| 1 |
|
| /* |
| 2 |
|
| * matrix.c - misc. routines for manipulating matrices and vectors. |
| 3 |
|
| * |
| 4 |
|
| * (C) Copyright 2001 by NetGroup A/S. All rights reserved. |
| 5 |
|
| * |
| 6 |
|
| * $Log$ |
| 7 |
|
| * Revision 1.1 2006/06/20 15:57:22 djburke |
| 8 |
|
| * Hopefully a saner way to build Basic/MatrixOps |
| 9 |
|
| * |
| 10 |
|
| * Revision 1.1 2005/01/08 09:22:57 zowie |
| 11 |
|
| * Added non-symmetric matrices to eigens; updated version to 2.4.2cvs. |
| 12 |
|
| * |
| 13 |
|
| * Revision 1.1.1.1 2001/07/06 13:39:35 kneth |
| 14 |
|
| * Initial import of code. |
| 15 |
|
| * |
| 16 |
|
| * |
| 17 |
|
| * |
| 18 |
|
| * The matrices and vectors are indexed in C-style, i.e. from 0 to |
| 19 |
|
| * N-1. A matrix is assumed to be declared as double **, and it is |
| 20 |
|
| * allocated by MatrixAlloc. |
| 21 |
|
| * |
| 22 |
|
| * |
| 23 |
|
| * References: |
| 24 |
|
| * [1] Numerical Recipes in C, 2nd edition, |
| 25 |
|
| * W.H. Press, S.A. Teukolsky, W.T. Vitterling, and B.P. Flannery, |
| 26 |
|
| * Cambridge University Press, 1992. |
| 27 |
|
| * [2] Numerical Analysis, |
| 28 |
|
| * D. Kincaid and W. Cheney, |
| 29 |
|
| * Brooks/Cole Publishing Company, 1991. |
| 30 |
|
| * [3] The C Programming Language, 2nd edition, |
| 31 |
|
| * B.W. Kernighan and D.M. Ritchie, |
| 32 |
|
| * Prentice Hall, 1988. |
| 33 |
|
| * [4] Advanced Engineering Mathematics, 6th edition, |
| 34 |
|
| * E. Kreyszig, |
| 35 |
|
| * Wiley and Sons, 1988. |
| 36 |
|
| * |
| 37 |
|
| */ |
| 38 |
|
| |
| 39 |
|
| #include <stdio.h> |
| 40 |
|
| #include <stdlib.h> |
| 41 |
|
| #include <math.h> |
| 42 |
|
| |
| 43 |
|
| #ifndef TINY |
| 44 |
|
| # define TINY 1.0e-18 |
| 45 |
|
| #endif |
| 46 |
|
| |
| 47 |
|
| #include "sslib.h" |
| 48 |
|
| #include "matrix.h" |
| 49 |
|
| |
| 50 |
|
| |
| 51 |
|
| /* |
| 52 |
|
| * MatrixAlloc allocates storage for a square matrix with dimension |
| 53 |
|
| * n*n. An error message is printed, if it was impossible to allocate |
| 54 |
|
| * the neccesary space, [3]. |
| 55 |
|
| * |
| 56 |
|
| */ |
| 57 |
|
| |
| 58 |
20 |
| double **MatrixAlloc(const int n) { |
| 59 |
|
| |
| 60 |
|
| double **matrix; |
| 61 |
|
| int i; |
| 62 |
|
| |
| 63 |
20 |
| matrix=(double **)calloc(n, sizeof(double *)); |
| 64 |
20 |
▶ | if (matrix==NULL) |
Branch: 1 branch
| Branch | True | False | Total |
| if (matrix==NULL) |
0 |
20 |
20 |
|
| 65 |
0 |
| SSLerror("No memory available in routine MatrixAlloc"); |
| 66 |
|
| else |
| 67 |
74 |
▶ | for(i=0; i<n; i++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) { |
54 |
20 |
74 |
|
| 68 |
54 |
| matrix[i]=(double *)calloc(n, sizeof(double)); |
| 69 |
54 |
▶ | if (matrix[i]==NULL) |
Branch: 1 branch
| Branch | True | False | Total |
| if (matrix[i]==NULL) |
0 |
54 |
54 |
|
| 70 |
0 |
| SSLerror("No memory available in routine MatrixAlloc"); |
| 71 |
|
| } /* for i=1..n */ |
| 72 |
20 |
| return matrix; |
| 73 |
|
| } /* MatrixAlloc */ |
| 74 |
|
