oxiblas-sparse 0.2.2

Sparse matrix support for OxiBLAS
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
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
//! Coordinate (COO) format for sparse matrices.
//!
//! COO stores matrix data as triplets (row, col, value).
//! This format is ideal for:
//! - Matrix construction/assembly
//! - Adding entries incrementally
//! - Converting to CSR/CSC formats
//!
//! For an m×n matrix with nnz non-zeros:
//! - `row_indices` has length nnz
//! - `col_indices` has length nnz
//! - `values` has length nnz

use oxiblas_core::scalar::{Field, Real, Scalar};

/// Error type for COO matrix operations.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum CooError {
    /// Mismatched array lengths.
    LengthMismatch {
        /// Number of rows.
        rows_len: usize,
        /// Number of columns.
        cols_len: usize,
        /// Number of values.
        values_len: usize,
    },
    /// Row index out of bounds.
    InvalidRowIndex {
        /// The invalid index.
        index: usize,
        /// Number of rows.
        nrows: usize,
    },
    /// Column index out of bounds.
    InvalidColumnIndex {
        /// The invalid index.
        index: usize,
        /// Number of columns.
        ncols: usize,
    },
}

impl core::fmt::Display for CooError {
    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
        match self {
            Self::LengthMismatch {
                rows_len,
                cols_len,
                values_len,
            } => {
                write!(
                    f,
                    "Length mismatch: rows={rows_len}, cols={cols_len}, values={values_len}"
                )
            }
            Self::InvalidRowIndex { index, nrows } => {
                write!(f, "Row index {index} out of bounds for {nrows} rows")
            }
            Self::InvalidColumnIndex { index, ncols } => {
                write!(f, "Column index {index} out of bounds for {ncols} columns")
            }
        }
    }
}

impl std::error::Error for CooError {}

/// Coordinate format sparse matrix.
///
/// Stores entries as triplets (row, col, value).
/// May contain duplicate entries - these are summed during conversion.
#[derive(Debug, Clone)]
pub struct CooMatrix<T: Scalar> {
    /// Number of rows.
    nrows: usize,
    /// Number of columns.
    ncols: usize,
    /// Row indices.
    row_indices: Vec<usize>,
    /// Column indices.
    col_indices: Vec<usize>,
    /// Values.
    values: Vec<T>,
}

impl<T: Scalar + Clone> CooMatrix<T> {
    /// Creates a new COO matrix from raw components.
    ///
    /// # Arguments
    ///
    /// * `nrows` - Number of rows
    /// * `ncols` - Number of columns
    /// * `row_indices` - Row index for each entry
    /// * `col_indices` - Column index for each entry
    /// * `values` - Value for each entry
    ///
    /// # Errors
    ///
    /// Returns an error if the input is invalid.
    pub fn new(
        nrows: usize,
        ncols: usize,
        row_indices: Vec<usize>,
        col_indices: Vec<usize>,
        values: Vec<T>,
    ) -> Result<Self, CooError> {
        // Validate lengths match
        let len = values.len();
        if row_indices.len() != len || col_indices.len() != len {
            return Err(CooError::LengthMismatch {
                rows_len: row_indices.len(),
                cols_len: col_indices.len(),
                values_len: len,
            });
        }

        // Validate indices
        for &row in &row_indices {
            if row >= nrows {
                return Err(CooError::InvalidRowIndex { index: row, nrows });
            }
        }

        for &col in &col_indices {
            if col >= ncols {
                return Err(CooError::InvalidColumnIndex { index: col, ncols });
            }
        }

        Ok(Self {
            nrows,
            ncols,
            row_indices,
            col_indices,
            values,
        })
    }

    /// Creates a COO matrix without validation.
    ///
    /// # Safety
    ///
    /// The caller must ensure:
    /// - All arrays have the same length
    /// - All row indices are < nrows
    /// - All column indices are < ncols
    #[inline]
    pub unsafe fn new_unchecked(
        nrows: usize,
        ncols: usize,
        row_indices: Vec<usize>,
        col_indices: Vec<usize>,
        values: Vec<T>,
    ) -> Self {
        Self {
            nrows,
            ncols,
            row_indices,
            col_indices,
            values,
        }
    }

    /// Creates an empty COO matrix.
    pub fn new_empty(nrows: usize, ncols: usize) -> Self {
        Self {
            nrows,
            ncols,
            row_indices: Vec::new(),
            col_indices: Vec::new(),
            values: Vec::new(),
        }
    }

    /// Creates an empty COO matrix with pre-allocated capacity.
    pub fn with_capacity(nrows: usize, ncols: usize, capacity: usize) -> Self {
        Self {
            nrows,
            ncols,
            row_indices: Vec::with_capacity(capacity),
            col_indices: Vec::with_capacity(capacity),
            values: Vec::with_capacity(capacity),
        }
    }

    /// Returns the number of rows.
    #[inline]
    pub fn nrows(&self) -> usize {
        self.nrows
    }

    /// Returns the number of columns.
    #[inline]
    pub fn ncols(&self) -> usize {
        self.ncols
    }

    /// Returns the shape (nrows, ncols).
    #[inline]
    pub fn shape(&self) -> (usize, usize) {
        (self.nrows, self.ncols)
    }

    /// Returns the number of stored entries (may include duplicates).
    #[inline]
    pub fn len(&self) -> usize {
        self.values.len()
    }

    /// Returns true if the matrix has no entries.
    #[inline]
    pub fn is_empty(&self) -> bool {
        self.values.is_empty()
    }

    /// Returns a reference to the row indices.
    #[inline]
    pub fn row_indices(&self) -> &[usize] {
        &self.row_indices
    }

    /// Returns a reference to the column indices.
    #[inline]
    pub fn col_indices(&self) -> &[usize] {
        &self.col_indices
    }

    /// Returns a reference to the values.
    #[inline]
    pub fn values(&self) -> &[T] {
        &self.values
    }

    /// Returns a mutable reference to the values.
    #[inline]
    pub fn values_mut(&mut self) -> &mut [T] {
        &mut self.values
    }

    /// Adds an entry to the matrix.
    ///
    /// # Panics
    ///
    /// Panics if row >= nrows or col >= ncols.
    pub fn push(&mut self, row: usize, col: usize, value: T) {
        assert!(row < self.nrows, "Row index {row} out of bounds");
        assert!(col < self.ncols, "Column index {col} out of bounds");

        self.row_indices.push(row);
        self.col_indices.push(col);
        self.values.push(value);
    }

    /// Adds an entry without bounds checking.
    ///
    /// # Safety
    ///
    /// row must be < nrows and col must be < ncols.
    #[inline]
    pub unsafe fn push_unchecked(&mut self, row: usize, col: usize, value: T) {
        self.row_indices.push(row);
        self.col_indices.push(col);
        self.values.push(value);
    }

    /// Returns an iterator over entries as (row, col, value).
    pub fn iter(&self) -> impl Iterator<Item = (usize, usize, &T)> {
        self.row_indices
            .iter()
            .zip(self.col_indices.iter())
            .zip(self.values.iter())
            .map(|((&row, &col), val)| (row, col, val))
    }

    /// Returns a mutable iterator over entries.
    pub fn iter_mut(&mut self) -> impl Iterator<Item = (usize, usize, &mut T)> {
        self.row_indices
            .iter()
            .zip(self.col_indices.iter())
            .zip(self.values.iter_mut())
            .map(|((&row, &col), val)| (row, col, val))
    }

    /// Converts to CSR format, summing duplicate entries.
    pub fn to_csr(&self) -> crate::csr::CsrMatrix<T>
    where
        T: Scalar<Real = T> + Field + Real,
    {
        crate::convert::coo_to_csr(self)
    }

    /// Converts to CSC format, summing duplicate entries.
    pub fn to_csc(&self) -> crate::csc::CscMatrix<T>
    where
        T: Scalar<Real = T> + Field + Real,
    {
        crate::convert::coo_to_csc(self)
    }

    /// Converts to dense matrix.
    pub fn to_dense(&self) -> oxiblas_matrix::Mat<T>
    where
        T: Field + bytemuck::Zeroable,
    {
        let mut dense: oxiblas_matrix::Mat<T> = oxiblas_matrix::Mat::zeros(self.nrows, self.ncols);

        for (row, col, val) in self.iter() {
            let current = dense[(row, col)].clone();
            dense[(row, col)] = current + val.clone();
        }

        dense
    }

    /// Creates a COO matrix from a dense matrix.
    pub fn from_dense(dense: &oxiblas_matrix::MatRef<'_, T>) -> Self
    where
        T: Field,
    {
        let (nrows, ncols) = dense.shape();
        let mut row_indices = Vec::new();
        let mut col_indices = Vec::new();
        let mut values = Vec::new();

        let eps = <T as Scalar>::epsilon();

        for i in 0..nrows {
            for j in 0..ncols {
                let val = dense[(i, j)].clone();
                if Scalar::abs(val.clone()) > eps {
                    row_indices.push(i);
                    col_indices.push(j);
                    values.push(val);
                }
            }
        }

