OpenChess

After the Cool Iris 13 version I realized that there were some bugs, then by fixing these bugs I realized that I could possibly improve a little more to Cool Iris 13, with the support of github copilot I have been able to create this latest version.

The changes I've made are:

futility_margin: Increased from 126 to 150 and reduced the noTtCutNode factor from 42 to 35, making pruning less aggressive.
Reduction:
Base value reduced from 1560 to 1460
Increased reductionScale threshold from 791 to 852
Adjusted the delta factor to be more accurate (1024 instead of 945)
futility_move_count:
For Upgrading Positions: Increased from 3 to 4 Base Moves
For positions that do not improve: changed from (3 + depth * depth) / 2 to (3 + depth * depth)
stat_bonus:
Increased depth multiplier from 334 to 384
Increased the ceiling from 1538 to 1792
These changes should result in:

A deeper search by reducing the aggressiveness of pruning
Better evaluation of movements when considering more positions
Greater weight to the history of movements to improve the order

Note that these changes will cause the engine:

Look deeper but slower
Be less likely to prune movements prematurely
Consider more variants in complex positions.

Enjoy!

https://pixeldrain.com/u/MGUMxUED

The updates of SpecTral and Dark SisTer can be found along with a large collection of engines, books, dvds, etc on my telegram channel: https://t.me/chessenginesac

Thank you very much!

Thank you very much Anton101 !

You are welcome!

Sorry but...New Cool Iris 13.97!!!!

What's new?

File search.cpp:
I have reduced the margin of futility in the futility_margin function to favor the search for tactics. Now, the margin is calculated as 200*d, allowing for deeper exploration in tactically rich positions.

File evaluate.cpp:
I've adjusted the simple_eval feature to include additional mobility bonuses for pawns, knights, bishops, queen, king, and rooks. This encourages more active and tactical positions, valuing the mobility of these pieces more.

These changes are designed to make the chess engine more valuable tactical positions and improve its search ability in complex situations.

It improved significantly in test suite with MULTIPV by 5, compared to Cool Iris 13.90

For server games or blitz control time tests I suggest the maximum multipv in 1 or 2, for better results.

I've added an experience file that you can start working with.

I use GitHub Copilot, Sonar Qube and Chatgpt 4to for analysis, debugging, test and compilations.

Download Telegram channel https://t.me/chessenginesac

https://immortalchess.pw/index.php?topi ... rdseen#new

Interesting!