Komodo 11.3 released

[lkaufman]

Now i'm confused i thought contempt 0 was correct against SF And Houdini

That's right, where did you read otherwise? Perhaps you misinterpreted something I wrote? Note there is a difference between what is "fair" and what is "best".

[lkaufman]

Is default contempt of 16 adequate value against K11.2.2 in selfplay (assuming the latter use C=10) ?

Based on the estimated elo diff. and the formula in readme the optimum value for this test is Contempt = 1. But of course that wouldn't be fair if you use 10 in 11.2.2. A setting of Contempt = 13 would be the closest equivalent to 10 in the older version so it would be "fair". We raised it to 16 partly because SF now uses 20, so at least when we play Stockfish we are no longer handicapped, and against Houdini we're only very slightly handicapped at default levels.

can a search be done with 2 different comtempt values ... and then the lines be displayed ... seems like that would be an interesting engine that could do that ...

I suppose you could run two instances of your GUI with Komodo Contempt set to different values, and compare, assuming you have at least a dual core machine.

[lkaufman]

I don't think that SF contempt 20 is the same as komodo 20, also the latest dev version of sf uses 12+logarythmic dynamic contempt

Regarding SF "dynamic contempt", am I misunderstanding something? It seems to be saying that if you are winning, don't trade pieces, which is of course ridiculous. At least that's what it looked like a couple weeks ago.

Well, ridiculous or not, but it gains elo in selfplay and passed LTC SPRT easier than it did pass STC.
I think the main thing why it works can be formulated like this "no need to go to really advantageous endgame if you can wait and go for even more advantageous endgame".
Also it tries to trade away everything when position becomes bad, this is also maybe a place where elo gain comes from.

It makes no sense to make so many trades in a bad position, unless those trades result in a drawn endgame, which is the exception rather than the rule.

There are a lot of things that don't really make sense but that make engines play stronger...

That comment is absolutely true. I think most of the elo gains since my Rybka 3 days have been from trusting test results over common sense. But in the case at hand, I must say that Komodo has long had an algorithm which is almost the opposite of SF "Dynamic Contempt", from which Komodo gained Elo! I guess the devil is in the details.

[Jesse Gersenson]

Hi Larry!

Automatic backward analysis of games doesn't work at two different PCs of mine neither in Fritz14 nor in Shredder13- GUI, at both it's x64.exe, one running Win7Pro, the other one Win8.

Engine doesn't start computing after GUI- command is given.

This was a bug and it has been fixed. Larry will announce details later today.

Login and re-download Komodo 11.3 and you will get the updated binary.

[Hurnavich]

Hi, this part

We raised it to 16 partly because SF now uses 20, so at least when we play Stockfish we are no longer handicapped, and against Houdini we're only very slightly handicapped at default levels.


So if SF at 20 then K needs to be 16 not 0 but then we say 0 is best against SF

So if SF is 20 we need to be 16 for K ? and if SF is 0 then we need to be 0?

That's why i'm confused.

[Nordlandia]

In the previous version C=10 was recommended in direct selfplay.

I guess C=16 is a tad optimistic, maybe C=13 since it's close equivalent to C=10. This case K11.3 vs K11.3

[peter]

Thank, Jesse, now it works well.

[lkaufman]

Hi, this part

We raised it to 16 partly because SF now uses 20, so at least when we play Stockfish we are no longer handicapped, and against Houdini we're only very slightly handicapped at default levels.


So if SF at 20 then K needs to be 16 not 0 but then we say 0 is best against SF

So if SF is 20 we need to be 16 for K ? and if SF is 0 then we need to be 0?

That's why i'm confused.

For best results against SF, use 0. Using the default of 16 is probably more or less comparable to 20 in SF on average, so I just meant that comparing ratings and results of SF using 20 and Komodo using 16 would be reasonably fair. In other words, when the top three engines play each other, they are all slightly weakened by using default Contempt values. In that sense, it is "fair".

[Hurnavich]

Thanks Larry clears that up many thanks.

[Ovyron]

I want to thank Ulysses P. here on the forum for suggesting we just we just automatically extend any subscriptions to make sure the buyers get some reasonable number (we set it to 3) Komodo versions, even if the year period expires. It was a great idea. Fair, simple, and takes some pressure of of us. Thanks.

Mark

I want to show my public appreciation to you and Larry Kaufman for giving me the engine for free for this! That was truly unexpected and I'm very glad that trying to help pays off.

I'll fully support Komodo and wish you great success in your endeavors.