| 1 |
|
| 2 |
/******************************************************* |
| 3 |
* |
| 4 |
* Copyright (c) 2003-2008 by University of Queensland |
| 5 |
* Earth Systems Science Computational Center (ESSCC) |
| 6 |
* http://www.uq.edu.au/esscc |
| 7 |
* |
| 8 |
* Primary Business: Queensland, Australia |
| 9 |
* Licensed under the Open Software License version 3.0 |
| 10 |
* http://www.opensource.org/licenses/osl-3.0.php |
| 11 |
* |
| 12 |
*******************************************************/ |
| 13 |
|
| 14 |
|
| 15 |
/**************************************************************/ |
| 16 |
|
| 17 |
/* Some utility routines: */ |
| 18 |
|
| 19 |
/**************************************************************/ |
| 20 |
|
| 21 |
#include "Finley.h" |
| 22 |
#include "Util.h" |
| 23 |
|
| 24 |
#ifdef _OPENMP |
| 25 |
#include <omp.h> |
| 26 |
#endif |
| 27 |
|
| 28 |
/**************************************************************/ |
| 29 |
|
| 30 |
/* returns true if any of the values in the short array values is not equalt to Zero */ |
| 31 |
|
| 32 |
bool_t Finley_Util_anyNonZeroDouble(dim_t N, double* values) { |
| 33 |
dim_t q; |
| 34 |
for (q=0;q<N;++q) if (ABS(values[q])>0) return TRUE; |
| 35 |
return FALSE; |
| 36 |
} |
| 37 |
/**************************************************************/ |
| 38 |
|
| 39 |
/* gathers double values out from in by index: */ |
| 40 |
|
| 41 |
/* out(1:numData,1:len)=in(1:numData,index(1:len)) */ |
| 42 |
|
| 43 |
void Finley_Util_Gather_double(dim_t len,index_t* index,dim_t numData,double* in, double * out){ |
| 44 |
dim_t s,i; |
| 45 |
for (s=0;s<len;s++) { |
| 46 |
for (i=0;i<numData;i++) { |
| 47 |
out[INDEX2(i,s,numData)]=in[INDEX2(i,index[s],numData)]; |
| 48 |
} |
| 49 |
} |
| 50 |
} |
| 51 |
|
| 52 |
/**************************************************************/ |
| 53 |
|
| 54 |
|
| 55 |
/* gathers maybelong values out from in by index: */ |
| 56 |
|
| 57 |
/* out(1:numData,1:len)=in(1:numData,index(1:len)) */ |
| 58 |
|
| 59 |
void Finley_Util_Gather_int(dim_t len,index_t* index,dim_t numData, index_t* in, index_t * out){ |
| 60 |
dim_t s,i; |
| 61 |
for (s=0;s<len;s++) { |
| 62 |
for (i=0;i<numData;i++) { |
| 63 |
out[INDEX2(i,s,numData)]=in[INDEX2(i,index[s],numData)]; |
| 64 |
} |
| 65 |
} |
| 66 |
} |
| 67 |
|
| 68 |
/**************************************************************/ |
| 69 |
|
| 70 |
/* adds a vector in into out using and index. */ |
| 71 |
|
| 72 |
/* out(1:numData,index[p])+=in(1:numData,p) where p = {k=1...len , index[k]<upperBound}*/ |
| 73 |
|
| 74 |
|
