genetic algorithm updates

continuing the thread from the previous post...

i'm on generation lucky thirteen, and i have doubled the gene pool size to 32 rulesets.  i've also automated the testing - scoring - averaging process, which was the bit that was so time consuming and required the majority of the mindless repetitive human input.  now i have a routine taking care of that for me, although i had to fork it off the SystemClock instead of the TempoClock to avoid overloading the scheduler.  also, it still takes about half a minute to run through each game, which means about 15 minutes for each generation.  again, this isn't because the processor is slow, but because the turns actually take miniscule (but not zero) amounts of time.  i did this because i can't figure out how to model the system in any way that does not involve forked scheduled routines.  at least im not still sitting in front of my computer pressing buttons every few seconds like a trained monkey.  well... that's debatable.

also, i have been writing out data files representing the genomes for each generation, but i discovered today that the first 10 files had been corrupted.  i have fixed the problem, and was able to save my breeders for generation 11.

i am interested in the possibility of sonifying rulesets as computer music.  obviously i'm interested in sonifying them as vocal music and instrumental music, but today i was thinking about how i'd do it with a computer as the source.  i think the concept that hurt my head when thinking about computer sonification before was that i assumed i'd let each unit resolve to a sound.  a more attractive approach, at least to me, would be to let only significant units (ie those with transitions) resolve to audible sounds, and let everything else resolve to blocks of silence.

Leave a Reply

Your email address will not be published. Required fields are marked *