numrs2 0.3.3

A Rust implementation inspired by NumPy for numerical computing (NumRS2)
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
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314

/// Tests for activation functions
use numrs2::nn::activation::*;
use scirs2_core::ndarray::{array, Array1, Array2};

const EPSILON: f64 = 1e-6;

#[test]
fn test_relu_basic() {
    let x = array![-2.0, -1.0, 0.0, 1.0, 2.0];
    let y = relu(&x.view()).expect("relu failed");
    let expected = array![0.0, 0.0, 0.0, 1.0, 2.0];

    for (&actual, &expected) in y.iter().zip(expected.iter()) {
        let actual: f64 = actual;
        let expected: f64 = expected;
        assert!((actual - expected).abs() < EPSILON);
    }
}

#[test]
fn test_relu_2d() {
    let x = array![[-1.0, 0.0, 1.0], [2.0, -2.0, 3.0]];
    let y = relu_2d(&x.view()).expect("relu_2d failed");
    let expected = array![[0.0, 0.0, 1.0], [2.0, 0.0, 3.0]];

    assert_eq!(y.shape(), expected.shape());
    for (&actual, &expected) in y.iter().zip(expected.iter()) {
        let actual: f64 = actual;
        let expected: f64 = expected;
        assert!((actual - expected).abs() < EPSILON);
    }
}

#[test]
fn test_relu_all_positive() {
    let x = array![1.0, 2.0, 3.0, 4.0, 5.0];
    let y = relu(&x.view()).expect("relu failed");

    // All values should remain unchanged
    for (&actual, &expected) in y.iter().zip(x.iter()) {
        let actual: f64 = actual;
        let expected: f64 = expected;
        assert!((actual - expected).abs() < EPSILON);
    }
}

#[test]
fn test_relu_all_negative() {
    let x = array![-5.0, -4.0, -3.0, -2.0, -1.0];
    let y = relu(&x.view()).expect("relu failed");

    // All values should be zero
    for &val in y.iter() {
        let val: f64 = val;
        assert!(val.abs() < EPSILON);
    }
}

#[test]
fn test_leaky_relu_basic() {
    let x = array![-2.0, -1.0, 0.0, 1.0, 2.0];
    let alpha = 0.01_f64;
    let y = leaky_relu(&x.view(), alpha).expect("leaky_relu failed");

    assert!((y[0] - (-0.02_f64)).abs() < EPSILON); // -2 * 0.01 = -0.02
    assert!((y[1] - (-0.01_f64)).abs() < EPSILON); // -1 * 0.01 = -0.01
    assert!(y[2].abs() < EPSILON); // 0
    assert!((y[3] - 1.0_f64).abs() < EPSILON); // 1
    assert!((y[4] - 2.0_f64).abs() < EPSILON); // 2
}

#[test]
fn test_leaky_relu_negative_alpha_error() {
    let x = array![1.0, 2.0, 3.0];
    let result = leaky_relu(&x.view(), -0.1_f64);

    assert!(result.is_err());
}

#[test]
fn test_sigmoid_basic() {
    let x = array![0.0, 1.0, -1.0, 2.0, -2.0];
    let y = sigmoid(&x.view()).expect("sigmoid failed");

    // sigmoid(0) = 0.5
    assert!((y[0] - 0.5_f64).abs() < EPSILON);

    // sigmoid(x) should be in (0, 1)
    for &val in y.iter() {
        let val: f64 = val;
        assert!(val > 0.0_f64 && val < 1.0_f64);
    }

    // sigmoid(-x) = 1 - sigmoid(x)
    let x_neg = array![-1.0];
    let x_pos = array![1.0];
    let y_neg = sigmoid(&x_neg.view()).expect("sigmoid failed");
    let y_pos = sigmoid(&x_pos.view()).expect("sigmoid failed");

    assert!((y_neg[0] + y_pos[0] - 1.0_f64).abs() < EPSILON);
}

#[test]
fn test_tanh_basic() {
    let x = array![0.0, 1.0, -1.0, 2.0, -2.0];
    let y = tanh(&x.view()).expect("tanh failed");

    // tanh(0) = 0
    let y0: f64 = y[0];
    assert!(y0.abs() < EPSILON);

    // tanh(x) should be in (-1, 1)
    for &val in y.iter() {
        let val: f64 = val;
        assert!(val > -1.0_f64 && val < 1.0_f64);
    }

    // tanh is odd: tanh(-x) = -tanh(x)
    assert!((y[1] + y[2]).abs() < EPSILON);
    assert!((y[3] + y[4]).abs() < EPSILON);
}

#[test]
fn test_gelu_basic() {
    let x = array![0.0, 1.0, -1.0];
    let y = gelu(&x.view()).expect("gelu failed");

    // GELU(0) ≈ 0
    let y0: f64 = y[0];
    assert!(y0.abs() < 0.01_f64);

    // GELU should be smooth and monotonic
    assert!(y[1] > y[0]); // GELU(1) > GELU(0)
    assert!(y[2] < y[0]); // GELU(-1) < GELU(0)
}

#[test]
fn test_swish_basic() {
    let x = array![0.0, 1.0, -1.0, 2.0];
    let y = swish(&x.view()).expect("swish failed");

    // Swish(0) ≈ 0
    let y0: f64 = y[0];
    assert!(y0.abs() < 0.01_f64);

    // For positive x, swish(x) should be close to x for large x
    // swish(2) should be close to 2 (≈1.76, so tolerance needs to be ~0.3)
    assert!((y[3] - 2.0_f64).abs() < 0.3_f64);
}