| 75 |
void Finley_Util_AddScatter(dim_t len,index_t* index,dim_t numData,double* in,double * out, index_t upperBound){ |
| 76 |
dim_t i,s; |
| 77 |
for (s=0;s<len;s++) { |
| 78 |
for(i=0;i<numData;i++) { |
| 79 |
if( index[s]<upperBound ) { |
| 80 |
out[INDEX2(i,index[s],numData)]+=in[INDEX2(i,s,numData)]; |
| 81 |
} |
| 82 |
} |
| 83 |
} |
| 84 |
} |
| 85 |
|
| 86 |
/* multiplies two matrices */ |
| 87 |
|
| 88 |
/* A(1:A1,1:A2)=B(1:A1,1:B2)*C(1:B2,1:A2) */ |
| 89 |
|
| 90 |
void Finley_Util_SmallMatMult(dim_t A1,dim_t A2, double* A, dim_t B2, double*B, double* C) { |
| 91 |
dim_t i,j,s; |
| 92 |
for (i=0;i<A1*A2;i++) A[i]=0; |
| 93 |
for (i=0;i<A1;i++) { |
| 94 |
for (j=0;j<A2;j++) { |
| 95 |
for (s=0;s<B2;s++) { |
| 96 |
A[INDEX2(i,j,A1)]+=B[INDEX2(i,s,A1)]*C[INDEX2(s,j,B2)]; |
| 97 |
} |
| 98 |
} |
| 99 |
} |
| 100 |
} |
| 101 |
|
| 102 |
/* multiplies a two sets of matries: */ |
| 103 |
|
| 104 |
/* A(1:A1,1:A2,i)=B(1:A1,1:B2,i)*C(1:B2,1:A2,i) i=1,len */ |
| 105 |
|
| 106 |
void Finley_Util_SmallMatSetMult(dim_t len,dim_t A1,dim_t A2, double* A, dim_t B2, double*B, double* C) { |
| 107 |
dim_t q,i,j,s; |
| 108 |
for (i=0;i<A1*A2*len;i++) A[i]=0; |
| 109 |
for (q=0;q<len;q++) { |
| 110 |
for (i=0;i<A1;i++) { |
| 111 |
for (j=0;j<A2;j++) { |
| 112 |
for (s=0;s<B2;s++) { |
| 113 |
A[INDEX3(i,j,q,A1,A2)]+=B[INDEX3(i,s,q,A1,B2)]*C[INDEX3(s,j,q,B2,A2)]; |
| 114 |
} |
| 115 |
} |
| 116 |
} |
| 117 |
} |
| 118 |
} |
| 119 |
/* inverts the set of dim x dim matrices A(:,:,1:len) with dim=1,2,3 */ |
| 120 |
/* the determinante is returned. */ |
| 121 |
|
| 122 |
void Finley_Util_InvertSmallMat(dim_t len,dim_t dim,double* A,double *invA, double* det){ |
| 123 |
dim_t q; |
| 124 |
register double D,A11,A12,A13,A21,A22,A23,A31,A32,A33; |
| 125 |
|
| 126 |
switch(dim) { |
| 127 |
case 1: |
| 128 |
for (q=0;q<len;q++) { |
| 129 |
D=A[q]; |
| 130 |
if (ABS(D) > 0 ){ |
| 131 |
det[q]=D; |
| 132 |
D=1./D; |
| 133 |
invA[q]=D; |
| 134 |
} else { |
| 135 |
Finley_setError(ZERO_DIVISION_ERROR,"__FILE__: Non-regular matrix"); |
| 136 |
return; |
| 137 |
} |
| 138 |
} |
| 139 |
break; |
| 140 |
|
| 141 |
case 2: |
| 142 |
for (q=0;q<len;q++) { |
| 143 |
A11=A[INDEX3(0,0,q,2,2)]; |
| 144 |
A12=A[INDEX3(0,1,q,2,2)]; |
| 145 |
A21=A[INDEX3(1,0,q,2,2)]; |
| 146 |
A22=A[INDEX3(1,1,q,2,2)]; |
| 147 |
|
| 148 |
D = A11*A22-A12*A21; |
| 149 |
if (ABS(D) > 0 ){ |
| 150 |
det[q]=D; |
| 151 |
D=1./D; |
| 152 |
invA[INDEX3(0,0,q,2,2)]= A22*D; |
| 153 |
invA[INDEX3(1,0,q,2,2)]=-A21*D; |
| 154 |
invA[INDEX3(0,1,q,2,2)]=-A12*D; |
| 155 |
invA[INDEX3(1,1,q,2,2)]= A11*D; |
