Attachment 'jgd_solve.patch'

Download

   1 # HG changeset patch
   2 # User Clement Pernet <clement.pernet@gmail.com>
   3 # Date 1224064381 -7200
   4 # Node ID 27b3032e924cbab41eb79da139e41d02f2533c6e
   5 # Parent  7d525bf8b16f4c51f6e45e0b09dda5cdf3eae34c
   6 * fixed bug in permutation applytrans
   7 * added test_zero
   8 * added license in parity
   9 * added linear system solving using pluq
  10 * added appropriate test suite for solve
  11 
  12 diff -r 7d525bf8b16f -r 27b3032e924c src/packedmatrix.c
  13 --- a/src/packedmatrix.c	Sat Oct 11 10:16:18 2008 +0200
  14 +++ b/src/packedmatrix.c	Wed Oct 15 11:53:01 2008 +0200
  15 @@ -984,6 +984,45 @@ permutation *mzd_col_block_rotate(packed
  16    return P;
  17  }
  18  
  19 +
  20 +
  21 +
  22 +int mzd_test_zero(packedmatrix *A)
  23 +{
  24 +  /* Could be improved: stopping as the first non zero value is found (status!=0)*/
  25 +  size_t mb = A->nrows;
  26 +  size_t nb = A->ncols;
  27 +  size_t Aoffset = A->offset;
  28 +  size_t nbrest = (nb + Aoffset) % RADIX;
  29 +  int status=0;
  30 +  if (nb + Aoffset >= RADIX) {
  31 +          // Large A
  32 +    word mask_begin = RIGHT_BITMASK(RADIX-Aoffset);
  33 +    if (Aoffset == 0)
  34 +      mask_begin = ~mask_begin;
  35 +    word mask_end = LEFT_BITMASK(nbrest);
  36 +    size_t i;
  37 +    for (i=0; i<mb; ++i) {
  38 +        status |= A->values [A->rowswap [i]] & mask_begin;
  39 +        size_t j;
  40 +        for ( j = 1; j < A->width-1; ++j)
  41 +            status |= A->values [A->rowswap [i] + j];
  42 +        status |= A->values [A->rowswap [i] + A->width - 1] & mask_end;
  43 +    }
  44 +  } else {
  45 +          // Small A
  46 +    word mask = ((ONE << nb) - 1) ;
  47 +    mask <<= (RADIX-nb-Aoffset);
  48 +
  49 +    size_t i;
  50 +    for (i=0; i < mb; ++i) {
  51 +        status |= A->values [A->rowswap [i]] & mask;
  52 +    }
  53 +  }
  54 +  
  55 +  return !status;
  56 +}
  57 +
  58  void mzd_apply_p_left(packedmatrix *A, permutation *P) {
  59    assert(A->offset == 0);
  60    size_t i;
  61 @@ -995,8 +1034,8 @@ void mzd_apply_p_left(packedmatrix *A, p
  62  
  63  void mzd_apply_p_left_trans(packedmatrix *A, permutation *P) {
  64    assert(A->offset == 0);
  65 -  size_t i;
  66 -  for (i=0; i<P->length; i++) {
  67 +  int i;
  68 +  for (i=P->length-1; i>=0; i--) {
  69      if(P->values[i] != i) 
  70        mzd_row_swap(A, i, P->values[i]);
  71    }
  72 @@ -1004,8 +1043,8 @@ void mzd_apply_p_left_trans(packedmatrix
  73  
  74  void mzd_apply_p_right_trans(packedmatrix *A, permutation *P) {
  75    assert(A->offset == 0);
  76 -  size_t i;
  77 -  for (i=0; i<P->length; i++) {
  78 +  int i;
  79 +  for (i = P->length-1;i>=0; i--) {
  80      if(P->values[i] != i) 
  81        mzd_col_swap(A, i, P->values[i]);
  82    }
  83 @@ -1014,7 +1053,7 @@ void mzd_apply_p_right(packedmatrix *A, 
  84  void mzd_apply_p_right(packedmatrix *A, permutation *P) {
  85    assert(A->offset == 0);
  86    size_t i;
  87 -  for (i=0; 0<P->length; i++) {
  88 +  for (i=0; i<P->length; i++) {
  89      if(P->values[i] != i) 
  90        mzd_col_swap(A, i, P->values[i]);
  91    }
  92 diff -r 7d525bf8b16f -r 27b3032e924c src/packedmatrix.h
  93 --- a/src/packedmatrix.h	Sat Oct 11 10:16:18 2008 +0200
  94 +++ b/src/packedmatrix.h	Wed Oct 15 11:53:01 2008 +0200
  95 @@ -762,6 +762,17 @@ static inline void mzd_clear_bits(const 
  96    }
  97  }
  98  
  99 +
 100 +
 101 +/**
 102 + * \brief Zero test for matrix.