        Self {
            nrows,
            ncols,
            row_indices,
            col_indices,
            values,
        }
    }

    /// Removes duplicate entries by summing them.
    pub fn sum_duplicates(&mut self)
    where
        T: Field,
    {
        if self.is_empty() {
            return;
        }

        // Sort by (row, col)
        let mut indices: Vec<usize> = (0..self.len()).collect();
        indices.sort_by_key(|&i| (self.row_indices[i], self.col_indices[i]));

        let mut new_rows = Vec::with_capacity(self.len());
        let mut new_cols = Vec::with_capacity(self.len());
        let mut new_vals = Vec::with_capacity(self.len());

        let mut prev_row = self.row_indices[indices[0]];
        let mut prev_col = self.col_indices[indices[0]];
        let mut acc = self.values[indices[0]].clone();

        for &idx in &indices[1..] {
            let row = self.row_indices[idx];
            let col = self.col_indices[idx];

            if row == prev_row && col == prev_col {
                acc = acc + self.values[idx].clone();
            } else {
                if Scalar::abs(acc.clone()) > T::epsilon() {
                    new_rows.push(prev_row);
                    new_cols.push(prev_col);
                    new_vals.push(acc);
                }
                prev_row = row;
                prev_col = col;
                acc = self.values[idx].clone();
            }
        }

        // Push last entry
        if Scalar::abs(acc.clone()) > T::epsilon() {
            new_rows.push(prev_row);
            new_cols.push(prev_col);
            new_vals.push(acc);
        }

        self.row_indices = new_rows;
        self.col_indices = new_cols;
        self.values = new_vals;
    }

    /// Scales all values by a scalar.
    pub fn scale(&mut self, alpha: T) {
        for val in &mut self.values {
            *val = val.clone() * alpha.clone();
        }
    }

    /// Returns a scaled copy of this matrix.
    pub fn scaled(&self, alpha: T) -> Self {
        let mut result = self.clone();
        result.scale(alpha);
        result
    }

    /// Reserves capacity for additional entries.
    pub fn reserve(&mut self, additional: usize) {
        self.row_indices.reserve(additional);
        self.col_indices.reserve(additional);
        self.values.reserve(additional);
    }

    /// Clears all entries but keeps dimensions.
    pub fn clear(&mut self) {
        self.row_indices.clear();
        self.col_indices.clear();
        self.values.clear();
    }
}

/// Builder for constructing COO matrices incrementally.
#[derive(Debug, Clone)]
pub struct CooMatrixBuilder<T: Scalar> {
    matrix: CooMatrix<T>,
}

impl<T: Scalar + Clone> CooMatrixBuilder<T> {
    /// Creates a new builder.
    pub fn new(nrows: usize, ncols: usize) -> Self {
        Self {
            matrix: CooMatrix::new_empty(nrows, ncols),
        }
    }

    /// Creates a new builder with pre-allocated capacity.
    pub fn with_capacity(nrows: usize, ncols: usize, capacity: usize) -> Self {
        Self {
            matrix: CooMatrix::with_capacity(nrows, ncols, capacity),
        }
    }

    /// Adds an entry to the matrix.
    pub fn add(&mut self, row: usize, col: usize, value: T) -> &mut Self {
        self.matrix.push(row, col, value);
        self
    }

    /// Adds a diagonal entry.
    pub fn add_diagonal(&mut self, index: usize, value: T) -> &mut Self {
        self.matrix.push(index, index, value);
        self
    }

    /// Adds entries for a dense block.
    pub fn add_block(
        &mut self,
        start_row: usize,
        start_col: usize,
        block: &oxiblas_matrix::MatRef<'_, T>,
    ) -> &mut Self
    where
        T: Field,
    {
        let (nrows, ncols) = block.shape();
        let eps = <T as Scalar>::epsilon();

        for i in 0..nrows {
            for j in 0..ncols {
                let val = block[(i, j)].clone();
                if Scalar::abs(val.clone()) > eps {
                    self.matrix.push(start_row + i, start_col + j, val);
                }
            }
        }
        self
    }

    /// Builds the COO matrix.
    pub fn build(self) -> CooMatrix<T> {
        self.matrix
    }

    /// Builds a CSR matrix, summing duplicates.
    pub fn build_csr(self) -> crate::csr::CsrMatrix<T>
    where
        T: Scalar<Real = T> + Field + Real,
    {
        self.matrix.to_csr()
    }