| 156 |
} else { |
| 157 |
Finley_setError(ZERO_DIVISION_ERROR,"__FILE__: Non-regular matrix"); |
| 158 |
return; |
| 159 |
} |
| 160 |
} |
| 161 |
break; |
| 162 |
|
| 163 |
case 3: |
| 164 |
for (q=0;q<len;q++) { |
| 165 |
A11=A[INDEX3(0,0,q,3,3)]; |
| 166 |
A21=A[INDEX3(1,0,q,3,3)]; |
| 167 |
A31=A[INDEX3(2,0,q,3,3)]; |
| 168 |
A12=A[INDEX3(0,1,q,3,3)]; |
| 169 |
A22=A[INDEX3(1,1,q,3,3)]; |
| 170 |
A32=A[INDEX3(2,1,q,3,3)]; |
| 171 |
A13=A[INDEX3(0,2,q,3,3)]; |
| 172 |
A23=A[INDEX3(1,2,q,3,3)]; |
| 173 |
A33=A[INDEX3(2,2,q,3,3)]; |
| 174 |
|
| 175 |
D = A11*(A22*A33-A23*A32)+ A12*(A31*A23-A21*A33)+A13*(A21*A32-A31*A22); |
| 176 |
if (ABS(D) > 0 ){ |
| 177 |
det[q] =D; |
| 178 |
D=1./D; |
| 179 |
invA[INDEX3(0,0,q,3,3)]=(A22*A33-A23*A32)*D; |
| 180 |
invA[INDEX3(1,0,q,3,3)]=(A31*A23-A21*A33)*D; |
| 181 |
invA[INDEX3(2,0,q,3,3)]=(A21*A32-A31*A22)*D; |
| 182 |
invA[INDEX3(0,1,q,3,3)]=(A13*A32-A12*A33)*D; |
| 183 |
invA[INDEX3(1,1,q,3,3)]=(A11*A33-A31*A13)*D; |
| 184 |
invA[INDEX3(2,1,q,3,3)]=(A12*A31-A11*A32)*D; |
| 185 |
invA[INDEX3(0,2,q,3,3)]=(A12*A23-A13*A22)*D; |
| 186 |
invA[INDEX3(1,2,q,3,3)]=(A13*A21-A11*A23)*D; |
| 187 |
invA[INDEX3(2,2,q,3,3)]=(A11*A22-A12*A21)*D; |
| 188 |
} else { |
| 189 |
Finley_setError(ZERO_DIVISION_ERROR,"__FILE__: Non-regular matrix"); |
| 190 |
return; |
| 191 |
} |
| 192 |
} |
| 193 |
break; |
| 194 |
|
| 195 |
} |
| 196 |
return; |
| 197 |
} |
| 198 |
|
| 199 |
/* sets the derterminate of a set of dim x dim matrices A(:,:,1:len) with dim=1,2,3 */ |
| 200 |
|
| 201 |
void Finley_Util_DetOfSmallMat(dim_t len,dim_t dim,double* A, double* det){ |
| 202 |
dim_t q; |
| 203 |
register double A11,A12,A13,A21,A22,A23,A31,A32,A33; |
| 204 |
|
| 205 |
switch(dim) { |
| 206 |
case 1: |
| 207 |
for (q=0;q<len;q++) { |
| 208 |
det[q]=A[q]; |
| 209 |
} |
| 210 |
break; |
| 211 |
|
| 212 |
case 2: |
| 213 |
for (q=0;q<len;q++) { |
| 214 |
A11=A[INDEX3(0,0,q,2,2)]; |
| 215 |
A12=A[INDEX3(0,1,q,2,2)]; |
| 216 |
A21=A[INDEX3(1,0,q,2,2)]; |
| 217 |
A22=A[INDEX3(1,1,q,2,2)]; |
| 218 |
|
| 219 |
det[q] = A11*A22-A12*A21; |
| 220 |
} |
| 221 |
break; |
| 222 |
|
| 223 |
case 3: |
| 224 |
for (q=0;q<len;q++) { |
| 225 |
A11=A[INDEX3(0,0,q,3,3)]; |
| 226 |
A21=A[INDEX3(1,0,q,3,3)]; |
| 227 |
A31=A[INDEX3(2,0,q,3,3)]; |
| 228 |
A12=A[INDEX3(0,1,q,3,3)]; |
| 229 |
A22=A[INDEX3(1,1,q,3,3)]; |
| 230 |
A32=A[INDEX3(2,1,q,3,3)]; |
| 231 |
A13=A[INDEX3(0,2,q,3,3)]; |
| 232 |
A23=A[INDEX3(1,2,q,3,3)]; |
| 233 |
A33=A[INDEX3(2,2,q,3,3)]; |
| 234 |
|
| 235 |
det[q] = A11*(A22*A33-A23*A32)+ A12*(A31*A23-A21*A33)+A13*(A21*A32-A31*A22); |
