spin-sim 0.2.0

Ising model Monte Carlo: Metropolis, Gibbs, Wolff, Swendsen-Wang, parallel tempering, Houdayer ICM
Documentation 
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

/// Streaming autocorrelation accumulator using a ring buffer.
///
/// Computes the normalized autocorrelation function Γ(δ) for a per-temperature
/// time series without storing the full history. Memory is O(max_lag × n_temps).
pub struct AutocorrAccum {
    max_lag: usize,
    ring_len: usize,
    n_temps: usize,
    /// Ring buffer of recent values, shape [n_temps][max_lag + 1].
    ring: Vec<Vec<f32>>,
    /// Running sum of o, shape [n_temps].
    sum_o: Vec<f64>,
    /// Running sum of o², shape [n_temps].
    sum_o2: Vec<f64>,
    /// Running sum of o(t)·o(t−δ), shape [n_temps][max_lag+1].
    sum_prod: Vec<Vec<f64>>,
    /// Total number of values pushed so far.
    n_recorded: usize,
    /// Current position in the ring buffer.
    ring_pos: usize,
}

impl AutocorrAccum {
    pub fn new(max_lag: usize, n_temps: usize) -> Self {
        let ring_len = max_lag + 1;
        Self {
            max_lag,
            ring_len,
            n_temps,
            ring: (0..n_temps).map(|_| vec![0.0f32; ring_len]).collect(),
            sum_o: vec![0.0; n_temps],
            sum_o2: vec![0.0; n_temps],
            sum_prod: (0..n_temps).map(|_| vec![0.0; max_lag + 1]).collect(),
            n_recorded: 0,
            ring_pos: 0,
        }
    }

    #[allow(clippy::needless_range_loop)]
    pub fn push(&mut self, values: &[f64]) {
        let pos = self.ring_pos;
        let ml = self.max_lag;
        let ring_len = self.ring_len;
        for t in 0..self.n_temps {
            let o = values[t] as f32;
            self.ring[t][pos] = o;
            self.sum_o[t] += o as f64;
            self.sum_o2[t] += (o as f64) * (o as f64);

            let n_back = self.n_recorded.min(ml);
            for delta in 0..=n_back {
                let idx = if pos >= delta {
                    pos - delta
                } else {
                    pos + ring_len - delta
                } % ring_len;
                self.sum_prod[t][delta] += o as f64 * self.ring[t][idx] as f64;
            }
        }
        self.n_recorded += 1;
        self.ring_pos = (pos + 1) % ring_len;
    }

    pub fn finish(&self) -> Vec<Vec<f64>> {
        let m = self.n_recorded as f64;
        (0..self.n_temps)
            .map(|t| {
                let mean = self.sum_o[t] / m;
                let var = self.sum_o2[t] / m - mean * mean;
                (0..=self.max_lag)
                    .map(|delta| {
                        let count = self.n_recorded.saturating_sub(delta) as f64;
                        if count <= 0.0 || var <= 0.0 {
                            return if delta == 0 { 1.0 } else { 0.0 };
                        }
                        (self.sum_prod[t][delta] / count - mean * mean) / var
                    })
                    .collect()
            })
            .collect()
    }
}

pub fn sokal_tau(gamma: &[f64]) -> f64 {
    let mut tau = 0.5;
    for (w, &g) in gamma.iter().enumerate().skip(1) {
        tau += g;
        if w as f64 >= 5.0 * tau {
            return tau;
        }
    }
    tau
}

#[cfg(test)]
mod tests {
    use super::AutocorrAccum;

    fn brute_force_gamma(series: &[f64], max_lag: usize) -> Vec<f64> {
        let n = series.len() as f64;
        let mean = series.iter().sum::<f64>() / n;
        let var = series.iter().map(|x| x * x).sum::<f64>() / n - mean * mean;
        (0..=max_lag)
            .map(|delta| {
                let count = series.len().saturating_sub(delta) as f64;
                if count <= 0.0 || var <= 0.0 {
                    return if delta == 0 { 1.0 } else { 0.0 };
                }
                let prod = (delta..series.len())
                    .map(|i| series[i] * series[i - delta])
                    .sum::<f64>();
                (prod / count - mean * mean) / var
            })
            .collect()
    }

    #[test]
    fn max_lag_wraparound_uses_distinct_history_slot() {
        let max_lag = 4;
        let series = [1.0, 2.0, 3.0, 4.0, 5.0];
        let mut acc = AutocorrAccum::new(max_lag, 1);
        for &value in &series {
            acc.push(&[value]);
        }

        let gamma = acc.finish();
        let expected = brute_force_gamma(&series, max_lag);

        for (got, want) in gamma[0].iter().zip(expected.iter()) {
            assert!((got - want).abs() < 1e-9, "got {got}, want {want}");
        }
    }
}