Tiny Holes Advance Quantum Computing

Nick writes "Worldwide, scientists are racing to develop computers that exploit the quantum mechanical properties of atoms - quantum computers. One strategy for making them involves packaging individual atoms on a chip so that laser beams can read quantum data. Scientists at Ohio State University have taken a step toward the development of quantum computers by making tiny holes that contain nothing at all. The holes - dark spots in an egg carton-shaped surface of laser light - could one day cradle atoms for quantum computing."

Wow

[Anonymous Coward]

Scientists at Ohio State University have taken a step toward the development of quantum computers by making tiny holes that contain nothing at all.

Now I know people often criticise Slashdot for having summaries that contain obscure terms without explaining them, but I think it's going a little overboard to explain what a hole is :)

Re:Wow

[Vo0k]

Actually, most of holes on Earth are full on air. Even void isn't quite empty. If you have a couple of atoms forming a particle, the space between them isn't quite empty either - they partially overlap, the uncertainity principle says they "partially are" there. The idea is about making small holes with REAL void - no particles, no photons, no "with a little probability, there" electrons, just total null. Not quite easy. I, for one, can't quite imagine how are they going to stop neutrinos from entering that space...

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:Great principle

[stevok]

Not exactly. Quantum computers can simulate classical computers with no problems. That's one of the tenets of quantum computation. I would love to see a 747 parallel park in Manhattan. Also, the fact that quantum computers can factor large integers efficiently necessarily implies that they can do other NP-complete problems efficiently, such as the traveling salesman problem. If we can ever get more than seven qubits to behave, we'll be amazed by the things quantum computers can do. But, alas, scientists have only implemented Shor's Algorithm for factoring integers on one number. 15. And hot damn, they got the factors right, 3 and 5. Yes, IAWAUGTOQC (I am writing an undergrad thesis on quantum computation).

Someone please clarify

[karvind]

From the article: "We're pretty sure we can trap atoms -- the first step towards making a quantum memory chip," Lafyatis said. A working computer based on the design is many years away, though, he cautioned. In fact, Christandl suspects that they are at least two years away from being able to isolate one atom per trap -- the physical arrangement required for a true quantum memory device.

1. What is the working principle behind this (mechanism of trapping) ?

2. Are these experiments performed at room temperature ?

3. How do they ensure they have trapped one "desired" atom and not more atoms and not some other impurity?

4. How is the laser prevented from interfering with lattice (non-desirable interactions) ?

5. What is the decoherence [wikipedia.org] time which governs if you can really do any computation before the result is lost ?

This is indeed an important step forward. But alas the student is graduating in august and I hope there is someone to followup on this work:

Theoretically, if they release the atoms above the chip in just the right way, the atoms will fall into the traps. They hope to be able to perform that final test before Christandl graduates in August.

Related article

[c0ldfusi0n]

Physicists could soon be creating black holes in the laboratory [sciam.com]

When shall we get pet dark holes?
Imagine cleaning the house with one of these around!

The Law.

[k96822]

...and this is why Moore's Law will continue, even though Moore himself says that it won't. Never underestimate the cleverness of the Human.

Re:Great principle

[Urkki]

• Quantum computers can simulate classical computers with no problems

So, what kind of scale are we talking about here? To simulate, say, a million-transistor CPU and a megabyte of RAM, how many qubits would you need? About as many as you need transistors, or radically less?

If the answer is millions, then I think my comparison to a jumbo jet is valid, as we're probably about as far from a quantum computer simulating even a 4004 with hundreds of bytes of RAM, than we're from ubiquitous flying cars replacing jumbos ;-)

Re:The Law.

[Quiet_Desperation]

Never underestimate the cleverness of the Human.

*cough*fusionpower*cough* The eternally "just around the corner" technology.

