Saturday, February 06, 2021

Don't Fall for Quantum Hype

[This is a transcript of the video embedded below.]

Quantum technology is presently amazingly popular. The United States and the United Kingdom have made it a „national initiative”, the European Union has a quantum technology “flagship.” India has a “national mission”, and China has announced they’ll put quantum technology into their next 5 year plan. What is “quantum technology” and what impact will it have on our lives? That’s what we will talk about today.


The quantum initiatives differ somewhat from nation to nation, but they usually contain research programs on four key topics that I will go through in this video. That’s: quantum computing, the quantum internet, quantum metrology, and quantum simulations.

We’ll start with quantum computing.

Quantum computing is one of the most interesting developments in the foundations of physics right now. I have talked about quantum computing in more detail in an earlier video, so check this out for more. In brief, quantum computers can speed up certain types of calculations dramatically. A quantum computer can do this because it does not work with “bits” that have values of either 0 or 1, but with quantum bits – “qbits” for short – that can be entangled, and can take on any value in between 0 and 1.

It’s not an accident that I say “between” instead of “both”, I think this describes the mathematics more accurately. Either way, of course, these are just attempts to put equations into words and the words will in the best case give you a rough idea of what’s really going on. But the bottom line is that you can process much more information with qbits than with normal bits. The consequence is that quantum computers can do certain calculations much faster than conventional computers. This speed-up only works for certain types of calculations though. So, quantum computers are special purpose machines.

The theory behind quantum computing is well understood and uncontroversial. Quantum computers already exist and so far they work as predicted. The problem with quantum computers is that for them to become commercially useful, you need to be able to bring a large number of qbits into controllable quantum states, and that’s really, really difficult.

Estimates say, the number we need to reach is roughly a million, details depend on the quality of qbits and the problem you are trying to solve. The status of research is presently at about 50 qbits. Yes, that’s a good start, but it’s a long way to go to a million and there’s no reason to expect anything resembling Moore’s will help us here, because we’re already working on the limit.

So, the major question for quantum computing is not “does it work”. We know it works. The question is “Will it scale”?

To me the situation for quantum computing today looks similar to the situation for nuclear fusion 50 years ago. 50 years ago, physicists understood how nuclear fusion works just fine, and they had experimentally checked that their theories were correct. The problem was “just” to make the technology large and still efficient enough to actually be useful. And, as you all know, that’s still the problem today.

Now, I am positive that we will eventually use both nuclear fusion and quantum computing in everyday life. But keep in mind that technology enthusiasts tend to be overly optimistic in their predictions for how long it will take for technology to become useful.

The Quantum Internet

The quantum internet refers to information transmitted with quantum effects. This means most importantly, the quantum internet uses quantum cryptography as a security protocol. Quantum cryptography is a method to make information transfer secure by exploiting the fact that in quantum mechanics, a measurement irreversibly changes the state of a quantum particle. This means if you encode a message suitably with quantum particles, you can tell whether it has been intercepted by a hacker, because the hacker’s measurement would change the behavior of the particles. That doesn’t prevent hacking, but it means you’d know when it happens.

I made an entire video about how quantum cryptography works, so check this out if you want to know more. Today I just want to draw your attention to two pointes that the headlines tend to get wrong.

First, you cannot transfer information faster than the speed of light with the quantum internet or with any other quantum effect. That quantum mechanics respects the speed of light limit is super-basic knowledge that you’d think every science writer knows about. Unfortunately, this is not the case. You see this over and over again in the headlines, that the quantum internet can supposedly beat the speed of light limit. It cannot. That’s just wrong.

And no, this does not depend on your interpretation of quantum mechanics, it’s wrong either way you look at it. No, this is not what Einstein meant with “spooky action at a distance”. It’s really just wrong. Quantum mechanics does not allow you to send information faster than the speed of light.

This isn’t the major issue I have with the coverage of the quantum internet though, because that’s obviously wrong and really what do you expect from the Daily Mail. No, the major issue I have is that almost all of the of the articles mislead the audience about the relevance of the quantum internet.

