CMSIS DSP Lib
Fork of mbed-dsp by
cmsis_dsp/MatrixFunctions/arm_mat_inverse_f32.c@1:fdd22bb7aa52, 2012-11-28 (annotated)
- Committer:
- emilmont
- Date:
- Wed Nov 28 12:30:09 2012 +0000
- Revision:
- 1:fdd22bb7aa52
- Child:
- 2:da51fb522205
DSP library code
Who changed what in which revision?
User | Revision | Line number | New contents of line |
---|---|---|---|
emilmont | 1:fdd22bb7aa52 | 1 | /* ---------------------------------------------------------------------- |
emilmont | 1:fdd22bb7aa52 | 2 | * Copyright (C) 2010 ARM Limited. All rights reserved. |
emilmont | 1:fdd22bb7aa52 | 3 | * |
emilmont | 1:fdd22bb7aa52 | 4 | * $Date: 15. February 2012 |
emilmont | 1:fdd22bb7aa52 | 5 | * $Revision: V1.1.0 |
emilmont | 1:fdd22bb7aa52 | 6 | * |
emilmont | 1:fdd22bb7aa52 | 7 | * Project: CMSIS DSP Library |
emilmont | 1:fdd22bb7aa52 | 8 | * Title: arm_mat_inverse_f32.c |
emilmont | 1:fdd22bb7aa52 | 9 | * |
emilmont | 1:fdd22bb7aa52 | 10 | * Description: Floating-point matrix inverse. |
emilmont | 1:fdd22bb7aa52 | 11 | * |
emilmont | 1:fdd22bb7aa52 | 12 | * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0 |
emilmont | 1:fdd22bb7aa52 | 13 | * |
emilmont | 1:fdd22bb7aa52 | 14 | * Version 1.1.0 2012/02/15 |
emilmont | 1:fdd22bb7aa52 | 15 | * Updated with more optimizations, bug fixes and minor API changes. |
emilmont | 1:fdd22bb7aa52 | 16 | * |
emilmont | 1:fdd22bb7aa52 | 17 | * Version 1.0.10 2011/7/15 |
emilmont | 1:fdd22bb7aa52 | 18 | * Big Endian support added and Merged M0 and M3/M4 Source code. |
emilmont | 1:fdd22bb7aa52 | 19 | * |
emilmont | 1:fdd22bb7aa52 | 20 | * Version 1.0.3 2010/11/29 |
emilmont | 1:fdd22bb7aa52 | 21 | * Re-organized the CMSIS folders and updated documentation. |
emilmont | 1:fdd22bb7aa52 | 22 | * |
emilmont | 1:fdd22bb7aa52 | 23 | * Version 1.0.2 2010/11/11 |
emilmont | 1:fdd22bb7aa52 | 24 | * Documentation updated. |
emilmont | 1:fdd22bb7aa52 | 25 | * |
emilmont | 1:fdd22bb7aa52 | 26 | * Version 1.0.1 2010/10/05 |
emilmont | 1:fdd22bb7aa52 | 27 | * Production release and review comments incorporated. |
emilmont | 1:fdd22bb7aa52 | 28 | * |
emilmont | 1:fdd22bb7aa52 | 29 | * Version 1.0.0 2010/09/20 |
emilmont | 1:fdd22bb7aa52 | 30 | * Production release and review comments incorporated. |
emilmont | 1:fdd22bb7aa52 | 31 | * -------------------------------------------------------------------- */ |
emilmont | 1:fdd22bb7aa52 | 32 | |
emilmont | 1:fdd22bb7aa52 | 33 | #include "arm_math.h" |
emilmont | 1:fdd22bb7aa52 | 34 | |
emilmont | 1:fdd22bb7aa52 | 35 | /** |
emilmont | 1:fdd22bb7aa52 | 36 | * @ingroup groupMatrix |
emilmont | 1:fdd22bb7aa52 | 37 | */ |
emilmont | 1:fdd22bb7aa52 | 38 | |
emilmont | 1:fdd22bb7aa52 | 39 | /** |
emilmont | 1:fdd22bb7aa52 | 40 | * @defgroup MatrixInv Matrix Inverse |
emilmont | 1:fdd22bb7aa52 | 41 | * |
emilmont | 1:fdd22bb7aa52 | 42 | * Computes the inverse of a matrix. |
emilmont | 1:fdd22bb7aa52 | 43 | * |
emilmont | 1:fdd22bb7aa52 | 44 | * The inverse is defined only if the input matrix is square and non-singular (the determinant |
emilmont | 1:fdd22bb7aa52 | 45 | * is non-zero). The function checks that the input and output matrices are square and of the |
emilmont | 1:fdd22bb7aa52 | 46 | * same size. |
emilmont | 1:fdd22bb7aa52 | 47 | * |
emilmont | 1:fdd22bb7aa52 | 48 | * Matrix inversion is numerically sensitive and the CMSIS DSP library only supports matrix |
emilmont | 1:fdd22bb7aa52 | 49 | * inversion of floating-point matrices. |
emilmont | 1:fdd22bb7aa52 | 50 | * |
emilmont | 1:fdd22bb7aa52 | 51 | * \par Algorithm |
emilmont | 1:fdd22bb7aa52 | 52 | * The Gauss-Jordan method is used to find the inverse. |
emilmont | 1:fdd22bb7aa52 | 53 | * The algorithm performs a sequence of elementary row-operations till it |
emilmont | 1:fdd22bb7aa52 | 54 | * reduces the input matrix to an identity matrix. Applying the same sequence |
emilmont | 1:fdd22bb7aa52 | 55 | * of elementary row-operations to an identity matrix yields the inverse matrix. |
emilmont | 1:fdd22bb7aa52 | 56 | * If the input matrix is singular, then the algorithm terminates and returns error status |
emilmont | 1:fdd22bb7aa52 | 57 | * <code>ARM_MATH_SINGULAR</code>. |
emilmont | 1:fdd22bb7aa52 | 58 | * \image html MatrixInverse.gif "Matrix Inverse of a 3 x 3 matrix using Gauss-Jordan Method" |
emilmont | 1:fdd22bb7aa52 | 59 | */ |
emilmont | 1:fdd22bb7aa52 | 60 | |
emilmont | 1:fdd22bb7aa52 | 61 | /** |
emilmont | 1:fdd22bb7aa52 | 62 | * @addtogroup MatrixInv |
emilmont | 1:fdd22bb7aa52 | 63 | * @{ |
emilmont | 1:fdd22bb7aa52 | 64 | */ |
emilmont | 1:fdd22bb7aa52 | 65 | |
emilmont | 1:fdd22bb7aa52 | 66 | /** |
emilmont | 1:fdd22bb7aa52 | 67 | * @brief Floating-point matrix inverse. |
emilmont | 1:fdd22bb7aa52 | 68 | * @param[in] *pSrc points to input matrix structure |
emilmont | 1:fdd22bb7aa52 | 69 | * @param[out] *pDst points to output matrix structure |
emilmont | 1:fdd22bb7aa52 | 70 | * @return The function returns |
emilmont | 1:fdd22bb7aa52 | 71 | * <code>ARM_MATH_SIZE_MISMATCH</code> if the input matrix is not square or if the size |
emilmont | 1:fdd22bb7aa52 | 72 | * of the output matrix does not match the size of the input matrix. |
