Author: Dann Corbit
Date: 18:34:44 10/11/05
Go up one level in this thread
On October 11, 2005 at 21:10:59, chandler yergin wrote: >On October 11, 2005 at 20:58:29, Dann Corbit wrote: > >>On October 11, 2005 at 20:47:54, chandler yergin wrote: >> >>>On October 11, 2005 at 20:34:28, Michael Yee wrote: >>> >>>>On October 11, 2005 at 19:52:46, chandler yergin wrote: >>>> >>>>>On October 11, 2005 at 19:43:59, Michael Yee wrote: >>>>> >>>>>>On October 11, 2005 at 17:46:02, chandler yergin wrote: >>>>>> >>>>>>>I apologize if I have offended anyone. >>>>>>>A case of "Apples & Oranges" again.. >>>>>>>Misunderstandings.. >>>>>>>Stephen, and a few others here, know I seldom voice "Opinions"; >>>>>>>I Post facts.. >>>>>>>That way... if someone differs from the Post, they should attack the Fact, >>>>>>>not the Poster. >>>>>>>It doesn't always work that way... >>>>>>>Everything I have Posted about Computer Chess Programs..I gave the Link >>>>>>>Everything else I have Posted about Engines, Analysis Modules, is directly >>>>>>>from the Chessbase Manual. It therefore refers to the Top Commercial >>>>>>>Programs only.. Fritz, & Shredder >>>>>>>Many of you that Program your own Engines 'tweak' them as you desire, >>>>>>>and have a lot of fun. >>>>>>>Players mostly use the Top Commercial Programs to assist in their play >>>>>>>and improve their expertise. >>>>>>>Thanks to a recent Post by Dan H. some of the confusion has been cleared up. >>>>>>>Chessbase does use Mini/Max and so does indeed search every legal move >>>>>>>for every position. >>>>>>>So the "Apples & Oranges" now are just the difference between the Top >>>>>>>Commercial Programs and the others. >>>>>>>I find it very intersting that some of the 'lesser' Engines are really >>>>>>>kicking Butt! Congrats to all! Keep it up... >>>>>>>A Swift Kick only hurts for a little while.. >>>>>>>;) >>>>>>>Chan >>>>>> >>>>>> >>>>>>The posts from Dan don't say that chessbase uses minimax and searches every >>>>>>legal move for every position. >>>>> >>>>>I know he didn't say it.. I said it! >>>>> >>>>> In fact, he said something almost the >>>>>>opposite--that if a program worked that way, it would be crushed. >>>>> >>>>>Just plain wrong..sorry. >>>>>> >>>>>>You actually supplied evidence of the answer in one of your other posts (where >>>>>>you provide a description of the analysis output). The analysis window reported: >>>>>> >>>>>>search depth = 12 (selective depth 40) >>>>>>positions searched = 4.2 x 10^7 >>>>> >>>>>I was using an example.. the type of info the Analysis Module window shows. >>>>> >>>>>> >>>>>>If the analysis module really searched every position up to depth 12 >>>>>>(approximately 20 moves, 20 replies to each of those moves, 20 replies to each >>>>>>of those replies, etc.), then it would have searched >>>>>> >>>>>>20^12 = 4.1 x 10^15 positions >>>>>> >>>>>>This is way more than what was reported in the analysis window. (And the 20 is a >>>>>>very low estimate of average moves per position anyway.) Instead, >>>>>> >>>>>>4.32^12 = 4.22 x 10^7 >>>>>> >>>>>>shows a branching factor that's more like 4--i.e., the program was not searching >>>>>>every move in every position. >>>>>> >>>>>>Michael >>>>> Michael, Please read the Chessbase Manual! >>>>>Do you have Chessbase? Do you have Fritz or Shredder? >>>>>If so Start the engine look at the Analysis Window as I have requested. >>>>>You will find what I said. EVERY Legal move in the position is evaluated. >>>>>It's the way it works. Like it or not. >>>> >>>>Some commercial engines I have include an old fritz 7, shredder 9 uci, fruit >>>>2.2, and gandalf 6. I've analyzed positions before (e.g., during some of the >>>>recent World Championship games from Argentina). What I see are depths, >>>>selective depths, and node counts that are not consistent with a program that >>>>searches every legal move in every position. >>>> >>>>You're right that engines do look at each move in the root (initial) position >>>>(as Dann noted in another post). If that's what you meant, then I agree. But you >>>>seem to referring to *all* positions (not just the root). >>> >>>No Michael, and as Shakespeare said.. "Aye, there lies the rub.." >>>Please re-read my Post carefully. >>>Well.. here.. >>>"Start your Engine for a position or at any part of a game. >>>Look at the analysis window >>>What do you see? >>>The analysis module should show the following: >>>The name of the Engine >>>The search depth (brute force selective) Example "12/40" means that all >>>variations were >>>searched to a depth of 12 ply, while some promising continuations were checked >>>down to 40 ply. >>>Next should show the move currently being investigated. Example f4-d6 (3/47) >>>meaning number 3 of 47 legal moves in the position. >>>Next it will show the speed at which the program is running. >>>Example: 403kN/s means it is looking at 403,000 nodes (= positions) per second. >>>This is normal on a 400MHz processor. >>>The main variation for example shows "=(0.00)", then the best sequence of moves >>>the program has found so far, the amount of time it has spent computing on the >>>position,(1 min. 46 sec) and the number of positions it has examined >>>(41937kN =41,973,000) >>>The evaluation expressed in units of a pawn, always from the point of view of >>>White >>>"+0.53) means the program thinks White has an advantage of about half a pawn; >>>" (-3.52" indicates Black is more than a piece up. If Mate is found the Program >>>stops calculating and displays the Mate. (Mate in 6)." >>> >>>What if anything is incorrect in the above? >> >>There is nothing wrong in the statements above. The problem is in your failure >>to understand what they mean. > >If I didn't understand it, I wouldn't have posted it. >To have a meaningful dialog about anything, both parties should have the >same consensus about the meaning. >I Posted my meaning; was looking for constructive comments by those that had a >different view. I sure got them.. but instead of commenting on specifics, >they were just personal attacks, with nothing constructive, >I appreciate your review! Please add where you believe I err. >Thanks and sincere Best Wishes, >Chan You err in thinking that chess programs use mini/max to search. They use a more sophisticated variant called alpha/beta which cuts the branching factor from 20 to around 6. Then they trim that further with lots of speculation that typically results in a branching factor of under 3. Notice that this does not mean that it seaches half as many nodes, but that in a 17 ply search it is searching less than one trillionth as many nodes as it would with a full width search. You err in thinking that a ply is the same as fullmove. A ply is 1/2 of a full move and constitues one turn for a player of a given color. You err in thinking that professional chess programs examine every node during a search. They search a microscopic fraction of the total number of nodes. Take a bathtub full of water. Now take an eyedropper and get some of the water. Squeeze gently to eek out the smallest drop you can create. On a 17 ply search the ratio between nodes examined and nodes ignored is less than the ratio of water in the bathtub to that drop you just made. >>>>You yourself computed a while ago an estimate of the number of move sequences >>>>that can happen from the start position. Given the nodes per second you see in >>>>the analysis windows, how could the engine possibly be following all move >>>>sequences (even taking into account transpositions)? >>> >>>That was Dann, not me.. >>>> >>>>Michael
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.