semiflow 0.15.0-beta

no_std Rust PDE solver: evolution equations via Chernoff approximation of operator semigroups (heat, diffusion, Schrodinger)
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

// Tests for point_eval.rs — moved from point_eval.rs (batch H6).
use super::*;
use crate::{
    diffusion::DiffusionChernoff, grid::Grid1D, grid_fn::GridFn1D, shift1d::ShiftChernoff1D,
};

fn make_1d_grid() -> Grid1D<f64> {
    Grid1D::new(-5.0_f64, 5.0, 32).unwrap()
}

fn make_gaussian_1d(grid: Grid1D<f64>) -> GridFn1D<f64> {
    let values: alloc::vec::Vec<f64> = (0..grid.n)
        .map(|i| {
            let x = grid.x_at(i);
            (-x * x).exp()
        })
        .collect();
    GridFn1D { values, grid }
}

// Backend A: eval_at returns without error for n=1.
#[test]
fn diffusion_eval_at_n1_ok() {
    let grid = make_1d_grid();
    let kernel = DiffusionChernoff::new(
        |_: f64| 1.0_f64,
        |_: f64| 0.0_f64,
        |_: f64| 0.0_f64,
        1.0_f64,
        grid,
    );
    let src = make_gaussian_1d(grid);
    let val = kernel.eval_at(0.01_f64, &src, &[0.0_f64], 1);
    assert!(val.is_ok(), "Backend A: eval_at n=1 must succeed");
}

// Backend B: eval_at returns without error for n=1.
#[test]
fn shift1d_eval_at_n1_ok() {
    let grid = make_1d_grid();
    // ShiftChernoff1D::new(a, b, c, c_norm_bound, grid) — 5 args.
    let kernel = ShiftChernoff1D::new(
        |_: f64| 1.0_f64,
        |_: f64| 0.0_f64,
        |_: f64| 0.0_f64,
        0.0_f64,
        grid,
    );
    let src = make_gaussian_1d(grid);
    let val = kernel.eval_at(0.01_f64, &src, &[0.0_f64], 1);
    assert!(val.is_ok(), "Backend B: eval_at n=1 must succeed");
}

// Guard: n_steps=0 returns DomainViolation.
#[test]
fn eval_at_zero_steps_errors() {
    let grid = make_1d_grid();
    let kernel = DiffusionChernoff::new(
        |_: f64| 1.0_f64,
        |_: f64| 0.0_f64,
        |_: f64| 0.0_f64,
        1.0_f64,
        grid,
    );
    let src = make_gaussian_1d(grid);
    let val = kernel.eval_at(0.01_f64, &src, &[0.0_f64], 0);
    assert!(
        matches!(val, Err(SemiflowError::DomainViolation { .. })),
        "n_steps=0 must return DomainViolation"
    );
}

// Guard: empty x slice returns DomainViolation for 1-D backend.
#[test]
fn eval_at_empty_x_errors() {
    let grid = make_1d_grid();
    let kernel = DiffusionChernoff::new(
        |_: f64| 1.0_f64,
        |_: f64| 0.0_f64,
        |_: f64| 0.0_f64,
        1.0_f64,
        grid,
    );
    let src = make_gaussian_1d(grid);
    let val = kernel.eval_at(0.01_f64, &src, &[], 1);
    assert!(
        matches!(val, Err(SemiflowError::DomainViolation { .. })),
        "empty x must return DomainViolation"
    );
}

// G_BATCH_POINTEVAL: eval_at_batch is 0-ULP identical to mapping eval_at.
#[test]
fn g_batch_pointeval_byte_identical_to_scalar() {
    let grid = make_1d_grid();
    let kernel = DiffusionChernoff::new(
        |_: f64| 1.0_f64,
        |_: f64| 0.0_f64,
        |_: f64| 0.0_f64,
        1.0_f64,
        grid,
    );
    let src = make_gaussian_1d(grid);
    let tau = 0.01_f64;
    let n_steps = 4_u32;
    let xs: &[&[f64]] = &[&[-1.0], &[0.0], &[1.0]];
    let batch = kernel.eval_at_batch(tau, &src, xs, n_steps).unwrap();
    for (i, x) in xs.iter().enumerate() {
        let scalar = kernel.eval_at(tau, &src, x, n_steps).unwrap();
        assert_eq!(
            batch[i].to_bits(),
            scalar.to_bits(),
            "G_BATCH_POINTEVAL: index {i} batch {:.16e} != scalar {:.16e}",
            batch[i],
            scalar,
        );
    }
}

// sample_gridfn2d: centre node returns values[centre_idx] exactly.
#[test]
fn sample_gridfn2d_centre_node() {
    use crate::{grid2d::Grid2D, grid_fn2d::GridFn2D};
    let gx = Grid1D::new(-1.0_f64, 1.0, 4).unwrap();
    let gy = Grid1D::new(-1.0_f64, 1.0, 4).unwrap();
    let grid = Grid2D::new(gx, gy);
    let values: alloc::vec::Vec<f64> = (0..16).map(f64::from).collect();
    let state = GridFn2D { values, grid };
    // Centre of grid: x_at(1), y_at(1) → index 1*4+1 = 5 (value 5.0)
    let cx = gx.x_at(1);
    let cy = gy.x_at(1);
    let sampled = sample_gridfn2d(&state, cx, cy);
    assert_eq!(
        sampled.to_bits(),
        5.0_f64.to_bits(),
        "sample_gridfn2d at node centre must return exact node value"
    );
}