Komodo 12 and MCTS

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
shrapnel
Posts: 1159
Joined: Fri Nov 02, 2012 8:43 am
Location: New Delhi, India

Re: Komodo 12 and MCTS

Post by shrapnel » Sat May 26, 2018 4:27 am

Hurnavich wrote:
Fri May 25, 2018 8:03 pm
For some reason in Fritz gui I cannot get MCTS to run
Hi Hurnavich bro
No loss if you can't run MCTS on Komodo. It is anyway almost useless because Komodo lacks NN (Neural Network) evaluation.
So, Positional Play will not improve any more than what Komodo already does.
To illustrate, think of the Upper part of the Human Body, from the waist upwards as Positional Play.
Let one leg be MCTS and the other be NN.
For good Positional Play (good balance in this example), both legs are required.
Using only MCTS, without NN, is equivalent to standing on only one leg; and we all know how stable that is ! :)
So, just stick with just normal Komodo which is an excellent alpha-beta Engine no doubt and don't bother with the MCTS part, which is just Marketing aimed at fooling the uninitiated.
Rgds
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Asus ROG Strix OC 11 GB Geforce 2080 Ti

pohl4711
Posts: 1068
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

Re: Komodo 12 and MCTS

Post by pohl4711 » Sat May 26, 2018 9:11 am

shrapnel wrote:
Sat May 26, 2018 4:27 am
Hurnavich wrote:
Fri May 25, 2018 8:03 pm
For some reason in Fritz gui I cannot get MCTS to run
Hi Hurnavich bro
No loss if you can't run MCTS on Komodo. It is anyway almost useless because Komodo lacks NN (Neural Network) evaluation.
So, Positional Play will not improve any more than what Komodo already does.
To illustrate, think of the Upper part of the Human Body, from the waist upwards as Positional Play.
Let one leg be MCTS and the other be NN.
For good Positional Play (good balance in this example), both legs are required.
Using only MCTS, without NN, is equivalent to standing on only one leg; and we all know how stable that is ! :)
So, just stick with just normal Komodo which is an excellent alpha-beta Engine no doubt and don't bother with the MCTS part, which is just Marketing aimed at fooling the uninitiated.
Rgds
100% agree. Komodo MCTS positional play is not better than Komodo. And the tactics are much weaker. So, there is no reason to use Komodo MCTS.
I believe, ChessBase asked the Komodo-team to do a MCTS-version, because they will release Komodo 12 + the FritzGUI as a DVD-Chessbase product. Chessbase just try to jump on the Leela-hype-train and sell some more copies.

CositasBuenas
Posts: 105
Joined: Tue Aug 03, 2010 5:36 pm

Re: Komodo 12 and MCTS

Post by CositasBuenas » Sat May 26, 2018 3:40 pm

Would you please prove ALL your allegations and accusations and give the computer chess community some credible evidence. If not, I wouldn't mind you refraining from further slander.

pohl4711
Posts: 1068
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

Re: Komodo 12 and MCTS

Post by pohl4711 » Sat May 26, 2018 5:23 pm

CositasBuenas wrote:
Sat May 26, 2018 3:40 pm
Would you please prove ALL your allegations and accusations and give the computer chess community some credible evidence. If not, I wouldn't mind you refraining from further slander.
Thats what you can read in this Thread by Komodo-Team:
“it is really a second engine, and I understand that ChessBase will treat it as an independent engine when they release their version of Komodo 12 (soon)“
“. MCTS does not depend on having a neural network, and the initial MCTS schemes had no such network at all. Komodo already knows a lot about chess. So we have come up with other ways to estimate initial winning chances and select moves to search/expand. MCTS is simply another way to search a game tree“

Thats, what Komodo MCTS do: another way to search the game tree with the Komodo-knowledge (evaluation). Where should better positional play come from? If anybody says so, he should give a proof, because why should another way to search the game tree change the style of positional play??
I played some dozend games vs. Wasp 3 with Komodo MCTS: lot of draws. Playing style identical to Komodo Alpha/Beta search. And definitly nothing in common with LC Zero. NOTHING. Just Komodo. Only with much weaker tactics.

menniepals
Posts: 224
Joined: Wed Mar 08, 2006 7:31 pm
Location: Houston, Texas

Re: Komodo 12 and MCTS

Post by menniepals » Sat May 26, 2018 10:08 pm

When will the android version be released?

mjlef
Posts: 1424
Joined: Thu Mar 30, 2006 12:08 pm
Contact:

