 |
| Herb[Romulus2] 2006-06-05 14:44:21 | I just noticed that it is pulled from the listings. I was going to ask anyway if there was something wrong with its definition, as I haven't produced any useful result of this type with the current client version from a-c that is. Even test queues with close to the best results definitions didn't come above 1.19% Just curious  |
| Stephen Brooks 2006-06-05 16:37:14 | The decay algorithm was changed to a more accurate one in 4.43, which means actually lower yields in a lot of situations. So DecayRotB and PhaseRotD should be having percentages in the new scheme. Also I found it seemed to have longer simulations than everything else and so was taking more CPU cycles of the project than the newer (more interesting) two. |
| Herb[Romulus2] 2006-06-06 06:47:49 | I've seen now several times identical results derived from different trial algorythms by different users or even myself producing the current 2nd best result in PhaseRotD one time with gen=4 and one time with gen=8. This suspects that the possible variations may have gone less by the new algorithm? However previously one could get with the same design always different results, this seems to have gone now, correct? |
| Stephen Brooks 2006-06-06 11:23:10 | In DecayRotB I have made it "deterministic" so the score should be entirely determined by the genome with no random fluctuations. That option hasn't been set for PhaseRotD. The new "algorithm" that changed the scores in 4.43 from 4.42 has to do with the way pion decays are handled, and changes the actual value of the scores. |
| Herb[Romulus2] 2006-06-06 12:59:47 | So that means anything still produced with older clients isn't accurate enough anymore and should be upgraded immediately? <note to calendar: update the 3 off-line boxes next Monday!> |
