apbase 0.1.3

High-performance automated map-generation pipeline for precision agriculture
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
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237

//! Thin wrappers around the cross-variogram and covariate-screening kernels.

use std::ffi::c_int;

use ndarray::Array1;

use crate::_native::{len_c_int, lib, require_len, to_c_int};
use crate::common::exceptions::Result;

/// Fit a cross-variogram model: `(model_values, rho, n_pairs)`.
#[allow(clippy::too_many_arguments)]
pub(crate) fn run_fit_cross_variogram(
    xa: &[f64],
    ya: &[f64],
    za: &[f64],
    xb: &[f64],
    yb: &[f64],
    zb: &[f64],
    max_colocation_dist: f64,
    n_lags: i64,
    max_pairs: i64,
    max_distance: f64,
    n_threads: i64,
) -> Result<(Array1<f64>, f64, i64)> {
    const CONTEXT: &str = "error fitting cross-variogram";
    require_len(ya.len(), xa.len(), "ya", CONTEXT)?;
    require_len(za.len(), xa.len(), "za", CONTEXT)?;
    require_len(yb.len(), xb.len(), "yb", CONTEXT)?;
    require_len(zb.len(), xb.len(), "zb", CONTEXT)?;
    let na = len_c_int(xa.len(), CONTEXT)?;
    let nb = len_c_int(xb.len(), CONTEXT)?;
    let n_lags = to_c_int(n_lags, CONTEXT)?;
    let max_pairs = to_c_int(max_pairs, CONTEXT)?;
    let thread_count = to_c_int(n_threads, CONTEXT)?;
    let mut model_info = [0.0f64; 5];
    let mut rho = 0.0f64;
    let mut n_pairs: c_int = 0;

    lib::load();
    // SAFETY: side A arrays hold `na` values, side B arrays `nb`, model_info 5.
    unsafe {
        lib::fit_cross_variogram_(
            xa.as_ptr(),
            ya.as_ptr(),
            za.as_ptr(),
            xb.as_ptr(),
            yb.as_ptr(),
            zb.as_ptr(),
            &max_colocation_dist,
            &n_lags,
            &max_pairs,
            &max_distance,
            &thread_count,
            model_info.as_mut_ptr(),
            &mut rho,
            &mut n_pairs,
            &na,
            &nb,
        );
    }
    Ok((Array1::from(model_info.to_vec()), rho, i64::from(n_pairs)))
}

/// Screen one covariate: `(decision, p_value, admissible_fraction, status)`.
#[allow(clippy::too_many_arguments)]
pub(crate) fn run_screen_covariate(
    model_00: &[f64],
    model_uu: &[f64],
    model_0u: &[f64],
    rho: f64,
    n_pairs: i64,
    alpha: f64,
    min_abs_rho: f64,
    max_violation_fraction: f64,
) -> Result<(bool, f64, f64, i32)> {
    const CONTEXT: &str = "error screening covariate";
    require_len(model_00.len(), 5, "model_00", CONTEXT)?;
    require_len(model_uu.len(), 5, "model_uu", CONTEXT)?;
    require_len(model_0u.len(), 5, "model_0u", CONTEXT)?;
    let n_pairs = to_c_int(n_pairs, CONTEXT)?;
    let mut decision: c_int = 0;
    let mut p_value = 0.0f64;
    let mut admissible_fraction = 0.0f64;
    let mut status: c_int = 0;

    lib::load();
    // SAFETY: the three model vectors hold 5 values each; outputs are scalars.
    unsafe {
        lib::screen_covariate_(
            model_00.as_ptr(),
            model_uu.as_ptr(),
            model_0u.as_ptr(),
            &rho,
            &n_pairs,
            &alpha,
            &min_abs_rho,
            &max_violation_fraction,
            &mut decision,
            &mut p_value,
            &mut admissible_fraction,
            &mut status,
        );
    }
    Ok((decision != 0, p_value, admissible_fraction, status))
}

#[cfg(test)]
mod tests {
    //! Checks of the cross-variogram and covariate-screening kernel wrappers.

    use super::*;
    use crate::_native::test_support::{assert_rejected, assert_status, grid_points};
    use crate::common::exceptions::{CROSS_VARIOGRAM_STATUS_ERRORS, ErrorKind};

    const MODEL_SPHERICAL: f64 = 1.0;

    fn fisher_p_value_reference(rho: f64, n_pairs: i64) -> f64 {
        let z_fisher = 0.5 * ((1.0 + rho) / (1.0 - rho)).ln();
        let z_stat = z_fisher * ((n_pairs - 3) as f64).sqrt();
        // SAFETY: erf is a pure libm function.
        let normal_cdf = 0.5 * (1.0 + unsafe { lib::erf(z_stat.abs() / 2f64.sqrt()) });
        2.0 * (1.0 - normal_cdf)
    }

    fn models(cross_partial_sill: f64) -> ([f64; 5], [f64; 5], [f64; 5]) {
        (
            [MODEL_SPHERICAL, 0.0, 1.0, 10.0, 0.0],
            [MODEL_SPHERICAL, 0.0, 1.0, 10.0, 0.0],
            [MODEL_SPHERICAL, 0.0, cross_partial_sill, 10.0, 0.0],
        )
    }