| 236 |
} |
| 237 |
break; |
| 238 |
|
| 239 |
} |
| 240 |
return; |
| 241 |
} |
| 242 |
/* returns the normalized vector Normal[dim,len] orthogonal to A(:,0,q) and A(:,1,q) in the case of dim=3 */ |
| 243 |
/* or the vector A(:,0,q) in the case of dim=2 */ |
| 244 |
|
| 245 |
void Finley_NormalVector(dim_t len, dim_t dim, dim_t dim1, double* A,double* Normal) { |
| 246 |
dim_t q; |
| 247 |
register double A11,A12,CO_A13,A21,A22,CO_A23,A31,A32,CO_A33,length,invlength; |
| 248 |
|
| 249 |
switch(dim) { |
| 250 |
case 1: |
| 251 |
for (q=0;q<len;q++) Normal[q] =1; |
| 252 |
break; |
| 253 |
case 2: |
| 254 |
for (q=0;q<len;q++) { |
| 255 |
A11=A[INDEX3(0,0,q,2,dim1)]; |
| 256 |
A21=A[INDEX3(1,0,q,2,dim1)]; |
| 257 |
length = sqrt(A11*A11+A21*A21); |
| 258 |
if (! length>0) { |
| 259 |
Finley_setError(ZERO_DIVISION_ERROR,"__FILE__: area equals zero."); |
| 260 |
return; |
| 261 |
} else { |
| 262 |
invlength=1./length; |
| 263 |
Normal[INDEX2(0,q,2)]=A21*invlength; |
| 264 |
Normal[INDEX2(1,q,2)]=-A11*invlength; |
| 265 |
} |
| 266 |
} |
| 267 |
break; |
| 268 |
case 3: |
| 269 |
for (q=0;q<len;q++) { |
| 270 |
A11=A[INDEX3(0,0,q,3,dim1)]; |
| 271 |
A21=A[INDEX3(1,0,q,3,dim1)]; |
| 272 |
A31=A[INDEX3(2,0,q,3,dim1)]; |
| 273 |
A12=A[INDEX3(0,1,q,3,dim1)]; |
| 274 |
A22=A[INDEX3(1,1,q,3,dim1)]; |
| 275 |
A32=A[INDEX3(2,1,q,3,dim1)]; |
| 276 |
CO_A13=A21*A32-A31*A22; |
| 277 |
CO_A23=A31*A12-A11*A32; |
| 278 |
CO_A33=A11*A22-A21*A12; |
| 279 |
length=sqrt(CO_A13*CO_A13+CO_A23*CO_A23+CO_A33*CO_A33); |
| 280 |
if (! length>0) { |
| 281 |
Finley_setError(ZERO_DIVISION_ERROR,"__FILE__: area equals zero."); |
| 282 |
return; |
| 283 |
} else { |
| 284 |
invlength=1./length; |
| 285 |
Normal[INDEX2(0,q,3)]=CO_A13*invlength; |
| 286 |
Normal[INDEX2(1,q,3)]=CO_A23*invlength; |
| 287 |
Normal[INDEX2(2,q,3)]=CO_A33*invlength; |
| 288 |
} |
| 289 |
|
| 290 |
} |
| 291 |
break; |
| 292 |
|
| 293 |
} |
| 294 |
return; |
| 295 |
} |
| 296 |
|
| 297 |
/* return the length of the vector which is orthogonal to the vectors A(:,0,q) and A(:,1,q) in the case of dim=3 */ |
| 298 |
/* or the vector A(:,0,q) in the case of dim=2 */ |
| 299 |
|
| 300 |
void Finley_LengthOfNormalVector(dim_t len, dim_t dim, dim_t dim1, double* A,double* length) { |
| 301 |
dim_t q; |
| 302 |
double A11,A12,CO_A13,A21,A22,CO_A23,A31,A32,CO_A33; |
| 303 |
|
| 304 |
switch(dim) { |
| 305 |
case 1: |
| 306 |
for (q=0;q<len;q++) length[q] =1; |
| 307 |
break; |
| 308 |
case 2: |
| 309 |
for (q=0;q<len;q++) { |
| 310 |
A11=A[INDEX3(0,0,q,2,dim1)]; |
| 311 |