| |
| 75 |
|
| |
| 76 |
|
| /* |
| 77 |
|
| * VectorAlloc allocated space for an n-dimensional vector of the type |
| 78 |
|
| * double *, [3]. It can be freed by VectorFree. |
| 79 |
|
| * |
| 80 |
|
| */ |
| 81 |
|
| |
| 82 |
12 |
| double *VectorAlloc(const int n) { |
| 83 |
|
| |
| 84 |
|
| double *temp; |
| 85 |
|
| |
| 86 |
12 |
| temp=(double *)calloc(n, sizeof(double)); |
| 87 |
12 |
▶ | if (temp==NULL) |
Branch: 1 branch
| Branch | True | False | Total |
| if (temp==NULL) |
0 |
12 |
12 |
|
| 88 |
0 |
| SSLerror("No memory available in routine VectorAlloc"); |
| 89 |
12 |
| return temp; |
| 90 |
|
| } /* VectorAlloc */ |
| 91 |
|
| |
| 92 |
|
| |
| 93 |
|
| /* |
| 94 |
|
| * IntVectorAlloc is similar to VectorAlloc, except that the base type is |
| 95 |
|
| * integers (int) instead of reals (double), [3]. |
| 96 |
|
| * |
| 97 |
|
| */ |
| 98 |
|
| |
| 99 |
7 |
| int *IntVectorAlloc(const int n) { |
| 100 |
|
| |
| 101 |
|
| int *temp; |
| 102 |
|
| |
| 103 |
7 |
| temp=(int *)calloc(n, sizeof(int)); |
| 104 |
7 |
▶ | if (temp==NULL) |
Branch: 1 branch
| Branch | True | False | Total |
| if (temp==NULL) |
0 |
7 |
7 |
|
| 105 |
0 |
| SSLerror("No memory available in routine IntVectorAlloc"); |
| 106 |
7 |
| return temp; |
| 107 |
|
| } /* IntVectorAlloc */ |
| 108 |
|
| |
| 109 |
|
| |
| 110 |
|
| /* |
| 111 |
|
| * SSL_ComplexMatrixAlloc allocates space for a nxn matrix with complex elements. |
| 112 |
|
| * |
| 113 |
|
| */ |
| 114 |
|
| |
| 115 |
0 |
| complex double **SSL_ComplexMatrixAlloc(const int n) { |
| 116 |
|
| |
| 117 |
|
| int i; |
| 118 |
|
| complex double **temp; |
| 119 |
|
| |
| 120 |
0 |
| temp=(complex double **)calloc(n, sizeof(complex double *)); |
| 121 |
0 |
▶ | if (temp==NULL) |
Branch: 1 branch
| Branch | True | False | Total |
| if (temp==NULL) |
0 |
0 |
0 |
|
| 122 |
0 |
| SSLerror("No memory available in routine SSL_ComplexMatrixAlloc"); |
| 123 |
|
| else { |
| 124 |
0 |
▶ | for(i=0; i<n; i++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) { |
0 |
0 |
0 |
|
| 125 |
0 |
| temp[i]=(complex double *)calloc(n, sizeof(complex double)); |
| 126 |
0 |
▶ | if (temp[i]==NULL) |
Branch: 1 branch
| Branch | True | False | Total |
| if (temp[i]==NULL) |
0 |
0 |
0 |
|
| 127 |
0 |
| SSLerror("No memory available in routine SSL_ComplexMatrixAlloc"); |
| 128 |
|
| } /* for i=1..n */ |
| 129 |
|
| } /* if else */ |
| 130 |
0 |
| return temp; |
| 131 |
|
| } /* SSL_ComplexMatrixAlloc */ |
| 132 |
|
| |
| 133 |
|
| |
| 134 |
|
| /* |
| 135 |
|
| * SSL_ComplexVectorAlloc allocates a vector of dimension n with complex |
| 136 |
|
| * elements. |
| 137 |
|
| * |
| 138 |
|
| */ |
| 139 |
|
| |
| 140 |
0 |
| complex double *SSL_ComplexVectorAlloc(const int n) { |
| 141 |
|
| |
| 142 |
|
| complex double *temp; |
| 143 |
|
| |
| 144 |
0 |
| temp=(complex double *)calloc(n, sizeof(complex double)); |
| 145 |
0 |
▶ | if (temp==NULL) |
Branch: 1 branch
| Branch | True | False | Total |
| if (temp==NULL) |
