Saturday, March 11, 2017

Five years to commercial quantum computer

The Economist mag reports:
A BATHING cap that can watch individual neurons, allowing others to monitor the wearer’s mind. A sensor that can spot hidden nuclear submarines. A computer that can discover new drugs, revolutionise securities trading and design new materials. A global network of communication links whose security is underwritten by unbreakable physical laws. Such — and more — is the promise of quantum technology.
No, none of this is happening.
Quantum mechanics replaced wholesale the centuries-old notion of a clockwork, deterministic universe with a reality that deals in probabilities rather than certainties — one where the very act of measurement affects what is measured. Along with that upheaval came a few truly mind-bending implications, such as the fact that particles are fundamentally neither here nor there but, until pinned down, both here and there at the same time: they are in a “superposition” of here-there-ness. The theory also suggested that particles can be spookily linked: do something to one and the change is felt instantaneously by the other, even across vast reaches of space. This “entanglement” confounded even the theory’s originators.

It is exactly these effects that show such promise now: the techniques that were refined in a bid to learn more about the quantum world are now being harnessed to put it to good use. Gizmos that exploit superposition and entanglement can vastly outperform existing ones—and accomplish things once thought to be impossible.
No, particles are not "here and there at the same time", and it never happens that "change is felt instantaneously".
Other aspects of quantum theory permit messaging without worries about eavesdroppers. ...

The advantageous interplay between odd quantum effects reaches its zenith in quantum computers. Rather than the 0s and 1s of standard computing, a quantum computer’s bits are in superpositions of both, and each “qubit” is entangled with every other. ...

Google said last week that such machines are only five years from commercial exploitability. This week IBM, which already runs a publicly accessible, rudimentary quantum computer, announced expansion plans. ...

Fortunately for quantum technologists, the remaining challenges are mostly engineering ones, rather than scientific. And today’s quantum-enhanced gizmos are just the beginning. What is most exciting about quantum technology is its as yet untapped potential. ... For much of the 20th century “quantum” has, in the popular consciousness, simply signified “weird”. In the 21st, it will come to mean “better”.
I would be in favor of giving a Nobel Prize to anyone who can demonstrate quantum supremacy. I guess the quantum computer advocates are not even going for the prize, because they imply that the big problems have already been solved, and only an engineering problem remains.

Quantum mechanics does not even help with protection from eavesdroppers, in spite of the claims. This is all a scam. In the 21st, quantum will mean scam.

If you don't believe me, wait five years and look for those commercial exploits. Research is always at least a couple of years ahead of commercialization, so Google is essential saying that quantum supremacy will be proved in the next 2-3 years.

The fallacy in all this is the belief that changes can be transmitted instantaneously, or that a particle can be in two places at once. If you believe those, it is not much more to believe that communications can bypass eavesdroppers and that particles can do simultaneous computations in parallel universes.

If quantum mechanics was invented in 1925, and quantum computing is just an engineering problem, why didn't anyone realize that until about 25 years ago?

The answer, I'm afraid, is that the founders of quantum mechanics understood the theory better than today's physicists.

Here is quantum computing, in a nutshell. Suppose I want to test a million numbers for some property. I toss a coin, splitting the world into two universes, one where I see heads and one where I see tails. I do it again, splitting those 2 universes into 4. I continue for 20 tosses, keeping a record of the sequence of heads and tails. Now I have a 20-bit number and I test the property for that number. Since 220 is about a million, there are a million copies of me in parallel universes, each testing a different number. Now I destroy the coin, and all those million coalesce into one, and if we all agree on the outcome, then it must be the same for all million. That is how quantum computers do parallel computation.

Scott Aaronson would complain that this explanation is oversimplified to the point of being wrong, because it leads you to expect exponential speedups, and that is not always possible. He likes the subject because it generates new classes of complexities for abstract theorists like him.

Okay, he is right that the computation in parallel worlds cannot be so cleanly separated and combined. I accept that. But the example is essentially right because the alleged speedups come from computations on entangled states, and you have avoid the measurement that collapses the entanglement.

It is as if they take the Schroedinger Cat metaphor too literally. They know that as soon as they open the box, they will see a dead cat, but somehow they think that a half-alive half-dead cat is going to do some meaningful work for them.


  1. Taken seriously, Schroedinger's cat is not and never has been a probability, it's a cat. A cat is a creature, a probability is numerical abstraction, and is utterly dependent upon other numbers acquired by some form of measurement, thus it is a second hand calculation, much like a velocity or an average, and has no ability to affect or inform causality, or produce another universe in violation of all conservation of mass and energy.
    If boxes must be employed to illustrate any kind of outcome, think of Reality as much more like a six sided die acting as a box containing six different kinds of cats. When you roll the die, The consequence of one side being up means by default that the other five sides were NOT also up. While an event may be represented by a number, an event is not a number, and can't be reified into a number. While a map can be sometimes useful, the map is NOT the territory. I can rip the map in half, I can't do that to the territory, and if I change the map in any way, the territory does not also change sympathetically as it is not informed by the map. Pretending that a altered representation causes the actuality to change accordingly is called magic, and has nothing to do with physics. Reality is all about the 'this OR that', and not the 'this AND that AND that AND that AND that unto infinitum'.

  2. You are missing the obvious objection from Davies: "generic entangled state of more than about 400 particles will have a quantum state with more components than I_universe, evolving in a Hilbert space with more dimensions than I_universe." Although the argument is a "bit" more complicated, the basic point is that one easily see how outrageous a QC would be. After that observation, it's basically fringe science. No one should take it seriously.