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

//! # Log Trait and Implementations
//!
//! This module provides the `Log` trait and its implementations for performing element-wise logarithm operations on matrices.
//!
//! ## Log Trait
//!
//! The `Log` trait defines two methods:
//!
//! - `log(&mut self, source: Matrix<ViewMem<Self::Item>, Self::Dim>)`: Computes the element-wise logarithm of the `source` matrix and stores the result in `self`.
//! - `log_assign(&mut self)`: Computes the element-wise logarithm of `self` in-place.
//!
//! ## Implementations
//!
//! The `Log` trait is implemented for `Matrix<SM, D>` where:
//! - `T` is the element type and must implement the `Num` trait.
//! - `D` is the dimension type and must implement the `DimTrait`.
//! - `SM` is the storage type and must implement the `ToViewMutMemory` trait with `Item = T`.
//!
//! The implementations handle both 1-dimensional and multi-dimensional matrices.
//!
//! ### 1-Dimensional Matrices
//!
//! For 1-dimensional matrices, the logarithm is computed using the `log_1d_cpu` function, which performs the operation using CPU.
//!
//! ### Multi-Dimensional Matrices
//!
//! For multi-dimensional matrices, the logarithm is computed by iterating over the first dimension and recursively calling the `log` or `log_assign` method on the sub-matrices.
//!
//! ## Functions
//!
//! - `log_1d_cpu<T, DM, SM, D1, D2>(dest: Matrix<DM, D1>, source: Matrix<SM, D2>)`: Computes the element-wise logarithm of `source` and stores the result in `dest` for 1-dimensional matrices using CPU.
//! - `log_1d_cpu_assign<T, M, D>(dest: Matrix<M, D>)`: Computes the element-wise logarithm of `dest` in-place for 1-dimensional matrices using CPU.
//!
//! ## Testing
//!
//! The module includes unit tests to verify the correctness of the `Log` trait implementations. The tests cover different scenarios, such as 1-dimensional and 2-dimensional matrices, and compare the results against ex//!
//! ## Example
//!
//! ```rust
//! use zenu_matrix::{matrix::{OwnedMatrix, ToViewMatrix}, matrix_impl::OwnedMatrixDyn, operation::log::Log};
//!
//! let mut a = OwnedMatrixDyn::from_vec(vec![1.0, 2.0, 3.0], [3]);
//! a.log_assign();
//!
//! let mut b = OwnedMatrixDyn::from_vec(vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0], [2, 3]);
//! let c = OwnedMatrixDyn::from_vec(vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0], [2, 3]);
//! b.log(c.to_view());
//! ```

use crate::{
    dim::{Dim1, DimTrait},
    index::Index0D,
    matrix::{IndexAxisDyn, IndexAxisMutDyn, MatrixBase, ToViewMatrix, ToViewMutMatrix},
    matrix_impl::{matrix_into_dim, Matrix},
    memory::{ToViewMutMemory, View, ViewMut},
    memory_impl::ViewMem,
    num::Num,
};

/// Trait for performing element-wise logarithm operations on matrices.
pub trait Log: ToViewMutMatrix + MatrixBase {
    /// Computes the element-wise logarithm of `source` and stores the result in `self`.
    ///
    /// # Panics
    ///
    /// Panics if the shapes of `self` and `source` do not match.
    fn log(&mut self, source: Matrix<ViewMem<Self::Item>, Self::Dim>);

    /// Computes the element-wise logarithm of `self` in-place.
    fn log_assign(&mut self);
}

impl<T, D, SM> Log for Matrix<SM, D>
where
    T: Num,
    D: DimTrait,
    SM: ToViewMutMemory<Item = T>,
{
    fn log(&mut self, source: Matrix<ViewMem<T>, D>) {
        if self.shape().slice() != source.shape().slice() {
            panic!("shape mismatch");
        }

        if self.shape().len() == 1 {
            log_1d_cpu(self.to_view_mut(), source.to_view());
        } else {
            for i in 0..self.shape()[0] {
                let mut dest = self.index_axis_mut_dyn(Index0D::new(i));
                let source = source.index_axis_dyn(Index0D::new(i));
                dest.log(source);
            }
        }
    }

    fn log_assign(&mut self) {
        if self.shape().len() == 1 {
            log_1d_cpu_assign(self.to_view_mut());
        } else {
            for i in 0..self.shape()[0] {
                let mut dest = self.index_axis_mut_dyn(Index0D::new(i));
                Log::log_assign(&mut dest);
            }
        }
    }
}

