Author: Vincent Diepeveen
Date: 15:42:59 07/04/03
Go up one level in this thread
On July 04, 2003 at 16:33:50, Uri Blass wrote: >On July 04, 2003 at 15:44:53, Vincent Diepeveen wrote: > >>On July 04, 2003 at 11:38:09, Andrew Williams wrote: >> >>>On July 04, 2003 at 11:18:58, Vincent Diepeveen wrote: >>> >>>>On July 03, 2003 at 13:57:02, Dann Corbit wrote: >>>> >>>>>On July 03, 2003 at 12:28:05, Ralph Stoesser wrote: >>>>> >>>>>>Dear chess programmers, >>>>>> >>>>>>What are your personal experiences with the MTD(f) search introduced by Aske >>>>>>Plaat some years ago? >>>>> >>>>>It does not work for me as well as it does for some others. >>>>> >>>>>I think success will depend very much on your particular engine. >>>>> >>>>>Andrew Williams has a successful implementation. >>>> >>>>Claims to have a successful implementation is more near the truth. >>>> >>> >>>I am a bit surprised to read this. I sincerely hope you're not claiming that I'm >>>lying about my implementation? >> >>No i just said that you *claim* to have a successful implementation. >> >>I didn't say a word more or less than that. I would be the last in the world to >>suggest you are a liar as everyone knows you are honest. >> >>I did imply however that i doubt that your implementation of MTD will use less >>nodes on average when all the participants of the world champs 2003 would get >>rid of the PVS that they use and use MTD instead. > >1)I do not think that the question if you have succesful implementation is >dependent on what other do. >Succesful implementation is something that is better than what you did >before doing it. > >2)You cannot know what other are going to do. >You even cannot know the list of the participants because some participants may >decide only later if to participate so >"all the participants of the world champs 2003 would get rid of the PVS... " has >no basis. > >Uri We can get lengthy discussions, but did you read what i wrote about Aske Plaat? He could *never* have concluded that at a 512 processor origin MTD worked for him. It is impossible that it *ever* works there for sure. Even those who have implemented MTD agree. Now second thing is. Some programs MTD won't work for sure because they use Pawn =1000. Ok end of story for those programs. They are forced to use PVS. then we keep left with a big group of programs pawn=100. carefully skipping fritz of course which is doing some sort of combination PVS with sometimes skipping plies using a single bound. Though that isn't exactly MTD, it sure isn't using enough to call it MTD. Now we know from all the commercial guys that they have extensively experimented with all search algorithms. In fact sometimes i see discussions about algorithms where i know some commercial guys experimented with years ago. There is no exceptions in the commercial ranges. If MTD would tie for them, they would use it. So we can scrap another 7 engines from the world champs. What do we get left with? Just a few amateur engines. From those at most SOS is using MTD. Now SOS is not exactly known for its big evaluation, so we can see SOS as an exception engine in this sense that its iterations won't change soon in score a lot. It either sees something or it doesn't see it. In its eyes something is 'very good' or 'very bad'. A clear vague grey area there is for those who either use mobility or some knowledge or whatever that gives the engine either some differentiation in score or positional ideas, none of them see MTD working. SOS really is the exception here. So what proof do we have that MTD actually *does* work? The proof we have tells us that the commercial programmers all have found out that it does *not* work for them. That's all we know. That's also the truth about MTD. Note that ASKE plaat in his research didn't use a qsearch at all. He used random trees. You find that good science? Best regards, Vincent
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.