Monday, June 22, 2015

misidentification

So there formed a problem with my program, and I didn't catch it because it was in a area that was pretty solid. So my program has the possiblility to die... ie no growth in population fitness. This could happen if the population zero is bad. The problem was that all the runs had the same population zero... or so I thought. I manipulated my popZero class so that its population is not mixed and is in order... this is not a problem, its like when you eat do you start from left to right, up to down bla bla bla... I found that some of the individuals read from the PopZero file spawned  genes from nothing. I saw very closely and it turns out that the scanner reads the line, but sometimes forgets what line its on and reads a gene above or below the line. for example the file is
a b c d e f
g h i j k l
m n o p q
 and you want the g line. the scanner might be like oh  here--- g h i j k l
but it might make a mistake and be like --- g h c d p l
Its hard to read just one line it seams. So this error introduces randomness in the program form the start.
I think its cool... in a ha ha weird way. The thing is now how to fix it. I guess read a line over and over but Ill work it out.

Ps. I was tired of editing the file for the runs so I made a ui that asks for all the params, the runs it itself. and when the run is done it ask if you want a rerun.  Work as smart as you have to


P.s.s
I found the problem, turns out that the way I was saving the float arrays where mixing randomly. I still dont know why, but I changed to strings being passed, Ill see how it goes

No comments:

Post a Comment