Author: Dan Honeycutt
Date: 10:21:12 02/18/05
Go up one level in this thread
This is not as simple a task as one might think. What you describe is essentially what I did to create Simon - I took my program Bruja, replaced the search and eval with very simple versions and stripped out other "whistles and bells to leave a basic functional engine. What I thought should be a few hours work turned out to take several days. When done I did very little testing - no need since it consisted of either very simple functions or functions taken directly from Bruja. That led to a bug that should qualify me for some sort of award. In Bruja's search, if it checks the opponent, it immediately generates moves for the next ply. It can then return a mate in 1 score if there are no moves or use the number of moves to decide how to extend. For Simon I wanted all the moves generated in one place, so I moved the check evasion to the next ply and adjusted the return score to mate in 0. Of course that should have been mated in 0. So Simon, if it found a way to mate the opponent, would avoid it like the plague and if it found a way to get itself mated would play right into it. Best Regards Dan H.
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.