Author: Tom Likens
Date: 09:05:37 12/29/03
Go up one level in this thread
On December 28, 2003 at 22:18:04, Anthony Cozzie wrote: [-snip-] >In my personal opinion: The time of such static tricks has passed. When people >ran on a 386 and struggled to get 5 ply, extensions && pruning were critical. >Top programs nowadays get 14 ply. The simple fact of the matter is that >computers are almost perfect tactically. More depth now is purely for >positional benefit. Right now I'm trying to _reduce_ my extensions, not add >more :) > >Thorsten and Ed have both said that Rebel plays better with the reductions off. >The only engine on ICC that uses his reductions is Chompster, and I've seen >chompster make errors on a 10 ply search that Zappa catches with a 6 ply search. > >That all said, your time is your own to spend :) > >anthony Hello Anthony, It's ironic, but the reason that most programs reach depths of 14 ply is because they are forward-pruning the heck out of the tree using the null-move heuristic. Of course, hardware *is* faster, but if you turned off the null-move most programs would not reach 14 ply in the middlegame and they would play weaker (IMHO). Some programs would reach the 14-ply anyway, *but* I'm guessing a large percentage of them would be using other forms of forward pruning (futility, razoring etc.) and/or other extensions. It takes awhile to reach depths of 14+ plies in the middlegame with a straight alpha-beta search (i.e. no extensions, reductions or elimination of branches). Can I prove any of this, no not really (I'm at work without access to any chess program). But I strongly suspect it. I intend to experiment with this idea and see if anything comes of it. One, at least to me, interesting thing I've noticed is that reducing the least promising (i.e. boring) moves in my initially very naive way, did *not* decrease the program's tactical strength. In fact, it solved a number of test suites much faster and in less time. But it did weaken the program positionally. This surprised me, because calling the evaluation function at all nodes slowed things down by about 20%. I don't know if this idea has merit, but I think it's an interesting problem, that may have a decent payoff. If not then I'm guessing I'll have a better understanding of what is going on in my engine. Either way, if I do learn anything I'll pass it on (yep, this is a hobby for me as well) ;-) regards, --tom
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.