It’s not explicitly lying, but it’s lying by omission. Here is a recent example from Don Lincoln who does exactly this, and pretty much every article you’ll read about the quantum internet goes somewhat like this.

First, they will tell you that quantum computers, if they reach a sufficiently large number of qbits, can break the security protocols that are currently being used on the internet quickly, which is a huge problem for national security and privacy. Second, they will tell you that the quantum internet is safe from hacking by quantum computers.

Now, these two statements separately are entirely correct. But there’s an important piece of information missing between them, which is that we have security protocols that do not require quantum technology but are safe from quantum computers nevertheless. They are just presently not in use. These security protocols that, for all we currently know, cannot be broken even by quantum computers are, somewhat confusingly, called “post-quantum cryptography” or, in somewhat better terminology, quantum-safe cryptography.

This means that we do not need the quantum internet to be safe from quantum computers. We merely need to update the current security protocols, and this update is already under way. For some reason the people who work on quantum things don’t like draw attention to that.

Quantum metrology

Quantum metrology is a collection of techniques to improve measurements by help of quantum effects. The word “metrology” means that this research is about measurement; it’s got nothing to do with meteorology, different thing entirely. Quantum metrology has recently seen quite a few research developments that I expect to become useful soon in areas like medicine or material science. That’s because one of the major benefits of quantum measurements is that they can make do with very few particles, and that means minimal damage to the sample.

Personally I think quantum metrology is the most promising part of the quantum technology package and the one that we’re most likely to encounter in new applications soon.

I made a video especially about quantum metrology earlier, so check this out for more detail.

Quantum Simulations

Quantum simulations are a scientifically extremely interesting development that I think has been somewhat underappreciated. In a quantum simulation you try to understand a complicated system whose properties you cannot calculate, by reproducing its behavior as good as you can with a different quantum system that you can control better, so you can learn more about it.

This is actually something I have worked on myself for some years, in particular the possibility that you can simulate black holes with superfluids. I will tell you more about this some other time, for today let me just say that I think this is a rather dramatic shift in the foundations of physics because it allows you to take out mathematics as the middleman. Instead of modeling a system with mathematics, either with a pen on paper or with computer code, you model it directly with another system without having to write down equations in one form or another.

Now, quantum simulations are really cool from the perspective of basic research, because they allow you to learn a great deal. You can for example simulate particles similar to the Higgs or certain types of neutrinos, and learn something about their behavior, which you couldn’t do in any other way.

However, quantum simulations are unlikely to have technological impact any time soon, and, what’s worse, they have been oversold by some people in the community. Especially all the talk about simulating wormholes is nonsense. These simulated “wormholes” have nothing in common with actual wormholes that, in case you missed it, we have good reason to think do not exist in the first place. I am highlighting the wormhole myth because to my shock I saw it appear in a white house report. So, quantum simulations are cool for the most part, but if someone starts babbling about wormholes, that is not serious science.

I hope this quick summary helps you make sense of all the quantum stuff in the headlines.

46 comments:

  1. Nicely explained as usual, thanks!

    Minor error: "but are save from" => safe

    ReplyDelete
  2. Wow! Thank you for this excellent and nicely pointed analysis of quantum information research. Particle and theoretical physics are not the only research games where exceptionally entertaining cheerleaders can push boring and less immediately profitable nuts-and-bolts players right off the field.

    ReplyDelete
  3. In cases of quantum functionalities being at the service of quantum technologies instead of the other way around, I have my doubts anyway.

    ReplyDelete
  4. This means that we do not need the quantum internet to be save from quantum computers. We merely need to update the current security protocols, and this update is already under way. For some reason the people who work on quantum things don’t like draw attention to that.?

    save —> safe

    ReplyDelete
  5. I think there is some confusion over announcement about traveling through a wormhole. This is not about traversable wormholes, such as portrayed in the movie “Contact” and the TV program “Deep Space Nine.” This is with Einstein-Rosen bridges. These exist with completely entangled black holes. Say we have a vast number of EPR pairs that we nudge (very carefully) into separate regions of space then these generate their own event horizons. The two resulting black holes are entangled. We might then be able to send one state of another EPR pair through the black hole, where it emerges near the second in a quantum teleportation. Of course, to make this work you must send a classical signal to tell the receiver, usually called Bob, how Alice prepared the state so he can set his receiver or detector properly. Otherwise, all Bob gets is quantum noise.

