tf_idf_vectorizer/utils/math/vector/
math.rs

1use std::{cmp::Ordering, ops::{AddAssign, MulAssign}, ptr};
2
3use num::Num;
4
5use super::ZeroSpVec;
6
7impl<N> ZeroSpVec<N> 
8where
9    N: Num + AddAssign + MulAssign
10{
11    /// ドット積を計算するメソッド
12    ///
13    /// # Arguments
14    /// * `other` - 他のベクトル
15    /// 
16    /// # Returns
17    /// * `N` - ドット積の結果
18    #[inline]
19    pub fn dot<R>(&self, other: &Self) -> R
20    where
21        R: Num + AddAssign,
22        N: Into<R> + Copy,
23    {
24        debug_assert_eq!(
25            self.len(),
26            other.len(),
27            "Vectors must be of the same length to compute dot product."
28        );
29    
30        let mut result = R::zero();
31        let self_nnz = self.nnz();
32        let other_nnz = other.nnz();
33    
34        if self_nnz == 0 || other_nnz == 0 {
35            return result;
36        }
37    
38        unsafe {
39            let self_inds = std::slice::from_raw_parts(self.ind_ptr(), self_nnz);
40            let self_vals = std::slice::from_raw_parts(self.val_ptr(), self_nnz);
41            let other_inds = std::slice::from_raw_parts(other.ind_ptr(), other_nnz);
42            let other_vals = std::slice::from_raw_parts(other.val_ptr(), other_nnz);
43            
44            let mut i = 0;
45            let mut j = 0;
46            
47            while i < self_nnz && j < other_nnz {
48                let s_ind = *self_inds.get_unchecked(i);
49                let o_ind = *other_inds.get_unchecked(j);
50                match s_ind.cmp(&o_ind) {
51                    Ordering::Equal => {
52                        result += self_vals[i].into() * other_vals[j].into();
53                        i += 1;
54                        j += 1;
55                    }
56                    Ordering::Less => i += 1,
57                    Ordering::Greater => j += 1,
58                }
59            }
60        }
61    
62        result
63    }
64
65    #[inline]
66    pub fn norm_sq<R>(&self) -> R
67    where
68        R: Num + AddAssign + Copy,
69        N: Into<R> + Copy,
70    {
71        let mut result = R::zero();
72        let self_nnz = self.nnz();
73    
74        if self_nnz == 0 {
75            return result;
76        }
77    
78        unsafe {
79            let self_val_ptr = self.val_ptr();
80    
81            for i in 0..self_nnz {
82                let val: R = (*self_val_ptr.add(i)).into();
83                result += val * val;
84            }
85        }
86    
87        result
88    }
89
90    /// アダマール積を計算するメソッド
91    /// 
92    /// # Arguments
93    /// * `other` - 他のベクトル
94    /// 
95    /// # Returns
96    /// * `ZeroSpVec<N>` - アダマール積の結果
97    #[inline]
98    pub fn hadamard(&self, other: &Self) -> Self {
99        debug_assert_eq!(
100            self.len(),
101            other.len(),
102            "Vectors must be of the same length to compute hadamard product."
103        );
104
105        let min_nnz = self.nnz().min(other.nnz());
106        let mut result: ZeroSpVec<N> = ZeroSpVec::with_capacity(min_nnz);
107        result.len = self.len();
108
109        // nnz == 0 ならゼロ埋めで返す
110        if self.nnz() == 0 {
111            result.len = self.len();
112            return result;
113        }
114
115        unsafe {
116            let mut i = 0;
117            let mut j = 0;
118            while i < self.nnz() && j < other.nnz() {
119                let self_ind = ptr::read(self.ind_ptr().add(i));
120                let other_ind = ptr::read(other.ind_ptr().add(j));
121                if self_ind == other_ind {
122                    // 同じインデックスの要素を掛け算して加算
123                    let value = ptr::read(self.val_ptr().add(i)) * ptr::read(other.val_ptr().add(j));
124                    result.raw_push(self_ind, value);
125                    i += 1;
126                    j += 1;
127                } else if self_ind < other_ind {
128                    i += 1;
129                } else {
130                    j += 1;
131                }
132            }
133        }
134        result
135    }
136}
tf_idf_vectorizer/utils/math/vector/math.rs

tf_idf_vectorizer/utils/math/vector/
math.rs