emilmont | 1:fdd22bb7aa52 | 73 | * If the input matrix is found to be singular (non-invertible), then the function returns |
emilmont | 1:fdd22bb7aa52 | 74 | * <code>ARM_MATH_SINGULAR</code>. Otherwise, the function returns <code>ARM_MATH_SUCCESS</code>. |
emilmont | 1:fdd22bb7aa52 | 75 | */ |
emilmont | 1:fdd22bb7aa52 | 76 | |
emilmont | 1:fdd22bb7aa52 | 77 | arm_status arm_mat_inverse_f32( |
emilmont | 1:fdd22bb7aa52 | 78 | const arm_matrix_instance_f32 * pSrc, |
emilmont | 1:fdd22bb7aa52 | 79 | arm_matrix_instance_f32 * pDst) |
emilmont | 1:fdd22bb7aa52 | 80 | { |
emilmont | 1:fdd22bb7aa52 | 81 | float32_t *pIn = pSrc->pData; /* input data matrix pointer */ |
emilmont | 1:fdd22bb7aa52 | 82 | float32_t *pOut = pDst->pData; /* output data matrix pointer */ |
emilmont | 1:fdd22bb7aa52 | 83 | float32_t *pInT1, *pInT2; /* Temporary input data matrix pointer */ |
emilmont | 1:fdd22bb7aa52 | 84 | float32_t *pInT3, *pInT4; /* Temporary output data matrix pointer */ |
emilmont | 1:fdd22bb7aa52 | 85 | float32_t *pPivotRowIn, *pPRT_in, *pPivotRowDst, *pPRT_pDst; /* Temporary input and output data matrix pointer */ |
emilmont | 1:fdd22bb7aa52 | 86 | uint32_t numRows = pSrc->numRows; /* Number of rows in the matrix */ |
emilmont | 1:fdd22bb7aa52 | 87 | uint32_t numCols = pSrc->numCols; /* Number of Cols in the matrix */ |
emilmont | 1:fdd22bb7aa52 | 88 | |
emilmont | 1:fdd22bb7aa52 | 89 | #ifndef ARM_MATH_CM0 |
emilmont | 1:fdd22bb7aa52 | 90 | |
emilmont | 1:fdd22bb7aa52 | 91 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
emilmont | 1:fdd22bb7aa52 | 92 | |
emilmont | 1:fdd22bb7aa52 | 93 | float32_t Xchg, in = 0.0f, in1; /* Temporary input values */ |
emilmont | 1:fdd22bb7aa52 | 94 | uint32_t i, rowCnt, flag = 0u, j, loopCnt, k, l; /* loop counters */ |
emilmont | 1:fdd22bb7aa52 | 95 | arm_status status; /* status of matrix inverse */ |
emilmont | 1:fdd22bb7aa52 | 96 | |
emilmont | 1:fdd22bb7aa52 | 97 | #ifdef ARM_MATH_MATRIX_CHECK |
emilmont | 1:fdd22bb7aa52 | 98 | |
emilmont | 1:fdd22bb7aa52 | 99 | |
emilmont | 1:fdd22bb7aa52 | 100 | /* Check for matrix mismatch condition */ |
emilmont | 1:fdd22bb7aa52 | 101 | if((pSrc->numRows != pSrc->numCols) || (pDst->numRows != pDst->numCols) |
emilmont | 1:fdd22bb7aa52 | 102 | || (pSrc->numRows != pDst->numRows)) |
emilmont | 1:fdd22bb7aa52 | 103 | { |
emilmont | 1:fdd22bb7aa52 | 104 | /* Set status as ARM_MATH_SIZE_MISMATCH */ |
emilmont | 1:fdd22bb7aa52 | 105 | status = ARM_MATH_SIZE_MISMATCH; |
emilmont | 1:fdd22bb7aa52 | 106 | } |
emilmont | 1:fdd22bb7aa52 | 107 | else |
emilmont | 1:fdd22bb7aa52 | 108 | #endif /* #ifdef ARM_MATH_MATRIX_CHECK */ |
emilmont | 1:fdd22bb7aa52 | 109 | |
emilmont | 1:fdd22bb7aa52 | 110 | { |
emilmont | 1:fdd22bb7aa52 | 111 | |
emilmont | 1:fdd22bb7aa52 | 112 | /*-------------------------------------------------------------------------------------------------------------- |
emilmont | 1:fdd22bb7aa52 | 113 | * Matrix Inverse can be solved using elementary row operations. |
emilmont | 1:fdd22bb7aa52 | 114 | * |
emilmont | 1:fdd22bb7aa52 | 115 | * Gauss-Jordan Method: |
emilmont | 1:fdd22bb7aa52 | 116 | * |
emilmont | 1:fdd22bb7aa52 | 117 | * 1. First combine the identity matrix and the input matrix separated by a bar to form an |
emilmont | 1:fdd22bb7aa52 | 118 | * augmented matrix as follows: |
emilmont | 1:fdd22bb7aa52 | 119 | * _ _ _ _ |
emilmont | 1:fdd22bb7aa52 | 120 | * | a11 a12 | 1 0 | | X11 X12 | |
emilmont | 1:fdd22bb7aa52 | 121 | * | | | = | | |
emilmont | 1:fdd22bb7aa52 | 122 | * |_ a21 a22 | 0 1 _| |_ X21 X21 _| |
emilmont | 1:fdd22bb7aa52 | 123 | * |
emilmont | 1:fdd22bb7aa52 | 124 | * 2. In our implementation, pDst Matrix is used as identity matrix. |
emilmont | 1:fdd22bb7aa52 | 125 | * |
emilmont | 1:fdd22bb7aa52 | 126 | * 3. Begin with the first row. Let i = 1. |
emilmont | 1:fdd22bb7aa52 | 127 | * |
emilmont | 1:fdd22bb7aa52 | 128 | * 4. Check to see if the pivot for row i is zero. |
emilmont | 1:fdd22bb7aa52 | 129 | * The pivot is the element of the main diagonal that is on the current row. |
emilmont | 1:fdd22bb7aa52 | 130 | * For instance, if working with row i, then the pivot element is aii. |
emilmont | 1:fdd22bb7aa52 | 131 | * If the pivot is zero, exchange that row with a row below it that does not |
emilmont | 1:fdd22bb7aa52 | 132 | * contain a zero in column i. If this is not possible, then an inverse |
emilmont | 1:fdd22bb7aa52 | 133 | * to that matrix does not exist. |
emilmont | 1:fdd22bb7aa52 | 134 | * |
emilmont | 1:fdd22bb7aa52 | 135 | * 5. Divide every element of row i by the pivot. |
emilmont | 1:fdd22bb7aa52 | 136 | * |
emilmont | 1:fdd22bb7aa52 | 137 | * 6. For every row below and row i, replace that row with the sum of that row and |
emilmont | 1:fdd22bb7aa52 | 138 | * a multiple of row i so that each new element in column i below row i is zero. |
emilmont | 1:fdd22bb7aa52 | 139 | * |
emilmont | 1:fdd22bb7aa52 | 140 | * 7. Move to the next row and column and repeat steps 2 through 5 until you have zeros |
emilmont | 1:fdd22bb7aa52 | 141 | * for every element below and above the main diagonal. |
emilmont | 1:fdd22bb7aa52 | 142 | * |
emilmont | 1:fdd22bb7aa52 | 143 | * 8. Now an identical matrix is formed to the left of the bar(input matrix, pSrc). |
emilmont | 1:fdd22bb7aa52 | 144 | * Therefore, the matrix to the right of the bar is our solution(pDst matrix, pDst). |
emilmont | 1:fdd22bb7aa52 | 145 | *----------------------------------------------------------------------------------------------------------------*/ |
emilmont | 1:fdd22bb7aa52 | 146 | |
emilmont | 1:fdd22bb7aa52 | 147 | /* Working pointer for destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 148 | pInT2 = pOut; |
emilmont | 1:fdd22bb7aa52 | 149 | |
emilmont | 1:fdd22bb7aa52 | 150 | /* Loop over the number of rows */ |
emilmont | 1:fdd22bb7aa52 | 151 | rowCnt = numRows; |
emilmont | 1:fdd22bb7aa52 | 152 | |
emilmont | 1:fdd22bb7aa52 | 153 | /* Making the destination matrix as identity matrix */ |
emilmont | 1:fdd22bb7aa52 | 154 | while(rowCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 155 | { |
emilmont | 1:fdd22bb7aa52 | 156 | /* Writing all zeroes in lower triangle of the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 157 | j = numRows - rowCnt; |
emilmont | 1:fdd22bb7aa52 | 158 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 159 | { |
emilmont | 1:fdd22bb7aa52 | 160 | *pInT2++ = 0.0f; |
emilmont | 1:fdd22bb7aa52 | 161 | j--; |
emilmont | 1:fdd22bb7aa52 | 162 | } |
emilmont | 1:fdd22bb7aa52 | 163 | |
emilmont | 1:fdd22bb7aa52 | 164 | /* Writing all ones in the diagonal of the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 165 | *pInT2++ = 1.0f; |
emilmont | 1:fdd22bb7aa52 | 166 | |
emilmont | 1:fdd22bb7aa52 | 167 | /* Writing all zeroes in upper triangle of the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 168 | j = rowCnt - 1u; |
emilmont | 1:fdd22bb7aa52 | 169 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 170 | { |
emilmont | 1:fdd22bb7aa52 | 171 | *pInT2++ = 0.0f; |
emilmont | 1:fdd22bb7aa52 | 172 | j--; |
emilmont | 1:fdd22bb7aa52 | 173 | } |
emilmont | 1:fdd22bb7aa52 | 174 | |
emilmont | 1:fdd22bb7aa52 | 175 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 176 | rowCnt--; |
emilmont | 1:fdd22bb7aa52 | 177 | } |
emilmont | 1:fdd22bb7aa52 | 178 | |
emilmont | 1:fdd22bb7aa52 | 179 | /* Loop over the number of columns of the input matrix. |
emilmont | 1:fdd22bb7aa52 | 180 | All the elements in each column are processed by the row operations */ |
emilmont | 1:fdd22bb7aa52 | 181 | loopCnt = numCols; |
emilmont | 1:fdd22bb7aa52 | 182 | |
emilmont | 1:fdd22bb7aa52 | 183 | /* Index modifier to navigate through the columns */ |
emilmont | 1:fdd22bb7aa52 | 184 | l = 0u; |
emilmont | 1:fdd22bb7aa52 | 185 | |
emilmont | 1:fdd22bb7aa52 | 186 | while(loopCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 187 | { |
emilmont | 1:fdd22bb7aa52 | 188 | /* Check if the pivot element is zero.. |
emilmont | 1:fdd22bb7aa52 | 189 | * If it is zero then interchange the row with non zero row below. |
emilmont | 1:fdd22bb7aa52 | 190 | * If there is no non zero element to replace in the rows below, |
emilmont | 1:fdd22bb7aa52 | 191 | * then the matrix is Singular. */ |
emilmont | 1:fdd22bb7aa52 | 192 | |
emilmont | 1:fdd22bb7aa52 | 193 | /* Working pointer for the input matrix that points |
emilmont | 1:fdd22bb7aa52 | 194 | * to the pivot element of the particular row */ |
emilmont | 1:fdd22bb7aa52 | 195 | pInT1 = pIn + (l * numCols); |
emilmont | 1:fdd22bb7aa52 | 196 | |
emilmont | 1:fdd22bb7aa52 | 197 | /* Working pointer for the destination matrix that points |
emilmont | 1:fdd22bb7aa52 | 198 | * to the pivot element of the particular row */ |
emilmont | 1:fdd22bb7aa52 | 199 | pInT3 = pOut + (l * numCols); |
emilmont | 1:fdd22bb7aa52 | 200 | |
emilmont | 1:fdd22bb7aa52 | 201 | /* Temporary variable to hold the pivot value */ |
emilmont | 1:fdd22bb7aa52 | 202 | in = *pInT1; |
emilmont | 1:fdd22bb7aa52 | 203 | |
emilmont | 1:fdd22bb7aa52 | 204 | /* Destination pointer modifier */ |
emilmont | 1:fdd22bb7aa52 | 205 | k = 1u; |
emilmont | 1:fdd22bb7aa52 | 206 | |
emilmont | 1:fdd22bb7aa52 | 207 | /* Check if the pivot element is zero */ |
emilmont | 1:fdd22bb7aa52 | 208 | if(*pInT1 == 0.0f) |
emilmont | 1:fdd22bb7aa52 | 209 | { |
emilmont | 1:fdd22bb7aa52 | 210 | /* Loop over the number rows present below */ |
emilmont | 1:fdd22bb7aa52 | 211 | i = numRows - (l + 1u); |
emilmont | 1:fdd22bb7aa52 | 212 | |
emilmont | 1:fdd22bb7aa52 | 213 | while(i > 0u) |
emilmont | 1:fdd22bb7aa52 | 214 | { |
emilmont | 1:fdd22bb7aa52 | 215 | /* Update the input and destination pointers */ |
emilmont | 1:fdd22bb7aa52 | 216 | pInT2 = pInT1 + (numCols * l); |
emilmont | 1:fdd22bb7aa52 | 217 | pInT4 = pInT3 + (numCols * k); |
emilmont | 1:fdd22bb7aa52 | 218 | |
emilmont | 1:fdd22bb7aa52 | 219 | /* Check if there is a non zero pivot element to |
emilmont | 1:fdd22bb7aa52 | 220 | * replace in the rows below */ |
emilmont | 1:fdd22bb7aa52 | 221 | if(*pInT2 != 0.0f) |
emilmont | 1:fdd22bb7aa52 | 222 | { |
emilmont | 1:fdd22bb7aa52 | 223 | /* Loop over number of columns |
emilmont | 1:fdd22bb7aa52 | 224 | * to the right of the pilot element */ |
emilmont | 1:fdd22bb7aa52 | 225 | j = numCols - l; |
emilmont | 1:fdd22bb7aa52 | 226 | |