    This is a long way from sending people or large objects. This would be a sort of extreme version of the Star Trek transporter, which I honestly doubt will ever exist. Just preparing entangled black holes is a daunting task; indeed, it is probably impossible FAPP. I do though think we can arrange for quantum analogues of quantum black holes that do these things.

    Quantum computers really solve just one class of problems. These are problems with linear algebra, such as Fourier analysis or Gram-Schmidt analyses. Linear algebra is a vast area and is a backbone of mathematics. As a result, this means quantum computers could cover a huge range of algorithms. Yet many algorithms are outside that domain and quantum computers will have limited applicability. For this reason, practical quantum computer will I think emerge as processors that are used according to some master permuter of information in a standard classical computer. Computer before long may be a complex of different architectures, neural networks, spintronics and in time a quantum processor that performs limited operations.

    With quantum computing and quantum metrology, I have been kicking around in my mind an idea on how quantum computing and metrology for small numbers of qubits in effect can overlap. Can Hadamard gates and CNOTs by useful in quantum measurement or metrology? In one sense I think so, for measurements act in ways similar to a Hadamard gate, or should I say the opposite, and that quantum metrological techniques could employ some elementary quantum computing.

    There has been a lot of hyper over “China has quantum supremacy.” They set up an optical quantum computer that could compute the combinatorics of light paths. They won the game of quantum computing of light path combinatorics. This is not really quantum supremacy. I do not think it is any more than Google claiming it over a year ago.

    ReplyDelete
    Replies
    1. "Computer before long may be a complex of different architectures, neural networks, spintronics and in time a quantum processor that performs limited operations."
      Progammed - if you could still call it that - by a bunch of specialists focusing on solving differnt tasks on their part of the architecture. These heterogeneous systems would be vastly superior to anything we call "computer" now - but unfortunately, we would have to take their "solutions" as some kind of oracle, without any reasonable chance of verifying them.

      Delete
    2. The funny think about quantum computers is they solve in effect linear algebra problems. If there is one thing we probably understand very well it is linear algebra. The theory of quantum states and operators in QM is linear algebra, but with a bit of complexity if the spectra is infinite. So we might well say that in isolation this part is fairly straight forwards.

      Where things get a bit odd is a heterogenous architecture requires some sort of permuter that can analyze "this algorithm is linear algebra so off to the q-chip you go, this one is an extremal problem so off to the AI-neural net it goes and so forth. This means this requires some universal categorical system of algorithms. I have no proof of this, but I gravely suspect this is a problem related to Turing's halting problem. Such a universal decision procedure probably does not exist.

      Delete
  6. During the entire history of Moore's law, computers were working on the limit of what was technologically feasible at the time. So this statement "there’s no reason to expect anything resembling Moore’s will help us here, because we’re already working on the limit" is not really a good justification for not expecting progress. We've seen an incredible amount of progress in the last 20 years, although as you point out, we are still very far from what we need for useful quantum computing.

    ReplyDelete
    Replies
    1. Dr. Shor, I've always appreciated your reasoned approach on all of this. Despite the fact that you were the one who literally started the ball rolling in the rapid expansion of what has become a vast industry, you have always served as a voice of reason and assessment when advising others, including me back when I worked with the US DoD. Thank you for that as much as for your amazing algorithm, since it is often the quieter voices who add the greatest value.

      Delete
    2. This comment has been removed by the author.

      Delete
    3. Jay,

      When I ask "does it scale" I am, needless to say, not referring to "technological limitations" as Peter erroneously implies, but to economical limitations. As long as you can put your qubits together in one fridge it'll be fine. After that? We'll see.

