The importance of "Dynamism" in Komodo

[Laskos]

I was curious what does this mean:

The other new option, "Dynamism", gives the user full control over the style of Komodo; a low setting makes it prioritize static advantages like material and pawn structure, while a high setting prioritizes short-term ("dynamic") advantages like mobility and threats to the king. Although the default (100) is believed to be best for actual play, a lower value around 80 seems to produce more realistic evaluations in most positions so may be better for opening analysis.

I decided to use testsuites of different character to check for this setting. STS 1-15 - mostly positional in character, and Arasan 18 - mostly tactical.

STS: 5s/move

Code: Select all

Dynamism  Solved

 80        1346            
100        1367    ------
120        1341

The optimum is at the value of Dynamism 100.

Arasan: 5s/move

Code: Select all

Dynamism  Solved
100         126
140         155  
180         159
190         174    ------
200         171

The optimum is the value of Dynamism 190. The improvement compared to the default number of solved positions is huge, in fact we will see that combined with "King Safety", "Dynamism = 190" makes a tactical monster of Komodo.

So, in tactical problems, high values of Dynamism are favored. Maybe Larry's value of 80 is applicable to very quiet, positional problems?

Then, I also tested several hard tactical suites, and it is possible that Komodo 9.42 Dynamism=190; King Safety=160 (or similar values, try them) is the new King of tactics, maybe at least on a par with Houdini 4 Tactical, way above SF Matefinder. The testsuites were Arasan 18, HardTalkchess2015, DieRybka with about 5s/move.

[Laskos]

I was curious what does this mean:

The other new option, "Dynamism", gives the user full control over the style of Komodo; a low setting makes it prioritize static advantages like material and pawn structure, while a high setting prioritizes short-term ("dynamic") advantages like mobility and threats to the king. Although the default (100) is believed to be best for actual play, a lower value around 80 seems to produce more realistic evaluations in most positions so may be better for opening analysis.

I decided to use testsuites of different character to check for this setting. STS 1-15 - mostly positional in character, and Arasan 18 - mostly tactical.

STS: 5s/move

Code: Select all

Dynamism  Solved

 80        1346            
100        1367    ------
120        1341

The optimum is at the value of Dynamism 100.

Arasan: 5s/move

Code: Select all

Dynamism  Solved
100         126
140         155  
180         159
190         174    ------
200         171

The optimum is the value of Dynamism 190. The improvement compared to the default number of solved positions is huge, in fact we will see that combined with "King Safety", "Dynamism = 190" makes a tactical monster of Komodo.

So, in tactical problems, high values of Dynamism are favored. Maybe Larry's value of 80 is applicable to very quiet, positional problems?

Then, I also tested several hard tactical suites, and it is possible that Komodo 9.42 Dynamism=190; King Safety=160 (or similar values, try them) is the new King of tactics, maybe at least on a par with Houdini 4 Tactical, way above SF Matefinder. The testsuites were Arasan 18, HardTalkchess2015, DieRybka with about 5s/move.

Komodo D 190 KS 160 beats Houdini 4 Tactical in new Arasan 19 beta suite:
5s/move, 4 cores:
K: 178/250
H: 172/250

My impression is that by varying "Dynamism" from say 60 to 190, Komodo became both the strongest positional engine and the strongest tactical engine. Great feature!

[lkaufman]

That's very interesting Kai. Using huge values for Dynamism and/or King Safety makes evaluations of ordinary positions very unrealistic, but I can believe it helps tactics. I think Dynamism = 80 produces the most realistic evaluations in ordinary opening/early middlegame positions, but that doesn't seem to correlate with best play, and probably correlates even less with tactical acumen. Most tactical puzzles involve sacrifice, so any change that devalues material tends to help them. There is a lot to learn here I think.

[melajara]

That's very interesting, indeed. I'm wondering if those settings could be auto-tuned by Komodo itself, in a kind of metalevel or "introspective" mode where Komodo would apply different settings given the "feeling" of the position it gets down its analysis, e.g. dynamism = 80 in quiet positions and 160 in the middle of a set of exchanges. Ditto for King safety.

Just a suggestion.

[shrapnel]

Interesting.
What Contempt value did you use ?

[Nordlandia]

Is default parameters optimal in case late-endgame analysis?

I'm not sure whether 80 or 100 is best in this position originating from WCC 1972. Also maybe some other parameters may be adjusted for altered move output !?

[d]8/pp4p1/3kpp2/1P6/8/P3P1K1/8/2B5 w - - 0 36

[lkaufman]

Is default parameters optimal in case late-endgame analysis?

I'm not sure whether 80 or 100 is best in this position originating from WCC 1972. Also maybe some other parameters may be adjusted for altered move output !?

[d]8/pp4p1/3kpp2/1P6/8/P3P1K1/8/2B5 w - - 0 36

Of course nono-default values of Dynamism will work better in specific positions. Perhaps there is some general rule we could use that would help, for example reduce (or increase) Dynamism in late endgames, but we haven't found it yet. Or it could be varied based on specific characteristics of the position, as suggested in one response here. There are just too many possibilities to test; if anyone comes up with evidence that the default is wrong in some specific class of positions (must be rigorously defined), let us know.

[Laskos]

Interesting.
What Contempt value did you use ?

Default values, meaning analysis contempt 0, right?

[lkaufman]

Yes, contempt defaults to zero in analysis or infinite mode.

[mjlef]

Yes, contempt defaults to zero in analysis or infinite mode.

To be clear, if Contempt for White is turned off, in analysis mode Contempt is automatically set to zero. If Contempt for White is turned on, whatever you have Contempt set to gets applied to white with black getting the same value with the opposite sign. This allows people to use Contempt in analysis mode without messing up the analysis as they switch sides.