A21=A[INDEX3(1,0,q,2,dim1)]; |
| 312 |
length[q] = sqrt(A11*A11+A21*A21); |
| 313 |
} |
| 314 |
break; |
| 315 |
case 3: |
| 316 |
for (q=0;q<len;q++) { |
| 317 |
A11=A[INDEX3(0,0,q,3,dim1)]; |
| 318 |
A21=A[INDEX3(1,0,q,3,dim1)]; |
| 319 |
A31=A[INDEX3(2,0,q,3,dim1)]; |
| 320 |
A12=A[INDEX3(0,1,q,3,dim1)]; |
| 321 |
A22=A[INDEX3(1,1,q,3,dim1)]; |
| 322 |
A32=A[INDEX3(2,1,q,3,dim1)]; |
| 323 |
CO_A13=A21*A32-A31*A22; |
| 324 |
CO_A23=A31*A12-A11*A32; |
| 325 |
CO_A33=A11*A22-A21*A12; |
| 326 |
length[q]=sqrt(CO_A13*CO_A13+CO_A23*CO_A23+CO_A33*CO_A33); |
| 327 |
} |
| 328 |
break; |
| 329 |
|
| 330 |
} |
| 331 |
return; |
| 332 |
} |
| 333 |
|
| 334 |
/* inverts the map map of length len */ |
| 335 |
/* there is no range checking! */ |
| 336 |
/* at output Map[invMap[i]]=i for i=0:lenInvMap */ |
| 337 |
|
| 338 |
void Finley_Util_InvertMap(dim_t lenInvMap, index_t* invMap,dim_t lenMap, index_t* Map) { |
| 339 |
dim_t i; |
| 340 |
for (i=0;i<lenInvMap;i++) invMap[i]=0; |
| 341 |
for (i=0;i<lenMap;i++) { |
| 342 |
if (Map[i]>=0) invMap[Map[i]]=i; |
| 343 |
} |
| 344 |
} |
| 345 |
|
| 346 |
/* orders a Finley_Util_ValueAndIndex array by value */ |
| 347 |
/* it is assumed that n is large */ |
| 348 |
|
| 349 |
int Finley_Util_ValueAndIndex_compar(const void *arg1 , const void *arg2 ) { |
| 350 |
Finley_Util_ValueAndIndex *e1,*e2; |
| 351 |
e1=(Finley_Util_ValueAndIndex*) arg1; |
| 352 |
e2=(Finley_Util_ValueAndIndex*) arg2; |
| 353 |
if (e1->value < e2->value) return -1; |
| 354 |
if (e1->value > e2->value) return 1; |
| 355 |
if (e1->index < e2->index) return -1; |
| 356 |
if (e1->index > e2->index) return 1; |
| 357 |
return 0; |
| 358 |
} |
| 359 |
|
| 360 |
void Finley_Util_sortValueAndIndex(dim_t n,Finley_Util_ValueAndIndex* array) { |
| 361 |
/* OMP : needs parallelization !*/ |
| 362 |
qsort(array,n,sizeof(Finley_Util_ValueAndIndex),Finley_Util_ValueAndIndex_compar); |
| 363 |
} |
| 364 |
|
| 365 |
|
| 366 |
/**************************************************************/ |
| 367 |
|
| 368 |
/* calculates the minimum value from a dim X N integer array */ |
| 369 |
|
| 370 |
index_t Finley_Util_getMinInt(dim_t dim,dim_t N,index_t* values) { |
| 371 |
dim_t i,j; |
| 372 |
index_t out,out_local; |
| 373 |
out=INDEX_T_MAX; |
| 374 |
if (values!=NULL && dim*N>0 ) { |
| 375 |
out=values[0]; |
| 376 |
#pragma omp parallel private(out_local) |
| 377 |
{ |
| 378 |
out_local=out; |
| 379 |
#pragma omp for private(i,j) schedule(static) |
| 380 |
for (j=0;j<N;j++) { |
| 381 |
for (i=0;i<dim;i++) out_local=MIN(out_local,values[INDEX2(i,j,dim)]); |
