Wednesday, November 17, 2010

Possible functions to evolve using reaction networks

How would simulated evolution resolve the following challenges:

1) Measure the variance (noise) of a signal
2) Measure the frequency of a signal -- can be related to (1)
3) Reverse of (1), i.e. increase noise based on deterministic signal (without affecting mean)
4) Reverse of (2), i.e. increase frequency based on amplitude of an input signal
5) Control the width of a bimodal distribution based on a deterministic signal
6) Adapt to an external pattern of events (i.e. learn cause-delay-effect relationships) purely through the reaction network -- this would probably be a combination of signal processing + memory