emilmont | 1:fdd22bb7aa52 | 227 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 228 | { |
emilmont | 1:fdd22bb7aa52 | 229 | /* Exchange the row elements of the input matrix */ |
emilmont | 1:fdd22bb7aa52 | 230 | Xchg = *pInT2; |
emilmont | 1:fdd22bb7aa52 | 231 | *pInT2++ = *pInT1; |
emilmont | 1:fdd22bb7aa52 | 232 | *pInT1++ = Xchg; |
emilmont | 1:fdd22bb7aa52 | 233 | |
emilmont | 1:fdd22bb7aa52 | 234 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 235 | j--; |
emilmont | 1:fdd22bb7aa52 | 236 | } |
emilmont | 1:fdd22bb7aa52 | 237 | |
emilmont | 1:fdd22bb7aa52 | 238 | /* Loop over number of columns of the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 239 | j = numCols; |
emilmont | 1:fdd22bb7aa52 | 240 | |
emilmont | 1:fdd22bb7aa52 | 241 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 242 | { |
emilmont | 1:fdd22bb7aa52 | 243 | /* Exchange the row elements of the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 244 | Xchg = *pInT4; |
emilmont | 1:fdd22bb7aa52 | 245 | *pInT4++ = *pInT3; |
emilmont | 1:fdd22bb7aa52 | 246 | *pInT3++ = Xchg; |
emilmont | 1:fdd22bb7aa52 | 247 | |
emilmont | 1:fdd22bb7aa52 | 248 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 249 | j--; |
emilmont | 1:fdd22bb7aa52 | 250 | } |
emilmont | 1:fdd22bb7aa52 | 251 | |
emilmont | 1:fdd22bb7aa52 | 252 | /* Flag to indicate whether exchange is done or not */ |
emilmont | 1:fdd22bb7aa52 | 253 | flag = 1u; |
emilmont | 1:fdd22bb7aa52 | 254 | |
emilmont | 1:fdd22bb7aa52 | 255 | /* Break after exchange is done */ |
emilmont | 1:fdd22bb7aa52 | 256 | break; |
emilmont | 1:fdd22bb7aa52 | 257 | } |
emilmont | 1:fdd22bb7aa52 | 258 | |
emilmont | 1:fdd22bb7aa52 | 259 | /* Update the destination pointer modifier */ |
emilmont | 1:fdd22bb7aa52 | 260 | k++; |
emilmont | 1:fdd22bb7aa52 | 261 | |
emilmont | 1:fdd22bb7aa52 | 262 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 263 | i--; |
emilmont | 1:fdd22bb7aa52 | 264 | } |
emilmont | 1:fdd22bb7aa52 | 265 | } |
emilmont | 1:fdd22bb7aa52 | 266 | |
emilmont | 1:fdd22bb7aa52 | 267 | /* Update the status if the matrix is singular */ |
emilmont | 1:fdd22bb7aa52 | 268 | if((flag != 1u) && (in == 0.0f)) |
emilmont | 1:fdd22bb7aa52 | 269 | { |
emilmont | 1:fdd22bb7aa52 | 270 | status = ARM_MATH_SINGULAR; |
emilmont | 1:fdd22bb7aa52 | 271 | |
emilmont | 1:fdd22bb7aa52 | 272 | break; |
emilmont | 1:fdd22bb7aa52 | 273 | } |
emilmont | 1:fdd22bb7aa52 | 274 | |
emilmont | 1:fdd22bb7aa52 | 275 | /* Points to the pivot row of input and destination matrices */ |
emilmont | 1:fdd22bb7aa52 | 276 | pPivotRowIn = pIn + (l * numCols); |
emilmont | 1:fdd22bb7aa52 | 277 | pPivotRowDst = pOut + (l * numCols); |
emilmont | 1:fdd22bb7aa52 | 278 | |
emilmont | 1:fdd22bb7aa52 | 279 | /* Temporary pointers to the pivot row pointers */ |
emilmont | 1:fdd22bb7aa52 | 280 | pInT1 = pPivotRowIn; |
emilmont | 1:fdd22bb7aa52 | 281 | pInT2 = pPivotRowDst; |
emilmont | 1:fdd22bb7aa52 | 282 | |
emilmont | 1:fdd22bb7aa52 | 283 | /* Pivot element of the row */ |
emilmont | 1:fdd22bb7aa52 | 284 | in = *(pIn + (l * numCols)); |
emilmont | 1:fdd22bb7aa52 | 285 | |
emilmont | 1:fdd22bb7aa52 | 286 | /* Loop over number of columns |
emilmont | 1:fdd22bb7aa52 | 287 | * to the right of the pilot element */ |
emilmont | 1:fdd22bb7aa52 | 288 | j = (numCols - l); |
emilmont | 1:fdd22bb7aa52 | 289 | |
emilmont | 1:fdd22bb7aa52 | 290 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 291 | { |
emilmont | 1:fdd22bb7aa52 | 292 | /* Divide each element of the row of the input matrix |
emilmont | 1:fdd22bb7aa52 | 293 | * by the pivot element */ |
emilmont | 1:fdd22bb7aa52 | 294 | in1 = *pInT1; |
emilmont | 1:fdd22bb7aa52 | 295 | *pInT1++ = in1 / in; |
emilmont | 1:fdd22bb7aa52 | 296 | |
emilmont | 1:fdd22bb7aa52 | 297 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 298 | j--; |
emilmont | 1:fdd22bb7aa52 | 299 | } |
emilmont | 1:fdd22bb7aa52 | 300 | |
emilmont | 1:fdd22bb7aa52 | 301 | /* Loop over number of columns of the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 302 | j = numCols; |
emilmont | 1:fdd22bb7aa52 | 303 | |
emilmont | 1:fdd22bb7aa52 | 304 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 305 | { |
emilmont | 1:fdd22bb7aa52 | 306 | /* Divide each element of the row of the destination matrix |
emilmont | 1:fdd22bb7aa52 | 307 | * by the pivot element */ |
emilmont | 1:fdd22bb7aa52 | 308 | in1 = *pInT2; |
emilmont | 1:fdd22bb7aa52 | 309 | *pInT2++ = in1 / in; |
emilmont | 1:fdd22bb7aa52 | 310 | |
emilmont | 1:fdd22bb7aa52 | 311 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 312 | j--; |
emilmont | 1:fdd22bb7aa52 | 313 | } |
emilmont | 1:fdd22bb7aa52 | 314 | |
emilmont | 1:fdd22bb7aa52 | 315 | /* Replace the rows with the sum of that row and a multiple of row i |
emilmont | 1:fdd22bb7aa52 | 316 | * so that each new element in column i above row i is zero.*/ |
emilmont | 1:fdd22bb7aa52 | 317 | |
emilmont | 1:fdd22bb7aa52 | 318 | /* Temporary pointers for input and destination matrices */ |
emilmont | 1:fdd22bb7aa52 | 319 | pInT1 = pIn; |
emilmont | 1:fdd22bb7aa52 | 320 | pInT2 = pOut; |
emilmont | 1:fdd22bb7aa52 | 321 | |
emilmont | 1:fdd22bb7aa52 | 322 | /* index used to check for pivot element */ |
emilmont | 1:fdd22bb7aa52 | 323 | i = 0u; |
emilmont | 1:fdd22bb7aa52 | 324 | |
emilmont | 1:fdd22bb7aa52 | 325 | /* Loop over number of rows */ |
emilmont | 1:fdd22bb7aa52 | 326 | /* to be replaced by the sum of that row and a multiple of row i */ |
emilmont | 1:fdd22bb7aa52 | 327 | k = numRows; |
emilmont | 1:fdd22bb7aa52 | 328 | |
emilmont | 1:fdd22bb7aa52 | 329 | while(k > 0u) |
emilmont | 1:fdd22bb7aa52 | 330 | { |
emilmont | 1:fdd22bb7aa52 | 331 | /* Check for the pivot element */ |
emilmont | 1:fdd22bb7aa52 | 332 | if(i == l) |
emilmont | 1:fdd22bb7aa52 | 333 | { |
emilmont | 1:fdd22bb7aa52 | 334 | /* If the processing element is the pivot element, |
emilmont | 1:fdd22bb7aa52 | 335 | only the columns to the right are to be processed */ |
emilmont | 1:fdd22bb7aa52 | 336 | pInT1 += numCols - l; |
emilmont | 1:fdd22bb7aa52 | 337 | |
emilmont | 1:fdd22bb7aa52 | 338 | pInT2 += numCols; |
emilmont | 1:fdd22bb7aa52 | 339 | } |
emilmont | 1:fdd22bb7aa52 | 340 | else |
emilmont | 1:fdd22bb7aa52 | 341 | { |
emilmont | 1:fdd22bb7aa52 | 342 | /* Element of the reference row */ |
emilmont | 1:fdd22bb7aa52 | 343 | in = *pInT1; |
emilmont | 1:fdd22bb7aa52 | 344 | |
emilmont | 1:fdd22bb7aa52 | 345 | /* Working pointers for input and destination pivot rows */ |
emilmont | 1:fdd22bb7aa52 | 346 | pPRT_in = pPivotRowIn; |
emilmont | 1:fdd22bb7aa52 | 347 | pPRT_pDst = pPivotRowDst; |
emilmont | 1:fdd22bb7aa52 | 348 | |
emilmont | 1:fdd22bb7aa52 | 349 | /* Loop over the number of columns to the right of the pivot element, |
emilmont | 1:fdd22bb7aa52 | 350 | to replace the elements in the input matrix */ |
emilmont | 1:fdd22bb7aa52 | 351 | j = (numCols - l); |
emilmont | 1:fdd22bb7aa52 | 352 | |
emilmont | 1:fdd22bb7aa52 | 353 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 354 | { |
emilmont | 1:fdd22bb7aa52 | 355 | /* Replace the element by the sum of that row |
emilmont | 1:fdd22bb7aa52 | 356 | and a multiple of the reference row */ |
emilmont | 1:fdd22bb7aa52 | 357 | in1 = *pInT1; |
emilmont | 1:fdd22bb7aa52 | 358 | *pInT1++ = in1 - (in * *pPRT_in++); |
emilmont | 1:fdd22bb7aa52 | 359 | |
emilmont | 1:fdd22bb7aa52 | 360 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 361 | j--; |
emilmont | 1:fdd22bb7aa52 | 362 | } |
emilmont | 1:fdd22bb7aa52 | 363 | |
emilmont | 1:fdd22bb7aa52 | 364 | /* Loop over the number of columns to |
emilmont | 1:fdd22bb7aa52 | 365 | replace the elements in the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 366 | j = numCols; |
emilmont | 1:fdd22bb7aa52 | 367 | |
emilmont | 1:fdd22bb7aa52 | 368 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 369 | { |
emilmont | 1:fdd22bb7aa52 | 370 | /* Replace the element by the sum of that row |
emilmont | 1:fdd22bb7aa52 | 371 | and a multiple of the reference row */ |
emilmont | 1:fdd22bb7aa52 | 372 | in1 = *pInT2; |
emilmont | 1:fdd22bb7aa52 | 373 | *pInT2++ = in1 - (in * *pPRT_pDst++); |
emilmont | 1:fdd22bb7aa52 | 374 | |
emilmont | 1:fdd22bb7aa52 | 375 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 376 | j--; |
emilmont | 1:fdd22bb7aa52 | 377 | } |
emilmont | 1:fdd22bb7aa52 | 378 | |
emilmont | 1:fdd22bb7aa52 | 379 | } |
emilmont | 1:fdd22bb7aa52 | 380 | |
emilmont | 1:fdd22bb7aa52 | 381 | /* Increment the temporary input pointer */ |
emilmont | 1:fdd22bb7aa52 | 382 | pInT1 = pInT1 + l; |
emilmont | 1:fdd22bb7aa52 | 383 | |
emilmont | 1:fdd22bb7aa52 | 384 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 385 | k--; |
emilmont | 1:fdd22bb7aa52 | 386 | |
emilmont | 1:fdd22bb7aa52 | 387 | /* Increment the pivot index */ |
emilmont | 1:fdd22bb7aa52 | 388 | i++; |
emilmont | 1:fdd22bb7aa52 | 389 | } |
emilmont | 1:fdd22bb7aa52 | 390 | |
emilmont | 1:fdd22bb7aa52 | 391 | /* Increment the input pointer */ |
emilmont | 1:fdd22bb7aa52 | 392 | pIn++; |
emilmont | 1:fdd22bb7aa52 | 393 | |
emilmont | 1:fdd22bb7aa52 | 394 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 395 | loopCnt--; |
emilmont | 1:fdd22bb7aa52 | 396 | |
emilmont | 1:fdd22bb7aa52 | 397 | /* Increment the index modifier */ |
emilmont | 1:fdd22bb7aa52 | 398 | l++; |
emilmont | 1:fdd22bb7aa52 | 399 | } |
emilmont | 1:fdd22bb7aa52 | 400 | |
emilmont | 1:fdd22bb7aa52 | 401 | |
emilmont | 1:fdd22bb7aa52 | 402 | #else |
emilmont | 1:fdd22bb7aa52 | 403 | |
emilmont | 1:fdd22bb7aa52 | 404 | /* Run the below code for Cortex-M0 */ |
emilmont | 1:fdd22bb7aa52 | 405 | |
emilmont | 1:fdd22bb7aa52 | 406 | float32_t Xchg, in = 0.0f; /* Temporary input values */ |
emilmont | 1:fdd22bb7aa52 | 407 | uint32_t i, rowCnt, flag = 0u, j, loopCnt, k, l; /* loop counters */ |
emilmont | 1:fdd22bb7aa52 | 408 | arm_status status; /* status of matrix inverse */ |
emilmont | 1:fdd22bb7aa52 | 409 | |
emilmont | 1:fdd22bb7aa52 | 410 | #ifdef ARM_MATH_MATRIX_CHECK |
emilmont | 1:fdd22bb7aa52 | 411 | |
emilmont | 1:fdd22bb7aa52 | 412 | /* Check for matrix mismatch condition */ |
emilmont | 1:fdd22bb7aa52 | 413 | if((pSrc->numRows != pSrc->numCols) || (pDst->numRows != pDst->numCols) |
emilmont | 1:fdd22bb7aa52 | 414 | || (pSrc->numRows != pDst->numRows)) |
emilmont | 1:fdd22bb7aa52 | 415 | { |