0 |
0 |
0 |
|
| 146 |
0 |
| SSLerror("No memory available in routine SSL_ComplexVectorAlloc"); |
| 147 |
0 |
| return temp; |
| 148 |
|
| } /* SSL_ComplexVectorAlloc */ |
| 149 |
|
| |
| 150 |
|
| |
| 151 |
|
| /* |
| 152 |
|
| * MatrixMul computes the product between two square matrices, [4, pp. |
| 153 |
|
| * 357-358]. Both matrices are assumed to have the dimension n*n. A and B are |
| 154 |
|
| * input, and res is the output, i.e. the routine computes res=A*B. |
| 155 |
|
| * |
| 156 |
|
| */ |
| 157 |
|
| |
| 158 |
9 |
| void MatrixMul(const int n, double **res, double **A, double **B) { |
| 159 |
|
| |
| 160 |
|
| int i, j, k; |
| 161 |
|
| double x; |
| 162 |
|
| |
| 163 |
33 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
24 |
9 |
33 |
|
| 164 |
96 |
▶ | for(j=0; j<n; j++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) { |
72 |
24 |
96 |
|
| 165 |
72 |
| x=0.0; |
| 166 |
312 |
▶ | for(k=0; k<n; k++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(k=0; k<n; k++) |
240 |
72 |
312 |
|
| 167 |
240 |
| x+=A[i][k]*B[k][j]; |
| 168 |
72 |
| res[i][j]=x; |
| 169 |
|
| } /* for j=1..n */ |
| 170 |
9 |
| } /* MatrixMul */ |
| 171 |
|
| |
| 172 |
|
| |
| 173 |
|
| /* |
| 174 |
|
| * Transpose is simply doing as the name says, i.e. transposing the |
| 175 |
|
| * matrix A, [4, pp. 368-370]. A is assumed to be a square matrix. |
| 176 |
|
| * |
| 177 |
|
| */ |
| 178 |
|
| |
| 179 |
3 |
| void Transpose(const int n, double **res, double **A) { |
| 180 |
|
| |
| 181 |
|
| int i, j; |
| 182 |
|
| |
| 183 |
11 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
8 |
3 |
11 |
|
| 184 |
32 |
▶ | for(j=0; j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) |
24 |
8 |
32 |
|
| 185 |
24 |
| res[j][i]=A[i][j]; |
| 186 |
3 |
| } /* Transpose */ |
| 187 |
|
| |
| 188 |
|
| |
| 189 |
|
| /* |
| 190 |
|
| * MatrixFree, VectorFree, IntVectorFree, SSL_ComplexMatrixFree, and |
| 191 |
|
| * SSL_ComplexVectorFree free the space used by the objects, [3]. |
| 192 |
|
| * |
| 193 |
|
| */ |
| 194 |
|
| |
| 195 |
18 |
| void MatrixFree(const int n, double **matrix) { |
| 196 |
|
| |
| 197 |
|
| int i; |
| 198 |
|
| |
| 199 |
66 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
48 |
18 |
66 |
|
| 200 |
48 |
| free((void *)matrix[i]); |
| 201 |
18 |
| free((void *)matrix); |
| 202 |
18 |
| } /* MatrixFree */ |
| 203 |
|
| |
| 204 |
0 |
| void SSL_ComplexMatrixFree(const int n, complex double **matrix) { |
| 205 |
|
| |
| 206 |
|
| int i; |
| 207 |
|
| |
| 208 |
0 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
0 |
0 |
0 |
|
| 209 |
0 |
| free((void *)matrix[i]); |
| 210 |
0 |
| free((void *)matrix); |
| 211 |
0 |
| } /* SSL_ComplexMatrixFree */ |
| 212 |
|
| |
| 213 |
9 |
| void VectorFree(const int n, double *vector) { |
| 214 |
|
| |
| 215 |
9 |
| free((void *)vector); |
| 216 |
9 |
| } /* VectorFree */ |
| 217 |
|
| |
| 218 |
6 |
| void IntVectorFree(const int n, int *vector) { |
| 219 |
|
| |
| 220 |
6 |