/// Computes the element-wise logarithm of `source` and stores the result in `dest` for 1-dimensional matrices using CPU.
fn log_1d_cpu<T, DM, SM, D1, D2>(dest: Matrix<DM, D1>, source: Matrix<SM, D2>)
where
    T: Num,
    DM: ViewMut<Item = T>,
    SM: View<Item = T>,
    D1: DimTrait,
    D2: DimTrait,
{
    let source_stride = dest.stride()[0];
    let dest_stride = source.stride()[0];
    let mut dest: Matrix<DM, Dim1> = matrix_into_dim(dest);
    let source: Matrix<SM, Dim1> = matrix_into_dim(source);
    let dest = dest.as_mut_slice();
    let source = source.as_slice();
    for i in 0..source.len() {
        dest[i * dest_stride] = source[i * source_stride].ln();
    }
}

/// Computes the element-wise logarithm of `dest` in-place for 1-dimensional matrices using CPU.
fn log_1d_cpu_assign<T, M, D>(dest: Matrix<M, D>)
where
    T: Num,
    M: ViewMut<Item = T>,
    D: DimTrait,
{
    let dest_stride = dest.stride()[0];
    let mut dest: Matrix<M, Dim1> = matrix_into_dim(dest);
    let dest = dest.as_mut_slice();
    for i in 0..dest.len() {
        dest[i * dest_stride] = dest[i * dest_stride].ln();
    }
}

#[cfg(test)]
mod log {
    use crate::{
        matrix::{OwnedMatrix, ToViewMatrix},
        matrix_impl::OwnedMatrixDyn,
        operation::asum::Asum,
    };

    use super::Log;

    #[test]
    fn log_1d_assign() {
        let mut a = OwnedMatrixDyn::from_vec(vec![1.0, 2.0, 3.0], [3]);
        a.log_assign();
        let ans = OwnedMatrixDyn::from_vec(vec![0.0, 0.6931471805599453, 1.0986122886681098], [3]);
        let diff = a - ans;
        let diff_asum = diff.asum();
        assert!(diff_asum < 1.0e-10);
    }

    #[test]
    fn log_1d() {
        let mut a = OwnedMatrixDyn::from_vec(vec![1.0, 2.0, 3.0], [3]);
        let b = OwnedMatrixDyn::from_vec(vec![1.0, 2.0, 3.0], [3]);
        a.log(b.to_view());
        let ans = OwnedMatrixDyn::from_vec(vec![0.0, 0.6931471805599453, 1.0986122886681098], [3]);
        let diff = a - ans;
        let diff_asum = diff.asum();
        assert!(diff_asum < 1.0e-10);
    }

    #[test]
    fn log_2d() {
        let mut a = OwnedMatrixDyn::from_vec(vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0], [2, 3]);
        let b = OwnedMatrixDyn::from_vec(vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0], [2, 3]);
        a.log(b.to_view());
        let ans = OwnedMatrixDyn::from_vec(
            vec![
                0.0,
                0.6931471805599453,
                1.0986122886681098,
                1.3862943611198906,
                1.6094379124341003,
                1.791759469228055,
            ],
            [2, 3],
        );
        let diff = a - ans;
        let diff_asum = diff.asum();
        assert!(diff_asum < 1.0e-10);
    }

    #[test]
    fn log_2d_assign() {
        let mut a = OwnedMatrixDyn::from_vec(vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0], [2, 3]);
        a.log_assign();
        let ans = OwnedMatrixDyn::from_vec(
            vec![
                0.0,
                0.6931471805599453,
                1.0986122886681098,
                1.3862943611198906,
                1.6094379124341003,
                1.791759469228055,
            ],
            [2, 3],
        );
        let diff = a - ans;
        let diff_asum = diff.asum();
        assert!(diff_asum < 1.0e-10);
    }
}