Print

Print


Hey,

> How can we provide customized (not randomly generated) initial
> population for a genetic algorithm in ECJ?

The initial population may be loaded from a file, for example:

pop.subpop.0.file = <filename>

specifies the the entire population 0 will be read from a file named
<filename>

> And one more thing, how can the execution time of the algorithm be
> measured?

In linux at least, you may launch ECJ with

time java ec.Evolve ....

which will return all sorts of time statistics.  I am not sure how to do
this in Windows, but there should be a simple solution (one simple
approach may be to slightly extend the ec.Evolve class to compute this
information).

Regards,

Liviu.