      Delete
    4. Sabine,
      FWIW I find your blog post excellent as compared to 99% of any attempted popularization I've read on this topic. It do however reads as if you believe there's a technological limitation: "does it scales" is exactly what experts say when they mean "is it just a technical problem". If you were actually referring to an economical question, that sounds like an original argument you should explain in clear.
      In any case, I'd be more than happy to hear Peter Shor's take on whether he accepts or not the evidences of quantum supremacy presented by the google team. Wouldn't you?

      Delete
    5. Jay,

      I have never heard the question "Does it scale?" to refer to anything but questions of profit and economics, see eg here. I have no problem whatsoever with Peter laying out his opinions about quantum supremacy, should he desire to do so; I don't know what makes you think otherwise.

      Delete
    6. This comment has been removed by the author.

      Delete
    7. I think you can make a good case that Google's experiment demonstrated quantum supremacy, even if you could simulate the behavior of their 53-qubit computer with 2.5 days of compute time on a 250 petabyte machine (the largest-memory computing machine currently in existence).

      Delete
    8. Jay,

      "Well you sort of described the idea yourself in "As long as you can put your qubits together in one fridge it'll be fine. After that?": that's the same question as "does it scale"."

      Yes, I "sort of" described what I meant in the hope that it would further your comprehension, though, seeing your response, that seems to have been in vain.

      Delete
    9. This comment has been removed by the author.

      Delete
  7. Last year there was an intriguing article about 2-D materials and a peculiar emergent effect of potential puddles containing patterns of distributed electrons in-between. It was observed at room temperatures and was somewhat missed by the hype machine (as far as I can tell). It seemed especially interesting as depending on parameters it may open many possibilities for computing in general (also quantum computing and metrology) as it potentially allows to represent a state by fewer electrons, hence helping to reduce thermodynamic losses (even not mentioning possible usages for noise reduction/stabilization for quantum systems, etc.)

    The article is from phys.org, "Physicists may have accidentally discovered a new state of matter", https://phys.org/news/2020-02-physicists-accidentally-state.html.
    I thought that may be of interest as few people seem to cover it.

    ReplyDelete
  8. If it is possible to build a true universal computer with quantum technology, it would probably not be a Turing device, or follow that model, but incorporate principles from quantum metrology. It would be more like an analog computer, and programming it would be akin to setting up initial conditions. You'd think this would only be applicable to "computing" the behavior of quantum objects, but, since everything is a quantum object, it should be possible to model anything. Exactly how you'd go about this, I have no idea. But quantum computing doesn't seem to get you to a universal computer unless you use something other than the Turing model. Quantum metrology might actually get you there, by a somewhat circuitous route.

    ReplyDelete
  9. If I ever get a hold of a good quantum computer, I will call it "Qeniac" and mine bitcoins and become super-rich. The only problem is that it will probably dim the lights of Philadelphia when its running!

    ReplyDelete
  10. Thank you a sober look at the way quantum information technologies are portrayed. Interesting that you consider quantum metrology to be the most promising. I agree. An example that you could have mentioned is LIGO.

    ReplyDelete
  11. Another one to add to the "quantum list" - although Vadim's comment nearly touched it, is "quantum material science". Never heard of it? well, me neither - and yet, there are some interesting developments which involve such diverse topics as quasi-particles in "exotic" materials, research in the physics of catalysts for more effective and selective chemistry, and "spintronics". If this doesn't sound sexy enough for you: it will lead to much better VR and TV screens ;-) Here is just an excerpt from phys.org:
    https://phys.org/news/2021-02-ghost-particle-ml-full-quantum.html
    https://phys.org/news/2021-02-quantum-tunneling-graphene-advances-age.html
    https://phys.org/news/2021-02-potential-quantum-material-spintronic-technologies.html
    https://phys.org/news/2021-02-bimeronium-member-topological-textures-family.html
    https://phys.org/news/2021-02-inductance-based-quantum-effect-potential.html
    https://phys.org/news/2021-02-nanolight.html
    and this is not even an exhausitve list of a weeks(!) worth of research in material science on a whole new level. This why "quantum material science" would be 2nd on my list, right after quantum metrology.