| 382 |
} |
| 383 |
#pragma omp critical |
| 384 |
out=MIN(out_local,out); |
| 385 |
} |
| 386 |
} |
| 387 |
return out; |
| 388 |
} |
| 389 |
|
| 390 |
/* calculates the maximum value from a dim X N integer array */ |
| 391 |
|
| 392 |
index_t Finley_Util_getMaxInt(dim_t dim,dim_t N,index_t* values) { |
| 393 |
dim_t i,j; |
| 394 |
index_t out,out_local; |
| 395 |
out=-INDEX_T_MAX; |
| 396 |
if (values!=NULL && dim*N>0 ) { |
| 397 |
out=values[0]; |
| 398 |
#pragma omp parallel private(out_local) |
| 399 |
{ |
| 400 |
out_local=out; |
| 401 |
#pragma omp for private(i,j) schedule(static) |
| 402 |
for (j=0;j<N;j++) { |
| 403 |
for (i=0;i<dim;i++) out_local=MAX(out_local,values[INDEX2(i,j,dim)]); |
| 404 |
} |
| 405 |
#pragma omp critical |
| 406 |
out=MAX(out_local,out); |
| 407 |
} |
| 408 |
} |
| 409 |
return out; |
| 410 |
} |
| 411 |
/**************************************************************/ |
| 412 |
|
| 413 |
/* calculates the minimum value from a dim X N integer array */ |
| 414 |
|
| 415 |
index_t Finley_Util_getFlaggedMinInt(dim_t dim,dim_t N,index_t* values, index_t ignore) { |
| 416 |
dim_t i,j; |
| 417 |
index_t out,out_local; |
| 418 |
out=INDEX_T_MAX; |
| 419 |
if (values!=NULL && dim*N>0 ) { |
| 420 |
out=values[0]; |
| 421 |
#pragma omp parallel private(out_local) |
| 422 |
{ |
| 423 |
out_local=out; |
| 424 |
#pragma omp for private(i,j) schedule(static) |
| 425 |
for (j=0;j<N;j++) { |
| 426 |
for (i=0;i<dim;i++) if (values[INDEX2(i,j,dim)]!=ignore) out_local=MIN(out_local,values[INDEX2(i,j,dim)]); |
| 427 |
} |
| 428 |
#pragma omp critical |
| 429 |
out=MIN(out_local,out); |
| 430 |
} |
| 431 |
} |
| 432 |
return out; |
| 433 |
} |
| 434 |
|
| 435 |
/* calculates the maximum value from a dim X N integer array */ |
| 436 |
|
| 437 |
index_t Finley_Util_getFlaggedMaxInt(dim_t dim,dim_t N,index_t* values, index_t ignore) { |
| 438 |
dim_t i,j; |
| 439 |
index_t out,out_local; |
| 440 |
out=-INDEX_T_MAX; |
| 441 |
if (values!=NULL && dim*N>0 ) { |
| 442 |
out=values[0]; |
| 443 |
#pragma omp parallel private(out_local) |
| 444 |
{ |
| 445 |
out_local=out; |
| 446 |
#pragma omp for private(i,j) schedule(static) |
| 447 |
for (j=0;j<N;j++) { |
| 448 |
for (i=0;i<dim;i++) if (values[INDEX2(i,j,dim)]!=ignore) out_local=MAX(out_local,values[INDEX2(i,j,dim)]); |
| 449 |
} |
| 450 |
#pragma omp critical |
| 451 |
out=MAX(out_local,out); |
| 452 |
} |
| 453 |
} |
| 454 |
return out; |
| 455 |
} |
| 456 |
|
| 457 |
/* set the index of the positive entries in mask. The length of index is returned. */ |
| 458 |
|
| 459 |