Hey, I tease mankind. :)

Re:Great principle

[stevok]

Like the article said, the issue isn't processor speed, it's algorithm time as a function of input size, i.e. logN. Factoring integers takes an exponential amount of time on classical computers. The best known classical algorithm (called GNFS) is O(exp((logN)^{1/3}(loglogN)^{2/3})), whereas Shor's algorithm can factor N in O((logN)^3) time. But, Shor takes roughly 2^N qubits to factor N. So, if we're talking about factoring a 200 digit RSA number, that's a whole crapload of qubits to control. Many orders of magnitude more than we can control now. In short, you're absolutely right about quantum computers being completely impractical until there are some huge breakthroughs in engineering and physics. This is why I love being a math major. We don't have to worry about silly things like actually building a quantum computer. We just sit around and daydream about how a quantum computer would work, then when we've got it all figured out, we blame the physicists and engineers for not building one.

Re:Wow

[madaxe42]

You can actually guarantee that it will be empty, by creating wave functions that overlap in such a fashion that the probability of a particle being in that space is, in fact, 0, or, by creating wavefunctions which when combined state that the probability of there not being something in that location is infinite. Picture two asymptotic curves joining at a vertical axis, mirrored.

There are a lot of extremely odd quantum effects which aren't physically possible, in any classical or comprehensible universe, however do happen. For instance, it's possible to create a negative temperature. Not negative, as in minus 22 farenheit, but negative, as in below absolute zero!

This happens when you rapidly invert the polarity of a magnetic field in which is contained a bose-einstein condensate - in the time that it takes for the condensate to re-align it's spin, it has a rapid change from a negative temperature to a positive temperature once more. The energy of a negative temperature is, actually, greater than that of an infinite positive temperature!

Anyway, enough quantum rambling. If you don't believe me, look here [google.com].

Re:Factoring is NOT known to be NP-complete

[ortholattice]

While I have not read the paper you mention completely enough yet to understand its argument, let me point out the obvious fact that nature, by definition, "simulates itself," i.e. is its own computer. Now, the processes involved are extremely complex; just to simulate the processes going on inside of a single atom can take years of computation time on an ordinary computer, yet it happens essentially instantly in actuality. Is the problem of simulating nature on a submicroscopic level NP-complete? I'm not sure, but certainly it's way beyond the realm of what present-day computers can do. The question is, can we somehow harness this incredible built-in computational power to solve more general problems?

Re:Great principle

[Old Telco Guy]

Indeed! So you could set up an NP complete problem, say a routing problem, as a superposition of all solution sets, and as long as you set things up so that the only state that survives is the shortest/best/whatever route, you'd be doing NP complete problems in linear time (albeit with a lot of quantum "memory"). This means you could have a chess/go machine that played a perfect game with no heuristics necessary. Telephone routing would be perfect, business routing (airlines, FAA airways, mail, whatever) would be routed with perfect efficiency, etc. Cool.

See the Animation (.mov and .wmv)

[Anonymous Coward]

Here are posted movies of the experiment

http://researchnews.osu.edu/archive/eggcarton.htm [osu.edu]

Re:Factoring is NOT known to be NP-complete

[Planesdragon]

While I have not read the paper you mention completely enough yet to understand its argument, let me point out the obvious fact that nature, by definition, "simulates itself," i.e. is its own computer.

You're misusing that first word.

A "simulation" is a testable model of something, usually created for a specific kind of testing, that specifically is NOT the thing itself. By way of example, consider "simulating" adding numbers on a computer chip. Most of the time you wouldn't bother doing it, because it's easier just to actually add them.

But you could "simulate", oh, a computer chip running a very-complex program, just by having it do something that's needlessly complex. (Like, oh, performing random operations on a random number of a random size.)

When you start dealing with Quantum Mechanics, it's important to stop every now and again, and remember that what we have for QM is a *simulation* -- i.e., in certain fundamental ways it's simply wrong, but the wrongness is OK because we don't need to know everything about how a process works for that process to work, or even to come up with a new process.