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

use kdtree::distance::squared_euclidean;
use kdtree::KdTree;
use num_traits::{Float, NumOps, Zero};
use std::collections::BTreeMap;
use std::error::Error;

pub(crate) fn det_correlation_matrix<T: Clone + NumOps + Zero>(mat: &[Vec<T>]) -> T {
    let dim = mat[0].len();
    let num = mat.len();
    let mut cor = Vec::new();
    for i in 0..num {
        let mut column = Vec::new();
        for j in 0..num {
            let mut c = T::zero();
            for k in 0..dim {
                c = c + mat[i][k].clone() * mat[j][k].clone();
            }
            column.push(c);
        }
        cor.push(column);
    }
    det(&cor)
}

pub(crate) fn dot<T: Clone + NumOps + Zero>(x: &[T], y: &[T]) -> T {
    x.iter()
        .zip(y.iter())
        .map(|(a, b)| a.clone() * b.clone())
        .fold(T::zero(), |sum, c| sum + c)
}

pub(crate) fn sub<T: Clone + NumOps + Zero>(x: &[T], y: &[T]) -> Vec<T> {
    x.iter()
        .zip(y.iter())
        .map(|(a, b)| a.clone() - b.clone())
        .collect()
}

pub(crate) fn min_max_index_each_axis<T: Clone + Zero + PartialOrd>(
    points: &[Vec<T>],
) -> Vec<(usize, usize)> {
    let dim = points[0].len();
    let mut min_index = vec![0; dim];
    let mut max_index = vec![0; dim];
    let mut min = vec![T::zero(); dim];
    let mut max = vec![T::zero(); dim];
    for (index, point) in points.iter().enumerate() {
        for (j, element) in point.iter().enumerate() {
            if index == 0 || *element < min[j] {
                min[j] = element.clone();
                min_index[j] = index;
            }
            if index == 0 || *element > max[j] {
                max[j] = element.clone();
                max_index[j] = index;
            }
        }
    }
    min_index.into_iter().zip(max_index.into_iter()).collect()
}

pub(crate) fn is_same_dimension<T>(points: &[Vec<T>]) -> bool {
    if points.len() == 0 {
        return true;
    }
    let dim = points[0].len();
    if points.iter().skip(1).find(|p| p.len() != dim).is_some() {
        false
    } else {
        true
    }
}

pub(crate) fn det<T: NumOps + Clone>(m: &[Vec<T>]) -> T {
    let row_dim = m.len();
    let column_dim = m[0].len();
    assert_eq!(row_dim, column_dim);
    match column_dim {
        1 => m[0][0].clone(),
        2 => det_2x2(m),
        3 => det_3x3(m),
        4 => det_4x4(m),
        _ => {
            unimplemented!("matrix size should be less than 4 dim to calcurate determinant for now")
        }
    }
}

fn det_2x2<T: NumOps + Clone>(m: &[Vec<T>]) -> T {
    m[0][0].clone() * m[1][1].clone() - m[0][1].clone() * m[1][0].clone()
}

#[rustfmt::skip]
fn det_3x3<T: NumOps+Clone>(m: &[Vec<T>]) -> T {
    m[0][0].clone() * (m[1][1].clone() * m[2][2].clone() - m[1][2].clone() * m[2][1].clone())
  - m[1][0].clone() * (m[0][1].clone() * m[2][2].clone() - m[0][2].clone() * m[2][1].clone())
  + m[2][0].clone() * (m[0][1].clone() * m[1][2].clone() - m[0][2].clone() * m[1][1].clone())
}

#[rustfmt::skip]
fn det_4x4<T: NumOps + Clone>(m: &[Vec<T>]) -> T {
    m[0][0].clone()
        * (   m[1][1].clone() * m[2][2].clone() * m[3][3].clone()
            + m[1][2].clone() * m[2][3].clone() * m[3][1].clone()
            + m[1][3].clone() * m[2][1].clone() * m[3][2].clone()
            - m[1][3].clone() * m[2][2].clone() * m[3][1].clone()
            - m[1][2].clone() * m[2][1].clone() * m[3][3].clone()
            - m[1][1].clone() * m[2][3].clone() * m[3][2].clone())
  - m[1][0].clone()
        * (   m[0][1].clone() * m[2][2].clone() * m[3][3].clone()
            + m[0][2].clone() * m[2][3].clone() * m[3][1].clone()
            + m[0][3].clone() * m[2][1].clone() * m[3][2].clone()
            - m[0][3].clone() * m[2][2].clone() * m[3][1].clone()
            - m[0][2].clone() * m[2][1].clone() * m[3][3].clone()
            - m[0][1].clone() * m[2][3].clone() * m[3][2].clone())
  + m[2][0].clone()
        * (   m[0][1].clone() * m[1][2].clone() * m[3][3].clone()
            + m[0][2].clone() * m[1][3].clone() * m[3][1].clone()
            + m[0][3].clone() * m[1][1].clone() * m[3][2].clone()
            - m[0][3].clone() * m[1][2].clone() * m[3][1].clone()
            - m[0][2].clone() * m[1][1].clone() * m[3][3].clone()
            - m[0][1].clone() * m[1][3].clone() * m[3][2].clone())
  - m[3][0].clone()
        * (   m[0][1].clone() * m[1][2].clone() * m[2][3].clone()
            + m[0][2].clone() * m[1][3].clone() * m[2][1].clone()
            + m[0][3].clone() * m[1][1].clone() * m[2][2].clone()
            - m[0][3].clone() * m[1][2].clone() * m[2][1].clone()
            - m[0][2].clone() * m[1][1].clone() * m[2][3].clone()
            - m[0][1].clone() * m[1][3].clone() * m[2][2].clone())
}

pub fn remove_nearby_points<T>(
    points: &[Vec<T>],
    squared_distance: T,
) -> Result<Vec<Vec<T>>, kdtree::ErrorKind>
where
    T: Float,
{
    let mut points_map = BTreeMap::new();
    let dim = points[0].len();
    let mut kdtree = KdTree::new(dim);
    for (i, point) in points.iter().enumerate() {
        points_map.insert(i, point);
        kdtree.add(point, i)?;
    }
    let mut live_indices = Vec::new();
    while let Some((id, point)) = points_map.iter().next() {
        let mut remove_list = Vec::new();
        live_indices.push(*id);
        for (_near_point_distance, near_point_id) in
            kdtree.within(&point, squared_distance, &squared_euclidean)?
        {
            remove_list.push(*near_point_id);
        }
        for key in remove_list {
            points_map.remove(&key);
        }
    }
    Ok(live_indices
        .into_iter()
        .map(|i| points[i].to_vec())
        .collect())
}

#[test]
fn det_4x4_test() {
    let r1 = vec![1., 2., 3., 4.];
    let r2 = vec![5., 6., 7., 8.];
    let r3 = vec![9., 10., 11., 12.];
    let r4 = vec![13., 14., 15., 16.];
    assert_eq!(det(&[r1, r2, r3, r4]), 0.);
    let r1 = vec![1., 2., 3., 1.];
    let r2 = vec![6., 6., 7., 1.];
    let r3 = vec![9., 11., 11., 1.];
    let r4 = vec![15., 14., 15., 1.];
    assert_eq!(det(&[r1, r2, r3, r4]), -4.);
    let r1 = vec![1., 1., 1., 1.];
    let r2 = vec![1., 2., 2., 1.];
    let r3 = vec![1., 2., 3., 1.];
    let r4 = vec![1., 1., 1., 2.];
    assert_eq!(det(&[r1, r2, r3, r4]), 1.);
}