xref: /aosp_15_r20/external/libaom/aom_dsp/mathutils.h (revision 77c1e3ccc04c968bd2bc212e87364f250e820521)
1 /*
2  * Copyright (c) 2017, Alliance for Open Media. All rights reserved.
3  *
4  * This source code is subject to the terms of the BSD 2 Clause License and
5  * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
6  * was not distributed with this source code in the LICENSE file, you can
7  * obtain it at www.aomedia.org/license/software. If the Alliance for Open
8  * Media Patent License 1.0 was not distributed with this source code in the
9  * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
10  */
11 
12 #ifndef AOM_AOM_DSP_MATHUTILS_H_
13 #define AOM_AOM_DSP_MATHUTILS_H_
14 
15 #include <assert.h>
16 #include <math.h>
17 #include <string.h>
18 
19 #include "aom_dsp/aom_dsp_common.h"
20 
21 static const double TINY_NEAR_ZERO = 1.0E-16;
22 
23 // Solves Ax = b, where x and b are column vectors of size nx1 and A is nxn
linsolve(int n,double * A,int stride,double * b,double * x)24 static inline int linsolve(int n, double *A, int stride, double *b, double *x) {
25   int i, j, k;
26   double c;
27   // Forward elimination
28   for (k = 0; k < n - 1; k++) {
29     // Bring the largest magnitude to the diagonal position
30     for (i = n - 1; i > k; i--) {
31       if (fabs(A[(i - 1) * stride + k]) < fabs(A[i * stride + k])) {
32         for (j = 0; j < n; j++) {
33           c = A[i * stride + j];
34           A[i * stride + j] = A[(i - 1) * stride + j];
35           A[(i - 1) * stride + j] = c;
36         }
37         c = b[i];
38         b[i] = b[i - 1];
39         b[i - 1] = c;
40       }
41     }
42     for (i = k; i < n - 1; i++) {
43       if (fabs(A[k * stride + k]) < TINY_NEAR_ZERO) return 0;
44       c = A[(i + 1) * stride + k] / A[k * stride + k];
45       for (j = 0; j < n; j++) A[(i + 1) * stride + j] -= c * A[k * stride + j];
46       b[i + 1] -= c * b[k];
47     }
48   }
49   // Backward substitution
50   for (i = n - 1; i >= 0; i--) {
51     if (fabs(A[i * stride + i]) < TINY_NEAR_ZERO) return 0;
52     c = 0;
53     for (j = i + 1; j <= n - 1; j++) c += A[i * stride + j] * x[j];
54     x[i] = (b[i] - c) / A[i * stride + i];
55   }
56 
57   return 1;
58 }
59 
60 ////////////////////////////////////////////////////////////////////////////////
61 // Least-squares
62 // Solves for n-dim x in a least squares sense to minimize |Ax - b|^2
63 // The solution is simply x = (A'A)^-1 A'b or simply the solution for
64 // the system: A'A x = A'b
65 //
66 // This process is split into three steps in order to avoid needing to
67 // explicitly allocate the A matrix, which may be very large if there
68 // are many equations to solve.
69 //
70 // The process for using this is (in pseudocode):
71 //
72 // Allocate mat (size n*n), y (size n), a (size n), x (size n)
73 // least_squares_init(mat, y, n)
74 // for each equation a . x = b {
75 //    least_squares_accumulate(mat, y, a, b, n)
76 // }
77 // least_squares_solve(mat, y, x, n)
78 //
79 // where:
80 // * mat, y are accumulators for the values A'A and A'b respectively,
81 // * a, b are the coefficients of each individual equation,
82 // * x is the result vector
83 // * and n is the problem size
least_squares_init(double * mat,double * y,int n)84 static inline void least_squares_init(double *mat, double *y, int n) {
85   memset(mat, 0, n * n * sizeof(double));
86   memset(y, 0, n * sizeof(double));
87 }
88 
89 // Round the given positive value to nearest integer
iroundpf(float x)90 static AOM_FORCE_INLINE int iroundpf(float x) {
91   assert(x >= 0.0);
92   return (int)(x + 0.5f);
93 }
94 
least_squares_accumulate(double * mat,double * y,const double * a,double b,int n)95 static inline void least_squares_accumulate(double *mat, double *y,
96                                             const double *a, double b, int n) {
97   for (int i = 0; i < n; i++) {
98     for (int j = 0; j < n; j++) {
99       mat[i * n + j] += a[i] * a[j];
100     }
101   }
102   for (int i = 0; i < n; i++) {
103     y[i] += a[i] * b;
104   }
105 }
106 
least_squares_solve(double * mat,double * y,double * x,int n)107 static inline int least_squares_solve(double *mat, double *y, double *x,
108                                       int n) {
109   return linsolve(n, mat, n, y, x);
110 }
111 
112 // Matrix multiply
multiply_mat(const double * m1,const double * m2,double * res,const int m1_rows,const int inner_dim,const int m2_cols)113 static inline void multiply_mat(const double *m1, const double *m2, double *res,
114                                 const int m1_rows, const int inner_dim,
115                                 const int m2_cols) {
116   double sum;
117 
118   int row, col, inner;
119   for (row = 0; row < m1_rows; ++row) {
120     for (col = 0; col < m2_cols; ++col) {
121       sum = 0;
122       for (inner = 0; inner < inner_dim; ++inner)
123         sum += m1[row * inner_dim + inner] * m2[inner * m2_cols + col];
124       *(res++) = sum;
125     }
126   }
127 }
128 
approx_exp(float y)129 static inline float approx_exp(float y) {
130 #define A ((1 << 23) / 0.69314718056f)  // (1 << 23) / ln(2)
131 #define B \
132   127  // Offset for the exponent according to IEEE floating point standard.
133 #define C 60801  // Magic number controls the accuracy of approximation
134   union {
135     float as_float;
136     int32_t as_int32;
137   } container;
138   container.as_int32 = ((int32_t)(y * A)) + ((B << 23) - C);
139   return container.as_float;
140 #undef A
141 #undef B
142 #undef C
143 }
144 #endif  // AOM_AOM_DSP_MATHUTILS_H_
145