Author: George Tsavdaris
Date: 15:52:58 12/28/03
Go up one level in this thread
On December 28, 2003 at 17:43:19, Dimitris Poulos wrote:
>On December 28, 2003 at 14:29:17, Kurt Utzinger wrote:
>
>>**************************************************************
>>Rapid 15m+10s Tournament with The King 3.23 blitz
>>(a new personality only for fast time control games)
>
>Why do you say it is only for fast time control games? You have tested it in
>longer time controls and doesn't have that good results? To me it would be weird
>if it played worse in longer time controls, since it uses the same engine:
>TheKing. At least it should play the same worse as any other TheKing setting
>compared to a fast time control.
You may be right and an engine A good at short time controls, be also good in
long t.c. This means the following diagram:
Relative Strength of engine A***
1|
|
|
|
|
|
|.....................
|
O======================
time/move
or
Relative Strength of engine A***
1|
| . .
| .
| .
| .
| .
| .
| .
| .
|.
|
O=========================
time/move
---------------------------------------------------------------------------
But it may be this way:
Relative Strength of engine A***
1|
|.
| .
| .
| .
| .
| .
| .
| .
| . .
O===================================
time/move
This means that the engine A is not so succesfull in long time control, as it
is in blitz. You may say this is very unlikely but here is a test by me:
(***Relative Strength of engine A means, the points of the engine A after
playing on a specific round robin tournament with X other engines, divided
with the total number of games A played).
----------------------------------------------------------------------------
I created a tournament with 10 different CM8000 settings and played a 4-Round
Robin tournament in 8 different time controls (1'/g , 2'/g , 5'/g , 10'/g ,
20'/g , 60'/g , 40'/40 , 80'/40). 36 games for every engine at every time
control====288 total.
Here is the results of 4 engines:
Relative Strength of A1***
1|
|
| . . .
| . . . .
|
| .
|
|
|
|
|
O=====================================
t1 t2 t3 t4 t5 t6 t7 t8 time/move
Relative Strength of A2***
1|
|
|
|
| . .
|
| . . .
| .
| . .
|
|
O=====================================
t1 t2 t3 t4 t5 t6 t7 t8 time/move
Relative Strength of A3***
1|
|
|
|
|
| . .
|
| . . .
| . .
| .
|
O=====================================
t1 t2 t3 t4 t5 t6 t7 t8 time/move
Relative Strength of A4***
1|
|
|
| .
| . .
|
| . . .
| .
| .
|
|
O=====================================
t1 t2 t3 t4 t5 t6 t7 t8 time/move
As one can see, no clear result can be extracted from this data, but i have to
say that there is no straight line or an increasing line. Of course very little
games were played and as this is an interesting experiment i would make in the
future a bigger one.
There is nothing weird from my point of view if an engine A with parameters XX
plays worse in long t.c than that it plays in short t.c. Selective search
parameter is an important factor, but also other parameters change the strength
of the engine at various time controls.
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.