Author: Uri Blass
Date: 12:29:18 04/10/04
Go up one level in this thread
On April 10, 2004 at 14:36:02, Ed Schröder wrote:
>On April 09, 2004 at 21:39:56, Uri Blass wrote:
>
>>On April 09, 2004 at 15:47:30, Uri Blass wrote:
>
>>>I understand that you use non recursive search in Rebel and I asked some
>>>question that you did not reply until now
>>>
>>>see
>>>http://www.talkchess.com/forums/1/message.html?359230
>>>
>>>I have another question
>>>You claim that you call search only once.
>
>
>>>My question is if your search get parameters and what parameters does it get.
>>>Does it get initial depth when you have some iterate function that calls search
>>>again and again with increased depth?
>
>I think it makes little sense to explain how my search exactly works, there are
>many roads that lead to Rome, meaning in the end they are equal and do the very
>same.
>
>I don't understand how others do it calling search or q_search all the time, it
>looks time consuming if you ask me,
It also looks to me but recursive search is the only way that I learned.
I also don't understand how one can work
>with only one alpha and one beta while these are ply based.
alpha and beta are of course ply based but recursive search means that
only one alpha and one beta are in memory that is not hidden.
But I assume that in
>practice all is the very same and I never bothered about it.
>
>To explain how I do it I first must understand how you do it otherwise we end up
>in a Babel-like confusion as we already seem to have now.
>
>Ed
I do not think that you need to understand how other do it in order to explain
what you do.
The misunderstanding is not result of the fact that you did not understand how
other do it but because your explanation was not clear.
I have specific questions and I hope that you can give answers.
questions:
1)What parameters does search get and what does it return?
I guess that you get depth, alpha and beta and returns integer.
I guess that it does not give you the best move and that for every depth if you
want to search to that depth, you have another loop on the legal moves that
calls search and call it again in case of fail low or fail high.
2)How do you start Search.
Do you start it with ab[0]=beta,ab[1]=alpha?
3)what do you do after initializing the ab array?
I think that you need something like
While (search is not finished)
{
makemoves();
evaluate();
undomoves();
}
I think that it can be productive if you describe the loops of makemoves and
undomoves when it is seems to me that undomoves need to update the ab array in
cases that evaluate generated a score above alpha.
Uri
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.