TalkChess.com
Hosted by Your Move Chess & Games

Author Message
Larry Kaufman

Joined: 10 Jan 2010
Posts: 1228
Location: Maryland USA

Post subject: Re: When does a cut-node became an all-node?    Posted: Wed Mar 28, 2012 5:55 pm

diep wrote:
[snip]

lkaufman wrote:
 diep wrote: Reducing 1 ply is the mathematical optimum, basically reduction of 1 ply is still what's getting done in the end by most of todays engines, be it in a more risky way. Also that it's the mathematical optimum between risk and reduction of search tree doesn't mean that taking more risk isn't gonna work for you

I don't understand how there can be a mathematical optimum reduction, it should depend on how good your move ordering is. If it's good enough, you could reduce all moves after the first by 100 plies!

Well now you write as how we did do computerchess in the 90s. Namely using the extreme assumptions. We already know for a year or 13, that the fliprate (heh another computerchess statistic i introduced back then) of crafty was around a 5% and of Diep it was far lower, between 1% and 2% at the time. Probably that's different nowadays, especially crafty will have improved and with it most of the modern beancounters, but not *that much* better than it used to be.

So we can safely assume that reducing things 100 ply is too much of a risk. The mathematical optimum between risk and profit is then 1 ply. In how i wrote that down, that means simply this breaks even a lot quicker.

If you fiddle some with chances you'll also end up there.

Reducing 2 ply, simply is very very risky as compared to reducing 1 ply. Which doesn't mean that you shouldn't try doing that if you already get 20+ ply anyway

A pragmatic approach there isn't gonna get punished...

Three questions:
1. Could you please define "fliprate"?
2. What assumptions are made in your proof that the ideal LMR reduction is just 1 ply?
3. What is your opinion of the idea of increasing the null move reduction with increased depth? Stockfish pushes this to the extreme, adding another ply of reduction for each 4 plies of search. I imagine you will think it is a bad idea, but I don't like to assume such things.

Larry
 Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First
Subject Author Date/Time
Alcides Schulz Tue Mar 27, 2012 11:55 am
Vincent Diepeveen Tue Mar 27, 2012 12:37 pm
Lucas Braesch Tue Mar 27, 2012 1:03 pm
Edmund Moshammer Tue Mar 27, 2012 1:24 pm
Vincent Diepeveen Tue Mar 27, 2012 4:52 pm
Thomas Petzke Tue Mar 27, 2012 1:28 pm
Larry Kaufman Tue Mar 27, 2012 1:46 pm
Alcides Schulz Tue Mar 27, 2012 2:03 pm
Thomas Petzke Tue Mar 27, 2012 2:42 pm
Vincent Diepeveen Tue Mar 27, 2012 5:11 pm
Larry Kaufman Tue Mar 27, 2012 6:15 pm
Vincent Diepeveen Tue Mar 27, 2012 6:22 pm
Larry Kaufman Tue Mar 27, 2012 6:31 pm
Vincent Diepeveen Tue Mar 27, 2012 6:43 pm
Larry Kaufman Tue Mar 27, 2012 7:17 pm
Vincent Diepeveen Wed Mar 28, 2012 8:00 am
Larry Kaufman Wed Mar 28, 2012 2:08 pm
Vincent Diepeveen Wed Mar 28, 2012 5:07 pm
Re: When does a cut-node became an all-node? Larry Kaufman Wed Mar 28, 2012 5:55 pm
Vincent Diepeveen Wed Mar 28, 2012 8:33 pm
Larry Kaufman Wed Mar 28, 2012 9:16 pm
Vincent Diepeveen Wed Mar 28, 2012 9:37 pm
Ed Schroder Thu Mar 29, 2012 3:22 pm
Vincent Diepeveen Wed Mar 28, 2012 8:56 pm
Larry Kaufman Wed Mar 28, 2012 9:26 pm
Vincent Diepeveen Wed Mar 28, 2012 9:59 pm
Vincent Diepeveen Wed Mar 28, 2012 10:02 pm
Larry Kaufman Wed Mar 28, 2012 10:46 pm
Ricardo Barreira Wed Mar 28, 2012 8:57 pm
Vincent Diepeveen Wed Mar 28, 2012 9:22 pm
Ricardo Barreira Thu Mar 29, 2012 9:38 am
Vincent Diepeveen Thu Mar 29, 2012 9:45 am
Marco Costalba Tue Mar 27, 2012 6:37 pm
Robert Hyatt Tue Mar 27, 2012 5:19 pm
Vincent Diepeveen Tue Mar 27, 2012 5:35 pm
Steve Maughan Tue Mar 27, 2012 5:59 pm
Robert Hyatt Tue Mar 27, 2012 5:12 pm
Alcides Schulz Tue Mar 27, 2012 5:41 pm
Robert Hyatt Wed Mar 28, 2012 10:01 pm
Larry Kaufman Wed Mar 28, 2012 10:31 pm
Daniel Shawul Thu Mar 29, 2012 12:28 am
Larry Kaufman Thu Mar 29, 2012 1:51 am
Sam Hamilton Thu Mar 29, 2012 2:05 am
Larry Kaufman Thu Mar 29, 2012 2:37 am
Sam Hamilton Thu Mar 29, 2012 2:56 am
Vincent Diepeveen Thu Mar 29, 2012 9:36 am
Ed Schroder Thu Mar 29, 2012 4:01 pm
Daniel Shawul Thu Mar 29, 2012 10:53 am
Larry Kaufman Thu Mar 29, 2012 2:24 pm
Daniel Shawul Thu Mar 29, 2012 4:04 pm
Daniel Shawul Thu Mar 29, 2012 4:27 pm
Larry Kaufman Thu Mar 29, 2012 5:49 pm
Daniel Shawul Thu Mar 29, 2012 7:18 pm
Larry Kaufman Thu Mar 29, 2012 4:28 pm
Daniel Shawul Thu Mar 29, 2012 4:52 pm
Larry Kaufman Thu Mar 29, 2012 5:58 pm
Daniel Shawul Thu Mar 29, 2012 7:13 pm
Karlo Bala Jr. Thu Mar 29, 2012 8:39 pm
Daniel Shawul Thu Mar 29, 2012 9:02 pm
Karlo Bala Jr. Thu Mar 29, 2012 9:23 pm
Daniel Shawul Thu Mar 29, 2012 9:45 pm
Vincent Diepeveen Fri Mar 30, 2012 12:26 am
Robert Hyatt Thu Mar 29, 2012 8:34 pm
Alcides Schulz Thu Mar 29, 2012 3:05 am
Thomas Petzke Thu Mar 29, 2012 6:53 am

 Jump to: Select a forum Computer Chess Club Forums----------------Computer Chess Club: General TopicsComputer Chess Club: Tournaments and MatchesComputer Chess Club: Programming and Technical DiscussionsComputer Chess Club: Engine Origins Other Forums----------------Chess Thinkers ForumForum Help and Suggestions
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum