rlst 0.10.0

A Rust native linear algebra library.
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

//! A rank1 update of a two-dimensional array.

use crate::dense::layout::convert_1d_nd_from_shape;
use crate::dense::traits::{UnsafeRandom1DAccessByRef, UnsafeRandom1DAccessByValue};
use crate::dense::types::{c32, c64};

use crate::{Array, ChunkedAccess, DataChunk, RlstScalar, Shape, UnsafeRandomAccessByValue};

use super::empty_chunk;

/// Rank-1 Update of a two-dimensional array.
pub struct Rank1Array<
    Item: RlstScalar,
    ArrayImplU: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
    ArrayImplV: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
> {
    u: Array<Item, ArrayImplU, 1>,
    v: Array<Item, ArrayImplV, 1>,
}

impl<
        Item: RlstScalar,
        ArrayImplU: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
        ArrayImplV: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
    > Rank1Array<Item, ArrayImplU, ArrayImplV>
{
    /// Initialize a rank 1 update of the form `arr + uv^T`.
    ///
    /// The argument `v` is only transposed and not conjugate transposed.
    pub fn new(u: Array<Item, ArrayImplU, 1>, v: Array<Item, ArrayImplV, 1>) -> Self {
        Self { u, v }
    }
}

impl<Item: RlstScalar, ArrayImplU, ArrayImplV> ContainerTypeMarker
    for Rank1Array<Item, ArrayImplU, ArrayImplV>
{
    type Type = Item;
}

impl<
        Item: RlstScalar,
        ArrayImplU: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
        ArrayImplV: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
    > UnsafeRandomAccessByValue<2> for Rank1Array<Item, ArrayImplU, ArrayImplV>
{
    type Item = Item;

    #[inline(always)]
    unsafe fn get_value_unchecked(&self, multi_index: [usize; 2]) -> Self::Item {
        if coe::is_same::<Item, f64>() {
            let u_val: f64 = coe::coerce_static(self.u.get_value_unchecked([multi_index[0]]));
            let v_val: f64 = coe::coerce_static(self.v.get_value_unchecked([multi_index[1]]));
            coe::coerce_static(u_val * v_val)
        } else if coe::is_same::<Item, f32>() {
            let u_val: f32 = coe::coerce_static(self.u.get_value_unchecked([multi_index[0]]));
            let v_val: f32 = coe::coerce_static(self.v.get_value_unchecked([multi_index[1]]));
            coe::coerce_static(u_val * v_val)
        } else if coe::is_same::<Item, c32>() {
            let u_val: c32 = coe::coerce_static(self.u.get_value_unchecked([multi_index[0]]));
            let v_val: c32 = coe::coerce_static(self.v.get_value_unchecked([multi_index[1]]));
            coe::coerce_static(c32::new(
                u_val.re().mul_add(v_val.re(), -u_val.im() * v_val.im()),
                u_val.re().mul_add(v_val.im(), u_val.im() * v_val.re()),
            ))
        } else if coe::is_same::<Item, c64>() {
            let u_val: c64 = coe::coerce_static(self.u.get_value_unchecked([multi_index[0]]));
            let v_val: c64 = coe::coerce_static(self.v.get_value_unchecked([multi_index[1]]));
            coe::coerce_static(c64::new(
                u_val.re().mul_add(v_val.re(), -u_val.im() * v_val.im()),
                u_val.re().mul_add(v_val.im(), u_val.im() * v_val.re()),
            ))
        } else {
            panic!("Unknown type");
        }
    }
}

impl<
        Item: RlstScalar,
        ArrayImplU: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
        ArrayImplV: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
    > UnsafeRandom1DAccessByValue for Rank1Array<Item, ArrayImplU, ArrayImplV>
{
    type Item = Item;

    #[inline(always)]
    unsafe fn get_value_1d_unchecked(&self, index: usize) -> Self::Item {
        self.get_value_unchecked(convert_1d_nd_from_shape(index, self.shape()))
    }
}

impl<
        Item: RlstScalar,
        ArrayImplU: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
        ArrayImplV: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
    > Shape<2> for Rank1Array<Item, ArrayImplU, ArrayImplV>
{
    fn shape(&self) -> [usize; 2] {
        [self.u.shape()[0], self.v.shape()[0]]
    }
}

impl<
        Item: RlstScalar,
        ArrayImplU: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
        ArrayImplV: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
        const N: usize,
    > ChunkedAccess<N> for Rank1Array<Item, ArrayImplU, ArrayImplV>
{
    type Item = Item;

    fn get_chunk(&self, chunk_index: usize) -> Option<crate::DataChunk<Self::Item, N>> {
        let nelements = self.u.shape()[0] * self.v.shape()[0];
        if let Some(chunk) = empty_chunk::<N, Item>(chunk_index, nelements) {
            let mut output_data = [<Item as num::Zero>::zero(); N];
            let mut u_data = [<Item as num::Zero>::zero(); N];
            let mut v_data = [<Item as num::Zero>::zero(); N];
            let shape = self.shape();
            for (index, u_elem, v_elem) in
                itertools::izip!(0..chunk.valid_entries, u_data.iter_mut(), v_data.iter_mut(),)
            {
                let multi_index = convert_1d_nd_from_shape(index + chunk.start_index, shape);
                *u_elem = unsafe { self.u.get_value_unchecked([multi_index[0]]) };
                *v_elem = unsafe { self.v.get_value_unchecked([multi_index[1]]) };
            }

            for (out, &u_elem, &v_elem) in itertools::izip!(
                output_data.as_mut_slice(),
                u_data.as_slice(),
                v_data.as_slice()
            ) {
                *out = u_elem * v_elem;
            }

            Some(DataChunk::<Item, N> {
                data: output_data,
                start_index: chunk.start_index,
                valid_entries: chunk.valid_entries,
            })
        } else {
            None
        }
    }
}

impl<
        Item: RlstScalar,
        ArrayImplU: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
        ArrayImplV: UnsafeRandomAccessByValue<1, Item = Item> + Shape<1>,
    > Rank1Array<Item, ArrayImplU, ArrayImplV>
{
    /// Return a new rank1 array that implements `u x v^T`
    pub fn new_array(
        u: Array<Item, ArrayImplU, 1>,
        v: Array<Item, ArrayImplV, 1>,
    ) -> Array<Item, Rank1Array<Item, ArrayImplU, ArrayImplV>, 2> {
        Array::new(Rank1Array::new(u, v))
    }
}

#[cfg(test)]
mod test {

    use approx::assert_relative_eq;
    use rand::SeedableRng;

    use crate::{assert_array_relative_eq, prelude::*};

    #[test]
    fn test_rank1_f32() {
        let eps = 1E-5;
        let mut rng = rand::rngs::StdRng::seed_from_u64(0);

        let m = 50;
        let n = 25;

        let mut u = rlst_dynamic_array1!(f32, [m]);
        let mut v = rlst_dynamic_array1!(f32, [n]);

        let mut expected = rlst_dynamic_array2!(f32, [m, n]);
        let mut actual = rlst_dynamic_array2!(f32, [m, n]);
        let mut actual_chunked = rlst_dynamic_array2!(f32, [m, n]);

        u.fill_from_equally_distributed(&mut rng);
        v.fill_from_equally_distributed(&mut rng);

        for row_index in 0..m {
            for col_index in 0..n {
                expected[[row_index, col_index]] = u[[row_index]] * v[[col_index]]
            }
        }

        let rank1_array = rlst_rank1_array!(u.r(), v.r());

        actual.fill_from(rank1_array.r());
        actual_chunked.fill_from_chunked::<_, 13>(rank1_array.r());

        assert_array_relative_eq!(rank1_array, expected, eps);
        assert_array_relative_eq!(actual, expected, eps);
        assert_array_relative_eq!(actual_chunked, expected, eps);
    }

    #[test]
    fn test_rank1_f64() {
        let eps = 1E-14;
        let mut rng = rand::rngs::StdRng::seed_from_u64(0);

        let m = 50;
        let n = 25;

        let mut u = rlst_dynamic_array1!(f64, [m]);
        let mut v = rlst_dynamic_array1!(f64, [n]);

        let mut expected = rlst_dynamic_array2!(f64, [m, n]);
        let mut actual = rlst_dynamic_array2!(f64, [m, n]);
        let mut actual_chunked = rlst_dynamic_array2!(f64, [m, n]);

        u.fill_from_equally_distributed(&mut rng);
        v.fill_from_equally_distributed(&mut rng);

        for row_index in 0..m {
            for col_index in 0..n {
                expected[[row_index, col_index]] = u[[row_index]] * v[[col_index]]
            }
        }

        let rank1_array = rlst_rank1_array!(u.r(), v.r());

        actual.fill_from(rank1_array.r());
        actual_chunked.fill_from_chunked::<_, 13>(rank1_array.r());

        assert_array_relative_eq!(rank1_array, expected, eps);
        assert_array_relative_eq!(actual, expected, eps);
        assert_array_relative_eq!(actual_chunked, expected, eps);
    }

    #[test]
    fn test_rank1_c32() {
        let eps = 1E-5;
        let mut rng = rand::rngs::StdRng::seed_from_u64(0);

        let m = 50;
        let n = 25;

        let mut u = rlst_dynamic_array1!(c32, [m]);
        let mut v = rlst_dynamic_array1!(c32, [n]);

        let mut expected = rlst_dynamic_array2!(c32, [m, n]);
        let mut actual = rlst_dynamic_array2!(c32, [m, n]);
        let mut actual_chunked = rlst_dynamic_array2!(c32, [m, n]);

        u.fill_from_equally_distributed(&mut rng);
        v.fill_from_equally_distributed(&mut rng);

        for row_index in 0..m {
            for col_index in 0..n {
                expected[[row_index, col_index]] = u[[row_index]] * v[[col_index]]
            }
        }

        let rank1_array = rlst_rank1_array!(u.r(), v.r());

        actual.fill_from(rank1_array.r());
        actual_chunked.fill_from_chunked::<_, 13>(rank1_array.r());

        assert_array_relative_eq!(rank1_array, expected, eps);
        assert_array_relative_eq!(actual, expected, eps);
        assert_array_relative_eq!(actual_chunked, expected, eps);
    }

    #[test]
    fn test_rank1_c64() {
        let eps = 1E-14;
        let mut rng = rand::rngs::StdRng::seed_from_u64(0);

        let m = 50;
        let n = 25;

        let mut u = rlst_dynamic_array1!(c64, [m]);
        let mut v = rlst_dynamic_array1!(c64, [n]);

        let mut expected = rlst_dynamic_array2!(c64, [m, n]);
        let mut actual = rlst_dynamic_array2!(c64, [m, n]);
        let mut actual_chunked = rlst_dynamic_array2!(c64, [m, n]);

        u.fill_from_equally_distributed(&mut rng);
        v.fill_from_equally_distributed(&mut rng);

        for row_index in 0..m {
            for col_index in 0..n {
                expected[[row_index, col_index]] = u[[row_index]] * v[[col_index]]
            }
        }

        let rank1_array = rlst_rank1_array!(u.r(), v.r());

        actual.fill_from(rank1_array.r());
        actual_chunked.fill_from_chunked::<_, 13>(rank1_array.r());

        assert_array_relative_eq!(rank1_array, expected, eps);
        assert_array_relative_eq!(actual, expected, eps);
        assert_array_relative_eq!(actual_chunked, expected, eps);
    }
}