 |
| Universal Creations 2009-03-06 02:07:31 | The results are quite steady for some time. Is there going to be a optimalisation soon? I like to see some improvement in this project. Mpts are nice for the team-stats, but the results are the reason for this project. |
| Stephen Brooks 2009-03-10 17:57:25 | I was just thinking about that. I'm going to do another one with NO seeding, plus I've changed the way it handles the phases. There was something worrying me about the phase variables in that they should be "circular" but weren't previously allowed to wrap at the ends of the range. Now I've increased the range so it allows at least one wrap. Theroetically it should find the same sort of optimum as before. |
| Stephen Brooks 2009-03-10 17:58:48 | Oh, yeah, the alteration of the phase in Linac900Removable6c2 pretty much garauntees there'll be no seeding from previous optimisations. I think the seeding actually sucked some of the life out of the last two somehow. |
| tomaz 2009-03-11 09:04:35 | "I think the seeding actually sucked some of the life out of the last two somehow." Probably because of very limited number of seeds (one or two). Wondering how it would develop if there were reasonable selecton of about 50 seeds ? |
| Stephen Brooks 2009-03-20 15:03:45 | I think the problem is when I seed I put something in that may be so totally different from the existing "gene pool" that it can't "interbreed" with it! So eventually the original path just goes extinct. The Linac900Removable6c2 seems to be going quite well so far, though. |