| free((void *)vector); |
| 221 |
6 |
| } /* IntVectorFree */ |
| 222 |
|
| |
| 223 |
0 |
| void SSL_ComplexVectorFree(const int n, complex double *vector) { |
| 224 |
|
| |
| 225 |
0 |
| free((void *)vector); |
| 226 |
0 |
| } /* SSL_ComplexVectorFree */ |
| 227 |
|
| |
| 228 |
|
| |
| 229 |
|
| /* |
| 230 |
|
| * LUfact and LUsubst are plain LU decomposition and substitution routines, |
| 231 |
|
| * [1, pp. 43-50], [2, pp. 145-149]. The version here is Gaussian elimination |
| 232 |
|
| * with scaled row pivoting, and it is based on the algorithm given in [2]. |
| 233 |
|
| * |
| 234 |
|
| * LUfact_fixed and LUsubst_fixed are specialised versions of LUfact and |
| 235 |
|
| * LUsubst. They are used in OEE solvers. |
| 236 |
|
| * |
| 237 |
|
| * The parameters are used: |
| 238 |
|
| * n the dimension of the matrix. |
| 239 |
|
| * a the matrix; it contains both L and U at termination. |
| 240 |
|
| * p permutation index. |
| 241 |
|
| * b the constant vector, and at termination it will contain |
| 242 |
|
| * the solution. |
| 243 |
|
| * |
| 244 |
|
| */ |
| 245 |
|
| |
| 246 |
0 |
| void LUfact(const int n, double **a, int *p) { |
| 247 |
|
| |
| 248 |
|
| int i, j, k; /* counters */ |
| 249 |
|
| double z; /* temporary real */ |
| 250 |
|
| double *s; /* pivot elements */ |
| 251 |
|
| int not_finished; /* loop control var. */ |
| 252 |
|
| int i_swap; /* swap var. */ |
| 253 |
|
| double temp; /* another temp. real */ |
| 254 |
|
| |
| 255 |
0 |
| s=VectorAlloc(n); |
| 256 |
0 |
▶ | for(i=0; i<n; i++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) { |
0 |
0 |
0 |
|
| 257 |
0 |
| p[i]=i; |
| 258 |
0 |
| s[i]=0.0; |
| 259 |
0 |
▶ | for(j=0; j<n; j++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) { |
0 |
0 |
0 |
|
| 260 |
0 |
| z=fabs(a[i][j]); |
| 261 |
0 |
▶ | if (s[i]<z) |
Branch: 1 branch
| Branch | True | False | Total |
| if (s[i]<z) |
0 |
0 |
0 |
|
| 262 |
0 |
| s[i]=z; |
| 263 |
|
| } /* for j */ |
| 264 |
|
| } /* for i */ |
| 265 |
|
| |
| 266 |
0 |
▶ | for(k=0; k<(n-1); k++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(k=0; k<(n-1); k++) { |
0 |
0 |
0 |
|
| 267 |
0 |
| j=k-1; /* select j>=k so ... */ |
| 268 |
0 |
| not_finished=1; |
| 269 |
0 |
▶ | while (not_finished) { |
Branch: 1 branch
| Branch | True | False | Total |
| while (not_finished) { |
0 |
0 |
0 |
|
| 270 |
0 |
| j++; |
| 271 |
0 |
| temp=fabs(a[p[j]][k]/s[p[j]]); |
| 272 |
0 |
▶ | for(i=k; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=k; i<n; i++) |
0 |
0 |
0 |
|
| 273 |
0 |
▶ | if (temp>=(fabs(a[p[i]][k])/s[p[i]])) |
Branch: 1 branch
| Branch | True | False | Total |
| if (temp>=(fabs(a[p[i]][k])/s[p[i]])) |
0 |
0 |
0 |
|
| 274 |
0 |
| not_finished=0; /* end loop */ |
| 275 |
|
| } /* while */ |
| 276 |
0 |
| i_swap=p[k]; |
| 277 |
0 |
| p[k]=p[j]; |
| 278 |
0 |
| p[j]=i_swap; |
| 279 |
0 |
| temp=1.0/a[p[k]][k]; |
| 280 |
0 |
▶ | for(i=(k+1); i<n; i++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=(k+1); i<n; i++) { |
