Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: mtd/pvs hybrid?

Author: Andrew Williams

Date: 05:58:41 08/29/01

Go up one level in this thread


On August 29, 2001 at 08:25:26, Pat King wrote:

>On August 28, 2001 at 07:02:40, Andrew Williams wrote:
>
>>>On August 27, 2001 at 17:26:24, Andrew Williams wrote:
>>>
>>>My MTD(f) program checks any move which it is about to make, if it didn't
>>>generate it using the usual generate_moves() function. So any hash or killer
>>>move is checked to see if:
>>>
>>>(a) There is a piece on the from square.
>>>(b) The piece on the from square is capable of making the proposed move.
>>>(c) There are no intervening pieces which would prevent the move
>>>(d) etc (Manchester City have won again tonight, and are top of the League
>>>    and I've been celebrating, so I'm not able to recall the rest of the
>>>    conditions - hopefully you get the idea)
>>>
>>OK. Recovered now, so:
>>
>>(d) No friendly piece on the target square.
>>(e) If it's a capture, there must be an enemy piece on
>>    the target square...
>>(f) ... unless it's ep, in which case check for the pawn
>>    on the ep square
>>(g) If it's a promotion, make sure it's a pawn that's moving
>>    and that it's on the 7th rank now.
>>(h) etc etc
>>
>>Regards
>>
>>Andrew
>
>Yes, of course Zotron tests the legality of hash moves within the search. My
>assumption has been that the hash entry for the root position must always be
>valid, and this is the way I've passed my engine's move back to the calling
>functions. Works 99.999% of the time, and fails spectacularly 0.001% of the
>time. I've had a devil of a time catching that 0.001%.
>
>Pat

Ahhh. I see. Mine also uses the hash table to construct its PV.

Did you catch your error?

Andrew



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.