Slashdot Mirror
Computers That Solve Problems Without Being On
Iron Monkey writes: "Nature has this article about how quantum computers can theoretically solve problems without ever actually being turned on! Maybe California can use a few of these to solve their energy crisis - the ultimate in conservation."
Suppose you have a bomb with a trigger on its tip, so sensitive that a single photon hitting it will make it explode. Suppose further that you have some good bombs and some duds with stuck triggers that don't move. Can you tell which bombs are good and which aren't? Well sure, just hit it with a photon and see if it blows up. But can you do it without blowing up the bomb? Yes! (Well, probabilistically yes.)
Here's how it works: shoot a laser beam through a beamsplitter, bounce the two beams off of mirrors (one of which is attached to the trigger of a bomb) which redirect the beams back into another beamsplitter, with detectors on the other side. You won't understand the setup without looking at the picture -- I will be referring to it.
A laser beam is a coherent superposition of a bunch of photons. What happens if you turn the intensity down so much that you're getting single photons out? When the photon reaches the first beamsplitter, you might think that the photon either goes one way or the other. But quantum mechanics says that the photon enters a superposition state in which there is a 50% probability that it took the upper path and a 50% probability that took the lower path -- you won't know until you collapse the wavefunction by measuring which path it went through (by putting a detector along the path or something). (I'm being vague here: it's not that it went one way but you don't know until you measure it; rather, "which way it went" is simply undefined until you measure it, and if you don't measure it then it's never defined.)
So anyway, the photon hits the first beamsplitter and enters a superposition. If the bomb is a dud (fixed mirror), then this is an ordinary interferometer. At the second beamsplitter, the two beam wavefunctions (representing a single photon) interfere with themselves to produce a state with 100% probability of being detected at B and 0% probability of being detected at A. (You can see this from symmetry: the beam enters the system horizontally and has to come out horizontally too since the apparatus is symmetric.) This is precisely what you would expect classically with wave interference, by the way. Nothing too odd.
However, suppose that the bomb isn't a dud. Then the the impact of a photon on the mirror is free to move the trigger and set off the bomb, so the bomb serves as a measuring device! If it a photon hits it, the bomb will explode, so you have definitely measured that the photon took the lower path. What happens here? Like I said, there's a 50% probability that a measuring device (such as the bomb) inserted into the lower beam will measure a photon. If that happens, the bomb will explode. But what if the photon is measured to not take that path (by virtue of the bomb not exploding)? Then with 100% probability it took the upper path. When it hits the second beamsplitter, it's just as if it hit the first beamsplitter, since there's no interference from the other beam -- we know that nothing went that way. So with 50% probability it goes to detector A, and with 50% it goes to detector B.
The upshot: if the bomb is a dud, then you will get a photon at B with 100% probability. If the bomb is good, then you get an explosion with 50% probability, a photon at A with 25% probability, and a photon at B with 25% probability.
The point: if you get any photons at A, then you know for sure that the bomb was good. But you didn't actually ever send a photon to the bomb to find out! This is a "counterfactual" -- you are obtaining information about something that never happened (a photon hitting the bomb), but could have!!
Of course, you don't have an infallible scheme. Half the time you have a good bomb your test blows it up, and half the time it doesn't you can't tell whether it was good or not (because you got a photon at B). But 25% of the good bombs are provably good (without blowing them up). It turns out you can cascade this process to make the probability of detecting a good bomb as high as you want.
This effect is known as "quantum non-demolition" and has been experimentally verified (not with real bombs of course). You can use it to measure things using photons without destroying the photons (normally a photon is destroyed whenever it interacts with something -- it is absorbed).
I am intrigued/scadalized/cooled out by the article and (therefore) don't have the background to dispute it.
..or is this based on probability of some kind of tunneling right under the potential energy hill from start point to endpoint in a finite amount of time? (And is there such a "hill" in quantum computing, forgot to ask that too).
But what about:
-energy required to set up initial values so that the answer is all zeroes
-energy required for cooling or otherwise insulating/maintaining computer during the time it will be computing and simultaneously off
-who's going to police those qubits and tell them not to cannabilize energy from that environment (and presumably return it.. oops don't want to go there)
-would you need a reservoir of energy attached to the thing so that it would be theoretically possible for the system to go up the energy hump if calculation (in other universe required it)?
