Page 1 of 7
Alouette 0.0.1
Posted: Fri Mar 29, 2019 12:11 am
by Roland Chastain
Hello!
I started a new engine called
Alouette.
In its current state, the engine plays the first legal move that it finds.
It can play chess 960.
Before working on move evaluation, I would like to make the current program as clean as possible.
Comments and suggestions are welcome. It's written in Pascal Object and uses bitboards.
You can download
Windows binaries and source code here. I created a
Git repository.
Regards.
Roland
Re: Alouette 0.0.1
Posted: Fri Mar 29, 2019 2:26 am
by supersharp77
Roland Chastain wrote: ↑Fri Mar 29, 2019 12:11 am
Hello!
I started a new engine called
Alouette.
In its current state, the engine plays the first legal move that it finds.
It can play chess 960.
Before working on move evaluation, I would like to make the current program as clean as possible.
Comments and suggestions are welcome. It's written in Pascal Object and uses bitboards.
You can download
Windows binaries and source code here. I created a
Git repository.
Regards.
Roland
Thanks for your engine efforts..Bon Chance....AR
Re: Alouette 0.0.1
Posted: Fri Mar 29, 2019 6:08 am
by Modern Times
Roland Chastain wrote: ↑Fri Mar 29, 2019 12:11 am
Hello!
I started a new engine called
Alouette.
It can play chess 960.
Very cool
Re: Alouette 0.0.1
Posted: Fri Mar 29, 2019 7:57 am
by Sergio Martinez
Thank you very much Roland
Re: Alouette 0.0.1
Posted: Fri Mar 29, 2019 11:11 pm
by CMCanavessi
Roland Chastain wrote: ↑Fri Mar 29, 2019 12:11 am
Hello!
I started a new engine called
Alouette.
In its current state, the engine plays the first legal move that it finds.
It can play chess 960.
Before working on move evaluation, I would like to make the current program as clean as possible.
Comments and suggestions are welcome. It's written in Pascal Object and uses bitboards.
You can download
Windows binaries and source code here. I created a
Git repository.
Regards.
Roland
So right now it's similar to a random mover or is there some sort of "logic" in the move chosen?
Re: Alouette 0.0.1
Posted: Fri Mar 29, 2019 11:16 pm
by Roland Chastain
CMCanavessi wrote: ↑Fri Mar 29, 2019 11:11 pm
So right now it's similar to a random mover or is there some sort of "logic" in the move chosen?
Yes, it's similar to a random mover. The moves are in generation order. The first legal move is played.
Code: Select all
Rank Name Elo +/- Games Score Draws
1 Ram 2.0 [wb2uci] 130 136 14 67.9% 50.0%
2 Belofte 0.2.7 76 202 14 60.7% 7.1%
3 NEG 0.3 [wb2uci] 25 135 14 53.6% 50.0%
4 Belofte 0.2.8 0 70 14 50.0% 85.7%
5 Alouette 0.0.1 [32] 0 100 14 50.0% 71.4%
6 Iota 2015 -25 85 14 46.4% 78.6%
7 Alouette 0.0.1 [debug] -102 92 14 35.7% 71.4%
8 Alouette 0.0.1 [64] -102 92 14 35.7% 71.4%
Finished match
Re: Alouette 0.0.1
Posted: Sun Mar 31, 2019 2:27 pm
by Modern Times
I look forward to further development of this engine !
I did run some chess960 games. Most were OK but the GUI did report an illegal move in one of them.
Re: Alouette 0.0.1
Posted: Sun Mar 31, 2019 3:01 pm
by Roland Chastain
Modern Times wrote: ↑Sun Mar 31, 2019 2:27 pm
I look forward to further development of this engine !
I did run some chess960 games. Most were OK but the GUI did report an illegal move in one of them.
Thank you very much. Yes, there were some illegal moves. There was also a problem with castling notation. I believe it is fixed now. Please find Alouette 0.0.2 here:
https://sites.google.com/view/eschecs/alouette
Now the program has a beginning of logic, very simplistic. I hope there is no more bugs.
I would like to explore very simple ideas, and why not make a kind of toy engine for children, with personnalities.
Regards.
Roland
Re: Alouette 0.0.1
Posted: Sun Mar 31, 2019 3:41 pm
by Roland Chastain
Sorry, there are still illegal moves. Not made enough tests in FRC mode.
Regards.
Re: Alouette 0.0.1
Posted: Wed Apr 03, 2019 10:39 am
by Roland Chastain
Hello! I fixed a bug in the castling procedure. It was the origin of some illegal moves in FRC mode. I hope it is good now. Please find Alouette 0.0.3 still at
the same place.
Now I will try to make a little better move evaluation. If you have some very simple ideas to suggest, you are welcome.
Here is the current evaluation formula:
Code: Select all
function Evalue(const APos: TPosition; ACoup: integer): integer;
...
begin
...
result :=
0
+ bonusRoque { bonus for castling }
+ bonusNombreCoups div 3 { bonus for the number of possible "second moves" }
- malusPiecesMenacees { malus for threatened pieces }
- Ord(NomCoup(ACoup) = AvantDernier); { malus for move repetition }
end;
There is no depth in evaluation. The function just evaluates the position after the move.