Author: Frank Schneider
Date: 05:33:08 12/30/98
Go up one level in this thread
On December 30, 1998 at 05:51:36, Frank Schneider wrote: >On December 30, 1998 at 04:29:49, Frank Schneider wrote: > >>Hi all programmers! >> >>I know it sounds very simple and unsound, but has anyone tried this: >> >>IF >>- a position is found in the transpositiontable and >>- it is an upper bound (all moves failed low) and >>- the hashvalue is < alpha and >>- hashdepth==depth-1 (tableentry from previous iteration) >>THEN >>only search the first n moves but do not store the result in the >>transpositiontable. >> >>The idea is based on the observation that most nodes are either >>a) nodes where only very few moves are searched until one causes a cutoff >>or >>b) nodes where all moves fail low. This is where the heuristic tries to >>save some time. >> >>I just ran a first test with a mixed 100 positions testsuite (all searched 6 >>ply) and got nice results: >>- more correct moves (??) >>- >20% less nodes >> >>I'll have a closer look at this.... >Here are some more results: > >correct / nodes |n=infinite |n=15 |n=10 |n=5 >------------------------------------------------------------------------------- >Standard #6 |23 / 8306851 |22 / 6616146 |25 / 6229700 |24 / 5852685 >WAC #6 |256 / 13428654 |254 / 11575072 |253 / 10977570 |253 / 10987063 >------------------------------------------------------------------------------- > Some more results: correct / nodes |n=infinite |n=15 |n=10 |n=5 -------------------------------------------------------------------------------- Standard #6 |23 / 8306851 |22 / 6616146 |25 / 6229700 |24 / 5852685 WAC #6 |256 / 13428654 |254 / 11575072 |253 / 10977570 |253 / 10987063 -------------------------------------------------------------------------------- Standard #7 |25 / 26399528 |20 / 20759570 | | WAC #7 |280 / 35352959 |274 / 30579448 | | -------------------------------------------------------------------------------- As expected it doesn't look good anymore. Are there possible refinements? >n is the number of moves searched before cutting. >Standard is a testsuite of 100 mixed positions (opening, middlegame, endgame, >positional and tactical positions). >#6 means every position was searched 6 iterations. > >I'll now try #7 but that may take some time... no, see above > >> >>Have a Happy New Year, >> >>Frank
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.