-do you ever even turn a quantum computer on anyway? seems a delicate enough computer could get by without humans intentionally putting power to it
-(back to the question of energy required to set up the computer): is there not a law which requires energy to create or change information, or is this a silly misconception. If so, are they not just taking care of energy expenditures before getting to the calculation stage, leaving a lot of the actual energy goings on in the unfathomable, unaccountable finite time span of computation? Or is there something else going on?
** It seems there is something more interesting going on, but there is neither mathematical meat for the professional nor a real explanation for the layman. The Royal Society Proceedings are too briefly abstracted to get anywhere. Could someone paraphrase or post the text of those proceedings or is this going to be an exercise in frustration?
In particular what exactly are the limits to the kind of probing you can accomplish which was mentioned in that abstract?
You are completely misunderstanding and need to read the original paper rather than the nature article. The strategy in the paper is not about saving energy. Quantum computers already use zero energy because they are reversible computers. The authors are not in the least bit confused about what they mean by off but it's not surprising you are confused because the Nature article is describing things in a stupid manner to be sensationalist.
--
-- SIGFPE
Quantum entanglement is nothing more than the superposition of multiparticle states that aren't simple tensor products of the individual particle states. That's it. In fact your description is just as inaccurate as his. If two particles are entangled it doesn't mean that knowing the state of one tells you the state of the other. Things can be slightly entangled in that you might be able to glean some information about one by observing the other.
---- SIGFPE
"Michison and Jozsa describe a scheme for probing all the possible states of a quantum computer, including that in which all the 'switches' are 'off' -- that is, in which the computer is not turned on."
Just because your register is zeroed doesn't mean the chip is off.
I'll accept that the computer is off if it's not using any energy.
Cheerio,
Link.
True, but calculating this proof will definately need one of those computers!
It took a bit of head scratching and squinting, but I did finally manage to figure out how the thing works.
It uses some of the same theory as I would expect the "quantum computer" to use. In this case, they use a bomb which has a trigger which is sensative to a single photon of light. A dud bomb will pass the photon, but a live bomb will stop it. The experiment shows how it is possible to detect, using quantum mechanics, whether a bomb is live or a dud, without actually exposing it to the photon of light, and thereby exploding it.
I'm sure this is a much simplified version of what they're planning for the computer, but it's described in terms that pretty much anyone can understand, even though it's not obvious at first.
Help find a cure for Gidget.
I always tended to solve my Windows problems by just not turning the box on.
Not only doesn't this post have anything to do with the article, it's just plain wrong. I got as far as
"Quantum computers are based on the concept of quantum entanglement, the ability of a quantum state to exist in a superposition of all of its mutually exclusive states"
and stopped reading. What he describes is just called superposition; quantum entanglement is when two particles' quantum states depend on each other in such a way that once you know the state of one, you also know the state of the other. To read more about it, look up the EPR paradox.
Don't spout off BS, and please, moderators, don't moderate it up!
Okay, let me give this a try...
It's not so much that it picks a *right* answer out of all possible answers, but rather, the impossible states collapse and you are left with a quantum superposition of all the possible states. (Or, to try to put it more simply, those answers that cannot exist cease to exist, leaving only those answers that can exist.)
The White House said: White House spokesman Ari Fleischer was adamant Monday when asked whether the president would ask Americans to stop using so much energy.
"The president believes that it's an American way of life, that it should be the goal of policy-makers to protect the American way of life. The American way of life is a blessed one."
I say: Iron Monkey, what are you smoking? It's UNamerican to conserve. It is my patriotic duty to be an energy hog. Who needs these new-fangled 'putin machines work'in without 'tricity?
How I wish I had a quantum girlfriend that could take care of my problems without me having to turn her on.. no foreplay necessary...
A Zen machine.
Milo
Will they come with a sticker "Infinite Monkeys Inside"?
And if you watch the screen while it's running, will this collapse the computer's state and break it?
One line blog. I hear that they're called Twitters now.
They've tried that before, the answer was 42.
But what was the question again?
Does this machine run on software that doesn't have to be written?
Problem solved.
--
Friends don't let friends use multiple inheritance.
I swear, anything that is published through Nature has to be the most questionable of all things. Unfortunately, this argument is missing the most important part of quantum computing, the collapsing of the states into the final results. Without that you end up with unknown states, which you can guess the probabilities for all the possible outcomes, which in the end makes you do the whole thing by hand anyway. Trust me, I've got plenty of quantum particles making up my body and I am the master and doing a whole lotta nothing. By their logic, I should not have failed differential equations.
Why, of course, it should be well-known that while nobody looks at a quantum computer, it can be on, off, or both, right?