Merge pull request #2 from Saphereye/dev/algo-refactor

dev/algo refactor

Cargo.toml

@@ -1,10 +1,11 @@
 [package]
 name = "shard"
-version = "1.0.0"
+version = "2.0.0"
 edition = "2021"
 authors = ["Saphereye <[email protected]>"]
 
 [dependencies]
 chess = "3.2.0"
-clap = { version = "4.5.18", features = ["derive"] }
+nom = "7.1.3"
 rand = "0.8.5"
+rayon = "1.10.0"

README.md

@@ -1,32 +1,24 @@
 # Shard Chess Engine
 ![](https://github.com/Saphereye/shard/blob/main/assets/demo.png)
+*Shard Vs Shard (Shard Wins)*
 
-# Roadmap
-### Improve Evaluation Function
-Enhance the evaluation function to consider more factors:
-- [ ] King safety
-- [ ] Pawn structure (isolated, doubled, passed pawns)
-- [ ] Piece mobility
-- [ ] Control of key squares (e.g., center control)
-- [ ] Rook on open files
-- [ ] Bishop pair advantage
+## Introduction
+Shard is a chess engine written in Rust. It is supports most UCI commands.
+The engine utilizes the Monte Carlo Tree Search algorithm.
 
-### Enhance Move Ordering
-Implement more advanced move ordering techniques:
-- [ ] History heuristic: Track the success of moves in the search tree and prioritize moves that have historically caused beta cutoffs.
-- [X] Killer move heuristic: Track moves that cause beta cutoffs at the same depth and prioritize them.
-- [X] Principal variation search: Use the best move from the previous iteration as the first move in the current iteration.
-
-### Improve Quiescence Search
-Extend the quiescence search to include more types of moves, such as:
-- [X] Checks
-- [ ] Threats (moves that create immediate threats)
-
-### Optimize Transposition Table
-- [ ] Use Zobrist hashing to efficiently hash board positions and implement a more sophisticated replacement scheme (e.g., always replace, depth- [ based replacement).
-
-### Implement Advanced Search Techniques
-- [X] **Iterative Deepening**: Perform a series of depth- [ limited searches, increasing the depth with each iteration. This helps with move ordering and time management.
-- [ ] **Null Move Pruning**: Assume the opponent will pass their turn and see if the current player can still achieve a good position. If so, prune the branch.
-- [ ] **Late Move Reductions**: Reduce the search depth for moves that are considered less likely to be good (e.g., moves that are not captures or checks).
-- [ ] **Aspiration Windows**: Use a narrow window around the expected score to speed up alpha- [ beta search. If the score falls outside the window, re- [ search with a wider window.
+The tree looks like this:
+```
+Node t:16396, n:36 (rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1) (None)
+    Node t:11048, n:8 (rnbqkbnr/pppppppp/8/8/6P1/8/PPPPPP1P/RNBQKBNR b KQkq - 0 1) (g2g4)
+        Node t:538, n:1 (rnbqkbnr/1ppppppp/8/p7/6P1/8/PPPPPP1P/RNBQKBNR w KQkq - 0 1) (a7a5)
+        Node t:-477, n:1 (rnbqkbnr/1ppppppp/p7/8/6P1/8/PPPPPP1P/RNBQKBNR w KQkq - 0 1) (a7a6)
+        Node t:-553, n:1 (rnbqkbnr/p1pppppp/8/1p6/6P1/8/PPPPPP1P/RNBQKBNR w KQkq - 0 1) (b7b5)
+        ...
+    Node t:7211, n:6 (rnbqkbnr/pppppppp/8/8/8/7P/PPPPPPP1/RNBQKBNR b KQkq - 0 1) (h2h3)
+        Node t:-137, n:1 (rnbqkbnr/1ppppppp/8/p7/8/7P/PPPPPPP1/RNBQKBNR w KQkq - 0 1) (a7a5)
+        Node t:-1526, n:1 (rnbqkbnr/1ppppppp/p7/8/8/7P/PPPPPPP1/RNBQKBNR w KQkq - 0 1) (a7a6)
+        ...
+    ...
+```
+Each 'Node' represents a position in the game. The 't' value is the total score of the position, and the 'n' value is the number of times the position has been visited.
+When a move is made, the engine will set the current position to the position of the move, and the tree will be updated accordingly from there.