Author: Rick Bischoff
Date: 12:59:15 09/09/03
Go up one level in this thread
Hello again, You aren't collecting the principal variation during your search? The principal variation is the "line" your search thinks represents the best play by both sides. You should implement it anyway just to better understand what is going on during your search. I would like to know how exactly you are getting the scores for these moves.. are you? A. making the move B. running score = alphabeta(depth = x,alpha = -INF,beta = INF) C. unmake the move Or... running alphabeta(depth = x,alpha = -INF,beta = INF) ... make move ... score = -alphabeta(x-1,-beta,-alpha) ... unmake move ... if (ply == 0) print " current move score is : " score ?? On September 09, 2003 at 14:25:36, Jose Santos wrote: >Hi, > >>When you are "scoring" each move, are you referring to alpha? If so, note >that it never decreases, so if a move is "no better" than a predecessor, they >will appear to have the same score. > >No, I'm not printing Alpha. I'm printing the real value of the move. If I take >the alpha beta cuts from mini-max I get the exact same values I posted. > >>Another thing to point out is you may want to print out the PV for each move >to get a better idea of what your evaluation is making your search do. > >What's the PV of a move ? The evaluation of just that move not considering >anything else ? >I've done that and the results were what I expected, they are diversified. >(I increase the pawn value by 10% per each row it crosses) > >G1-F3 Material: 0 CountMoves: 2 Development: 100 Value: 266 >G1-H3 Material: 0 CountMoves: 0 Development: 100 Value: 214 >B1-A3 Material: 0 CountMoves: 0 Development: 100 Value: 214 >B1-C3 Material: 0 CountMoves: 2 Development: 100 Value: 266 >H2-H3 Material: 100 CountMoves: -1 Development: -250 Value: -413 >H2-H4 Material: 200 CountMoves: 1 Development: -250 Value: -285 >G2-G3 Material: 100 CountMoves: 1 Development: 0 Value: 113 >G2-G4 Material: 200 CountMoves: 1 Development: 0 Value: 216 >F2-F3 Material: 100 CountMoves: -1 Development: 0 Value: 69 >F2-F4 Material: 200 CountMoves: 0 Development: 0 Value: 210 >E2-E3 Material: 100 CountMoves: 10 Development: 0 Value: 336 >E2-E4 Material: 200 CountMoves: 10 Development: 0 Value: 456 >D2-D3 Material: 100 CountMoves: 7 Development: 0 Value: 269 >D2-D4 Material: 200 CountMoves: 8 Development: 0 Value: 420 >C2-C3 Material: 100 CountMoves: 1 Development: 0 Value: 103 >C2-C4 Material: 200 CountMoves: 2 Development: 0 Value: 244 >B2-B3 Material: 100 CountMoves: 1 Development: 0 Value: 113 >B2-B4 Material: 200 CountMoves: 1 Development: 0 Value: 216 >A2-A3 Material: 100 CountMoves: -1 Development: -250 Value: -413 >A2-A4 Material: 200 CountMoves: 1 Development: -250 Value: -285 > >Regards, > > Jose Santos
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.