 103 + *
 104 + * \param A Input matrix.
 105 + *
 106 + */
 107 +int mzd_test_zero(packedmatrix *A) ;
 108 +
 109 +
 110  /**
 111   * Rotate zero columns to the end.
 112   *
 113 diff -r 7d525bf8b16f -r 27b3032e924c src/parity.h
 114 --- a/src/parity.h	Sat Oct 11 10:16:18 2008 +0200
 115 +++ b/src/parity.h	Wed Oct 15 11:53:01 2008 +0200
 116 @@ -5,6 +5,27 @@
 117   *
 118   * \author David Harvey
 119   */
 120 +
 121 +#ifndef PARITY_H
 122 +#define PARITY_H
 123 + /*******************************************************************
 124 + *
 125 + *            M4RI: Method of the Four Russians Inversion
 126 + *
 127 + *       Copyright (C) 2008 David Harvey
 128 + *
 129 + *  Distributed under the terms of the GNU General Public License (GPL)
 130 + *
 131 + *    This code is distributed in the hope that it will be useful,
 132 + *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 133 + *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 134 + *    General Public License for more details.
 135 + *
 136 + *  The full text of the GPL is available at:
 137 + *
 138 + *                  http://www.gnu.org/licenses/
 139 + *
 140 + ********************************************************************/
 141  
 142  /**
 143   * \brief Step for mixing two 64-bit words to compute their parity.
 144 @@ -98,3 +119,5 @@ static inline word parity64(word* buf)
 145  
 146     return MIX1(e1, e0);
 147  }
 148 +
 149 +#endif
 150 diff -r 7d525bf8b16f -r 27b3032e924c src/solve.c
 151 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
 152 +++ b/src/solve.c	Wed Oct 15 11:53:01 2008 +0200
 153 @@ -0,0 +1,151 @@
 154 + /*******************************************************************
 155 + *
 156 + *            M4RI: Method of the Four Russians Inversion
 157 + *
 158 + *       Copyright (C) 2008 Jean-Guillaume.Dumas@imag.fr
 159 + *
 160 + *  Distributed under the terms of the GNU General Public License (GPL)
 161 + *
 162 + *    This code is distributed in the hope that it will be useful,
 163 + *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 164 + *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 165 + *    General Public License for more details.
 166 + *
 167 + *  The full text of the GPL is available at:
 168 + *
 169 + *                  http://www.gnu.org/licenses/
 170 + *
 171 + ********************************************************************/
 172 +
 173 +#include "solve.h"
 174 +#include "strassen.h"
 175 +#include "pluq.h"
 176 +#include "trsm.h"
 177 +#include "permutation.h"
 178 +
 179 +/**
 180 + * \brief System solving with matrix.
 181 + *
 182 + * Solves A X = B where A, and B are input matrices.
 183 + * The solution X is stored inplace on B
 184 + * * 
 185 + *
 186 + * \param A Input matrix.
 187 + * \param B Input matrix, being overwritten by the solution matrix X
 188 + * \param cutoff Minimal dimension for Strassen recursion.