    ReplyDelete
    Replies
    1. Yeah, I've since realized it's a huge and growing promising field. I saw a couple of interesting articles and mind-blowing experiments concerning fiddling with light. Many were on the level of peculiarities (albeit many were highly promising from a layman's perspective).

      What struck me most in the aforementioned article is that, first of all, it's something that is already working on the kitchen-sink level at room temperatures (yeah-yeah, the devil is in the details, nonetheless), and has a very high potential of extension to different areas of research (while the hype is with the high pressure superconductors, which do not seem to be so flexible and not so easily generalizable, at least from a layman's perspective).

      One of the authors of research, Zachariah Hennighausen, mentions that there are a lot of issues to resolve in order to come up with something industrially scalable. He attempted to assemble, describe and integrate 2D material approaches in one recent paper, "Twistronics: A turning point in 2D quantum materials" (arxiv.org/abs/2101.04501). It still seems much more promising and viable than fusion (or high pressure superconductors) and extremely extensible. I think the strongest part in this area is that it's highly experimental and accessible (also highly marketable, a lot of cool videos :-)

      Delete
  12. Great post.

    Simulating a quantum system via another, although new here, is an old idea - there were such things as analog computers that did this for ordinary, non-quantum physical systems. I've never come across one though.

    ReplyDelete
  13. I do not typically recommend YouTube clips, but if Sabine will permit it this 22 second compendium of quotes from the movie Ant-Man and the Wasp is just too deliciously apt for this topic:

    https://youtu.be/a_7JkJD3Q9A

    ReplyDelete
  14. Thanks Sabine, I can now replace "analogues" with "Quantum Simulations" which I think is an analogue!

    ReplyDelete
  15. I think Sabine is too skeptical towards quantum technology. For example, she compares it with nuclear fusion. But as long as the field still has breakthroughs from undergraduates like Ewin Tang, putting in (huge amounts of) public money is exactly the right thing to do.

    She is also underestimating the quantum magic and focuses too much on doing well-known things better. For example, verifiable randomness would be sort of magic. If it really works, and if a sizable part of industry would be able to verify (and trust) it independently for themselves, it could enable better approaches for public call for bit. Such things can have a huge impact.

    ReplyDelete
    Replies
    1. Let's see in 20 years who has been underestimating what. I'll agree that "verifiable randomness" would be sort of magic, because you can't verify it.

      Delete
    2. This comment has been removed by the author.

      Delete
    3. Jay,

      The name of that function does not mean what you evidently think it means. Did you actually read the Wikipedia entry?

      Delete
    4. Classic example of a term with two meanings.

      Jay: Jakito is not talking about a specific application of cryptography here, but about the fact that there is no way to determine whether any random-looking signal really originated from random source – or the output of, well, basically any cryptographic encoder that's worth its money.

      Delete
    5. This comment has been removed by the author.

      Delete
    6. Looking back in 20 years is fair enough. We still want to see it in our lifetime, after all.

      This part got me thinking that getting useful stuff ready and (re)usable for a wider public without waiting too long could sometimes be a good idea. Because even after it is finished, it will still take a long time before it gets (re)used by others. Of course, there are all those other tasks with higher short term priority and clearer ways to success.

      The way to verify randomness will probably work in such a way that the parties which have the possibility to verify it must already be in the "interactive loop" before the randomness gets generated. And it will not be absolute certainty, but only a boost (amplification) in the credence that it is random. Such an "interactive loop" only seems realistic for a very small number of parties.

      Delete
  16. Ewin Tang's breakthrough, as far as Wikipedia tells it, is to show that the exponential speedup of a certain quantum algorithm over a classical algorithm has a classical analalogue! That's not a breakthrough in quantum computing but a breakthrough of classical computing. Wikipedia even says she has 'dequantised' the algorithm. Personally, I think this adjective is a little confusing here.

    ReplyDelete
    Replies
    1. For me, Ewin Tang's breakthrough is without doubt a success for quantum computing. The question for me is more the practical impact of that breakthrough (and similar other breakthroughs), and whether nuclear fusion had similar directly related breakthroughs (with practical impact?) that I just didn't notice.

