Introducing my chess engine: GM of Logic! (w/Gui)

[voyagerOne]

This is a Java applet... so you need to have Java installed in order to run it.

Engine Details:
Magic Bitboards.
MVV/LVA move ordering.
Iterative Deepening.
Fail Soft Alpha Beta.
Somewhat modest Evaluation.

Still need to do:
Advance pruning techniques...right now its only alphabeta.
TT
Opening book.
SEE.
Null Move Pruning
Etc.

Please let me know what you folk think.

GM of Logic


P.S. It runs much much faster in a stand alone applet. I believe it runs as a 32 bit version since its a web applet.

[smatovic]

Congratulations,
looks fast and i like the web-interface.

--
Srdja

[ZirconiumX]

It has a bug:

I set up the Ruy Lopez, and in reurn to 3.Bb5, it plays 3. ... Nf6. When I defend with 4. d3, the e-pawn magically disappears!

Is that what you mean by Magic Bitboards?

Matthew:out

[Sven]

It has a bug:

I set up the Ruy Lopez, and in reurn to 3.Bb5, it plays 3. ... Nf6. When I defend with 4. d3, the e-pawn magically disappears!

Is that what you mean by Magic Bitboards?

Matthew:out

I saw a similar thing whenever the engine makes a capture move but I think the piece reappears after a a couple of seconds. It may be just a temporary display problem.

I observed that the PV display is incorrect sometimes. That might indicate that the PV is not maintained correctly. Also the move strings are slightly non-standard, e.g. e1g1 instead of 0-0 (or O-O), or e4 instead of dxe4 for a "pawn takes pawn" capture. These are minor issues, of course.

Anyway, a nice Java engine, well done Bill! Do you plan to add time control as one of your next features?

Sven

[Dave_N]

Beats me! Nice App

[voyagerOne]

It has a bug:

I set up the Ruy Lopez, and in reurn to 3.Bb5, it plays 3. ... Nf6. When I defend with 4. d3, the e-pawn magically disappears!
Is that what you mean by Magic Bitboards?
Matthew:out

Yes, this why my engine is so strong...it proved that the Ruy Lopez is unsound and found a way to kill your pawn somehow

I am aware of this and its just a display latency issue...I just need to lower the thread sleep time. You should notice it will reappear after the computer move...

I observed that the PV display is incorrect sometimes. That might indicate that the PV is not maintained correctly. Also the move strings are slightly non-standard, e.g. e1g1 instead of 0-0 (or O-O), or e4 instead of dxe4 for a "pawn takes pawn" capture. These are minor issues, of course.

Yes, I am aware of this...in my next release I will "clean it up" and standardize it.

Anyway, a nice Java engine, well done Bill! Do you plan to add time control as one of your next features?

Thank you Sven!

This took many many hours.
Numerous cups of coffee.
And more curse words than variables in my code, to create this.

[cms271828]

Very nice engine so far

I've also got one on website, check it out if you want..(JChess3
www.colin-java.co.uk/progs

I didn't manage to beat yours, cause I blundered early on.
It seems to play pretty quick considering no null move or TT or SEE
My engine has a big flaw currently with the move ordering cause I also do MVV/LVA, and I play through all the captures, then the non-captures, I'm hoping there will be a big improvement once I got SEE running.

Out of interest, do you use QS, and how do you generate moves?

[voyagerOne]

Colin:

Your JChess is quite impressive...I like the layout, nice work!

Yes my engine uses QS.
I use Magics to generate all pseudo-legal moves at the beginning of each ply.

Let me know if you have any more questions...I will be more than happy to answer.

Bill

[Robert]

I managed to draw (3 fold repetition) but the engine dont understands the rule!

[cms271828]

Thanks..

I've just been reading your applet code, and I'm totally baffled
I don't see alpha and beta anywhere, but it seems to play chess!

I noticed everything is static too, and most your classes extend the Chess class.
I don't think thats a good idea, but I can see your trying to give each class full access to everything in Chess, so maybe its ok.

But its not what I was expecting to see, some of the code looks very complex, I see you've mastered bitboards too. You may want to convert to fixed shift magics at some point, they seem to be about 40% quicker, might make your program run a couple of % faster, let me know and I'll send you some code for it.

Good job anyway, always good to see different coding styles to ones own