0 |
0 |
0 |
|
| 281 |
0 |
| z=a[p[i]][k]*temp; |
| 282 |
0 |
| a[p[i]][k]=z; |
| 283 |
0 |
▶ | for(j=(k+1); j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=(k+1); j<n; j++) |
0 |
0 |
0 |
|
| 284 |
0 |
| a[p[i]][j]-=z*a[p[k]][j]; |
| 285 |
|
| } /* for i */ |
| 286 |
|
| } /* for k */ |
| 287 |
|
| |
| 288 |
0 |
| VectorFree(n, s); |
| 289 |
0 |
| } /* LUfact */ |
| 290 |
|
| |
| 291 |
0 |
| void LUsubst(const int n, double **a, int *p, double *b) { |
| 292 |
|
| |
| 293 |
|
| int i, j, k; /* counters */ |
| 294 |
|
| double sum; /* temporary sum variable */ |
| 295 |
|
| double *x; /* solution */ |
| 296 |
|
| |
| 297 |
0 |
| x=VectorAlloc(n); |
| 298 |
|
| |
| 299 |
0 |
▶ | for(k=0; k<(n-1); k++) /* forward subst */ |
Branch: 1 branch
| Branch | True | False | Total |
| for(k=0; k<(n-1); k++) /* forward subst */ |
0 |
0 |
0 |
|
| 300 |
0 |
▶ | for(i=(k+1); i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=(k+1); i<n; i++) |
0 |
0 |
0 |
|
| 301 |
0 |
| b[p[i]]-=a[p[i]][k]*b[p[k]]; |
| 302 |
|
| |
| 303 |
0 |
▶ | for(i=(n-1); i>=0; i--) { /* back subst */ |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=(n-1); i>=0; i--) { /* back subst */ |
0 |
0 |
0 |
|
| 304 |
0 |
| sum=b[p[i]]; |
| 305 |
0 |
▶ | for(j=(i+1); j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=(i+1); j<n; j++) |
0 |
0 |
0 |
|
| 306 |
0 |
| sum-=a[p[i]][j]*x[j]; |
| 307 |
0 |
| x[i]=sum/a[p[i]][i]; |
| 308 |
|
| } /* for i */ |
| 309 |
|
| |
| 310 |
0 |
▶ | for(i=0; i<n; i++) /* copy solution */ |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) /* copy solution */ |
0 |
0 |
0 |
|
| 311 |
0 |
| b[i]=x[i]; |
| 312 |
|
| |
| 313 |
0 |
| VectorFree(n, x); |
| 314 |
0 |
| } /* LUsubst */ |
| 315 |
|
| |
| 316 |
|
| |
| 317 |
|
| /* |
| 318 |
|
| * GaussSeidel is an implementation of the Gauss-Seidel method, which is an |
| 319 |
|
| * iterative method, [2, pp. 189-191]. The norm applied is the L1-norm. |
| 320 |
|
| * |
| 321 |
|
| * The parameters are: |
| 322 |
|
| * n the dimension. |
| 323 |
|
| * a the coefficient matrix. |
| 324 |
|
| * b the constant vector. |
| 325 |
|
| * x the initial guess, and at termination the solution. |
| 326 |
|
| * eps the precision. |
| 327 |
|
| * max_iter the maximal number of iterations allowed. |
| 328 |
|
| * |
| 329 |
|
| */ |
| 330 |
|
| |
| 331 |
0 |
| void GaussSeidel(const int n, double **a, double *b, double *x, double eps, |
| 332 |
|
| int max_iter) { |
| 333 |
|
| |
| 334 |
|
| int iter, i, j; /* counter */ |
| 335 |
|
| double sum; /* temporary real */ |
| 336 |
|
| double *x_old; /* old solution */ |
| 337 |
|
| double norm; /* L1-norm */ |
| 338 |
|
| |
| 339 |
0 |
| x_old=VectorAlloc(n); |
| 340 |
|
| |
| 341 |
0 |
| iter=0; |
| 342 |
|
| do { /* repeat until safisfying sol. */ |
| 343 |
0 |
| iter++; |
| 344 |
0 |
▶ | for(i=0; i<n; i++) /* copy old solution */ |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) /* copy old solution */ |
0 |
0 |
0 |
|
| 345 |