 189 + *
 190 + */
 191 +void mzd_solve_left (packedmatrix *A, packedmatrix *B, const int cutoff, const int InconsistencyCheck) 
 192 +{    
 193 +  if(A->ncols > B->nrows)
 194 +    m4ri_die("mzd_solve_left: A ncols (%d) need to be lower than B nrows (%d).\n", A->ncols, B->nrows);
 195 +  if (cutoff <= 0)
 196 +    m4ri_die("mzd_solve_left: cutoff must be > 0.\n");
 197 +
 198 +  _mzd_solve_left (A, B, cutoff, InconsistencyCheck);
 199 +}
 200 + 
 201 +void mzd_pluq_solve_left (packedmatrix *A, size_t rank, 
 202 +                          permutation *P, permutation *Q, 
 203 +                          packedmatrix *B, const int cutoff, const int InconsistencyCheck) 
 204 +{
 205 +  if(A->ncols > B->nrows)
 206 +    m4ri_die("mzd_pluq_solve_left: A ncols (%d) need to be lower than B nrows (%d).\n", A->ncols, B->nrows);
 207 +  if(P->length != A->nrows)
 208 +      m4ri_die("mzd_pluq_solve_left: A nrows (%d) need to match P size (%d).\n", A->nrows, P->length);
 209 +  if(Q->length != A->ncols)
 210 +      m4ri_die("mzd_pluq_solve_left: A ncols (%d) need to match Q size (%d).\n", A->ncols, P->length);
 211 +
 212 +  if (cutoff <= 0)
 213 +    m4ri_die("mzd_pluq_solve_left: cutoff must be > 0.\n");
 214 +
 215 +  _mzd_pluq_solve_left (A, rank, P, Q, B, cutoff, InconsistencyCheck);
 216 +}
 217 +
 218 +void _mzd_pluq_solve_left (packedmatrix *A, size_t rank, 
 219 +                           permutation *P, permutation *Q, 
 220 +                           packedmatrix *B, const int cutoff, const int InconsistencyCheck) 
 221 +{
 222 +        /** A is supposed to store L lower triangular and U upper triangular
 223 +         *  B is modified in place 
 224 +         *  (Bi's in the comments are just modified versions of B)
 225 +         *  PLUQ = A
 226 +         *  1°) P B2 = B1
 227 +         *  2°) L B3 = B2
 228 +         *  3°) U B4 = B3
 229 +         *  4°) Q B5 = B4
 230 +         */
 231 +    
 232 +        /* P B2 = B1 or B2 = P^T B1*/
 233 +    mzd_apply_p_left_trans(B, P);
 234 +    
 235 +        /* L B3 = B2 */
 236 +   
 237 +        /* view on the upper part of L */
 238 +    packedmatrix *LU = mzd_init_window(A,0,0,rank,rank);
 239 +    packedmatrix *Y1 = mzd_init_window(B,0,0,rank,B->ncols);
 240 +    _mzd_trsm_lower_left(LU, Y1, cutoff);
 241 +    
 242 +    if (InconsistencyCheck) {
 243 +          /* Check for inconsistency */
 244 +          /** FASTER without this check
 245 +           * 
 246 +           * update with the lower part of L 
 247 +           */
 248 +      packedmatrix *H = mzd_init_window(A,rank,0,A->nrows,rank);
 249 +      packedmatrix *Y2 = mzd_init_window(B,rank,0,B->nrows,B->ncols);
 250 +      mzd_addmul(Y2, H, Y1, cutoff);
 251 +          /*
 252 +           * test whether Y2 is the zero matrix
 253 +           */
 254 +      if(! mzd_test_zero(Y2) ) {
 255 +          printf("inconsistent system of size %d x %d\n", Y2->nrows, Y2->ncols);
 256 +          printf("Y2=");
 257 +          mzd_print_matrix(Y2);
 258 +      }
 259 +      mzd_free_window(H);
 260 +      mzd_free_window(Y2);
 261 +    }
 262 +        /* U B4 = B3 */
 263 +    _mzd_trsm_upper_left(LU, Y1, cutoff);
 264 +    mzd_free_window(LU);
 265 +    mzd_free_window(Y1);
 266 +
 267 +    if (! InconsistencyCheck) {
 268 +       /** Default is to set the indefined bits to zero 
 269 +         * if inconsistency has been checked then 
 270 +         *    Y2 bits are already all zeroes
 271 +         * thus this clearing is not needed
 272 +         */
 273 +      if (rank < B->nrows) mzd_clear_bits(B,rank,0, (B->nrows-rank)*B->ncols);
 274 +    }
 275 +    
 276 +        /* Q B5 = B4 or B5 = Q^T B4*/
 277 +    mzd_apply_p_left_trans(B, Q);
 278 +        /* P L U Q B5 = B1 */
 279 +}
 280 +
 281 +
 282 +void _mzd_solve_left (packedmatrix *A, packedmatrix *B, const int cutoff, const int InconsistencyCheck) 
 283 +{
 284 +        /**
 285 +         *  B is modified in place 
 286 +         *  (Bi's in the comments are just modified versions of B)
 287 +         *  1°) PLUQ = A
 288 +         *  2°) P B2 = B1
 289 +         *  3°) L B3 = B2
 290 +         *  4°) U B4 = B3
 291 +         *  5°) Q B5 = B4
 292 +         */
 293 +    permutation * P = mzp_init(A->nrows);
 294 +    permutation * Q = mzp_init(A->ncols);
 295 +    
 296 +        /* PLUQ = A */
 297 +    size_t rank = _mzd_pluq(A, P, Q, cutoff);  
 298 +        /* 2°, 3°, 4°, 5° */
 299 +    mzd_pluq_solve_left(A, rank, P, Q, B, cutoff, InconsistencyCheck);
 300 +
 301 +    mzp_free(P);
 302 +    mzp_free(Q);
 303 +}
 304 +
 305 diff -r 7d525bf8b16f -r 27b3032e924c src/solve.h
 306 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
 307 +++ b/src/solve.h	Wed Oct 15 11:53:01 2008 +0200
 308 @@ -0,0 +1,123 @@
 309 +/**
 310 + * \file solve.h
 311 + *
 312 + * \brief system solving with matrix routines.
 313 + *
 314 + * \author Jean-Guillaume Dumas <Jean-Guillaume.Dumas@imag.fr>
 315 + *
 316 + */
 317 +#ifndef SOLVE_H
 318 +#define SOLVE_H
 319 + /*******************************************************************
 320 + *
 321 + *            M4RI: Method of the Four Russians Inversion
 322 + *
 323 + *       Copyright (C) 2008 Jean-Guillaume.Dumas@imag.fr
 324 + *
 325 + *  Distributed under the terms of the GNU General Public License (GPL)
 326 + *
 327 + *    This code is distributed in the hope that it will be useful,
 328 + *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 329 + *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 330 + *    General Public License for more details.
 331 + *
 332 + *  The full text of the GPL is available at:
 333 + *
 334 + *                  http://www.gnu.org/licenses/
 335 + *
 336 + ********************************************************************/
 337 +
 338 +
 339 +#include "misc.h"
 340 +#include "packedmatrix.h"
 341 +#include "stdio.h"
 342 +
 343 +/**
 344 + * \brief System solving with matrix.
 345 + *
 346 + * Solves A X = B where A, and B are input matrices.
 347 + * The solution X is stored inplace on B
 348 + *
 349 + * \param A Input matrix.
 350 + * \param B Input matrix, being overwritten by the solution matrix X
 351 + * \param cutoff Minimal dimension for Strassen recursion.
 352 + * \param InconsistencyCheck decide wether or not to check for incosistency (faster without but output not defined if system is not consistent).
 353 + *
 354 + */
 355 +void mzd_solve_left (packedmatrix *A, packedmatrix *B, const int cutoff, const int InconsistencyCheck);
 356 +
 357 +
 358 +/**
 359 + * \brief System solving with matrix.
 360 + *
 361 + * Solves (P L U Q) X = B where 
 362 + * A is an input matrix supposed to store both:
 363 + *      an upper right triangular matrix U
 364 + *      a lower left unitary triangular matrix L
 365 + * P and Q are permutation matrices
 366 + * B is the input matrix to be solved.
