A simple genetic algorithm where each genome / individual is a single process.
The crossover implementation can only handle populations which are divisible by four. Other populations will also be processed, but mixed up with simple cloned individuals. If crossover isn't enabled there are no restrictions.
The gen_server which represents the indivdiual
The gen_server which holds the population and manages the individuals
A simple monitor which writes some statistics to a plain text file This file can be plot with
$ gnuplot simple.plotA simple monitor which can be used to trigger external events or write special formated statistics. The following example writes each 10 ticks the tickcount and the unix timestamp to the file special_statistics:
monitor_trigger:start("echo -n ~w~i~i >> special_statistics; date '+ %s' >> special_statistics", 10, format). For more details check the header of the module.
Before the first run (or after sourcecode modifications) you need to compile the .erl files. This can be done by hand
$ erlc environment.erl individual.erl monitor.erlor with the include makefile
$ makeTo start the simple default environment use
$ erl -s monitor start -s environment startand initialize evolution with
environment:tick(Count).where Count is the amount of iterations which should be calculated.
Results will be automatically saved in default_results.
The actual state could be requested with
environment:gene_pool().or
environment:fitness().For further details check the source code annotations.
Intel(R) Core(TM) i7 CPU 950 @ 3.07GHz (mostly idle):
100,000 cycles with default parameters (100 Individual, 8Bit Genome) are computed in less than 45 minutes.