      Delete
    2. Where's the quantum? The algorithm is designed to run on a classical computer.

      Delete
    3. The quantum is in the fact that the research leading to the breakthrough was in quantum computing. Note that the master thesis was supervised by Scott Aaronson, whose speciality is theoretical computer science related to quantum computing.

      Delete
  17. I loved this article. Keep doing the good work.
    Love you..

    ReplyDelete
  18. Einstein told the Pittsburgh Post-Gazette on December 29, 1934: “There is not the slightest indication that nuclear energy will ever be obtainable. It would mean that the atom would have to be shattered at will.”

    ReplyDelete
  19. It’s not true that you need a million qubits to get anything useful. Actually with a couple of hundred qubits you will get far superior computer than the fastest supercomputers in the world
    https://spectrum.ieee.org/tech-talk/computing/hardware/qubit-supremacy

    ReplyDelete
    Replies
    1. Joanna,

      What you say is incorrect. You misunderstand what "supremacy" means. It merely means you have a quantum computer that can do some things faster than a conventional computer. It does not mean that this computation is actually good for something. And since any molecule can "compute" something faster than a regular computer (eg its energy bands) that means absolutely nothing. Please do some more research before claiming you've understood it all.

      Delete
  20. About quantum computer, I wonder how useful they really could be.
    You can not read a qubit without destroying it (by collapsing it), and you can not copy a qubit for the same reason. So, in a quantum computer program, you can not use the "if" nor the "loop" construct. The program flow has to be linear from start to finish.
    Some quantum algorithms are known, like Shor's algorithm for integer factorization. But how much can you do without "if" or "loop"?

    ReplyDelete
  21. Regarding: "Now, quantum simulations are really cool from the perspective of basic research, because they allow you to learn a great deal. You can for example simulate particles similar to the Higgs or certain types of neutrinos, and learn something about their behavior, which you couldn’t do in any other way."

    There is a formal analogy between the Higgs mechanism and superconductivity. The historical record provides ample evidence that analogies between superconductivity and particle physics played an important heuristic role in the development of the Higgs model.

    But what has recently hit my hot button was the possibility that this analogy may be more than a formal one but actually a physical one. The Mexican hat potential and spontaneous symmetry breaking are present in both these mechanisms.

    It has recently been discovered that irradiating a superconductor with a laser will generate polaritons which inherit their Mexican hat potential from their superconducting electron feedstock. A highly probable slow light mixing cavity will maximize the light/matter quasiparticle environment that surrounds the superconductor. It has been experimentally verified that the polaritons that are produced by the superconductor will generate a tachyonic Higgs field. These quasiparticles are called cavity Higgs polaritons.

    This serendipity opens up a physical platform where Spontaneous symmetry breaking, Bose condensation, the Higgs field, and tachyonic condensation open up the door to a realization of the predictions of string theory such as black strings and bubbles of metastable AdS space. Generating a metastable bubble of AdS space would enable the possible experimental production of topological vortex-like defects such as the 'tHooft-Polyakov monopole. Furthermore, the radius of curvature of anti de Sitter space provides an extra length scale that could allow the study of the equations of motion in a limit where the masses of the Higgs field and the massive vector bosons are both vanishing. This alone might allow the study of how matter and forces behave in a new AdS based universe let alone allow for the availability of an experimental platform on which many of the posits of string theory can be physically tested in a real world rooted experimental system.

    ReplyDelete
  22. While quantum computation indeed can only give a speedup for a certain problem, the kind of program quantum computation can give a speed up is the important one like artificial intelligence, and if it manages to reach desktop computing, computer graphics.

    ReplyDelete

PLEASE READ THE COMMENT RULES BEFORE COMMENTING.

Comment moderation on this blog is turned on.
Submitted comments will only appear after manual approval, which can take up to 24 hours.
Comments posted as "Unknown" go straight to junk. You may have to click on the orange-white blogger icon next to your name to change to a different account.