Tactical Factor in Ktulu 9

[tomgdrums]

As I said in an earlier topic, I have become re-interested in Ktulu 9 due to some information that Frank Quisinksy had shared with me in a recent post.

Another discussion came up as to what the tactical factor setting may or may not do regarding Ktulu's performance. (Tactical factor is the only real setting one can change on the engine)

So I though I would try a little experiment that is going to be fairly unscientific but it is all I have the time to do.

This post is the first step. I have had Ktulu 9 (default settings) play three different timed matches against three different engines for a total of 9 matches. I am then going to have try a new tactical factor setting and have Ktulu play thru the same field with the same different time settings. I only picked three engines because they were the only three single processor engines I currently have, and it just seems correct to test this with SP engines as that is what Ktulu is.

The engines are Fruit 2.3.1, Hiarcs 12.1 and Rybka 3 (again all SP).

The results were as follows:


Fruit 2.3.1-Ktulu 9 G/10 (3: +1, =1, -1)

Fruit 2.3.1-Ktulu 9 G/5 (3: +1, =0, -2)

Ktulu 9-Fruit 2.3.1 G/1 (3: +3, =0, -0)

Ktulu ended up with 6 wins, 1 draw and 2 losses against Fruit

Ktulu 9-Hiarcs 12.1 G/10 (3: +0, =2, -1)

Hiarcs 12.1-Ktulu 9 (3: +0, =2, -1)

Ktulu 9-Hiarcs 12.1 (3: +1, =1, -1)

Ktulu ended up with 2 wins, 5 draws and 2 losses against Hiarcs


Ktulu 9-Rybka 3 G/10 (3: +0, =2, -1)

Rybka 3-Ktulu 9 G/5 (3: +2. =1, -0)

Ktulu 9-Rybka3 G/1 (3: +0, =0, -3)

Ktulu ended up with 0 wins, 3 draws and 5 losses against Rybka 3.

I am now going to do the entire thing again with Ktulu 9 having an increased tactical factor. The default is 30, I am going try 50.

[DomLeste]

Not enough games try 1+1" time control with more games.

test tactical factor 0 or 20

use test suite if need be

http://www.rybkachess.com/free/NoomenTestsuite2006.pgn

[hokaro]

using time control 1+1 for Ktulu is almost senseless... It is well-known that short time controls is not what Ktulu can play, especially that Ktulu have a time control bug, still unfixed...Even programmer Rahman Paidar recommends only long games with Ktulu...

why do you want to test Ktulu's strength in so short time controls, anyway? if you wanna analyze endgame, you need longer time controls, and there you will see the real power of Ktulu against other engines, especially Rybka...

regards
Hoshad