Revolution dev 1.0.1 040925

I have uploaded to the main branch the source code of the version that surpasses all the ones I have created, including yesterday’s, with some more functions. After thousands of games in SPRT tests, he always ends up passing with at least five elo points. We’ve added the following:

Experience Learning

• Experience Enabled: enables or disables the experience feature (default true).
• Experience File: name of the file where the experience data is stored (default experience.exp; legacy .bin files are converted automatically to this format).
• Experience Readonly: if true, no changes are written to the file.
• Experience Prior: uses stored experience to bias root move ordering.
• Experience Width: number of principal moves to consider (1–20).
• Experience Eval Weight: weighting of evaluation when ordering moves (0–10).
• Experience Min Depth: minimum depth required to store a move (4–64).
• Experience Max Moves: maximum number of moves saved per position (1–100).
• Experience Book: if true, use the experience file as an opening book.
• Experience Book Max Moves: limit of moves considered when using the experience book (1–100, default 100).
• Experience Book Min Depth: minimum depth required for a move to be used from the experience book (1–255, default 4)

UCI Options

Minimum Thinking Time

The Minimum Thinking Time option ensures the engine spends at least a specified number of milliseconds searching for a move, even if it finds a good one instantly. This avoids extremely fast, low-quality replies. Set it with:setoption name Minimum Thinking Time value

setoption name Minimum Thinking Time value <milliseconds>

setoption name Minimum Thinking Time value <milliseconds>

Falcon Net

Revolution can switch to an alternative neural network using the FalconFile option. To load the bundled 3.net file, send:

setoption name FalconFile value 3.net

setoption name FalconFile value 3.net

Last test SPRT

Clear summary:

Results of revolution vs revolution_dev_010925_v1.0.1 (10+0.1, 1t, 32MB, UHO_2024_8mvs_+085_+094.pgn): Elo: 25.05 +/- 23.55, nElo: 44.79 +/- 41.91 LOS: 98.19 %, DrawRatio: 56.06 %, PairsRatio: 1.76 Games: 264, Wins: 71, Losses: 52, Draws: 141, Points: 141.5 (53.60 %) Ptnml(0-2): [3, 18, 74, 31, 6], WL/DD Ratio: 0.61 LLR: 0.23 (7.9%) (-2.94, 2.94) [0.00, 2.50]

Results of revolution vs revolution_dev_010925_v1.0.1 (10+0.1, 1t, 32MB, UHO_2024_8mvs_+085_+094.pgn): Elo: 25.05 +/- 23.55, nElo: 44.79 +/- 41.91 LOS: 98.19 %, DrawRatio: 56.06 %, PairsRatio: 1.76 Games: 264, Wins: 71, Losses: 52, Draws: 141, Points: 141.5 (53.60 %) Ptnml(0-2): [3, 18, 74, 31, 6], WL/DD Ratio: 0.61 LLR: 0.23 (7.9%) (-2.94, 2.94) [0.00, 2.50]

• Who’s ahead: revolution (the first engine listed) by +25.05 Elo vs revolution_dev_010925_v1.0.1 at 10+0.1, 1 thread, 32 MB.
• Degree of certainty: The error is large ( ±23.55 ); the LLR=0.23 does not cross the SPRT thresholds (±2.94), so the test is not yet decided (although the LOS=98.19% is favorable).

How many more games to fine-tune the margin?

• For ±10 Elo : ~ 1,464 total games ( +1,200 more).
• For ±5 Elo : ~ 5,857 total ( +5,593 more).
• For ±6/±8/±12/±15 : ~4 067 / 2 288 / 1 017 / 651 totals (≈ +3 803 / +2 024 / +753 / +387).

Ruler:Naim=Nactual⋅(ANDactual/ANDaim)2N_\text{objetivo}=N_\text{actual}\cdot(E_\text{actual}/E_\text{objetivo})^2Naim​=Nactual​⋅(Andactual​/Andaim​)2with your numbersN=264,AND=23.55N=264, E=23.55N=264,AND=23.55.

Download the last version