emilmont | 1:fdd22bb7aa52 | 416 | /* Set status as ARM_MATH_SIZE_MISMATCH */ |
emilmont | 1:fdd22bb7aa52 | 417 | status = ARM_MATH_SIZE_MISMATCH; |
emilmont | 1:fdd22bb7aa52 | 418 | } |
emilmont | 1:fdd22bb7aa52 | 419 | else |
emilmont | 1:fdd22bb7aa52 | 420 | #endif /* #ifdef ARM_MATH_MATRIX_CHECK */ |
emilmont | 1:fdd22bb7aa52 | 421 | { |
emilmont | 1:fdd22bb7aa52 | 422 | |
emilmont | 1:fdd22bb7aa52 | 423 | /*-------------------------------------------------------------------------------------------------------------- |
emilmont | 1:fdd22bb7aa52 | 424 | * Matrix Inverse can be solved using elementary row operations. |
emilmont | 1:fdd22bb7aa52 | 425 | * |
emilmont | 1:fdd22bb7aa52 | 426 | * Gauss-Jordan Method: |
emilmont | 1:fdd22bb7aa52 | 427 | * |
emilmont | 1:fdd22bb7aa52 | 428 | * 1. First combine the identity matrix and the input matrix separated by a bar to form an |
emilmont | 1:fdd22bb7aa52 | 429 | * augmented matrix as follows: |
emilmont | 1:fdd22bb7aa52 | 430 | * _ _ _ _ _ _ _ _ |
emilmont | 1:fdd22bb7aa52 | 431 | * | | a11 a12 | | | 1 0 | | | X11 X12 | |
emilmont | 1:fdd22bb7aa52 | 432 | * | | | | | | | = | | |
emilmont | 1:fdd22bb7aa52 | 433 | * |_ |_ a21 a22 _| | |_0 1 _| _| |_ X21 X21 _| |
emilmont | 1:fdd22bb7aa52 | 434 | * |
emilmont | 1:fdd22bb7aa52 | 435 | * 2. In our implementation, pDst Matrix is used as identity matrix. |
emilmont | 1:fdd22bb7aa52 | 436 | * |
emilmont | 1:fdd22bb7aa52 | 437 | * 3. Begin with the first row. Let i = 1. |
emilmont | 1:fdd22bb7aa52 | 438 | * |
emilmont | 1:fdd22bb7aa52 | 439 | * 4. Check to see if the pivot for row i is zero. |
emilmont | 1:fdd22bb7aa52 | 440 | * The pivot is the element of the main diagonal that is on the current row. |
emilmont | 1:fdd22bb7aa52 | 441 | * For instance, if working with row i, then the pivot element is aii. |
emilmont | 1:fdd22bb7aa52 | 442 | * If the pivot is zero, exchange that row with a row below it that does not |
emilmont | 1:fdd22bb7aa52 | 443 | * contain a zero in column i. If this is not possible, then an inverse |
emilmont | 1:fdd22bb7aa52 | 444 | * to that matrix does not exist. |
emilmont | 1:fdd22bb7aa52 | 445 | * |
emilmont | 1:fdd22bb7aa52 | 446 | * 5. Divide every element of row i by the pivot. |
emilmont | 1:fdd22bb7aa52 | 447 | * |
emilmont | 1:fdd22bb7aa52 | 448 | * 6. For every row below and row i, replace that row with the sum of that row and |
emilmont | 1:fdd22bb7aa52 | 449 | * a multiple of row i so that each new element in column i below row i is zero. |
emilmont | 1:fdd22bb7aa52 | 450 | * |
emilmont | 1:fdd22bb7aa52 | 451 | * 7. Move to the next row and column and repeat steps 2 through 5 until you have zeros |
emilmont | 1:fdd22bb7aa52 | 452 | * for every element below and above the main diagonal. |
emilmont | 1:fdd22bb7aa52 | 453 | * |
emilmont | 1:fdd22bb7aa52 | 454 | * 8. Now an identical matrix is formed to the left of the bar(input matrix, src). |
emilmont | 1:fdd22bb7aa52 | 455 | * Therefore, the matrix to the right of the bar is our solution(dst matrix, dst). |
emilmont | 1:fdd22bb7aa52 | 456 | *----------------------------------------------------------------------------------------------------------------*/ |
emilmont | 1:fdd22bb7aa52 | 457 | |
emilmont | 1:fdd22bb7aa52 | 458 | /* Working pointer for destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 459 | pInT2 = pOut; |
emilmont | 1:fdd22bb7aa52 | 460 | |
emilmont | 1:fdd22bb7aa52 | 461 | /* Loop over the number of rows */ |
emilmont | 1:fdd22bb7aa52 | 462 | rowCnt = numRows; |
emilmont | 1:fdd22bb7aa52 | 463 | |
emilmont | 1:fdd22bb7aa52 | 464 | /* Making the destination matrix as identity matrix */ |
emilmont | 1:fdd22bb7aa52 | 465 | while(rowCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 466 | { |
emilmont | 1:fdd22bb7aa52 | 467 | /* Writing all zeroes in lower triangle of the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 468 | j = numRows - rowCnt; |
emilmont | 1:fdd22bb7aa52 | 469 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 470 | { |
emilmont | 1:fdd22bb7aa52 | 471 | *pInT2++ = 0.0f; |
emilmont | 1:fdd22bb7aa52 | 472 | j--; |
emilmont | 1:fdd22bb7aa52 | 473 | } |
emilmont | 1:fdd22bb7aa52 | 474 | |
emilmont | 1:fdd22bb7aa52 | 475 | /* Writing all ones in the diagonal of the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 476 | *pInT2++ = 1.0f; |
emilmont | 1:fdd22bb7aa52 | 477 | |
emilmont | 1:fdd22bb7aa52 | 478 | /* Writing all zeroes in upper triangle of the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 479 | j = rowCnt - 1u; |
emilmont | 1:fdd22bb7aa52 | 480 | while(j > 0u) |
emilmont | 1:fdd22bb7aa52 | 481 | { |
emilmont | 1:fdd22bb7aa52 | 482 | *pInT2++ = 0.0f; |
emilmont | 1:fdd22bb7aa52 | 483 | j--; |
emilmont | 1:fdd22bb7aa52 | 484 | } |
emilmont | 1:fdd22bb7aa52 | 485 | |
emilmont | 1:fdd22bb7aa52 | 486 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 487 | rowCnt--; |
emilmont | 1:fdd22bb7aa52 | 488 | } |
emilmont | 1:fdd22bb7aa52 | 489 | |
emilmont | 1:fdd22bb7aa52 | 490 | /* Loop over the number of columns of the input matrix. |
emilmont | 1:fdd22bb7aa52 | 491 | All the elements in each column are processed by the row operations */ |