 367 + * The solution X is stored inplace on B
 368 + * This version assumes that the matrices are at an even position on
 369 + * the RADIX grid and that their dimension is a multiple of RADIX.
 370 + *
 371 + * \param A Input upper/lower triangular matrices.
 372 + * \param rank is rank of A.
 373 + * \param P Input row permutation matrix.
 374 + * \param Q Input column permutation matrix.
 375 + * \param B Input matrix, being overwritten by the solution matrix X.
 376 + * \param cutoff Minimal dimension for Strassen recursion.
 377 + * \param InconsistencyCheck decide whether or not to check for incosistency (faster without but output not defined if system is not consistent).
 378 + * 
 379 + */
 380 +void mzd_pluq_solve_left (packedmatrix *A, size_t rank, 
 381 +                          permutation *P, permutation *Q, 
 382 +                          packedmatrix *B, const int cutoff, const int InconsistencyCheck);
 383 +
 384 +
 385 +
 386 +/**
 387 + * \brief System solving with matrix.
 388 + *
 389 + * Solves (P L U Q) X = B where 
 390 + * A is an input matrix supposed to store both:
 391 + *      an upper right triangular matrix U
 392 + *      a lower left unitary triangular matrix L
 393 + * P and Q are permutation matrices
 394 + * B is the input matrix to be solved.
 395 + * The solution X is stored inplace on B
 396 + * This version assumes that the matrices are at an even position on
 397 + * the RADIX grid and that their dimension is a multiple of RADIX.
 398 + *
 399 + * \param A Input upper/lower triangular matrices.
 400 + * \param rank is rank of A.
 401 + * \param P Input row permutation matrix.
 402 + * \param Q Input column permutation matrix.
 403 + * \param B Input matrix, being overwritten by the solution matrix X.
 404 + * \param cutoff Minimal dimension for Strassen recursion.
 405 + * \param InconsistencyCheck decide whether or not to check for incosistency (faster without but output not defined if system is not consistent).
 406 + *
 407 + * \internal
 408 + */
 409 +void _mzd_pluq_solve_left (packedmatrix *A, size_t rank, 
 410 +                           permutation *P, permutation *Q, 
 411 +                           packedmatrix *B, const int cutoff, const int InconsistencyCheck);
 412 +
 413 +/**
 414 + * \brief System solving with matrix.
 415 + *
 416 + * Solves A X = B where A, and B are input matrices.
 417 + * The solution X is stored inplace on B
 418 + * This version assumes that the matrices are at an even position on
 419 + * the RADIX grid and that their dimension is a multiple of RADIX.
 420 + *
 421 + * \param A Input matrix.
 422 + * \param B Input matrix, being overwritten by the solution matrix X.
 423 + * \param cutoff Minimal dimension for Strassen recursion.
 424 + * \param InconsistencyCheck decide whether or not to check for incosistency (faster without but output not defined if system is not consistent).
