I downloaded the stockfish uci engine and using python-chess wanted to limit its playing strength. When trying to set the UCI_Elo value it requires a value of at least 1320.
Is there a free open source uci engine which will allow me to set the elo less than 1320 (down to 800 would be nice)?
I appreciate that specifically limiting the depth of analysis would make a difference, but that's not what i'm after. To make things more awkward I'm using linux, and it would be nice not to have to compile the engine myself.
I downloaded the stockfish uci engine and using python-chess wanted to limit its playing strength. When trying to set the UCI_Elo value it requires a value of at least 1320.
Is there a free open source uci engine which will allow me to set the elo less than 1320 (down to 800 would be nice)?
I appreciate that specifically limiting the depth of analysis would make a difference, but that's not what i'm after. To make things more awkward I'm using linux, and it would be nice not to have to compile the engine myself.
LC0 with Maia network can go down to 1100 Lichess rating.
LC0 "trivial" backend (included with some compilations) is very weak with rating in some hundreds, but isn't calibrated to anything. It was designed to sped up initial network training instead of using random networks. But it certainly knows something about chess.
LC0 with Maia network can go down to 1100 Lichess rating.
LC0 "trivial" backend (included with some compilations) is very weak with rating in some hundreds, but isn't calibrated to anything. It was designed to sped up initial network training instead of using random networks. But it certainly knows something about chess.
LucasChess for the PC - free - has an excellent range of rated computer bots to play against, from beginner to top engine.
LucasChess for the PC - free - has an excellent range of rated computer bots to play against, from beginner to top engine.
Just for reference (and i should have thought a bit more about it) fairy-stockfish (https://fairy-stockfish.github.io/) seems to be perfect for me! Its the one lichess uses to "play with the computer".
Just for reference (and i should have thought a bit more about it) fairy-stockfish (https://fairy-stockfish.github.io/) seems to be perfect for me! Its the one lichess uses to "play with the computer".
Take a look at the CCRL rating list https://www.computerchess.org.uk/ccrl/4040/, and choose something at the bottom. But I think finding something at 800 level may be very difficult because it is likely that implementing even some basic algorithms may be sufficient for a rating of 1200, unless the developer intentionally makes the engine to play bad.
There are some old engines called crafty and phalanx. They have ways to reduce strength, but the strength reduction may not be calibrated to the Elo scale. They are (I believe) not UCI engines, but you can run them in xboard on linux.
Also, compiling is often very easy, especially if the developer says that the engine will work on linux. I can help with compiling or configuring xboard.
Take a look at the CCRL rating list https://www.computerchess.org.uk/ccrl/4040/, and choose something at the bottom. But I think finding something at 800 level may be very difficult because it is likely that implementing even some basic algorithms may be sufficient for a rating of 1200, unless the developer intentionally makes the engine to play bad.
There are some old engines called crafty and phalanx. They have ways to reduce strength, but the strength reduction may not be calibrated to the Elo scale. They are (I believe) not UCI engines, but you can run them in xboard on linux.
Also, compiling is often very easy, especially if the developer says that the engine will work on linux. I can help with compiling or configuring xboard.