[hcarvalhoalves]

> But here's the tricky part. The scientists can put the rubidium atom in superposition, so that it is simultaneously in that energetic state and not in the energetic state. It's on and off. Because of this, the photon both does and does not enter the mirror, mingle, and gain its polarization change. And the photon, by virtue of having both changed and not changed, carries that superposition information and can bring it to a different atom-based qubit.

I know almost nothing about Quantum Mechanics, but this sounds amazingly ingenious.

So what did they do, some sort of parallel universe transistor where the rubidium atom acts as the gate? If you assemble a processor out of this, will it compute all possible computations at the same time? And, last but not least... how do you make it converge to the computation you actually want?

Quantum computers make my head spin.

[tedsanders]

Yeah, one interpretation of what a quantum computer does is that it performs all computations in parallel. However... you run into problems when you try to read the output of this computer. In order to read the output you must measure a quantum state, and when this happens you only get to see one answer at random, not all of them. And that's not really useful. But if you can design an algorithm so that all of the parallel computations add together to form one answer (i.e., get all the wrong answer's probabilities to cancel out), you can get an exponential speed up. The key takeway is that quantum computers are not faster for general problems - quantum computers are only faster for problems where a special algorithm exists (like Shor's algorithm for factoring).