Skip to main content
3 votes
Accepted

More accurate estimate of chess games possible

Peruse this link. Also see link for possible duplicate. Sum-up of first link: Longest game under FIDE rules: 8848.5 (Fabel) Maximum mobility number: 218 moves Shannon's number, assuming 40 (full) ...
Hauke Reddmann's user avatar
3 votes
Accepted

What's the maximum number of pieces attacking a single square?

Assuming you mean pieces of just one color, 15 can be attacking one square: [fen "B3R3/7Q/3N1N2/2N3N1/3Kn3/2NP1PN1/3N1N2/4R3 w - - 0 1"] The wrong-colored bishop is the only one that can't ...
D M's user avatar
  • 19.7k
3 votes

Is the following position a fortress ? Or can White win?

This looks like a draw. First, the pawn ending (1.Qa8 Ka8 2.ba4 Kb8 3.Kd6 Kc8) is dead drawn: the bK stays on c8/d8/c7 and the wK cannot approach. 1.ba4 is an immediate stalemate, so White has to try ...
Evargalo's user avatar
  • 16.4k
3 votes

Humanlike chess engine that isn't a neural network

That would be a fun goal. Last year we have a discussion on "Ideal Chess Engine for Kids (for Practice)" in talkchess. I tried to make some of its requirements in my engine CDrill. For ...
ferdy's user avatar
  • 4,115
3 votes

Chess match between two smartphones

If you want to do it manually, You can install an engine of your choice onto each phone, start a match on each phone (one where the engine is playing white, and one where it is playing black) and ...
Ryley's user avatar
  • 416
3 votes

Why is this the best move?

10 . . . Nxe4 is best because it wins a pawn, while other moves only keep material equality (or worse, give away material). The move is a typical "desperado" sacrifice, introducing a line ...
Noam D. Elkies's user avatar
2 votes

What would the Best move for Black be in Algebraic Chess Notation?

Rc1+ leads to mate in 2 [fen "7k/1p4p1/p4b1p/3N3P/2p5/2rb4/PP2r3/K2R2R1 b - - 0 1"] 1...Rc1+ 2. Rxc1 Bxb2#
Brian Towers's user avatar
  • 98.5k
2 votes

Humanlike chess engine that isn't a neural network

I've got a quick idea here, so it might be complete garbage. What if we blend the traced evaluation with a static evaluation in shallower levels? Instead of only using the static evaluation score from ...
Sheyteo's user avatar
  • 101
2 votes

Humanlike chess engine that isn't a neural network

The best way to make a engine to make human moves is if you program it to have a small depth (5-6 sounds good)for middlegame and big depth for endgame(10-12) That way you will give your engine a ...
Cerise's user avatar
  • 271
2 votes

Transposition Table eviction policy

You're over-complicating it to the extent it doesn't make sense. You don't change the size of used RAM by "clearing" an entry. Clearing an entry means zeroing or changing the value to ...
Keith Downes's user avatar
2 votes

Changing the FEN position color

Here is another approach. It uses the python chess library to flip the side to move via a null move, which also removes the e.p. square if there is. I also added a condition that the side to move is ...
ferdy's user avatar
  • 4,115
1 vote

LLMs for chess games

This isn't really possible. The reasons are technical. LLMs are machine learning models, which means they are only as good as the training data you give them. It's why vanilla LLMs are notoriously bad ...
Allure's user avatar
  • 27.9k
1 vote

Transposition Table eviction policy

I do believe I have come up with an ugly but (hopefully) complete solution to your problem though. What you have to do is keep track of the first irreversible move in every line that happens in search ...
Hank's back rank's user avatar
1 vote

In a blitz game , Leela zero failed to win a winning 7 pieces endgame. Why?

As far as I can tell, the latest update in regards to tablebases is from August 2018. The support in v0.17.0 is partial only, only WDL tables are probed, but not DTZ. That means, that Leela is only ...
Nelson O's user avatar
  • 725
1 vote

What unusual engines do we have?

Leela Zero with one node is an interesting engine setting in Sadler's book The Silicon Road to Chess Improvement, as it doesn't do any search but relies purely on its neural net evaluation function. ...
qwr's user avatar
  • 3,902
1 vote

Are cyborg (human+computer) players really better than the best chess engines?

At this point in time in my opinion the human factor in a computer + human vs computer game might still be playing a difference, although I'm not aware of any experiments in this field done recently. ...
Andrea Nerla's user avatar
1 vote

Why is futility pruning causing an Elo drop in my engine?

As I see if you don't do any depth checking. Also if I am reading your code correctly, you do not check to see if the move is a capture. I could be wrong as you might already be checking this. But, ...
Valsmikor's user avatar
1 vote

How is Stockfish so fast?

Perft speed can be affected by: Coding language - for example, compiled vs interpreted. Bulk counting - whether you un-make leaf nodes - I believe SF doesn't do this, which makes it significantly ...
HTTP 410's user avatar
  • 597
1 vote

How can I make my chess engine more efficient?

If you are not yet using move ordering, I would strongly suggest you do so. Even some basic MVV-LVA helped increase my Python engine's speed by almost 500 percent. Here is some sample code if you want:...
Stockfischer's user avatar
1 vote

Are engine ratings directly comparable to human-level play?

From the Stockfish FAQ, it is basically impossible to compare human and engine ELO, which is highly dependent on rating pool (which other engines played), hardware, time control, and choice of opening ...
qwr's user avatar
  • 3,902

Only top scored, non community-wiki answers of a minimum length are eligible