World Draughts Forum

As a pure excercice I proposed one day to a GMI to try and win with white side in the following position : http://fmjd.org/dias2/save/12411147405.png During about 15 moves Damy confirmed after each white move that white could win but suddenly after a non optimum move from white side, Damy claimed t...

What you are talking about is called "playing against a fallible opponent." It has been the topic of several artificial intelligence papers. The way most a.i. researches handled their experiments was to design two different evaluation functions. 1. Normal Mode evaluation (NM) 2. "Fallible Opponent" ...

Remenber my "vicious" question about the "25 moves" rule. How Damy is working with this rule ? First of all, as any other programmer interested in db, I generated my endgame db without taking into account this rule. I am not sure it is the best solution but it is another subject. What you are refer...

Why do you want to generate the DTW db instead of simply the DTC db? Do you konw that in very very exceptionnal occasions the DTW give a wrong result when facing a program which has the DTC db ? That is not correct. A DTW database knows how many moves until the last move is made on the board. A DTC...

There are many good points being discussed here. As for programming a computer to play more like a human, in terms of knowledge, this was a goal of the 1970's chess program PARADISE which used a very different approach to the game. It "created plans" and not only searched. Plans were determined by i...

Hey Ed,

What position is the longest conversion win you have computed for 10x10 draughts so far?

Or is that a guarded secret until after your next tournament?

[img]images/smilies/icon_smile.gif[/img]

This discussion seems very interesting and I will be happy to have other feelings. For me the most interesting positions are really the theoritically draw positions. Assuming that the initial position is a draw position it is obvious that, in order to win, the opponent has to do a mistake. ... We r...

I believe I perfectly understand what you say but it seems that you miss my point. Let's take a position with only 10 pieces and I assume that you have build the DTW corresponding database. No doubt at all that your software is far better than a human for dealing with a winning position, but what a...

Software in checkers exceeds the strength of every player Does the checkers (or 10x10 international draugths) software strength exceed every player strength? Is this question really resolved ? Alex Moiseyev visited my house right before Christmas in 2004. For me, the timing was very unfortunate. I ...

Hi, Thank you for this explanation. Though I am not able to understand your examples because I do not know chekers rules, I believe I understand now your problem. It seems that that you look for some efficient algorithm to generate DTW databases in the same manner that we can generate DTC database....

I do not see your problem. Hi Gérard. The problem is, I must first do some forward move passes with "different logic" before I can switch to doing the "unmove" logic. But, as the example position shows, my "forward move logic" no longer applies. In my forward move generator, I would "solve" 25-30 a...

I have a question about computing Distance-To-Win databases for checkers (or Draughts) using an "Unmove Generator". Just so everyone knows, I've already done it using a forward move generator. And a quick reminder, DTW is different than "Distance To Conversion", just so we all are "on the same page"...

Ed, This is an impressive rate given that you used a single machine (I used a cluster of machines to compute my deeper depth perft numbers). Aart Hi Aart, I posted some code here for my attempted 100-bit engine for Intl Draughts. Hopefully the more experienced Draughts Programmers can take the idea...

I wrote a 64-bit move generator for the game of American Checkers that has been confirmed to be correct by Ed Gilbert and Aart Bik. I know it sounds strange to talk about 64 bits for a game with 32 squares, but take a look at this layout: XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX...

perft(18) = 3493881706141 perft(19) = 16114043592799 In light of our recent debugging session, I did the perft(18) by move as well, in case anyone needs to debug in the future. I like the speed number best @ 102,466,237 nodes/second. PARENT POSITION WITH RED TO MOVE XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX...