Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rebel 12 beta 1

Author: Marc van Hal

Date: 09:40:39 03/16/03

Go up one level in this thread


On March 15, 2003 at 13:18:36, Ed Schröder wrote:

>On March 15, 2003 at 08:52:52, Werner Schuele wrote:
>
>>Hi Ed,
>>beta 1 eng is really the best, I have got good results with my setting, but
>>using sel 90 instead 100 and strong squares and MISC 41 (WM-Test on AMD2000+
>>solving 59 positions, elo 2666):
>
>Hi Werner,
>
>These are indeed excellent settings, thanks. Is 2666 any good? I never had the
>patience the finish the WM test so I have no clue.
>
>Ed
>
>
>>* Rebel 12 (beta 1.01) (February 25,2003)
>>* [=M25=] - pl02: reduction ipv plw3
>>* [=M37=] - HK op HK extension



I am still not hapy with nul move
But here are my 2 cents and thoughts.
a sugestion could be lazy evaluation  on my machine the search becomes to slow
to look again at the right move
But a new cpu is comming soon.
the point is with nullmove it happens to often it skips the best candidates.


for that reason i still give attacking 125 and pins 150
and passed pawns TO 115
Intresting for Rebel is that this increases the search speed too.
Just check it.
to make the search more agresive so real nulmoves will lead to something.
good nul moves are unpin a piece prefent a pin
increase the kingsafety
Build up a pawn attack on the queen or kingside
But in a deeper sence this is playing for centre control!
or attacking a half open file.
Or a combo of both.
playing for strong squares many times means your willing to sac to get just that
strong square.
Besides creating a weaknes in the enemys camp and take us of it.
But how does a program evaluate what is a strong square?
In human terms I can explain it but not in computer terms.
But this is the same for pins.
in 30 seconds a move
I am still wondering what it will take to make Rebel play Nd2 in the Ruy
the most obvious is mobilety but that doesn't work.
I also had  personality who played Bc2 before Na5 wondering why it choose that
move.
it is obviously bad because of d5
Rebel prefers dxe5 or dxc5 which in a deepers sence is bad too and most of my
personalitys play d5 which is also not the best move. but relatively the best
move after Nd2
It plays d5 because of the increased valeu for passed pawns as far as I know.
I gave it a kingsafety of 200 to deal with the freeze check problem
The fake Bobby Fischer  personelty on ICC had a maximum kingsafety
It played in this way without book  because of just that.
Which for Chessmaster 5000 was 200.
With a selective search of 11 also maximum for Chessmaster 5000.



But a new cpu is comming soon.


>>* [=M40=] - PLMAX (z3) reduction
>>* [SMART] - mp=mp+w6_bonus[w6]
>>*
>>* P3-800  (5) 202-175 (+27)
>>* A-2000 (30)  14-6
>>*
>>*
>>[Personality = BETA1.ENG]    * Rebel 12 (Beta 1.01)
>>[Pawn Value = 100]
>>[Knight Value = 100]
>>[Bishop Value = 110]
>>[Rook Value = 100]
>>[Queen Value = 100]
>>[King Safety = 150]
>>[Mobility = 100]
>>[Pawn Structure = 100]
>>[Passed Pawns = 100]
>>[Pins = 100]
>>[Bishop Pair = 100]
>>[Chess Knowledge = 100]
>>[Strong Squares = 25]
>>[Attractiveness = 101]
>>[Attacking = 100]
>>[Strength of Play = 100]
>>[Draw Contempt Factor = 0.00]
>>[Selective Search = 90]
>>[Search Technique = NULLMOVE]
>>[Pruning = MISC_25]             * pl02: reduction ipv plw3
>>[Pruning = MISC_37]             * HK op HK extension
>>[Pruning = MISC_40]             * PLMAX (z3) reduction
>>[Pruning = MISC_41]             * Reduceer Speciaal
>>[ANTI-GM = OFF]
>>[EVALUATION = SMART]            * mp=mp+w6_bonus[w6]
>>[Extensions (remaining)= 3]
>>[Extensions (checks)= 0]
>>[Extensions (captures)= 0]



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.