emilmont | 1:fdd22bb7aa52 | 492 | loopCnt = numCols; |
emilmont | 1:fdd22bb7aa52 | 493 | |
emilmont | 1:fdd22bb7aa52 | 494 | /* Index modifier to navigate through the columns */ |
emilmont | 1:fdd22bb7aa52 | 495 | l = 0u; |
emilmont | 1:fdd22bb7aa52 | 496 | //for(loopCnt = 0u; loopCnt < numCols; loopCnt++) |
emilmont | 1:fdd22bb7aa52 | 497 | while(loopCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 498 | { |
emilmont | 1:fdd22bb7aa52 | 499 | /* Check if the pivot element is zero.. |
emilmont | 1:fdd22bb7aa52 | 500 | * If it is zero then interchange the row with non zero row below. |
emilmont | 1:fdd22bb7aa52 | 501 | * If there is no non zero element to replace in the rows below, |
emilmont | 1:fdd22bb7aa52 | 502 | * then the matrix is Singular. */ |
emilmont | 1:fdd22bb7aa52 | 503 | |
emilmont | 1:fdd22bb7aa52 | 504 | /* Working pointer for the input matrix that points |
emilmont | 1:fdd22bb7aa52 | 505 | * to the pivot element of the particular row */ |
emilmont | 1:fdd22bb7aa52 | 506 | pInT1 = pIn + (l * numCols); |
emilmont | 1:fdd22bb7aa52 | 507 | |
emilmont | 1:fdd22bb7aa52 | 508 | /* Working pointer for the destination matrix that points |
emilmont | 1:fdd22bb7aa52 | 509 | * to the pivot element of the particular row */ |
emilmont | 1:fdd22bb7aa52 | 510 | pInT3 = pOut + (l * numCols); |
emilmont | 1:fdd22bb7aa52 | 511 | |
emilmont | 1:fdd22bb7aa52 | 512 | /* Temporary variable to hold the pivot value */ |
emilmont | 1:fdd22bb7aa52 | 513 | in = *pInT1; |
emilmont | 1:fdd22bb7aa52 | 514 | |
emilmont | 1:fdd22bb7aa52 | 515 | /* Destination pointer modifier */ |
emilmont | 1:fdd22bb7aa52 | 516 | k = 1u; |
emilmont | 1:fdd22bb7aa52 | 517 | |
emilmont | 1:fdd22bb7aa52 | 518 | /* Check if the pivot element is zero */ |
emilmont | 1:fdd22bb7aa52 | 519 | if(*pInT1 == 0.0f) |
emilmont | 1:fdd22bb7aa52 | 520 | { |
emilmont | 1:fdd22bb7aa52 | 521 | /* Loop over the number rows present below */ |
emilmont | 1:fdd22bb7aa52 | 522 | for (i = (l + 1u); i < numRows; i++) |
emilmont | 1:fdd22bb7aa52 | 523 | { |
emilmont | 1:fdd22bb7aa52 | 524 | /* Update the input and destination pointers */ |
emilmont | 1:fdd22bb7aa52 | 525 | pInT2 = pInT1 + (numCols * l); |
emilmont | 1:fdd22bb7aa52 | 526 | pInT4 = pInT3 + (numCols * k); |
emilmont | 1:fdd22bb7aa52 | 527 | |
emilmont | 1:fdd22bb7aa52 | 528 | /* Check if there is a non zero pivot element to |
emilmont | 1:fdd22bb7aa52 | 529 | * replace in the rows below */ |
emilmont | 1:fdd22bb7aa52 | 530 | if(*pInT2 != 0.0f) |
emilmont | 1:fdd22bb7aa52 | 531 | { |
emilmont | 1:fdd22bb7aa52 | 532 | /* Loop over number of columns |
emilmont | 1:fdd22bb7aa52 | 533 | * to the right of the pilot element */ |
emilmont | 1:fdd22bb7aa52 | 534 | for (j = 0u; j < (numCols - l); j++) |
emilmont | 1:fdd22bb7aa52 | 535 | { |
emilmont | 1:fdd22bb7aa52 | 536 | /* Exchange the row elements of the input matrix */ |
emilmont | 1:fdd22bb7aa52 | 537 | Xchg = *pInT2; |
emilmont | 1:fdd22bb7aa52 | 538 | *pInT2++ = *pInT1; |
emilmont | 1:fdd22bb7aa52 | 539 | *pInT1++ = Xchg; |
emilmont | 1:fdd22bb7aa52 | 540 | } |
emilmont | 1:fdd22bb7aa52 | 541 | |
emilmont | 1:fdd22bb7aa52 | 542 | for (j = 0u; j < numCols; j++) |
emilmont | 1:fdd22bb7aa52 | 543 | { |
emilmont | 1:fdd22bb7aa52 | 544 | Xchg = *pInT4; |
emilmont | 1:fdd22bb7aa52 | 545 | *pInT4++ = *pInT3; |
emilmont | 1:fdd22bb7aa52 | 546 | *pInT3++ = Xchg; |
emilmont | 1:fdd22bb7aa52 | 547 | } |
emilmont | 1:fdd22bb7aa52 | 548 | |
emilmont | 1:fdd22bb7aa52 | 549 | /* Flag to indicate whether exchange is done or not */ |
emilmont | 1:fdd22bb7aa52 | 550 | flag = 1u; |
emilmont | 1:fdd22bb7aa52 | 551 | |
emilmont | 1:fdd22bb7aa52 | 552 | /* Break after exchange is done */ |
emilmont | 1:fdd22bb7aa52 | 553 | break; |
emilmont | 1:fdd22bb7aa52 | 554 | } |
emilmont | 1:fdd22bb7aa52 | 555 | |
emilmont | 1:fdd22bb7aa52 | 556 | /* Update the destination pointer modifier */ |
emilmont | 1:fdd22bb7aa52 | 557 | k++; |
emilmont | 1:fdd22bb7aa52 | 558 | } |
emilmont | 1:fdd22bb7aa52 | 559 | } |
emilmont | 1:fdd22bb7aa52 | 560 | |
emilmont | 1:fdd22bb7aa52 | 561 | /* Update the status if the matrix is singular */ |
emilmont | 1:fdd22bb7aa52 | 562 | if((flag != 1u) && (in == 0.0f)) |
emilmont | 1:fdd22bb7aa52 | 563 | { |
emilmont | 1:fdd22bb7aa52 | 564 | status = ARM_MATH_SINGULAR; |
emilmont | 1:fdd22bb7aa52 | 565 | |
emilmont | 1:fdd22bb7aa52 | 566 | break; |
emilmont | 1:fdd22bb7aa52 | 567 | } |
emilmont | 1:fdd22bb7aa52 | 568 | |
emilmont | 1:fdd22bb7aa52 | 569 | /* Points to the pivot row of input and destination matrices */ |
emilmont | 1:fdd22bb7aa52 | 570 | pPivotRowIn = pIn + (l * numCols); |
emilmont | 1:fdd22bb7aa52 | 571 | pPivotRowDst = pOut + (l * numCols); |
emilmont | 1:fdd22bb7aa52 | 572 | |
emilmont | 1:fdd22bb7aa52 | 573 | /* Temporary pointers to the pivot row pointers */ |
emilmont | 1:fdd22bb7aa52 | 574 | pInT1 = pPivotRowIn; |
emilmont | 1:fdd22bb7aa52 | 575 | pInT2 = pPivotRowDst; |
emilmont | 1:fdd22bb7aa52 | 576 | |
emilmont | 1:fdd22bb7aa52 | 577 | /* Pivot element of the row */ |
emilmont | 1:fdd22bb7aa52 | 578 | in = *(pIn + (l * numCols)); |
