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
//! Sorting operations for tensors.
//!
//! Provides `sort`, `argsort` (flat), and axis-based `sort_axis` / `argsort_axis`.
use crate::Scalar;
use crate::error::{CoreError, Result};
use super::Tensor;
impl<T: Scalar> Tensor<T> {
/// Sort all elements, returning a 1-D tensor in ascending order.
///
/// ```
/// # use scivex_core::Tensor;
/// let t = Tensor::from_vec(vec![3, 1, 4, 1, 5], vec![5]).unwrap();
/// assert_eq!(t.sort().as_slice(), &[1, 1, 3, 4, 5]);
/// ```
pub fn sort(&self) -> Tensor<T> {
let mut data = self.data.clone();
data.sort_unstable_by(|a, b| a.partial_cmp(b).unwrap_or(core::cmp::Ordering::Equal));
Tensor {
data,
shape: vec![self.numel()],
strides: vec![1],
}
}
/// Return indices that would sort all elements (flat), as a 1-D `Tensor<usize>`.
///
/// ```
/// # use scivex_core::Tensor;
/// let t = Tensor::from_vec(vec![30, 10, 20], vec![3]).unwrap();
/// assert_eq!(t.argsort().as_slice(), &[1, 2, 0]);
/// ```
pub fn argsort(&self) -> Tensor<usize> {
let mut indices: Vec<usize> = (0..self.numel()).collect();
indices.sort_unstable_by(|&a, &b| {
self.data[a]
.partial_cmp(&self.data[b])
.unwrap_or(core::cmp::Ordering::Equal)
});
Tensor {
data: indices,
shape: vec![self.numel()],
strides: vec![1],
}
}
/// Sort along a given axis, returning a new tensor with the same shape.
///
/// Each 1-D slice along `axis` is sorted independently in ascending order.
///
/// ```
/// # use scivex_core::Tensor;
/// let t = Tensor::from_vec(vec![3, 1, 4, 2], vec![2, 2]).unwrap();
/// let s = t.sort_axis(1).unwrap(); // sort each row
/// assert_eq!(s.as_slice(), &[1, 3, 2, 4]);
/// ```
pub fn sort_axis(&self, axis: usize) -> Result<Tensor<T>> {
if axis >= self.ndim() {
return Err(CoreError::AxisOutOfBounds {
axis,
ndim: self.ndim(),
});
}
let mut result = self.data.clone();
let outer: usize = self.shape[..axis].iter().product();
let axis_len = self.shape[axis];
let inner: usize = self.shape[axis + 1..].iter().product();
let mut slice_buf = vec![T::zero(); axis_len];
for o in 0..outer {
for i in 0..inner {
// Extract the 1-D slice along the axis
let base = o * axis_len * inner + i;
for (k, slot) in slice_buf.iter_mut().enumerate() {
*slot = self.data[base + k * inner];
}
// Sort the slice
slice_buf.sort_unstable_by(|a, b| {
a.partial_cmp(b).unwrap_or(core::cmp::Ordering::Equal)
});
// Write back
for (k, &val) in slice_buf.iter().enumerate() {
result[base + k * inner] = val;
}
}
}
Tensor::from_vec(result, self.shape.clone())
}
/// Return indices that would sort each slice along `axis`.
///
/// The result has the same shape as `self` but element type `usize`.
