 |
| [SG]Santas little helper 2004-10-30 11:33:16 | Hi at all, I just want to know why there are negative results possible and if they make sense why are in the results.dat only positive results? Thanks for info. [SG]Santas little helper |
| [SG]Santas little helper 2004-11-10 09:12:31 | Is there no informed human being out there ... ? :-/ |
| Stephen Brooks 2004-11-10 13:48:20 | No, just me. Surely those negative ones are really far too low to be of much use? |
| [TA]JonB 2004-11-10 15:18:27 | But, Stephen, if they aren't of any use, why does my client frequently run negative results 5 times to check them? Thankfully, it is fast, but seems like a waste of CPU. |
| AySz88 2004-11-10 20:37:55 | To answer the first part of your question: A negative result means that all or almost all of the particles died before reaching the goal. These negative results were created so we don't have tons of 0.001% designs at the beginning (I think less than 0.1% becomes a negative 0.something, and almost no particles becomes negative 1.something depending on how far most particles got). If there are only really low results, the optimizer can't really tell the difference between the results. |
| kitsura 2004-11-11 03:58:41 | quote: No its not a waste of CPU. After all all the good results originated from their lesser optimised counterparts. Like they say without evil there is no good. |
| [TA]JonB 2004-11-11 20:00:32 | No, I understand that the optimization process will frequently result in negative results. What I don't understand is why it runs them five times. If they are negative, I don't think there is any point in validating them. to follow the "evil/good" idea, should I do evil deeds five times just in case it might turn good? Too humanistic for me. |
| Stephen Brooks 2004-11-12 04:02:48 | The program really shouldn't re-check negative results 5 times unless there are no higher (therefore no positive) results in your results.dat file. Of course I think JonB frequently deletes his file, so I suppose between the time he deletes it and when the next samplefile arrives, it will recheck negative results. |