#[test]
fn test_mish_basic() {
    let x = array![0.0, 1.0, -1.0, 2.0];
    let y = mish(&x.view()).expect("mish failed");

    // Mish(0) ≈ 0
    let y0: f64 = y[0];
    assert!(y0.abs() < 0.01_f64);

    // Mish should be smooth
    assert!(y[1] > y[0]);
    assert!(y[2] < y[0]);
}

#[test]
fn test_elu_basic() {
    let x = array![-2.0, -1.0, 0.0, 1.0, 2.0];
    let alpha = 1.0_f64;
    let y = elu(&x.view(), alpha).expect("elu failed");

    // For x >= 0, ELU(x) = x
    let y2: f64 = y[2];
    assert!(y2.abs() < EPSILON); // ELU(0) = 0
    assert!((y[3] - 1.0_f64).abs() < EPSILON); // ELU(1) = 1
    assert!((y[4] - 2.0_f64).abs() < EPSILON); // ELU(2) = 2

    // For x < 0, ELU(x) = alpha * (exp(x) - 1)
    assert!(y[0] < 0.0_f64); // ELU(-2) < 0
    assert!(y[1] < 0.0_f64); // ELU(-1) < 0
    assert!(y[0] < y[1]); // ELU is monotonic
}

#[test]
fn test_selu_basic() {
    let x = array![-2.0, -1.0, 0.0, 1.0, 2.0];
    let y = selu(&x.view()).expect("selu failed");

    // SELU(0) ≈ 0
    let y2_selu: f64 = y[2];
    assert!(y2_selu.abs() < EPSILON);

    // For x > 0, SELU(x) = lambda * x
    const LAMBDA: f64 = 1.0507009873554804934193349852946;
    assert!((y[3] - LAMBDA * 1.0_f64).abs() < EPSILON);
    assert!((y[4] - LAMBDA * 2.0_f64).abs() < EPSILON);
}

#[test]
fn test_softmax_basic() {
    let x = array![1.0, 2.0, 3.0];
    let y = softmax(&x.view()).expect("softmax failed");

    // Sum should be 1
    let sum: f64 = y.iter().sum();
    assert!((sum - 1.0_f64).abs() < EPSILON);

    // All values should be positive
    for &val in y.iter() {
        let val: f64 = val;
        assert!(val > 0.0_f64);
    }

    // Larger inputs should have larger outputs
    assert!(y[0] < y[1]);
    assert!(y[1] < y[2]);
}

#[test]
fn test_softmax_2d() {
    let x = array![[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]];
    let y = softmax_2d(&x.view(), 1).expect("softmax_2d failed");

    // Each row should sum to 1
    for i in 0..y.nrows() {
        let row_sum: f64 = y.row(i).iter().sum();
        assert!((row_sum - 1.0_f64).abs() < EPSILON);
    }
}

#[test]
fn test_softmax_numerical_stability() {
    // Test with large values (should not overflow)
    let x = array![1000.0, 1001.0, 1002.0];
    let y = softmax(&x.view()).expect("softmax failed");

    // Should not contain NaN or infinity
    for &val in y.iter() {
        let val: f64 = val;
        assert!(val.is_finite());
    }

    // Sum should still be 1
    let sum: f64 = y.iter().sum();
    assert!((sum - 1.0_f64).abs() < EPSILON);
}

#[test]
fn test_log_softmax_basic() {
    let x = array![1.0, 2.0, 3.0];
    let y = log_softmax(&x.view()).expect("log_softmax failed");

    // All values should be negative (since log(p) where 0 < p < 1)
    for &val in y.iter() {
        let val: f64 = val;
        assert!(val < 0.0_f64);
    }

    // exp(log_softmax) should give softmax
    let softmax_from_log: Array1<f64> = y.mapv(|v| v.exp());
    let softmax_direct = softmax(&x.view()).expect("softmax failed");

    for (&a, &b) in softmax_from_log.iter().zip(softmax_direct.iter()) {
        let a: f64 = a;
        let b: f64 = b;
        assert!((a - b).abs() < EPSILON);
    }
}

#[test]
fn test_softplus_basic() {
    let x = array![0.0, 1.0, -1.0, 10.0];
    let y = softplus(&x.view()).expect("softplus failed");

    // softplus(0) = ln(2)
    assert!((y[0] - std::f64::consts::LN_2).abs() < 1e-5_f64);

    // softplus should always be positive
    for &val in y.iter() {
        let val: f64 = val;
        assert!(val > 0.0_f64);
    }

    // For large positive x, softplus(x) ≈ x
    assert!((y[3] - 10.0_f64).abs() < 0.01_f64);
}

#[test]
fn test_activation_empty_array() {
    let x: Array1<f64> = Array1::zeros(0);

    // Should handle empty arrays gracefully
    let y = relu(&x.view()).expect("relu failed");
    assert_eq!(y.len(), 0);
}

#[test]
fn test_activation_large_array() {
    let x = Array1::linspace(-100.0, 100.0, 10000);
    let y = relu(&x.view()).expect("relu failed");

    // Should handle large arrays without issues
    assert_eq!(y.len(), 10000);
}

#[test]
fn test_activation_f32() {
    // Test f32 versions
    let x = array![-1.0f32, 0.0, 1.0];
    let y = relu(&x.view()).expect("relu failed");

    assert!((y[0] - 0.0_f32).abs() < 1e-6_f32);
    assert!((y[1] - 0.0_f32).abs() < 1e-6_f32);
    assert!((y[2] - 1.0_f32).abs() < 1e-6_f32);
}