///
/// ```
/// # use scivex_core::Tensor;
/// let t = Tensor::from_vec(vec![3.0, 1.0, 4.0, 2.0], vec![2, 2]).unwrap();
/// let idx = t.argsort_axis(1).unwrap();
/// assert_eq!(idx.as_slice(), &[1, 0, 1, 0]);
/// ```
pub fn argsort_axis(&self, axis: usize) -> Result<Tensor<usize>> {
if axis >= self.ndim() {
return Err(CoreError::AxisOutOfBounds {
axis,
ndim: self.ndim(),
});
}
let outer: usize = self.shape[..axis].iter().product();
let axis_len = self.shape[axis];
let inner: usize = self.shape[axis + 1..].iter().product();
let numel = self.numel();
let mut result = vec![0usize; numel];
let mut idx_buf: Vec<usize> = (0..axis_len).collect();
for o in 0..outer {
for i in 0..inner {
let base = o * axis_len * inner + i;
// Reset index buffer
for (k, slot) in idx_buf.iter_mut().enumerate() {
*slot = k;
}
// Sort indices by comparing elements along the axis
idx_buf.sort_unstable_by(|&a, &b| {
let va = self.data[base + a * inner];
let vb = self.data[base + b * inner];
va.partial_cmp(&vb).unwrap_or(core::cmp::Ordering::Equal)
});
// Write back
for k in 0..axis_len {
result[base + k * inner] = idx_buf[k];
}
}
}
Tensor::from_vec(result, self.shape.clone())
}
}
#[cfg(test)]
#[allow(clippy::float_cmp)]
mod tests {
use super::*;
#[test]
fn test_sort_1d() {
let t = Tensor::from_vec(vec![3, 1, 4, 1, 5, 9], vec![6]).unwrap();
let s = t.sort();
assert_eq!(s.as_slice(), &[1, 1, 3, 4, 5, 9]);
assert_eq!(s.shape(), &[6]);
}
#[test]
fn test_argsort_1d() {
let t = Tensor::from_vec(vec![3, 1, 4, 1, 5, 9], vec![6]).unwrap();
let idx = t.argsort();
// Values at sorted indices should be ascending
let sorted: Vec<i32> = idx.as_slice().iter().map(|&i| t.as_slice()[i]).collect();
assert_eq!(sorted, &[1, 1, 3, 4, 5, 9]);
}
#[test]
fn test_sort_axis0_2d() {
// [[3, 1],
// [1, 4]]
let t = Tensor::from_vec(vec![3, 1, 1, 4], vec![2, 2]).unwrap();
let s = t.sort_axis(0).unwrap();
// Sort columns: col0=[3,1]->[1,3], col1=[1,4]->[1,4]
assert_eq!(s.as_slice(), &[1, 1, 3, 4]);
}
#[test]
fn test_sort_axis1_2d() {
// [[3, 1],
// [4, 2]]
let t = Tensor::from_vec(vec![3, 1, 4, 2], vec![2, 2]).unwrap();
let s = t.sort_axis(1).unwrap();
// Sort rows: row0=[3,1]->[1,3], row1=[4,2]->[2,4]
assert_eq!(s.as_slice(), &[1, 3, 2, 4]);
}
#[test]
fn test_argsort_axis() {
let t = Tensor::from_vec(vec![3.0_f64, 1.0, 4.0, 2.0], vec![2, 2]).unwrap();
let idx = t.argsort_axis(1).unwrap();
// Row 0: [3, 1] -> argsort = [1, 0]
// Row 1: [4, 2] -> argsort = [1, 0]
assert_eq!(idx.as_slice(), &[1, 0, 1, 0]);
assert_eq!(idx.shape(), &[2, 2]);
}
#[test]
fn test_flat_sort_from_2d() {
let t = Tensor::from_vec(vec![5, 2, 8, 1], vec![2, 2]).unwrap();
let s = t.sort();
assert_eq!(s.as_slice(), &[1, 2, 5, 8]);
assert_eq!(s.shape(), &[4]);
}
#[test]
fn test_sort_axis_out_of_bounds() {
let t = Tensor::from_vec(vec![1, 2, 3], vec![3]).unwrap();
assert!(t.sort_axis(1).is_err());
assert!(t.argsort_axis(1).is_err());
}
#[test]
fn test_already_sorted() {
let t = Tensor::from_vec(vec![1, 2, 3, 4, 5], vec![5]).unwrap();
let s = t.sort();
assert_eq!(s.as_slice(), &[1, 2, 3, 4, 5]);
}
#[test]
fn test_reverse_sorted() {
let t = Tensor::from_vec(vec![5, 4, 3, 2, 1], vec![5]).unwrap();
let s = t.sort();
assert_eq!(s.as_slice(), &[1, 2, 3, 4, 5]);
}
}