Since you’re here...

We hope you will consider supporting us today. We need your support to continue to exist, because good entries are more and more work time. Every reader contribution, however big or small, is so valuable. Support "Chess Engines Diary" even a small amount– and it only takes a minute. Thank you.
============================== My email: jotes@go2.pl



Amira 1.82 - new version chess engine

on
 


Amira - UCI chess engine
Author: FauziAkram

Amira 1.82 what's new?

Amira 1.82: Strategic Depth, Surgical Precision
Amira 1.82 is the culmination of extensive refinement following the 1.71 version. This version focuses on "polishing" the engine's core, introducing more precise evaluation weights and a more disciplined search algorithm. Amira 1.82 navigates complex tactical middlegames with greater composure and shows a marked improvement in conversion accuracy during technical endgames.

Key Features in This Release:
🧠 Precision Evaluation Model
The evaluation function has been fine-tuned to better handle the transition between the middlegame and the endgame, ensuring a more seamless strategic flow:

Tuned King Safety & Shelter: The parameters for king safety have been recalibrated. Amira is now more sensitive to "weak squares" around the king and better values the defensive contribution of a solid pawn shield.
Passed Pawn Maturity: Improvements to the PasserMyKingDistance and PasserEnemyKingDistance logic allow Amira to better judge the racing potential of passed pawns in the endgame.
Dynamic Outpost Evaluation: The engine's understanding of "Dominance" has been sharpened, rewarding knights and bishops that occupy central outposts more accurately based on their support by friendly pawns.
Refined Threat Assessment: The evaluate_threats_for_color function has been adjusted to better recognize hanging pieces and weak queen defenses, making the engine more opportunistic in quiet positions.
⚡ Optimized Search & Heuristics
The search architecture has undergone significant stability testing, leading to more efficient pruning and better move ordering:

Enhanced Late Move Reductions (LMR): The table-driven LMR has been tuned. Amira now applies reductions more intelligently to "quiet" moves that have historically failed to improve the position, allowing it to reach greater depths in critical lines.
Aspiration Window Refinement: The search now utilizes a more adaptive aspiration window. This prevents the engine from "chasing shadows" in volatile positions, leading to faster convergence on the best move.
Improved History Heuristic: The history scoring system—which tracks the success of quiet moves—has been recalibrated to better prioritize refutation moves across the search tree.
Aggressive SEE & Futility Pruning: Static Exchange Evaluation (SEE) thresholds have been tightened, allowing the engine to discard losing tactical sequences earlier and focus its "thinking time" on more promising candidates.
🛠 UCI & Performance
Stability Fixes: Various internal refactors to the bitboard utilities for improved performance across different hardware architectures.
Clock Discipline: Refined time management ensures the engine uses its time more efficiently in "sudden death" time controls while retaining its aggressive "panic mode" when the evaluation fluctuates.
Downloads & Compilation
The attached binary is a 64-bit executable optimized for Windows.

If you wish to compile the engine from source for maximum performance on your specific CPU, you can use the following command.

To compile using g++:

g++ -o Amira-1.82 main.cpp -std=c++17 -O3 -march=native -flto


Comments

Post a Comment