Think genetic programming. Current "populations" are in the 10^3-10^6
range. With a 64-qubit QC, you could have a population of sixteen
billion billion. Heck, with a 64-qubit QC, you could probably string
instructions together completely at random and wind up with the most
powerful solution immediately. Who needs evolutionary computation? As
for quantum neural networks, it seems to me that the number of neurons
simulated would simply equal 2^qubits or the amount of memory available
- practically speaking, the amount of memory available.
The real excitement comes when you consider the speed involved. The
human brain is composed of 10^11 neurons with 10^3 synapses each
(approx.) Assume one KB per synapse - you need 10^17 bytes of memory,
or a hundred thousand terabytes. The real kicker is that simulating a
single pulse of a real synapse probably won't take more than a thousand
operations. Real neurons do maybe a thousand pulses per second - but a
1-tHz computer could do a BILLION pulses a second, if it only had to
simulate one synapse. With a QC, each synapse is simulated in a
different "branch", so a QC simulating a neural network, if it works at
*all*, will work at a million times human speed. Subjectively, that's 1
year every 30 seconds.
--
sentience@pobox.com Eliezer S. Yudkowsky
http://tezcat.com/~eliezer/singularity.html
http://tezcat.com/~eliezer/algernon.html
Disclaimer: Unless otherwise specified, I'm not telling you
everything I think I know.