0 |
| x_old[i]=x[i]; |
| 346 |
0 |
| norm=0.0; /* do an iteration */ |
| 347 |
0 |
▶ | for(i=0; i<n; i++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) { |
0 |
0 |
0 |
|
| 348 |
0 |
| sum=-a[i][i]*x[i]; /* don't include term i=j */ |
| 349 |
0 |
▶ | for(j=0; j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) |
0 |
0 |
0 |
|
| 350 |
0 |
| sum+=a[i][j]*x[j]; |
| 351 |
0 |
| x[i]=(b[i]-sum)/a[i][i]; |
| 352 |
0 |
| norm+=fabs(x_old[i]-x[i]); |
| 353 |
|
| } /* for i */ |
| 354 |
0 |
▶ | } while ((iter<=max_iter) && (norm>=eps)); |
Branch: 2 branches
| Branch | True | False | Total |
| } while ((iter<=max_iter) && (norm>=eps)); |
0 |
0 |
0 |
| } while ((iter<=max_iter) && (norm>=eps)); |
0 |
0 |
0 |
|
| 355 |
|
| |
| 356 |
0 |
| VectorFree(n, x_old); |
| 357 |
0 |
| } /* GaussSeidel */ |
| 358 |
|
| |
| 359 |
|
| |
| 360 |
|
| /* |
| 361 |
|
| * Jacobi is an iterative equation solver, [2, pp. 185-189]. The algorithm |
| 362 |
|
| * can be optimised a bit, which is done in this implementation. The method |
| 363 |
|
| * is suitable for parallel computers. |
| 364 |
|
| * |
| 365 |
|
| * The arguments are the same as in GaussSeidel. |
| 366 |
|
| * |
| 367 |
|
| */ |
| 368 |
|
| |
| 369 |
0 |
| void Jacobi(const int n, double **a, double *b, double *x, double eps, |
| 370 |
|
| int max_iter) { |
| 371 |
|
| |
| 372 |
|
| double d; /* temporary real */ |
| 373 |
|
| int i, j, iter; /* counters */ |
| 374 |
|
| double **a_new; /* a is altered */ |
| 375 |
|
| double *b_new; /* b is altered */ |
| 376 |
|
| double *u; /* new solution */ |
| 377 |
|
| double norm; /* L1-norm */ |
| 378 |
|
| |
| 379 |
0 |
| a_new=MatrixAlloc(3); |
| 380 |
0 |
| b_new=VectorAlloc(3); |
| 381 |
0 |
| u=VectorAlloc(3); |
| 382 |
|
| |
| 383 |
0 |
▶ | for(i=0; i<n; i++) { /* the trick */ |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) { /* the trick */ |
0 |
0 |
0 |
|
| 384 |
0 |
| d=1.0/a[i][i]; |
| 385 |
0 |
| b_new[i]=d*b[i]; |
| 386 |
0 |
▶ | for(j=0; j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) |
0 |
0 |
0 |
|
| 387 |
0 |
| a_new[i][j]=d*a[i][j]; |
| 388 |
|
| } /* for i */ |
| 389 |
|
| |
| 390 |
0 |
| iter=0; |
| 391 |
|
| do { |
| 392 |
0 |
| iter++; |
| 393 |
0 |
| norm=0.0; |
| 394 |
0 |
▶ | for(i=0; i<n; i++) { /* update process */ |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) { /* update process */ |
0 |
0 |
0 |
|
| 395 |
0 |
| d=-a_new[i][i]*x[i]; /* don't include term i=j */ |
| 396 |
0 |
▶ | for(j=0; j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) |
0 |
0 |
0 |
|
| 397 |
0 |
| d+=a_new[i][j]*x[j]; |
| 398 |
0 |
| u[i]=b_new[i]-d; |
| 399 |
0 |
| norm=fabs(u[i]-x[i]); |
| 400 |
|
| } /* for i */ |
| 401 |
0 |
▶ | for(i=0; i<n; i++) /* copy solution */ |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) /* copy solution */ |
0 |
0 |
0 |
|
| 402 |
0 |
| x[i]=u[i]; |
| 403 |
0 |
▶ | } while ((iter<=max_iter) && (norm>=eps)); |
Branch: 2 branches
| Branch | True | False | Total |