    #[test]
    fn test_high_rho_small_n_is_rejected_by_significance() {
        let (model_00, model_uu, model_0u) = models(0.8);
        let (rho, n_pairs) = (0.8, 6);
        let (decision, p_value, admissible_fraction, status) =
            run_screen_covariate(&model_00, &model_uu, &model_0u, rho, n_pairs, 0.05, 0.3, 0.1).unwrap();
        let reference_p = fisher_p_value_reference(rho, n_pairs);
        assert_eq!(status, 0);
        assert!((p_value - reference_p).abs() <= 1e-9);
        assert!(reference_p > 0.05);
        assert!(!decision);
        assert_eq!(admissible_fraction, 1.0);
    }

    #[test]
    fn test_cauchy_schwarz_violation_is_rejected_even_when_significant() {
        let (model_00, model_uu, model_0u) = models(5.0);
        let (decision, p_value, admissible_fraction, status) =
            run_screen_covariate(&model_00, &model_uu, &model_0u, 0.99, 100, 0.05, 0.3, 0.1).unwrap();
        assert_eq!(status, 0);
        assert!(p_value < 0.05);
        assert!(admissible_fraction < 0.9);
        assert!(!decision);
    }

    #[test]
    fn test_insufficient_sample_reports_status_without_raising() {
        let (model_00, model_uu, model_0u) = models(0.8);
        let (decision, _, _, status) =
            run_screen_covariate(&model_00, &model_uu, &model_0u, 0.9, 3, 0.05, 0.3, 0.1).unwrap();
        assert_eq!(status, 1);
        assert!(!decision);
    }

    #[test]
    fn wrong_model_length_is_rejected_before_the_kernel() {
        let short = [1.0, 0.0, 1.0];
        let model = [MODEL_SPHERICAL, 0.0, 1.0, 10.0, 0.0];
        let error = run_screen_covariate(&short, &model, &model, 0.5, 10, 0.05, 0.3, 0.1).unwrap_err();
        assert_eq!(error.kind(), crate::common::ErrorKind::NativeExecutionError);
    }

    #[test]
    fn fit_cross_variogram_on_perfectly_correlated_colocated_samples() {
        let (x, y, za) = grid_points();
        let zb: Vec<f64> = za.iter().map(|v| 2.0 * v + 1.0).collect();
        let (model, rho, n_pairs) =
            run_fit_cross_variogram(&x, &y, &za, &x, &y, &zb, 0.0, 10, 0, 0.0, 1).unwrap();
        assert_eq!(n_pairs, 36);
        assert!((rho - 1.0).abs() <= 1e-12, "{rho}");
        assert!([1.0, 2.0, 3.0].contains(&model[0]), "{model}");
        assert!(model[3].is_finite() && model[3] > 0.0, "{model}");
    }

    #[test]
    fn fit_cross_variogram_mismatched_lengths_are_rejected_before_the_kernel() {
        let (x, y, z) = grid_points();
        let run =
            |ya: &[f64], zb: &[f64]| run_fit_cross_variogram(&x, ya, &z, &x, &y, zb, 0.0, 10, 0, 0.0, 1);
        assert_rejected(run(&y[1..], &z).unwrap_err(), "ya");
        assert_rejected(run(&y, &z[1..]).unwrap_err(), "zb");
    }

    #[test]
    fn screening_statuses_reach_the_documented_errors() {
        let table = CROSS_VARIOGRAM_STATUS_ERRORS;
        let (model_00, model_uu, model_0u) = models(0.8);
        let status = |m00: &[f64], muu: &[f64], m0u: &[f64], rho: f64, n_pairs: i64| {
            run_screen_covariate(m00, muu, m0u, rho, n_pairs, 0.05, 0.3, 0.1)
                .unwrap()
                .3
        };
        let sample = ErrorKind::CrossVariogramInsufficientSampleError;
        assert_status(status(&model_00, &model_uu, &model_0u, 0.9, 3), 1, table, sample);
        assert_status(
            status(&model_00, &model_uu, &model_0u, f64::NAN, 50),
            1,
            table,
            sample,
        );
        let bad_auto = [9.0, 0.0, 1.0, 10.0, 0.0];
        let auto = ErrorKind::CrossVariogramInvalidAutoModelError;
        assert_status(status(&bad_auto, &model_uu, &model_0u, 0.9, 50), 2, table, auto);
        assert_status(status(&model_00, &bad_auto, &model_0u, 0.9, 50), 2, table, auto);
        let cross = ErrorKind::CrossVariogramInvalidCrossModelError;
        assert_status(
            status(&model_00, &model_uu, &[0.0, 0.0, 0.8, 10.0, 0.0], 0.9, 50),
            3,
            table,
            cross,
        );
        assert_status(
            status(
                &model_00,
                &model_uu,
                &[MODEL_SPHERICAL, 0.0, 0.8, 0.0, 0.0],
                0.9,
                50,
            ),
            3,
            table,
            cross,
        );
    }
}