Author: Ulrich Tuerke
Date: 13:02:12 05/03/01
Go up one level in this thread
On May 03, 2001 at 15:41:01, Miguel A. Ballicora wrote: >On May 02, 2001 at 12:04:33, Ulrich Tuerke wrote: > >>On May 02, 2001 at 11:47:15, Gian-Carlo Pascutto wrote: >> >>>On May 02, 2001 at 11:36:08, Ulrich Tuerke wrote: >>> >>>>This way, you will have heavily extended root moves rather early in your move >>>>list. These moves must not always be good. >>>> >>>>Don't you regard this as a draw-back ? >>>> >>>>The advantage is perhaps that nice tactics (checking sacs ...) will be seen >>>>early. >>> >>>Well there's a reason why these moves are so heavily extended. Something >>>interesting must be happening there. Even if they aren't good (according >>>to the search) yet there's a good chance they'll change score quickly >>>soon. >>> >>>You do want to look those interesting moves first next iteration >>>don't you :) >> >>Quiet moves can be interesting too. Giving check can be a waste of time. >>After all, many are using the node count scheme. >>So I guess, it can't be bad. >> >>I do something different (estimates of scores). It's probably not better. > >What do you mean by estimates of scores? Do you estimate the score statically? >thx, As you certainly know, the problem is that search will in general give a true score for the first root move and only upper bounds for the remaining ones. I guess it doesn't make much sense to sort according to upper bounds. Therefore I perform a pure minmax (i.e. no alpha-beta) search in the 1st iteration. This way I get exact scores for all root moves. These values will be used when sorting the root moves. They will be updated using the results of the higher iterations, but of course not just overridden. Later, we can not afford to skip alpha-beta. So, I compare these exact values with the values/bounds returned by the search. For instance, an exact value from a higher iteration replaces always an exact value from the 1st iteration. Upper bounds from a higher iteration which are higher than the exact values from iteration 1 will not be used. However, upper bounds which are lower than the exact value are used to correct the exact values. Under equal values, I prefer the values of moves which had been the best in an earlier iteration. This doesn't work too bad. But I don't claim that it's better than Bob's method. Some extra bonus of my method is, that I get some useful data for the easy move algo. I'm not sure if I could make myself clear ? Uli > >Miguel > > >> >>Uli >> >>> >>>-- >>>GCP
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.