 425 + *
 426 + * \internal
 427 + */
 428 +void _mzd_solve_left (packedmatrix *A, packedmatrix *B, const int cutoff, const int InconsistencyCheck);
 429 +
 430 +
 431 +#endif // SOLVE_H
 432 diff -r 7d525bf8b16f -r 27b3032e924c testsuite/test_solve.c
 433 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
 434 +++ b/testsuite/test_solve.c	Wed Oct 15 11:53:01 2008 +0200
 435 @@ -0,0 +1,156 @@
 436 +#include <stdlib.h>
 437 +#include "m4ri.h"
 438 +#include "strassen.h"
 439 +#include "trsm.h"
 440 +#include "solve.h"
 441 +
 442 +int test_pluq_solve_left (int m, int n, int offsetA, int offsetB){
 443 +  packedmatrix* Abase = mzd_init (2048,2048);
 444 +  packedmatrix* Bbase = mzd_init (2048,2048);
 445 +  mzd_randomize (Abase);
 446 +  mzd_randomize (Bbase);
 447 +  packedmatrix* Bbasecopy = mzd_copy (NULL, Bbase);
 448 +
 449 +  packedmatrix* A = mzd_init_window (Abase, 0, offsetA, m, offsetA + m);
 450 +  packedmatrix* B = mzd_init_window (Bbase, 0, offsetB, m, offsetB + n);
 451 +  
 452 +  size_t i,j;
 453 +
 454 +  packedmatrix* W = mzd_init (B->nrows, B->ncols);
 455 +  for ( i=0; i<B->nrows; ++i)
 456 +      for ( j=0; j<B->ncols; ++j)
 457 +          mzd_write_bit(W,i,j, mzd_read_bit (B,i,j));
 458 +
 459 +  for (i=0; i<m; ++i){
 460 +    mzd_write_bit(A,i,i, 1);
 461 +  }  
 462 +
 463 +  mzd_solve_left(A, B, 2048, 1);
 464 +
 465 +/*  mzd_addmul(W, A, B, 2048); */
 466 +  packedmatrix *L = mzd_init(A->nrows,A->nrows);
 467 +  for ( i=0; i<m; ++i)
 468 +      for ( j=0; j<=i; ++j)
 469 +          if (mzd_read_bit (A,i,j)) 
 470 +              mzd_write_bit(L,i,j,1);
 471 +  
 472 +
 473 +  packedmatrix *U = mzd_init(A->nrows,A->ncols);
 474 +  for ( i=0; i<A->nrows; ++i)
 475 +      for ( j=i; j<A->ncols; ++j)
 476 +          if (mzd_read_bit (A,i,j)) 
 477 +              mzd_write_bit(U,i,j,1);
 478 +
 479 +  packedmatrix *X = mzd_init(B->nrows,B->ncols);
 480 +  for ( i=0; i<B->nrows; ++i)
 481 +      for ( j=0; j<B->ncols; ++j)
 482 +          mzd_write_bit(X,i,j, mzd_read_bit (B,i,j));
 483 +  packedmatrix *T = mzd_init(A->nrows,B->ncols);
 484 +  mzd_mul(T, U, X, 2048);
 485 +
 486 +  packedmatrix *H = mzd_init(B->nrows,B->ncols);
 487 +  mzd_mul(H, L, T, 2048);
 488 +
 489 +  packedmatrix *Z = mzd_init(B->nrows,B->ncols);
 490 +
 491 +  mzd_add(Z, W, H);
 492 +
 493 +  int status = 0;
 494 +  for ( i=0; i<m; ++i)
 495 +    for ( j=0; j<n; ++j){
 496 +      if (mzd_read_bit (Z,i,j)){
 497 +	status = 1;
 498 +      }
 499 +    }
 500 +/*
 501 +  // Verifiying that nothing has been changed around the submatrices
 502 +  mzd_addmul(W, A, B, 2048);
 503 +
 504 +  mzd_copy (B, W);
 505 +
 506 +  for ( i=0; i<2048; ++i)
 507 +    for ( j=0; j<2048/RADIX; ++j){
 508 +      if (Bbase->values [Bbase->rowswap[i] + j] != Bbasecopy->values [Bbasecopy->rowswap[i] + j]){
 509 +	status = 1;
 510 +      }
 511 +    }
 512 +*/
 513 +
 514 +  mzd_free(L);
 515 +  mzd_free(U);
 516 +  mzd_free(T);
 517 +  mzd_free(H);
 518 +  mzd_free(Z);
 519 +  mzd_free_window (A);
 520 +  mzd_free_window (B);
 521 +  mzd_free_window (W);
 522 +  mzd_free(Abase);
 523 +  mzd_free(Bbase);
 524 +  mzd_free(Bbasecopy);
 525 +
 526 +  if (!status)
 527 +    printf("passed\n");
 528 +  else
 529 +    printf("failed\n");
 530 +  return status;
 531 +}
 532 +
 533 +int main(int argc, char **argv) {
 534 +  int status = 0;
