deimos_numerics 0.17.0

Numerical methods and control systems analysis
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
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167

use super::compensated::{CompensatedField, CompensatedSum};
use core::fmt::Debug;
use faer::sparse::{SparseColMatRef, SparseRowMatRef};
use faer::{Index, Unbind};
use faer_traits::ComplexField;
use faer_traits::Conjugate;
use faer_traits::math_utils::zero;
use num_traits::Float;

/// Sparse matrix-vector product interface used by iterative solvers.
pub trait SparseMatVec<T: ComplexField + Copy>: Copy + Debug {
    /// Number of rows.
    fn nrows(&self) -> usize;

    /// Number of columns.
    fn ncols(&self) -> usize;

    /// Computes `out = self * rhs`.
    fn apply(&self, out: &mut [T], rhs: &[T]);

    /// Computes `out = self * rhs` with compensated accumulation.
    fn apply_compensated(&self, out: &mut [T], rhs: &[T]);
}

impl<T, I, ViewT> SparseMatVec<T> for SparseRowMatRef<'_, I, ViewT>
where
    T: CompensatedField,
    T::Real: Float,
    I: Index,
    ViewT: Conjugate<Canonical = T>,
{
    #[inline]
    fn nrows(&self) -> usize {
        self.symbolic().nrows().unbound()
    }

    #[inline]
    fn ncols(&self) -> usize {
        self.symbolic().ncols().unbound()
    }

    #[inline]
    fn apply(&self, out: &mut [T], rhs: &[T]) {
        let matrix = self.canonical();
        let nrows = matrix.nrows().unbound();
        let ncols = matrix.ncols().unbound();

        assert_eq!(out.len(), nrows);
        assert_eq!(rhs.len(), ncols);

        let row_ptr = matrix.row_ptr();
        let col_idx = matrix.col_idx();
        let values = matrix.val();

        out.fill(zero::<T>());
        for row in 0..nrows {
            let start = row_ptr[row].zx();
            let end = row_ptr[row + 1].zx();
            let mut sum = zero::<T>();

            for idx in start..end {
                sum += values[idx] * rhs[col_idx[idx].zx()];
            }

            out[row] = sum;
        }
    }

    #[inline]
    fn apply_compensated(&self, out: &mut [T], rhs: &[T]) {
        let matrix = self.canonical();
        let nrows = matrix.nrows().unbound();
        let ncols = matrix.ncols().unbound();

        assert_eq!(out.len(), nrows);
        assert_eq!(rhs.len(), ncols);

        let row_ptr = matrix.row_ptr();
        let col_idx = matrix.col_idx();
        let values = matrix.val();

        out.fill(zero::<T>());
        for row in 0..nrows {
            let start = row_ptr[row].zx();
            let end = row_ptr[row + 1].zx();
            let mut acc = CompensatedSum::<T>::default();

            for idx in start..end {
                acc.add(values[idx] * rhs[col_idx[idx].zx()]);
            }

            out[row] = acc.finish();
        }
    }
}

impl<T, I, ViewT> SparseMatVec<T> for SparseColMatRef<'_, I, ViewT>
where
    T: CompensatedField,
    T::Real: Float,
    I: Index,
    ViewT: Conjugate<Canonical = T>,
{
    #[inline]
    fn nrows(&self) -> usize {
        self.symbolic().nrows().unbound()
    }

    #[inline]
    fn ncols(&self) -> usize {
        self.symbolic().ncols().unbound()
    }

    #[inline]
    fn apply(&self, out: &mut [T], rhs: &[T]) {
        let matrix = self.canonical();
        let nrows = matrix.nrows().unbound();
        let ncols = matrix.ncols().unbound();

        assert_eq!(out.len(), nrows);
        assert_eq!(rhs.len(), ncols);

        let col_ptr = matrix.col_ptr();
        let row_idx = matrix.row_idx();
        let values = matrix.val();

        out.fill(zero::<T>());
        for col in 0..ncols {
            let rhs_value = rhs[col];
            let start = col_ptr[col].zx();
            let end = col_ptr[col + 1].zx();

            for idx in start..end {
                out[row_idx[idx].zx()] += values[idx] * rhs_value;
            }
        }
    }

    #[inline]
    fn apply_compensated(&self, out: &mut [T], rhs: &[T]) {
        let matrix = self.canonical();
        let nrows = matrix.nrows().unbound();
        let ncols = matrix.ncols().unbound();

        assert_eq!(out.len(), nrows);
        assert_eq!(rhs.len(), ncols);

        let col_ptr = matrix.col_ptr();
        let row_idx = matrix.row_idx();
        let values = matrix.val();
        let mut acc = vec![CompensatedSum::<T>::default(); nrows];

        for col in 0..ncols {
            let rhs_value = rhs[col];
            let start = col_ptr[col].zx();
            let end = col_ptr[col + 1].zx();

            for idx in start..end {
                acc[row_idx[idx].zx()].add(values[idx] * rhs_value);
            }
        }

        for (dst, acc) in out.iter_mut().zip(acc) {
            *dst = acc.finish();
        }
    }
}