Test cases and transposition tables!

The test cases was a great idea and it paid off! I detected a problem with the Zobrist keys in that en-passant was not correctly added/removed. I also detected a nasty (and embarrassing) castling bug. I omitted the break-statement after each case-statement when verifying that the castling was valid. This means that if white wants to castle king-side, queen side castling must not move through check and white must not prevent black to castling either side, either! A very silly bug! I only detected it after adding a large series of perft checks to my test cases. By squashing this bug, my engine plays much better chess already!

Transposition tables is turning out to be very annoying! I keep on getting them "working" only to find that in a duel (100 games, 3 minutes per side, different starting positions) with the previous version that is does badly... or at best, just as good. Shouldn't transposition tables increase the ELO with 200+ points? As a result I've put this on ice for awhile.

In the WRCC tournament, my engine lost a particular game by means of a long capturing sequence that ended in a check. As checks are not part of the quiescence search, my engine never saw the pending disaster. I am thinking of ways to incorporate checks into the quiescence search without slowing things down too much. But how?

I'm also going to attempt to incorporate null moves into the engine. After reading up on the concept a few months ago I was most intrigued with the concept and it would be interesting to see what effect it will have on my engine.

Vicki playing online

I've just checked on Vicki playing online at FICS. The engine is doing pretty well...actually too well. Vicki has a rating of over 2000 for standard and lightning play and 1897 for blitz (see my website for more exact figures - or better yet - log onto FICS and pay Vicki a visit. She plays under the "myvicki" handle).

I've shifted my approach when working on the code. Instead of hacking away and trying to squeeze a few ELO points in between work, family and social life (actually work and family only - sadly this blog is pretty much the limits of my social life :-(), I want to set definite milestones for the engine. The first important change is test cases. I've devoted an entire c-source file for test cases, which I'll run periodically to ensure that I did not break something while adding some nifty new feature. My "hacking"-approach is what cost me to play without a transposition table in last weekend's tournament. This needs to end...

Vicki tasted blood!

It is with great pride that I announce that Vicki is now playing on the Free Internet Chess Server (FICS) under the handle of "myvicki". Please pay her a visit! After all the fun of the WCRCC I am very motivated in getting Vicki to play decent chess!

Vicki is also competing in ChessWars XI under the promotion section. So far Vicki is not doing that bad, having drawn the first game. It would be interesting to see how she does later on.

Final standings...

The tournament was very exciting and Vicki actually won one game and drew another. For an engine that's missing serious components, I am quite surprised. However, I can't wait for the next tournament to start so that I can amaze a few people... grrr... :-)

Mediocre (blog here) did particularly well - the name really does not suit it well. It is far from being Mediocre and to think it is all written in Java! Well done to Jonatan Peterson. He's a real inspiration!

Just for fun, here is the game that Vicki won (my little engine was playing black). I hope to have many of these soon!!

Official tournament!

Hi all...

I haver entered Vicki into the "The 2007 Second Annual ACCA President's Tournament & 2007 World Computer Rapid Chess Championships". So far, 40 engines have entered and I fear them all! Although this will be a good exposure for Vicki.

I am having difficulties with the transposition table and it seems that Vicki may be the only engine participating without a transposition table at all. At least the improved move ordering is allowing fairly deep tree searches. Also the evaluation function has been improved somewhat and bug fixes have been made. I'll add this version of Vicki to the website soon.

I know Vicki will be obliterated in this tournament - but if I can only get one draw, I'll be very happy :-)

Hold thumbs!

Hello? Someone there?

Not a single post in June does not mean that I have not been busy! I have the new version of Vicki almost ready. Only a few more tweaks before release. My focus has been on the static evaluation code which was rather poor up till now. I separated the evaluation function into 3 sets, depending on opening, middle or endgame.

The transposition table is working... but I did expect a bit more... at least it is working. A few nasty bugs are still lurking in the code during draws which I still need to resolve.

I am also implementing better time control and rewriting the winboard section so that Vicki claims draws and end the game after losing/winning.

I'm also running tournaments to measure the development of the engine. This can be viewed at http://www.vicki.co.za/test.html and is updated real-time.

If only I had more time!

Transposition tables and tidying up.

I realised this morning that it has been quite a while since my last posting. I am currently working on a number of things in Vicki and my guess is that release 0.04a would be significantly stronger than the previous versions. I am somewhat reluctant to drop the alpha qualifier, but perhaps I will in the next release (version 0.05beta).

I currently have a structure called CHESSENGINE that I pass around (or the pointer to it) between functions. I realised that this parameter-passing may actually degrade performance, cause problems because of its size (I’m not completely sure, but isn’t there a limit of 64k on a "struct"s – even on modern operating systems?) and waste quite a few clock cycles to find the data element within this structure. Regardless, all I gain from this is the ability to have multiple engines in the same process as its state is completely described by this one chunk of data. I’m moving away from this and dumping everything as global variables and structures in a header file. Who wants multiple engines (of the same kind and strength) in the same process anyway?

I am also implementing an elementary transposition table and cleaning up the code as I go. At the moment I am having problems getting it to work, but I’m sure it is just a case of a silly bug lurking somewhere. Once this are done, Vicki will not be such a pushover anymore! ;-)

I’ll keep you all posted...