Author: Shep
Date: 12:02:39 04/19/01
Go up one level in this thread
On April 19, 2001 at 12:43:05, Christophe Theron wrote:
>On April 19, 2001 at 12:37:12, Dan Andersson wrote:
>
>>It would do to have a settings file or somesuch. And Switch it to the optimum at
>>once close to the match date. Or A gradual normalisation till the match takes
>>place.
There's an easy way which builds on the idea of creating a time-dependent
function.
Imagine the function
f(d) := 1 if 100 <= d <= 250
0 otherwise
(assuming that Kramnik gets the program on day 100 and the match starts at day
251)
Now for each eval term/function e(x), make the program return
e'(x) = e(x) + e(x) * f(d) * c(e), with c(e) being a constant value for each
eval term e
For example,
Value_of_knight_on_square_x gets multiplied by (1+f(d)*0.5)
The result:
1) You have a program which plays exactly like your original one outside of the
period where Kramnik prepares for it.
2) Within the period Kramnik is testing it, the program will play reproducably
the same chess.
Of course 2) is the sneakier part because Kramnik will never suspect the program
could alter its evaluation on a certain day.
---
Shep
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.