emilmont | 1:fdd22bb7aa52 | 579 | |
emilmont | 1:fdd22bb7aa52 | 580 | /* Loop over number of columns |
emilmont | 1:fdd22bb7aa52 | 581 | * to the right of the pilot element */ |
emilmont | 1:fdd22bb7aa52 | 582 | for (j = 0u; j < (numCols - l); j++) |
emilmont | 1:fdd22bb7aa52 | 583 | { |
emilmont | 1:fdd22bb7aa52 | 584 | /* Divide each element of the row of the input matrix |
emilmont | 1:fdd22bb7aa52 | 585 | * by the pivot element */ |
emilmont | 1:fdd22bb7aa52 | 586 | *pInT1++ = *pInT1 / in; |
emilmont | 1:fdd22bb7aa52 | 587 | } |
emilmont | 1:fdd22bb7aa52 | 588 | for (j = 0u; j < numCols; j++) |
emilmont | 1:fdd22bb7aa52 | 589 | { |
emilmont | 1:fdd22bb7aa52 | 590 | /* Divide each element of the row of the destination matrix |
emilmont | 1:fdd22bb7aa52 | 591 | * by the pivot element */ |
emilmont | 1:fdd22bb7aa52 | 592 | *pInT2++ = *pInT2 / in; |
emilmont | 1:fdd22bb7aa52 | 593 | } |
emilmont | 1:fdd22bb7aa52 | 594 | |
emilmont | 1:fdd22bb7aa52 | 595 | /* Replace the rows with the sum of that row and a multiple of row i |
emilmont | 1:fdd22bb7aa52 | 596 | * so that each new element in column i above row i is zero.*/ |
emilmont | 1:fdd22bb7aa52 | 597 | |
emilmont | 1:fdd22bb7aa52 | 598 | /* Temporary pointers for input and destination matrices */ |
emilmont | 1:fdd22bb7aa52 | 599 | pInT1 = pIn; |
emilmont | 1:fdd22bb7aa52 | 600 | pInT2 = pOut; |
emilmont | 1:fdd22bb7aa52 | 601 | |
emilmont | 1:fdd22bb7aa52 | 602 | for (i = 0u; i < numRows; i++) |
emilmont | 1:fdd22bb7aa52 | 603 | { |
emilmont | 1:fdd22bb7aa52 | 604 | /* Check for the pivot element */ |
emilmont | 1:fdd22bb7aa52 | 605 | if(i == l) |
emilmont | 1:fdd22bb7aa52 | 606 | { |
emilmont | 1:fdd22bb7aa52 | 607 | /* If the processing element is the pivot element, |
emilmont | 1:fdd22bb7aa52 | 608 | only the columns to the right are to be processed */ |
emilmont | 1:fdd22bb7aa52 | 609 | pInT1 += numCols - l; |
emilmont | 1:fdd22bb7aa52 | 610 | pInT2 += numCols; |
emilmont | 1:fdd22bb7aa52 | 611 | } |
emilmont | 1:fdd22bb7aa52 | 612 | else |
emilmont | 1:fdd22bb7aa52 | 613 | { |
emilmont | 1:fdd22bb7aa52 | 614 | /* Element of the reference row */ |
emilmont | 1:fdd22bb7aa52 | 615 | in = *pInT1; |
emilmont | 1:fdd22bb7aa52 | 616 | |
emilmont | 1:fdd22bb7aa52 | 617 | /* Working pointers for input and destination pivot rows */ |
emilmont | 1:fdd22bb7aa52 | 618 | pPRT_in = pPivotRowIn; |
emilmont | 1:fdd22bb7aa52 | 619 | pPRT_pDst = pPivotRowDst; |
emilmont | 1:fdd22bb7aa52 | 620 | |
emilmont | 1:fdd22bb7aa52 | 621 | /* Loop over the number of columns to the right of the pivot element, |
emilmont | 1:fdd22bb7aa52 | 622 | to replace the elements in the input matrix */ |
emilmont | 1:fdd22bb7aa52 | 623 | for (j = 0u; j < (numCols - l); j++) |
emilmont | 1:fdd22bb7aa52 | 624 | { |
emilmont | 1:fdd22bb7aa52 | 625 | /* Replace the element by the sum of that row |
emilmont | 1:fdd22bb7aa52 | 626 | and a multiple of the reference row */ |
emilmont | 1:fdd22bb7aa52 | 627 | *pInT1++ = *pInT1 - (in * *pPRT_in++); |
emilmont | 1:fdd22bb7aa52 | 628 | } |
emilmont | 1:fdd22bb7aa52 | 629 | /* Loop over the number of columns to |
emilmont | 1:fdd22bb7aa52 | 630 | replace the elements in the destination matrix */ |
emilmont | 1:fdd22bb7aa52 | 631 | for (j = 0u; j < numCols; j++) |
emilmont | 1:fdd22bb7aa52 | 632 | { |
emilmont | 1:fdd22bb7aa52 | 633 | /* Replace the element by the sum of that row |
emilmont | 1:fdd22bb7aa52 | 634 | and a multiple of the reference row */ |
emilmont | 1:fdd22bb7aa52 | 635 | *pInT2++ = *pInT2 - (in * *pPRT_pDst++); |
emilmont | 1:fdd22bb7aa52 | 636 | } |
emilmont | 1:fdd22bb7aa52 | 637 | |
emilmont | 1:fdd22bb7aa52 | 638 | } |
emilmont | 1:fdd22bb7aa52 | 639 | /* Increment the temporary input pointer */ |
emilmont | 1:fdd22bb7aa52 | 640 | pInT1 = pInT1 + l; |
emilmont | 1:fdd22bb7aa52 | 641 | } |
emilmont | 1:fdd22bb7aa52 | 642 | /* Increment the input pointer */ |
emilmont | 1:fdd22bb7aa52 | 643 | pIn++; |
emilmont | 1:fdd22bb7aa52 | 644 | |
emilmont | 1:fdd22bb7aa52 | 645 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 646 | loopCnt--; |
emilmont | 1:fdd22bb7aa52 | 647 | /* Increment the index modifier */ |
emilmont | 1:fdd22bb7aa52 | 648 | l++; |
emilmont | 1:fdd22bb7aa52 | 649 | } |
emilmont | 1:fdd22bb7aa52 | 650 | |
emilmont | 1:fdd22bb7aa52 | 651 | |
emilmont | 1:fdd22bb7aa52 | 652 | #endif /* #ifndef ARM_MATH_CM0 */ |
emilmont | 1:fdd22bb7aa52 | 653 | |
emilmont | 1:fdd22bb7aa52 | 654 | /* Set status as ARM_MATH_SUCCESS */ |
emilmont | 1:fdd22bb7aa52 | 655 | status = ARM_MATH_SUCCESS; |
emilmont | 1:fdd22bb7aa52 | 656 | |
emilmont | 1:fdd22bb7aa52 | 657 | if((flag != 1u) && (in == 0.0f)) |
emilmont | 1:fdd22bb7aa52 | 658 | { |
emilmont | 1:fdd22bb7aa52 | 659 | status = ARM_MATH_SINGULAR; |
emilmont | 1:fdd22bb7aa52 | 660 | } |
emilmont | 1:fdd22bb7aa52 | 661 | } |
emilmont | 1:fdd22bb7aa52 | 662 | /* Return to application */ |
emilmont | 1:fdd22bb7aa52 | 663 | return (status); |
emilmont | 1:fdd22bb7aa52 | 664 | } |
emilmont | 1:fdd22bb7aa52 | 665 | |
emilmont | 1:fdd22bb7aa52 | 666 | /** |
emilmont | 1:fdd22bb7aa52 | 667 | * @} end of MatrixInv group |
emilmont | 1:fdd22bb7aa52 | 668 | */ |