Quantum capacitance
Dean Esmay is excited about an article in quantum computation. To quote from the article:
As I explained in Dean's comments, this isn't anywhere near as exciting as the article makes it sound. Generally, reading the papers (or at least the abstracts) makes the actual results of the experiments clearer. The two papers mentioned in this article can be found here and here. Reading a qubit without collapsing its wavefunction is, to the best of our knowledge, physically impossible. It's not something you want in a quantum computer either, as the quantum algorithms won't work unless you collapse the wavefunction. This is best understood in the context of entanglement. Suppose that you have a three qubit register which is in an equal superposition of two values, 100 and 010. Both of these values are possible solutions to the problem you are solving, but 110 is not. Because these three qubits are entangled, once you read the first qubit, the others collapse into the appropriate state. So if your highest order qubit is read as 1, the second qubit collapses to 0, and the third is 0 as well. If your highest order qubit is read as 0, the second qubit is 1, and the third is 0. You read either 100 or 010. Now suppose you could read each qubit individually without collapsing the wavefunction. You could tell that the highest order qubit is half one and half zero, the next is half one and half zero, and the third is 0. Knowing only this, you might conclude that the solutions to the problem are 000, 100, 010, and 110. There is no way of telling, with just the above information, that 000 and 110 are not solutions, and that collapsing the superposition will only give you 100 or 010.
Of course, this experiment is impossible, as reading out a value of a qubit does collapse the wavefunction. There is a theory of doing non-destructive measurements, but these reduce down to ways of transferring information from a qubit to another quantum system, which is ultimately just two qubit operations. What is wanted for a non-demolition qubit measurement is something that collapses the wavefunction from a superposition, but does not disturb the probabilistic distribution of the states which results. Schemes using the resonant frequency of weakly coupled systems are one way of doing this, and Hakonen claims to have achieved it in the paper mentioned above. A friend (and former labmate) of mine has done something similar, as her paper shows. Personally, I've always favored very fast, strongly coupled measurements, using RSFQ superconducting electronics for example, instead. This gives you a very quick readout, on a picosecond timescale, collapsing the wavefunction quickly but measuring the result before it has time to change. The speed with which this can be done is part of the advantage.
Delsing and colleagues at Chalmers University began by embedding their Cooper-pair transistor in a resonant circuit. Next, they cooled the device down to millikelvin temperatures and measured how the phase of a radio-frequency signal changed when it was reflected from the circuit. Based on these measurements, the team was able to show that the device behaved like a quantum capacitor. Hakonen and co-workers in Helsinki and Moscow group employed a similar technique. Both teams found that the devices behaved as predicted by theory.
The effect could be used to read out quantum bits (qubits) in a reliable way because the quantum capacitance of the excited state of the qubit has the opposite sign to the ground state. These states could be used as the "1s" and "0s" in a quantum computer. Indeed Hakonen and colleagues have already used this approach to read the value of a qubit without changing its value — which is almost always a problem when measuring the quantum state of any system.
As I explained in Dean's comments, this isn't anywhere near as exciting as the article makes it sound. Generally, reading the papers (or at least the abstracts) makes the actual results of the experiments clearer. The two papers mentioned in this article can be found here and here. Reading a qubit without collapsing its wavefunction is, to the best of our knowledge, physically impossible. It's not something you want in a quantum computer either, as the quantum algorithms won't work unless you collapse the wavefunction. This is best understood in the context of entanglement. Suppose that you have a three qubit register which is in an equal superposition of two values, 100 and 010. Both of these values are possible solutions to the problem you are solving, but 110 is not. Because these three qubits are entangled, once you read the first qubit, the others collapse into the appropriate state. So if your highest order qubit is read as 1, the second qubit collapses to 0, and the third is 0 as well. If your highest order qubit is read as 0, the second qubit is 1, and the third is 0. You read either 100 or 010. Now suppose you could read each qubit individually without collapsing the wavefunction. You could tell that the highest order qubit is half one and half zero, the next is half one and half zero, and the third is 0. Knowing only this, you might conclude that the solutions to the problem are 000, 100, 010, and 110. There is no way of telling, with just the above information, that 000 and 110 are not solutions, and that collapsing the superposition will only give you 100 or 010.
Of course, this experiment is impossible, as reading out a value of a qubit does collapse the wavefunction. There is a theory of doing non-destructive measurements, but these reduce down to ways of transferring information from a qubit to another quantum system, which is ultimately just two qubit operations. What is wanted for a non-demolition qubit measurement is something that collapses the wavefunction from a superposition, but does not disturb the probabilistic distribution of the states which results. Schemes using the resonant frequency of weakly coupled systems are one way of doing this, and Hakonen claims to have achieved it in the paper mentioned above. A friend (and former labmate) of mine has done something similar, as her paper shows. Personally, I've always favored very fast, strongly coupled measurements, using RSFQ superconducting electronics for example, instead. This gives you a very quick readout, on a picosecond timescale, collapsing the wavefunction quickly but measuring the result before it has time to change. The speed with which this can be done is part of the advantage.
