kryst 3.2.1

Krylov subspace and preconditioned iterative solvers for dense and sparse linear systems, with shared and distributed memory parallelism.
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

#[cfg(feature = "metrics")]
mod enabled {
    use std::sync::atomic::{AtomicU64, Ordering};
    use std::time::{Duration, Instant};

    #[derive(Debug)]
    pub struct Counters {
        pub total_nanos: AtomicU64,
        pub count: AtomicU64,
        pub last_nanos: AtomicU64,
    }

    impl Counters {
        pub const fn new() -> Self {
            Self {
                total_nanos: AtomicU64::new(0),
                count: AtomicU64::new(0),
                last_nanos: AtomicU64::new(0),
            }
        }

        #[inline]
        pub fn snapshot(&self) -> (u64, u64, u64) {
            (
                self.total_nanos.load(Ordering::Relaxed),
                self.count.load(Ordering::Relaxed),
                self.last_nanos.load(Ordering::Relaxed),
            )
        }
    }

    pub struct SolveTimer<'a> {
        start: Instant,
        ctrs: &'a Counters,
    }

    impl<'a> SolveTimer<'a> {
        #[inline]
        pub fn start(ctrs: &'a Counters) -> Self {
            Self {
                start: Instant::now(),
                ctrs,
            }
        }

        #[inline]
        pub fn elapsed(&self) -> Duration {
            self.start.elapsed()
        }
    }

    impl<'a> Drop for SolveTimer<'a> {
        #[inline]
        fn drop(&mut self) {
            let dt = self.start.elapsed().as_nanos() as u64;
            self.ctrs.total_nanos.fetch_add(dt, Ordering::Relaxed);
            self.ctrs.count.fetch_add(1, Ordering::Relaxed);
            self.ctrs.last_nanos.store(dt, Ordering::Relaxed);
        }
    }
}

#[cfg(feature = "metrics")]
pub use enabled::*;

#[cfg(not(feature = "metrics"))]
mod disabled {
    use std::time::Duration;

    #[derive(Debug)]
    pub struct Counters;
    impl Default for Counters {
        fn default() -> Self {
            Self::new()
        }
    }

    impl Counters {
        pub const fn new() -> Self {
            Self
        }
        pub fn snapshot(&self) -> (u64, u64, u64) {
            (0, 0, 0)
        }
    }

    pub struct SolveTimer<'a> {
        _p: core::marker::PhantomData<&'a ()>,
    }
    impl<'a> SolveTimer<'a> {
        #[inline]
        pub fn start(_: &'a Counters) -> Self {
            Self {
                _p: core::marker::PhantomData,
            }
        }
        #[inline]
        pub fn elapsed(&self) -> Duration {
            Duration::ZERO
        }
    }
    impl<'a> Drop for SolveTimer<'a> {
        fn drop(&mut self) {}
    }
}

#[cfg(not(feature = "metrics"))]
pub use disabled::*;

/// Estimate the error of a preconditioner using random Rademacher probes.
///
/// The supplied closure should compute `M^{-1} A z` for an input vector `z`.
/// The function returns the maximum sampled value of `|| (I - M^{-1}A) z ||_2 / ||z||_2`.
pub fn proxy_operator_error<M: Fn(&[f64], &mut [f64])>(
    apply_m_inv_a: M,
    n: usize,
    samples: usize,
    seed: u64,
) -> f64 {
    let mut rng = oorandom::Rand64::new(seed.into());
    let mut z = vec![0.0f64; n];
    let mut w = vec![0.0f64; n];
    let mut worst = 0.0f64;
    for _ in 0..samples {
        for i in 0..n {
            z[i] = if rng.rand_u64() & 1 == 1 { 1.0 } else { -1.0 };
        }
        apply_m_inv_a(&z, &mut w);
        for i in 0..n {
            w[i] = z[i] - w[i];
        }
        let r = (w.iter().map(|x| x * x).sum::<f64>()).sqrt() / (n as f64).sqrt();
        worst = worst.max(r);
    }
    worst
}