Quantum Decoherence & Measurement Time

[michael879]

I was reading about quantum computing and I came across this:
http://en.wikipedia.org/wiki/Quantum_computing#Quantum_decoherence
It seems to suggest that fast logic gates don't decoher qubits while slow ones due. If this is true, it would seem like the term "measurement" in QM is a function of time. I might have misunderstood this but it seems like slow logic gates act as "measurements" while fast ones dont. I don't see any difference between sending a qubit through a logic gate and bouncing a photon off of it (or some other form of measurement). Does this mean that if you could measure a qubit fast enough it wouldn't decoher??

 

[f95toli]

There is something called "adiabatic quantumt computing" (which is what D-Wave are using for Orion) but I suspect that is not what you are referring to.

First of all, quantum gates do NOT cause a QC to decohere if they are properly implemented; quantum gates are used in systems which are designed to USE quantum coherence in some way. The only time a QC MUST decohere is when you read out the final result.
However, the problem is that a gate operation takes some time, let's say 10 ns. (note that a "gate" does not neccsarily refer to anything more than a certain sequence of pulses in quantum computing; it is usually not a circuit) .

Hence. if you have a coherence time of e.g. 1 microsecond that means that you can in theory perform 100 gate operations before the system collapses.
Now, there is something called quantum error correction algorihms which (in theory) can be used to build QCs that can operate for times much longer than the coherence times of the individual qubits. This means that the requirements on the qubits are somewhat relaxed. The bad news is that in order to implement these algorithms you need quite a few qubits, at least a few thousand, and you also need to be able to entangle all of them for the time it takes to perform the error correction.
This is way beyond what is possible at the moment but it DOES mean that there is a threshold beyond which we should be able to build practical QCs of just about any size.

This means that quantum correction is not really useful at the moment(I am not counting re-focusing pulses etc as QC) so instead we simply try to extend the coherence time of our qubits as much as possible and then we manipulate them using very short pulses before reading them out (there are also various "tricks" one can use such as non-demolition measurements etc).