Re: Komodo 12 and MCTS

Post by mjlef » Sun May 27, 2018 12:21 am

pohl4711 wrote:
Sat May 26, 2018 9:11 am
shrapnel wrote:
Sat May 26, 2018 4:27 am
Hurnavich wrote:
Fri May 25, 2018 8:03 pm
For some reason in Fritz gui I cannot get MCTS to run
Hi Hurnavich bro
No loss if you can't run MCTS on Komodo. It is anyway almost useless because Komodo lacks NN (Neural Network) evaluation.
So, Positional Play will not improve any more than what Komodo already does.
To illustrate, think of the Upper part of the Human Body, from the waist upwards as Positional Play.
Let one leg be MCTS and the other be NN.
For good Positional Play (good balance in this example), both legs are required.
Using only MCTS, without NN, is equivalent to standing on only one leg; and we all know how stable that is ! :)
So, just stick with just normal Komodo which is an excellent alpha-beta Engine no doubt and don't bother with the MCTS part, which is just Marketing aimed at fooling the uninitiated.
Rgds
100% agree. Komodo MCTS positional play is not better than Komodo. And the tactics are much weaker. So, there is no reason to use Komodo MCTS.
I believe, ChessBase asked the Komodo-team to do a MCTS-version, because they will release Komodo 12 + the FritzGUI as a DVD-Chessbase product. Chessbase just try to jump on the Leela-hype-train and sell some more copies.
Well, much of this is just totally wrong. Chessbase never asked us to do a Monte Carlo Tree Search. You give no evidence that a neural network is needed for Monte Carlo Tree Search and in fact, no neural network is needed, or was even in initial versions of MCTS. Please stop making things us. Just ask us instead of making unfounded claims. We are pretty friendly and would be happy to answer most questions.

Mark

lkaufman
Posts: 3680
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Komodo 12 and MCTS

Post by lkaufman » Sun May 27, 2018 1:38 am

pohl4711 wrote:
Sat May 26, 2018 5:23 pm
CositasBuenas wrote:
Sat May 26, 2018 3:40 pm
Would you please prove ALL your allegations and accusations and give the computer chess community some credible evidence. If not, I wouldn't mind you refraining from further slander.
Thats what you can read in this Thread by Komodo-Team:
“it is really a second engine, and I understand that ChessBase will treat it as an independent engine when they release their version of Komodo 12 (soon)“
“. MCTS does not depend on having a neural network, and the initial MCTS schemes had no such network at all. Komodo already knows a lot about chess. So we have come up with other ways to estimate initial winning chances and select moves to search/expand. MCTS is simply another way to search a game tree“

Thats, what Komodo MCTS do: another way to search the game tree with the Komodo-knowledge (evaluation). Where should better positional play come from? If anybody says so, he should give a proof, because why should another way to search the game tree change the style of positional play??
I played some dozend games vs. Wasp 3 with Komodo MCTS: lot of draws. Playing style identical to Komodo Alpha/Beta search. And definitly nothing in common with LC Zero. NOTHING. Just Komodo. Only with much weaker tactics.
We didn't even tell ChessBase about the MonteCarlo version until just days before we were to send them a prototype of Komodo 12. They were quite pleasantly surprised! While it is true that MCTS is a different search than normal alpha-beta, it is completely wrong to think that this means it won't play better positionally. That's like saying that a 20 ply search is only better than a 10 ply search due to tactics. But anyway, I agree that the Monte Carlo search in Komodo 12 is usually inferior to normal Komodo. However this may no longer be true when we release our first upgrade of Komodo 12. I really much prefer the style and positional play of the (current dev) MCTS version over an old Komodo alpha-beta version of similar rating. It has already proven to be tougher for human opponents than its rating would suggest, for very good reasons. Basically, it doesn't assume that it is playing against itself.
Komodo rules!

Hurnavich
Posts: 718
Joined: Wed Jun 19, 2013 6:14 pm

Re: Komodo 12 and MCTS

Post by Hurnavich » Sun May 27, 2018 4:26 am

Hi,
I'm still curious as to why MCTS is not working in fritz GUI
The GUI states that the engine is not supported
If the development of this MCTS engine improves
Then at some point I would like it to run in fritz
And on my system it does not
Yet all normal Komodo does

Thanks
"May your next game be your best"

CositasBuenas
Posts: 105
Joined: Tue Aug 03, 2010 5:36 pm

Re: Komodo 12 and MCTS

Post by CositasBuenas » Sun May 27, 2018 6:06 am

Komodo MCTS is working in Chessbase 14, HCE and Shredder GUI.

pohl4711
Posts: 1068
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

Re: Komodo 12 and MCTS

Post by pohl4711 » Sun May 27, 2018 8:31 am

mjlef wrote:
Sun May 27, 2018 12:21 am
You give no evidence that a neural network is needed for Monte Carlo Tree Search
I know. Because I never said something like this. All I say is, that the interesting, risky positional of LCZero comes from its neural network, not from the MCTS, which LCZero uses only, because the neural-net calculations are so slow, that using normal AlphaBeta with crunching billions of nodes, would not work properly.
I looked into some dozend games of Komodo MCTS and find it playing like Komodo: good, solid,.positional chess, but much weaker than Komodo 12, because no real AlphaBeta search is done (seems at one level to Wasp 3). And that is, what is to be expected, when MCTS is used without a neural net.
So, Komodo MCTS has nothing in common with LCZero, but the using of MCTS, which is not only useless without a neural net, but weakening. And so, I see no reason to use Komodo MCTS.
Thats all I said. And that is all I have to say about Komodo MCTS, until it uses a neural net.

And because, I believe, that the Komodoteam knows all this like me (better like me), I think, the whole MCTS-Komodo was done to suggest parallels to AlphaZero/LC Zero, which are not existing. And I dont like that.
That has nothing to do with Komodo itself. I like Komodos positional play and its solid evaluations very much and I used only Komodo for all my work on my SALC-openings. I believe, there is no better engine for analyzing, than Komodo. Now and since the last years.

Post Reply