Slashdot Mirror
German Scientists Create 5 qubit Quantum Register
CMan0 writes "In the University of Bonn, a team of scientists has built a 5 qubit register, using cesium atoms trapped by a laser-beam grid, The Register reports. They've been able to install an empty 5 bit register(i.e. all bits 0), change two of them to 1, and later read those 1s back. The next goal is to create an interaction between 2 bits. The full scientific article can be found here in PDF format."
This was covered on New Scientist and IndiaTimes a few days ago. Their articles:
-New Scientist
-IndiaTimes
I think it has to do with this.
Dark StateEast Coast Brewers
Several reasons - it's heavy, easier to be made neutral and easier to be trapped in a wave dipole trap (that's what they seem to be using). In a standing wave dipole trap, the first factor especially plays an important role in sustaining stability.
Plus, they've a discernible signature even in a spatially modulated environment and that helps.
So, we're what 10 - 20 years away from a QC that both gives you your answer and blue screens at the same time?
Atleast.
I would say maybe 50. It's not enough if you can get a system to do something - you need to make it reliable and scaleable.
We're still tackling the very basic problems in QC, and have a very very long way to go. Error correction is still a very big problem.
Some people, such as Alexei Kitaev, have done some pioneering work but it's still in its infancy. A long long way to go.
A qubit (for "quantum bit") is the basic information unit for a quantum computer (just as the classical bit is for classical computers). They are actually two-state quantum systems (just as ... Ok, I'm repeating myself :-))
The point is that quantum systems have properties which are not found in classical systems. For one, they cannot be just in the states "0" or "1" (in the usual notation for quantum states: |0> or |1>), but also in so called superpositions of those states. Such a superposition means that they are something like both states at the same time (remember Schrödinger's cat? That's exactly such a state, except that unlike atoms, cats cannot actually be brought into such a state). More importantly, such a superposition can extend over more than one qubit, in which case each single qubit doesn't have a defined quantum state at all, but only the whole set of qubit has. This is called entanglement.
Now, why is this so useful? Well, assume you create a set of, say, 8 qubits which are all "half zero, half one". And now you perform a normal calculation on them (but with quantum operations). Then you are actually performing the calculation on all 8-bit combinations, at the same time, i.e. for all numbers between 0 and 255. This remarkable effect is called quantum parallelism.
Now, of course there's a catch: You cannot read out more than one of the results (because reading out one destroys the superposition), and which one you get is essentially random. Ok, you now may think, I can effectively make the calculation just for one randomly selected number? So this is actually a disadvantage? Well, the point is that you can not just do "classical" calculations, but you can add operations which are not possible in classical computers. For example, there are several "half zero and half one" bit states, and you can do a quantum operation to convert one of them to |0> and one of them to |1>. Therefore you can extract properties of that result which depend not on just one of the results, but on several of them. And this allows you to actually reduce the numeric complexity of certain tasks. For example, you can search an unsorted database in O(sqrt(N)) time, instead of the classical O(N) time (N being the size of the database). The most famous algorithm is of course Shor's algorithm which allows factorizing large numbers in polynomial time, thus allowing to break public key encryption systems like PGP.
Now, there's not too much danger yet, since AFAIK the biggest number successfully factorized with a quantum computer is 15. But then, as long as 5 qubits are newsworthy, you cannot expect too much (imagine a message that someone managed to build a classical 5-bit computer!).
The Tao of math: The numbers you can count are not the real numbers.
From the theorist's perspectice it doesn't really matter how you implement this stuff - if it works, all implementations are equivalent.
At the current stage it is very reasonable to explore all possible routes to a QC (atoms, ions, photons, quantum dots, superconductors etc, a nice and readable uptodate overview is given in the Quantum Computation Roadmap): first, since it is not clear which will turn out to be most successful and second, because along the way lot of interesting physics can be expected from the coherent control of well isolated physical systems.But of course ther are (and will remain) technical advantages of certain implementations. I do not think that currently anybody knows what the most promising physical system is. Trapped ions are probably most advanced at the moment. Compared to them neutral atoms in optical lattices might two advantages: optical lattices appear to be rather "scalable", i.e., one might go beyond 5 qubits rather quickly, once complete coherent control has been demonstrated. (In a linear ion trap there will be difficulties to go beyond 10-20 ions, though very promising ways around these difficulties have also been demonstrated.) On the other hand, using neutral atoms (rather than charged ions) may make the qubits less susceptible to stray fields and other sources of decoherence.
