances of producing a viable product. I suppose there aren't any guarantees that it will be competitive with bog-standard computers, but it might be a reasonable gamble.
> maximising the chances of producing a viable product
That's the goal. The annealing QPU is a co-processor, like your GPU, like vector processors, or a DSP, etc. It doesn't need to compete with classical compute on the things classical compute is good at; it needs to compete on the things classical compute is bad at, or at the very least, bad at without throwing massive piles of money at it. There is a crossover point for optimization problems where we will be able to show a price/performance advantage over classical compute, which we term Quantum Advantage (vs. the more divisive term of Quantum Supremacy).
The trick at this point is formulating problems in such a way as to be something you can run on our hardware, which still requires a deep mathematical skillset. This is something we're building on... perhaps pay attention to our Qubits conference coming up next week - https://www.qubits.com/ - there should be some interesting announcements!
Since the discovery of quantum ECC, that's no longer a reason to believe a quantum computer can't be built, a reliable logical qubit can be built out of a number of less reliable physical qubits.