dim_t Finley_Util_packMask(dim_t N,index_t* mask,index_t* index) { |
| 460 |
dim_t out,k; |
| 461 |
out=0; |
| 462 |
/*OMP */ |
| 463 |
for (k=0;k<N;k++) { |
| 464 |
if (mask[k]>=0) { |
| 465 |
index[out]=k; |
| 466 |
out++; |
| 467 |
} |
| 468 |
} |
| 469 |
return out; |
| 470 |
} |
| 471 |
|
| 472 |
/* returns true if array contains value */ |
| 473 |
bool_t Finley_Util_isAny(dim_t N,index_t* array,index_t value) { |
| 474 |
bool_t out=FALSE; |
| 475 |
dim_t i; |
| 476 |
#pragma omp parallel for private(i) schedule(static) reduction(||:out) |
| 477 |
for (i=0;i<N;i++) out = out || (array[i]==value); |
| 478 |
return out; |
| 479 |
} |
| 480 |
/* calculates the cummultative sum in array and returns the total sum */ |
| 481 |
index_t Finley_Util_cumsum(dim_t N,index_t* array) { |
| 482 |
index_t out=0,tmp; |
| 483 |
dim_t i; |
| 484 |
#ifdef _OPENMP |
| 485 |
index_t *partial_sums=NULL, sum; |
| 486 |
partial_sums=TMPMEMALLOC(omp_get_max_threads(),index_t); |
| 487 |
#pragma omp parallel private(sum,i,tmp) |
| 488 |
{ |
| 489 |
sum=0; |
| 490 |
#pragma omp for schedule(static) |
| 491 |
for (i=0;i<N;++i) sum+=array[i]; |
| 492 |
partial_sums[omp_get_thread_num()]=sum; |
| 493 |
#pragma omp barrier |
| 494 |
#pragma omp master |
| 495 |
{ |
| 496 |
out=0; |
| 497 |
for (i=0;i<omp_get_max_threads();++i) { |
| 498 |
tmp=out; |
| 499 |
out+=partial_sums[i]; |
| 500 |
partial_sums[i]=tmp; |
| 501 |
} |
| 502 |
} |
| 503 |
#pragma omp barrier |
| 504 |
sum=partial_sums[omp_get_thread_num()]; |
| 505 |
#pragma omp for schedule(static) |
| 506 |
for (i=0;i<N;++i) { |
| 507 |
tmp=sum; |
| 508 |
sum+=array[i]; |
| 509 |
array[i]=tmp; |
| 510 |
} |
| 511 |
} |
| 512 |
TMPMEMFREE(partial_sums); |
| 513 |
#else |
| 514 |
for (i=0;i<N;++i) { |
| 515 |
tmp=out; |
| 516 |
out+=array[i]; |
| 517 |
array[i]=tmp; |
| 518 |
} |
| 519 |
#endif |
| 520 |
return out; |
| 521 |
} |
| 522 |
void Finley_Util_setValuesInUse(const index_t *values, const dim_t numValues, dim_t *numValuesInUse, index_t **valuesInUse, Paso_MPIInfo* mpiinfo) |
| 523 |
{ |
| 524 |
dim_t i; |
| 525 |
index_t lastFoundValue=INDEX_T_MIN, minFoundValue, local_minFoundValue, *newValuesInUse=NULL; |
| 526 |
register index_t itmp; |
| 527 |
bool_t allFound=FALSE; |
| 528 |
dim_t nv=0; |
| 529 |
|
| 530 |
while (! allFound) { |
| 531 |
/* |
| 532 |
* find smallest value bigger than lastFoundValue |
| 533 |
*/ |
| 534 |
minFoundValue=INDEX_T_MAX; |
| 535 |
#pragma omp parallel private(local_minFoundValue) |
| 536 |
{ |
| 537 |
local_minFoundValue=minFoundValue; |
| 538 |
