ndarray 0.4.6

An N-dimensional array for general elements and for numerics. Lightweight array views and slicing; views support chunking and splitting.
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

#![cfg(feature = "rblas")]
#![allow(deprecated)]

extern crate rblas;
extern crate num;
#[macro_use] extern crate ndarray;

use rblas::Gemm;
use rblas::attribute::Transpose;

use num::Float;

use ndarray::{
    OwnedArray,
    ArrayView,
    ArrayViewMut,
    arr2,
    Ix,
    ShapeError
};

use ndarray::blas::AsBlas;

#[test]
fn strided_matrix() {
    // smoke test, a matrix multiplication of uneven size
    let (n, m) = (45, 33);
    let mut a = OwnedArray::linspace(0., ((n * m) - 1) as f32, n as usize * m as usize ).into_shape((n, m)).unwrap();
    let mut b = OwnedArray::eye(m);
    let mut res = OwnedArray::zeros(a.dim());
    Gemm::gemm(&1., Transpose::NoTrans, &a.blas(), Transpose::NoTrans, &b.blas(),
               &0., &mut res.blas());
    assert_eq!(res, a);

    // matrix multiplication, strided
    let mut aprim = a.to_shared();
    aprim.islice(s![0..12, 0..11]);
    println!("{:?}", aprim.shape());
    println!("{:?}", aprim.strides());
    let mut b = OwnedArray::eye(aprim.shape()[1]);
    let mut res = OwnedArray::zeros(aprim.dim());
    Gemm::gemm(&1., Transpose::NoTrans, &aprim.blas(), Transpose::NoTrans, &b.blas(),
               &0., &mut res.blas());
    assert_eq!(res, aprim);

    // Transposed matrix multiply
    let (np, mp) = aprim.dim();
    let mut res = OwnedArray::zeros((mp, np));
    let mut b = OwnedArray::eye(np);
    Gemm::gemm(&1., Transpose::Trans, &aprim.blas(), Transpose::NoTrans, &b.blas(),
               &0., &mut res.blas());
    let mut at = aprim.clone();
    at.swap_axes(0, 1);
    assert_eq!(at, res);

    // strided, needs copy
    let mut abis = a.to_shared();
    abis.islice(s![0..12, ..;2]);
    println!("{:?}", abis.shape());
    println!("{:?}", abis.strides());
    let mut b = OwnedArray::eye(abis.shape()[1]);
    let mut res = OwnedArray::zeros(abis.dim());
    Gemm::gemm(&1., Transpose::NoTrans, &abis.blas(), Transpose::NoTrans, &b.blas(),
               &0., &mut res.blas());
    assert_eq!(res, abis);
}

#[test]
fn strided_view() {
    // smoke test, a matrix multiplication of uneven size
    let (n, m) = (45, 33);
    let mut a = OwnedArray::linspace(0., ((n * m) - 1) as f32, n as usize * m as usize ).into_shape((n, m)).unwrap();
    let mut b = OwnedArray::eye(m);
    let mut res = OwnedArray::zeros(a.dim());
    Gemm::gemm(&1.,
               Transpose::NoTrans, &a.blas_view_mut_checked().unwrap(),
               Transpose::NoTrans, &b.blas_view_mut_checked().unwrap(),
               &0., &mut res.blas_view_mut_checked().unwrap());
    assert_eq!(res, a);

    // matrix multiplication, strided
    let aprim = a.slice(s![0..12, 0..11]);
    let mut b = OwnedArray::eye(aprim.shape()[1]);
    let mut res = OwnedArray::zeros(aprim.dim());
    Gemm::gemm(&1.,
               Transpose::NoTrans, &aprim.bv(),
               Transpose::NoTrans, &b.blas(),
               &0., &mut res.blas());
    assert_eq!(res, aprim);

    // test out with matrices where lower axis is strided but has length 1
    let mut a3 = arr2(&[[1., 2., 3.]]);
    a3.swap_axes(0, 1);
    let mut b = OwnedArray::eye(a3.shape()[1]);
    let mut res = arr2(&[[0., 0., 0.]]);
    res.swap_axes(0, 1);
    Gemm::gemm(&1.,
               Transpose::NoTrans, &a3.bvm(),
               Transpose::NoTrans, &b.blas(),
               &0., &mut res.bvm());
    assert_eq!(res, a3);
}

#[test]
fn as_blas() {
    let mut a = OwnedArray::<f32, _>::zeros((4, 4));
    assert!(a.blas_view_mut_checked().is_ok());
    a.swap_axes(0, 1);
    assert!(a.blas_view_mut_checked().is_err());
    a.swap_axes(0, 1);

    {
        // increased row stride
        let mut b = a.slice_mut(s![..;2, ..]);
        assert!(b.blas_view_mut_checked().is_ok());
        b.bvm(); // no panic
    }
    {
        // inner dimension is not contig
        let mut b = a.slice_mut(s![.., ..;2]);
        assert!(b.blas_view_mut_checked().is_err());
    }
    {
        // inner dimension is length 1, is ok again
        let mut b = a.slice_mut(s![.., ..;4]);
        assert!(b.blas_view_mut_checked().is_ok());
        b.bvm();
    }
}

type Ix2 = (Ix, Ix);
fn dot<F>(a: ArrayView<F, Ix2>, b: ArrayView<F, Ix2>,
          c: &mut ArrayViewMut<F, Ix2>) -> Result<(), ShapeError>
    where F: Gemm + Float,
{
    let at = Transpose::NoTrans;
    let bt = Transpose::NoTrans;

    let ba = try!(a.blas_view_checked());
    let bb = try!(b.blas_view_checked());
    let mut bc = try!(c.blas_view_mut_checked());
    F::gemm(&F::one(),
            at, &ba,
            bt, &bb,
            &F::zero(), &mut bc);
    Ok(())
}

#[test]
fn test_dot() {
    let mut a = arr2(&[[1., 2.],
                       [0., 3.]]);

    let b = a.clone();
    let mut res = a.clone();
    res.assign_scalar(&0.);

    dot(a.view(), b.view(), &mut res.view_mut()).unwrap();
    println!("{:?}", res);

    a.swap_axes(0, 1);
    res.assign_scalar(&0.);

    let result = dot(a.view(), b.view(), &mut res.view_mut());
    assert!(result.is_err());
}