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

pub mod operations;
pub mod operators;

pub use operations::*;

use crate::{Matrix, Number, RankIndex, Tensor, TensorError};
use serde::{Deserialize, Serialize};
use std::collections::HashMap;

#[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
pub struct SparseTensor<T = f64>
where
    T: Number,
{
    sizes: Vec<usize>,
    elems: HashMap<Vec<usize>, T>,
    default: T,
}

impl<T> SparseTensor<T>
where
    T: Number,
{
    pub fn new(sizes: Vec<usize>) -> Self {
        Self {
            sizes,
            elems: HashMap::new(),
            default: T::default(),
        }
    }

    pub fn from(sizes: Vec<usize>, elems: HashMap<Vec<usize>, T>) -> Result<Self, TensorError> {
        for (index, _) in elems.iter() {
            if index.len() != sizes.len() {
                return Err(TensorError::RankMismatch);
            }
            for (rank, &d) in index.iter().enumerate() {
                if sizes[rank] <= d {
                    return Err(TensorError::OutOfRange);
                }
            }
        }
        Ok(Self {
            sizes,
            elems,
            default: T::default(),
        })
    }

    pub fn is_same_size(&self, other: &SparseTensor<T>) -> bool {
        self.sizes == other.sizes
    }

    pub fn total_size(&self) -> usize {
        self.sizes.iter().product()
    }

    pub fn not_1dimension_ranks(&self) -> usize {
        self.sizes.iter().filter(|&d| *d != 1).count()
    }

    pub fn reduce_1dimension_rank(&self) -> Self {
        let mut new_dims = vec![];
        for d in self.sizes.iter() {
            if *d != 1 {
                new_dims.push(*d);
            }
        }

        // TODO: parallelize
        let mut new_elems = HashMap::new();
        for (k, v) in self.elems.iter() {
            let mut new_k = vec![];
            for (i, d) in k.iter().enumerate() {
                if self.sizes[i] != 1 {
                    new_k.push(*d);
                }
            }
            new_elems.insert(new_k, *v);
        }

        Self {
            sizes: new_dims,
            elems: new_elems,
            default: self.default,
        }
    }

    pub fn to_vec(&self) -> Vec<T> {
        if self.rank() != 1 {
            panic!("SparseTensor::to_vec() is only available for rank 1 tensor.");
        }

        let mut vec = vec![T::default(); self.sizes[0]];
        for (k, v) in self.elems.iter() {
            vec[k[0]] = *v;
        }

        vec
    }

    pub fn to_mat(&self) -> Matrix<T> {
        if self.rank() != 2 {
            panic!("SparseTensor::to_mat() is only available for rank 2 tensor.");
        }

        let mut mat = Matrix::new(self.sizes[0], self.sizes[1]);
        for (k, v) in self.elems.iter() {
            mat[(k[0], k[1])] = *v;
        }
        mat
    }

    pub fn elems(&self) -> &HashMap<Vec<usize>, T> {
        &self.elems
    }

    pub fn elems_mut(&mut self) -> &mut HashMap<Vec<usize>, T> {
        &mut self.elems
    }

    pub fn eject(self) -> (Vec<usize>, HashMap<Vec<usize>, T>) {
        (self.sizes, self.elems)
    }
}

impl<T> Tensor<T> for SparseTensor<T>
where
    T: Number,
{
    fn rank(&self) -> usize {
        self.sizes.len()
    }

    fn size(&self, rank: RankIndex) -> usize {
        self.sizes[rank]
    }

    fn elem(&self, indices: &[usize]) -> T {
        self[indices]
    }

    fn elem_mut(&mut self, indices: &[usize]) -> &mut T {
        &mut self[indices]
    }
}

impl<T> From<Vec<T>> for SparseTensor<T>
where
    T: Number,
{
    fn from(vec: Vec<T>) -> Self {
        let sizes = vec![vec.len()];
        let elems = vec
            .into_iter()
            .enumerate()
            .map(|(i, v)| (vec![i], v))
            .collect();

        Self {
            sizes,
            elems,
            default: T::default(),
        }
    }
}