#pragma omp for private(i,itmp) schedule(static) |
| 539 |
for (i=0;i< numValues;i++) { |
| 540 |
itmp=values[i]; |
| 541 |
if ((itmp>lastFoundValue) && (itmp<local_minFoundValue)) local_minFoundValue=itmp; |
| 542 |
} |
| 543 |
#pragma omp critical |
| 544 |
minFoundValue=MIN(local_minFoundValue,minFoundValue); |
| 545 |
} |
| 546 |
#ifdef PASO_MPI |
| 547 |
local_minFoundValue=minFoundValue; |
| 548 |
MPI_Allreduce(&local_minFoundValue,&minFoundValue, 1, MPI_INT, MPI_MAX, mpiinfo->comm ); |
| 549 |
#endif |
| 550 |
|
| 551 |
/* if we found a new tag we need to add this too the valuesInUseList */ |
| 552 |
|
| 553 |
if (minFoundValue < INDEX_T_MAX) { |
| 554 |
newValuesInUse=MEMALLOC(nv+1,index_t); |
| 555 |
if (*valuesInUse!=NULL) { |
| 556 |
memcpy(newValuesInUse,*valuesInUse,sizeof(index_t)*nv); |
| 557 |
MEMFREE(*valuesInUse); |
| 558 |
} |
| 559 |
newValuesInUse[nv]=minFoundValue; |
| 560 |
*valuesInUse=newValuesInUse; |
| 561 |
newValuesInUse=NULL; |
| 562 |
nv++; |
| 563 |
lastFoundValue=minFoundValue; |
| 564 |
} else { |
| 565 |
allFound=TRUE; |
| 566 |
} |
| 567 |
} |
| 568 |
*numValuesInUse=nv; |
| 569 |
} |
| 570 |
|
| 571 |
|
| 572 |
#ifdef PASO_MPI |
| 573 |
void Finley_printDoubleArray( FILE *fid, dim_t n, double *array, char *name ) |
| 574 |
{ |
| 575 |
index_t i; |
| 576 |
|
| 577 |
if( name ) |
| 578 |
fprintf( fid, "%s [ ", name ); |
| 579 |
else |
| 580 |
fprintf( fid, "[ " ); |
| 581 |
for( i=0; i<(n<60 ? n : 60); i++ ) |
| 582 |
fprintf( fid, "%g ", array[i] ); |
| 583 |
if( n>=30 ) |
| 584 |
fprintf( fid, "... " ); |
| 585 |
fprintf( fid, "]\n" ); |
| 586 |
} |
| 587 |
void Finley_printIntArray( FILE *fid, dim_t n, int *array, char *name ) |
| 588 |
{ |
| 589 |
index_t i; |
| 590 |
|
| 591 |
if( name ) |
| 592 |
fprintf( fid, "%s [ ", name ); |
| 593 |
else |
| 594 |
fprintf( fid, "[ " ); |
| 595 |
for( i=0; i<(n<60 ? n : 60); i++ ) |
| 596 |
fprintf( fid, "%d ", array[i] ); |
| 597 |
if( n>=30 ) |
| 598 |
fprintf( fid, "... " ); |
| 599 |
fprintf( fid, "]\n" ); |
| 600 |
} |
| 601 |
void Finley_printMaskArray( FILE *fid, dim_t n, int *array, char *name ) |
| 602 |
{ |
| 603 |
index_t i; |
| 604 |
|
| 605 |
if( name ) |
| 606 |
fprintf( fid, "%s [ ", name ); |
| 607 |
else |
| 608 |
fprintf( fid, "[ " ); |
| 609 |
for( i=0; i<(n<60 ? n : 60); i++ ) |
| 610 |
if( array[i]!=-1 ) |
| 611 |
fprintf( fid, "%3d ", array[i] ); |
| 612 |
else |
| 613 |
fprintf( fid, " * " ); |
| 614 |
if( n>=30 ) |
| 615 |
fprintf( fid, "... " ); |
| 616 |
fprintf( fid, "]\n" ); |
| 617 |
} |
| 618 |
#endif |
Parent Directory
| 
Revision Log