 535 +
 536 +  printf("SolveLeft: small A even, small B odd  ... ");
 537 +  status += test_pluq_solve_left (2, 4, 0, 1);
 538 +
 539 +  printf("SolveLeft: small A even, small B even ... ");
 540 +  status += test_pluq_solve_left (10, 20, 0, 0);
 541 +  printf("SolveLeft: small A even, large B even ... ");
 542 +  status += test_pluq_solve_left (10, 80, 0, 0);
 543 +
 544 +  printf("SolveLeft: small A even, small B odd  ... ");
 545 +  status += test_pluq_solve_left (10, 20, 0, 15);
 546 +  printf("SolveLeft: small A even, large B odd  ... ");
 547 +  status += test_pluq_solve_left (10, 80, 0, 15);
 548 +
 549 +  printf("SolveLeft: small A odd, small B even  ... ");
 550 +  status += test_pluq_solve_left (10, 20, 15, 0);
 551 +  printf("SolveLeft: small A odd, large B even  ... ");
 552 +  status += test_pluq_solve_left (10, 80, 15, 0);
 553 +
 554 +  printf("SolveLeft: small A odd, small B odd   ... ");
 555 +  status += test_pluq_solve_left (10, 20, 15, 20);
 556 +  printf("SolveLeft: small A odd, large B odd   ... ");
 557 +  status += test_pluq_solve_left (10, 80, 15, 20);
 558 +
 559 +  printf("SolveLeft: large A even, small B even ... ");
 560 +  status += test_pluq_solve_left (70, 20, 0, 0);
 561 +  printf("SolveLeft: large A even, large B even ... ");
 562 +  status += test_pluq_solve_left (70, 80, 0, 0);
 563 +
 564 +  printf("SolveLeft: large A even, small B odd  ... ");
 565 +  status += test_pluq_solve_left (70, 10, 0, 15);
 566 +  printf("SolveLeft: large A even, large B odd  ... ");
 567 +  status += test_pluq_solve_left (70, 80, 0, 15);
 568 +
 569 +  printf("SolveLeft: large A odd, small B even  ... ");
 570 +  status += test_pluq_solve_left (70, 20, 15, 0);
 571 +  printf("SolveLeft: large A odd, large B even  ... ");
 572 +  status += test_pluq_solve_left (70, 80, 15, 0);
 573 +
 574 +  printf("SolveLeft: large A odd, small B odd   ... ");
 575 +  status += test_pluq_solve_left (70, 20, 15, 20);
 576 +  printf("SolveLeft: large A odd, large B odd   ... ");
 577 +  status += test_pluq_solve_left (70, 80, 15, 20);
 578 +
 579 +  printf("SolveLeft: larger A odd, larger B odd ... ");
 580 +  status += test_pluq_solve_left (770, 1600, 75, 89);
 581 +  printf("SolveLeft: larger A odd, larger B odd ... ");
 582 +  status += test_pluq_solve_left (1764, 1345, 198, 123);
 583 +
 584 +  
 585 +  if (!status) {
 586 +    printf("All tests passed.\n");
 587 +  }
 588 +
 589 +  m4ri_destroy_all_codes();
 590 +  return 0;
 591 +}
 592 

Attached Files

To refer to attachments on a page, use attachment:filename, as shown below in the list of files. Do NOT use the URL of the [get] link, since this is subject to change and can break easily.
  • [get | view] (2008-10-14 14:37:00, 30.0 KB) [[attachment:Capture-emacs@maruta.png]]
  • [get | view] (2008-10-14 15:39:57, 12.4 KB) [[attachment:f5.py]]
  • [get | view] (2008-10-29 13:41:03, 22.7 KB) [[attachment:f5.pyx]]
  • [get | view] (2008-10-23 05:36:34, 4.8 KB) [[attachment:gema.sage]]
  • [get | view] (2008-10-15 10:01:59, 18.7 KB) [[attachment:jgd_solve.patch]]
  • [get | view] (2008-10-15 09:30:20, 31.2 KB) [[attachment:m4ri_trsm_UL_LR.patch]]
  • [get | view] (2008-10-16 12:57:59, 10.8 KB) [[attachment:sage-view.el]]
 All files | Selected Files: delete move to page copy to page

You are not allowed to attach a file to this page.