Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question: When to Use Internal Iterative Deepening

Author: Robert Hyatt

Date: 07:18:45 02/13/01

Go up one level in this thread


On February 13, 2001 at 03:03:16, Tim Foden wrote:

>On February 13, 2001 at 00:16:24, Robert Hyatt wrote:
>
>>On February 12, 2001 at 23:00:36, William Bryant wrote:
>>
>>>On February 12, 2001 at 22:39:21, William Bryant wrote:
>>>
>>>>My understand is, Internal Iterataive Deepening is used when searching the
>>>>root PV node and you encounter a position for which there is no hash table
>>>>best move.  Specifically, this is done during the full width part of the
>>>>tree (rather than the PVS part of the search).
>>>>
>>>>My program at present tests for this by comparing beta to alpha.
>>>>ie if (beta == alpha+1) then this is the PVS part of the tree and IID should
>>>>be skipped.
>>>>
>>>>This doesn't help at all so I think I'm missing something in the implementation.
>>>>
>>>>Any comments or suggestions would be appreciated.
>>>>
>>>>William
>>>>wbryant@ix.netcom.com
>>>
>>>For Clarity,
>>>
>>>if ((there_is_no_best_move) && (beta != alpha + 1)) {
>>>	//do internal iterative deepening
>>>	}
>>>
>>>William
>>
>>
>>That is almost what I do...
>>
>>  if (tree->hash_move[ply]==0 && do_null && depth>=3*INCPLY) do {
>>    if (ply & 1) {
>>      if (alpha!=root_alpha || beta!=root_beta) break;
>>    }
>>    else {
>>      if (alpha!=-root_beta || beta!=-root_alpha) break;
>>    }
>>
>>Which says (a) no hash move;  (b) null-move is legal here (which means the
>>move at the previous ply was _not_ a null-move;  (c) 3 plies of depth left
>>to search (or more);  (d) alpha and beta are _exactly_ as they were at the
>>start of the iteration.
>
>I think I understand the reasons for (a), (c) and (d) above.  I'm note sure
>about (b) -- why do we want to make sure that the previous ply was not a null
>move?
>
>Cheers, Tim.


I don't want to waste the time just to do a null-move search.  IE IID is not
cheap and after a null-move, normal ordering usually works pretty well.  It is
worth testing to be sure, of course...



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.