    /// Builds a CSC matrix, summing duplicates.
    pub fn build_csc(self) -> crate::csc::CscMatrix<T>
    where
        T: Scalar<Real = T> + Field + Real,
    {
        self.matrix.to_csc()
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_coo_new() {
        let row_indices = vec![0, 1, 2];
        let col_indices = vec![0, 1, 2];
        let values = vec![1.0f64, 2.0, 3.0];

        let coo = CooMatrix::new(3, 3, row_indices, col_indices, values).unwrap();

        assert_eq!(coo.nrows(), 3);
        assert_eq!(coo.ncols(), 3);
        assert_eq!(coo.len(), 3);
    }

    #[test]
    fn test_coo_push() {
        let mut coo = CooMatrix::<f64>::new_empty(3, 3);

        coo.push(0, 0, 1.0);
        coo.push(1, 1, 2.0);
        coo.push(2, 2, 3.0);

        assert_eq!(coo.len(), 3);
        assert_eq!(coo.values(), &[1.0, 2.0, 3.0]);
    }

    #[test]
    fn test_coo_with_duplicates() {
        let row_indices = vec![0, 0, 1];
        let col_indices = vec![0, 0, 1];
        let values = vec![1.0f64, 2.0, 3.0];

        let mut coo = CooMatrix::new(2, 2, row_indices, col_indices, values).unwrap();
        coo.sum_duplicates();

        assert_eq!(coo.len(), 2);
        // (0,0) should be summed to 3.0
        let entries: Vec<_> = coo.iter().collect();
        assert_eq!(entries[0], (0, 0, &3.0));
        assert_eq!(entries[1], (1, 1, &3.0));
    }

    #[test]
    fn test_coo_iter() {
        let row_indices = vec![0, 1, 2];
        let col_indices = vec![0, 1, 2];
        let values = vec![1.0f64, 2.0, 3.0];

        let coo = CooMatrix::new(3, 3, row_indices, col_indices, values).unwrap();

        let entries: Vec<_> = coo.iter().collect();
        assert_eq!(entries.len(), 3);
        assert_eq!(entries[0], (0, 0, &1.0));
        assert_eq!(entries[1], (1, 1, &2.0));
        assert_eq!(entries[2], (2, 2, &3.0));
    }

    #[test]
    fn test_coo_scale() {
        let row_indices = vec![0, 1, 2];
        let col_indices = vec![0, 1, 2];
        let values = vec![1.0f64, 2.0, 3.0];

        let mut coo = CooMatrix::new(3, 3, row_indices, col_indices, values).unwrap();
        coo.scale(2.0);

        assert_eq!(coo.values(), &[2.0, 4.0, 6.0]);
    }

    #[test]
    fn test_coo_builder() {
        let mut builder = CooMatrixBuilder::<f64>::with_capacity(3, 3, 5);
        builder.add(0, 0, 1.0);
        builder.add(1, 1, 2.0);
        builder.add(2, 2, 3.0);
        builder.add_diagonal(0, 0.5); // Duplicate at (0,0)
        let csr = builder.build_csr();

        assert_eq!(csr.nnz(), 3);
        assert_eq!(csr.get(0, 0), Some(&1.5)); // 1.0 + 0.5
    }

    #[test]
    fn test_coo_empty() {
        let coo: CooMatrix<f64> = CooMatrix::new_empty(5, 5);
        assert!(coo.is_empty());
        assert_eq!(coo.nrows(), 5);
        assert_eq!(coo.ncols(), 5);
    }

    #[test]
    fn test_coo_clear() {
        let mut coo = CooMatrix::<f64>::new_empty(3, 3);
        coo.push(0, 0, 1.0);
        coo.push(1, 1, 2.0);

        assert_eq!(coo.len(), 2);
        coo.clear();
        assert!(coo.is_empty());
        assert_eq!(coo.nrows(), 3); // Dimensions preserved
    }

    #[test]
    fn test_coo_invalid_row() {
        let row_indices = vec![5];
        let col_indices = vec![0];
        let values = vec![1.0f64];

        let result = CooMatrix::new(3, 3, row_indices, col_indices, values);
        assert!(matches!(result, Err(CooError::InvalidRowIndex { .. })));
    }

    #[test]
    fn test_coo_invalid_col() {
        let row_indices = vec![0];
        let col_indices = vec![5];
        let values = vec![1.0f64];

        let result = CooMatrix::new(3, 3, row_indices, col_indices, values);
        assert!(matches!(result, Err(CooError::InvalidColumnIndex { .. })));
    }
}