| } while ((iter<=max_iter) && (norm>=eps)); |
0 |
0 |
0 |
| } while ((iter<=max_iter) && (norm>=eps)); |
0 |
0 |
0 |
|
| 404 |
|
| |
| 405 |
0 |
| MatrixFree(3, a_new); |
| 406 |
0 |
| VectorFree(3, b_new); |
| 407 |
0 |
| VectorFree(3, u); |
| 408 |
0 |
| } /* Jacobi */ |
| 409 |
|
| |
| 410 |
|
| |
| 411 |
|
| /* |
| 412 |
|
| * DotProd computes the dot product between two vectors. They are assumed to |
| 413 |
|
| * be of the same dimension. |
| 414 |
|
| * |
| 415 |
|
| */ |
| 416 |
|
| |
| 417 |
0 |
| double DotProd(const int n, double *u, double *v) { |
| 418 |
|
| |
| 419 |
|
| int i; /* counter */ |
| 420 |
0 |
| double sum=0.0; /* temporary real */ |
| 421 |
|
| |
| 422 |
0 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
0 |
0 |
0 |
|
| 423 |
0 |
| sum+=u[i]*v[i]; |
| 424 |
0 |
| return sum; |
| 425 |
|
| } /* DotProd */ |
| 426 |
|
| |
| 427 |
|
| |
| 428 |
|
| /* |
| 429 |
|
| * MatrixVecProd computes the matrix product between a matrix and a vector of |
| 430 |
|
| * the dimension n. The result is found in res. |
| 431 |
|
| * |
| 432 |
|
| */ |
| 433 |
|
| |
| 434 |
0 |
| void MatrixVecProd(const int n, double **A, double *v, double *res) { |
| 435 |
|
| |
| 436 |
|
| int i, j; /* counters */ |
| 437 |
|
| |
| 438 |
0 |
▶ | for(i=0; i<n; i++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) { |
0 |
0 |
0 |
|
| 439 |
0 |
| res[i]=0.0; |
| 440 |
0 |
▶ | for(j=0; j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) |
0 |
0 |
0 |
|
| 441 |
0 |
| res[i]+=A[i][j]*v[j]; |
| 442 |
|
| } /* for i */ |
| 443 |
0 |
| } /* MatrixVecProd */ |
| 444 |
|
| |
| 445 |
|
| |
| 446 |
|
| /* |
| 447 |
|
| * MatrixCopy copies the elements of the matrix A to the B. |
| 448 |
|
| * |
| 449 |
|
| */ |
| 450 |
|
| |
| 451 |
12 |
| void MatrixCopy(const int n, double **B, double **A) { |
| 452 |
|
| |
| 453 |
|
| int i, j; |
| 454 |
|
| |
| 455 |
44 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
32 |
12 |
44 |
|
| 456 |
128 |
▶ | for(j=0; j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) |
96 |
32 |
128 |
|
| 457 |
96 |
| B[i][j]=A[i][j]; |
| 458 |
12 |
| } /* MatrixCopy */ |
| 459 |
|
| |
| 460 |
|
| |
| 461 |
|
| /* |
| 462 |
|
| * L2VectorNorm computes the L2 or Eucleadian norm of a vector. |
| 463 |
|
| * |
| 464 |
|
| */ |
| 465 |
|
| |
| 466 |
0 |
| double L2VectorNorm(const int n, double *vec) { |
| 467 |
|
| |
| 468 |
|
| int i; |
| 469 |
0 |
| double norm=0.0; |
| 470 |
|
| |
| 471 |
0 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
0 |
0 |
0 |
|
| 472 |
0 |
| norm+=vec[i]*vec[i]; |
| 473 |
0 |
| return sqrt(norm); |
| 474 |
|
| } /* L2VectorNorm */ |
| 475 |
|
| |
| 476 |
|
| /* |
| 477 |
|
| * GSR is an implementation of the Gram-Schmidt Reorthonormalisation process, |
| 478 |
|
| * [2, pp. 246-250]. The n vectors are collected in a matrix, and the matrix is |
| 479 |
|
| * both input and output. The implementation is actually the modified algorithm |
| 480 |
|
