1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149

use super::*;

#[cfg(feature = "debug")]
fn prediction_accuracy_func(rms: f64) -> f64 {
    1.0 - 1.0 / (1.0 + ((10.0 - rms) / 3.0).exp())
}

#[cfg(feature = "debug")]
fn calculate_prediction_accuracy(rms: f64) -> f64 {
    (prediction_accuracy_func(rms) * 100.0) / prediction_accuracy_func(0.0)
}

pub fn self_play(
    engine: &mut impl ChessEngine,
    go_command: GoCommand,
    verbose: bool,
    move_limit: impl Into<Option<NumMoves>> + Copy,
) -> Result<()> {
    let move_limit = move_limit.into().unwrap_or(NumMoves::MAX);
    if move_limit == 0 {
        return Ok(());
    }
    let stating_fen = engine.get_board().get_fen();
    let mut time_taken_vec: Vec<f64> = Vec::new();
    let mut max_time_taken_fen = String::new();
    let mut prediction_score_vec = Vec::new();
    println_wasm!("{}", engine.get_board());
    if engine.get_board().is_game_over() {
        return Err(TimecatError::GameAlreadyOver);
    }
    let initial_num_moves = engine.get_board().get_num_moves();
    while !engine.get_board().is_game_over()
        && (engine.get_board().get_num_moves() as u64)
            < (initial_num_moves as u64) + (move_limit as u64)
    {
        let clock = Instant::now();
        if verbose {
            println_wasm!();
        }
        let response = engine.go(go_command, verbose);
        let Some(best_move) = response.get_best_move() else {
            return Err(TimecatError::BestMoveNotFound {
                fen: engine.get_board().get_fen(),
            });
        };
        let score = response.get_score();
        let time_elapsed = clock.elapsed();
        let best_move_san = best_move
            .stringify_move(engine.get_board().get_sub_board())
            .unwrap();
        let pv = get_pv_string(engine.get_board().get_sub_board(), response.get_pv());
        engine.get_board_mut().push_unchecked(best_move);
        if time_elapsed.as_secs_f64()
            > *time_taken_vec
                .iter()
                .max_by(|&x, &y| x.partial_cmp(y).unwrap())
                .unwrap_or(&0.0)
        {
            max_time_taken_fen = engine.get_board().get_fen();
        }
        time_taken_vec.push(time_elapsed.as_secs_f64());
        prediction_score_vec.push(score);
        let nps =
            (engine.get_num_nodes_searched() as u128 * 10u128.pow(9)) / time_elapsed.as_nanos();
        println_wasm!("\n{}\n", engine.get_board());
        println_info("Best Move", best_move_san);
        println_info("Score", score.stringify());
        println_info("Num Nodes Searched", engine.get_num_nodes_searched());
        println_info("PV Line", pv);
        println_info("Time Taken", time_elapsed.stringify());
        println_info("Nodes per second", format!("{} nodes/s", nps));
    }
    let mean = time_taken_vec.iter().sum::<f64>() / time_taken_vec.len() as f64;
    let std_err = (time_taken_vec
        .iter()
        .map(|x| (x - mean).powi(2))
        .sum::<f64>()
        / time_taken_vec.len() as f64)
        .sqrt();
    let max_time_taken = time_taken_vec
        .iter()
        .max_by(|a, b| a.partial_cmp(b).unwrap())
        .unwrap();
    let min_time_taken = time_taken_vec
        .iter()
        .min_by(|a, b| a.partial_cmp(b).unwrap())
        .unwrap();
    #[cfg(feature = "debug")]
    let max_abs_score = *prediction_score_vec.iter().max().unwrap();
    #[cfg(feature = "debug")]
    let min_abs_score = *prediction_score_vec.iter().min().unwrap();
    let prediction_score_rms = (prediction_score_vec
        .iter()
        .map(|&x| (x as f64).powi(2))
        .sum::<f64>()
        / prediction_score_vec.len() as f64)
        .sqrt();
    #[cfg(feature = "debug")]
    let prediction_accuracy = calculate_prediction_accuracy(prediction_score_rms);
    println_wasm!(
        "\n{}:\n\n{}",
        "Game PGN".colorize(INFO_MESSAGE_STYLE),
        engine.get_board().get_pgn(),
    );
    println_wasm!(
        "\n{}:\n\n[{}]",
        "Time taken for all moves".colorize(INFO_MESSAGE_STYLE),
        time_taken_vec
            .iter()
            .map(|x| (x * 1000.0).round() / 1000.0)
            .join(", "),
    );
    println_wasm!(
        "\n{}:\n\n[{}]\n",
        "Prediction Scores".colorize(INFO_MESSAGE_STYLE),
        prediction_score_vec
            .iter()
            .map(|&score| score.stringify())
            .join(", "),
    );
    if let GoCommand::Depth(depth) = go_command {
        println_info("Depth Searched", format!("{}", depth));
    } else if let GoCommand::MoveTime(time) = go_command {
        println_info("Time Searched Per Move", time.stringify());
    }
    println_info(
        "Time taken per move",
        format!("{:.3} \u{00B1} {:.3} s", mean, std_err),
    );
    println_info("Coefficient of Variation", format!("{:.3}", std_err / mean));
    println_info(
        "Prediction Score RMS",
        format!("{:.3}", prediction_score_rms),
    );
    #[cfg(feature = "debug")]
    println_info(
        "Prediction Accuracy",
        format!("{:.1} %", prediction_accuracy),
    );
    println_info("Max time taken", format!("{:.3} s", max_time_taken));
    println_info("Min time taken", format!("{:.3} s", min_time_taken));
    println_info("Max time taken by fen", max_time_taken_fen);
    #[cfg(feature = "debug")]
    println_info("Max prediction magnitude", max_abs_score.stringify());
    #[cfg(feature = "debug")]
    println_info("Min prediction magnitude", min_abs_score.stringify());
    engine.set_fen(&stating_fen)?;
    Ok(())
}