NMR quantum computing as demonstrated by IBM has many drawbacks.
For these reasons, liquid state NMR is not be considered to be scalable. Nevertheless, the NMR people have amazing control over the operations (logic gates) they can perform, and these ideas may (and have) fed back to other implementations. Moreover, there are attempts to overcome the mentioned difficulties (while keeping some advantages of NMR) by using nuclear spins in cold solids following Kane's proposal).First, there's not a single quantum system doing the computation, but rather some 10^20 molecules in the liquid - and you need so many to generate a detectable signal.
Second, the NMR quantum register cannot be properly initialized, rather it is in a nearly random state with only a slight enhancement of "0" over "1". This is part of the reason why so many systems are needed and it prevents the currently realized systmes from displaying any entanglement.
Finally, it is not clear how to scale such a system (increase the number of nuclear spins on a molecule): the larger that number, the more difficult it is to address individual qubits.
Just to expand on this post, you can treat |0> and |1> as vectors. Well actually they are vectors.
So |0> is [1,0] and |1> is [0,1]
So a "superposition" is simply A*|0> + B*|1>
= [A,B]
Nothing particulary fancy or anything.
The analogy I used to explain it to my dad is this:
Imagine I have a light bulb, with a dimmer switch. I could set this to a dimmer switch to anything in between on and off. Theoritically I could store an infinite amount of information in the dimmer switch. Imagine I took a large book, converted it to hex, and turned that into one long number. Then I prepended 0. to the front.
So you get "0.1939434....". Then I set the dimmer switch to that exact value.
But, if I want to look at the light, for some reason, I can only see if it's on or off. The chance I see it as being on is the same as the dimmer switch setting. (So if it's set to 0.5, then I have a 0.5 chance of seeing it as on, and 0.5 chance of seeing it's off).
I'm stretching this analogy a bit, but you can see that despite storing anything I want, I can still only read it as on or off.
So.. how do we use this usefully? We don't really know many practical uses, but what you can do is do calculations.
Say you put two of these lights in a room. Both are set to 0.5 brightness. With the case of the lights, the total brightness is now 1. So we've gone from having probability, to something definite. You are always going to see that as being on.
The analogy doesn't quite fit, but you can see how you can use the underlying probability to do calculations and get a definite answer.
That isn't even true with real quantum particles. You can manipulate force fields in order to skew the quantum wavefunctions, making it more likely for the outcome to be one option than another.
Yes, the behavior is random in the purest mathematical sense, but just because something is random doesn't mean it's unpredictable or uncontrollable.
Suppose I had a 12-sided die, which had the number 1 on each face except for a single face, which had the number 2 on it. Clearly, the outcome of the die toss is still randomly determined, even though the number 2 is only 1/11th as likely as the number 1. If I were betting on such a die, I would certainly bet on 1.
Manipulating the potential to change the quantum wavefunction is sort of analogous to changing the shape of the die. If I squash the die so that one axis is longer than the other, and the "2" face happens to fall on the end of the long axis, then I have dramatically reduced the probability of the die ever coming up 2. (Try tossing a book in the air and see how many times it lands perfectly on its spine. Possible, but very, very unlikely.) It could happen, but perhaps only one in a million times.
All theories that try to explain what we observe without entanglement have been disproven time and again. Bell's inequalities have been violated to 10 (or was it 50?) standard deviations and in various physical systems.
[1] One "regime" of quantum mechanics that has not been much explored (and where quantum computers would come in handy) is massive multi-particle entanglement: can hundreds, thousands or millions of particle be in strongly entangled states? Theory predicts it, few doubt it, quantum computers require it, but it has not yet been demonstrated. (If that was the intent of your comment, see this as a clarification, not an objection.)Quantum mechanics is is probably the most tested theory around, and entanglement is an integral and unavoidable part of this theory - I dont think there is particular need to "prove that it is for real"[1]. Einsteins dream of a complete, local realistic theory to describe nature is unlikely to come - nature just does not behave that way. And, frankly, I think it's much more fun that way...