| * as disucced in [2, p. 248]. Modified by the authors, so the vectors are |
| 481 |
|
| * normalised at the end. |
| 482 |
|
| * |
| 483 |
|
| */ |
| 484 |
|
| |
| 485 |
0 |
| void GSR(const int n, double **A) { |
| 486 |
|
| |
| 487 |
|
| int i, j, k; /* counters */ |
| 488 |
|
| double dot; |
| 489 |
|
| |
| 490 |
0 |
▶ | for(k=0; k<n; k++) { /* orthogonalisation */ |
Branch: 1 branch
| Branch | True | False | Total |
| for(k=0; k<n; k++) { /* orthogonalisation */ |
0 |
0 |
0 |
|
| 491 |
0 |
▶ | for(j=(k+1); j<n; j++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=(k+1); j<n; j++) { |
0 |
0 |
0 |
|
| 492 |
0 |
| dot=0.0; /* dot product <Aj, Ak> */ |
| 493 |
0 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
0 |
0 |
0 |
|
| 494 |
0 |
| dot+=A[i][j]*A[i][k]; |
| 495 |
0 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
0 |
0 |
0 |
|
| 496 |
0 |
| A[i][j]-=A[i][k]/dot; |
| 497 |
|
| } /* for j */ |
| 498 |
|
| } /* for k */ |
| 499 |
|
| |
| 500 |
0 |
▶ | for(k=0; k<n; k++) { /* normalisation */ |
Branch: 1 branch
| Branch | True | False | Total |
| for(k=0; k<n; k++) { /* normalisation */ |
0 |
0 |
0 |
|
| 501 |
0 |
| dot=0.0; /* Compute (L2 norm) */ |
| 502 |
0 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
0 |
0 |
0 |
|
| 503 |
0 |
| dot+=A[i][k]*A[i][k]; |
| 504 |
0 |
| dot=sqrt(dot); |
| 505 |
0 |
▶ | if (dot==0.0) |
Branch: 1 branch
| Branch | True | False | Total |
| if (dot==0.0) |
0 |
0 |
0 |
|
| 506 |
0 |
| SSLerror("Norm = 0 in routine GSR"); |
| 507 |
0 |
▶ | for(i=0; i<n; i++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) |
0 |
0 |
0 |
|
| 508 |
0 |
| A[i][k]/=dot; |
| 509 |
|
| } /* for k */ |
| 510 |
0 |
| } /* GSR */ |
| 511 |
|
| |
| 512 |
|
| |
| 513 |
|
| /* |
| 514 |
|
| * InversMatrix calculates the inverse matrix. The method is the solution |
| 515 |
|
| * of n linear set of equations which are solved by a LU factorisation. |
| 516 |
|
| * |
| 517 |
|
| */ |
| 518 |
|
| |
| 519 |
0 |
| void InversMatrix(const int n, double **b, double **ib) { |
| 520 |
|
| |
| 521 |
|
| double **a; |
| 522 |
|
| double *e; |
| 523 |
|
| int i,j; |
| 524 |
|
| int *p; |
| 525 |
|
| |
| 526 |
0 |
| a=MatrixAlloc(n); |
| 527 |
0 |
| e=VectorAlloc(n); |
| 528 |
0 |
| p=IntVectorAlloc(n); |
| 529 |
0 |
| MatrixCopy(n, a, b); |
| 530 |
0 |
| LUfact(n, a, p); |
| 531 |
0 |
▶ | for(i=0; i<n; i++) { |
Branch: 1 branch
| Branch | True | False | Total |
| for(i=0; i<n; i++) { |
0 |
0 |
0 |
|
| 532 |
0 |
▶ | for(j=0; j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) |
0 |
0 |
0 |
|
| 533 |
0 |
| e[j]=0.0; |
| 534 |
0 |
| e[i]=1.0; |
| 535 |
0 |
| LUsubst(n, a, p, e); |
| 536 |
0 |
▶ | for(j=0; j<n; j++) |
Branch: 1 branch
| Branch | True | False | Total |
| for(j=0; j<n; j++) |
0 |
0 |
0 |
|
| 537 |
0 |
| ib[j][i]=e[j]; |
| 538 |
|
| } /* for i=1..n */ |
| 539 |
|
| |
| 540 |
0 |
| MatrixFree(n, a); |
| 541 |
0 |
| VectorFree(n, e); |
| 542 |
0 |
| IntVectorFree(n, p); |
| 543 |
0 |
| } /* InversMatrix */ |
| 544 |
|
| |