1 #include <petsctaolinesearch.h> 2 #include <../src/tao/matrix/lmvmmat.h> 3 #include <../src/tao/unconstrained/impls/lmvm/lmvm.h> 4 #include <../src/tao/bound/impls/blmvm/blmvm.h> 5 6 /*------------------------------------------------------------*/ 7 #undef __FUNCT__ 8 #define __FUNCT__ "TaoSolve_BLMVM" 9 static PetscErrorCode TaoSolve_BLMVM(Tao tao) 10 { 11 PetscErrorCode ierr; 12 TAO_BLMVM *blmP = (TAO_BLMVM *)tao->data; 13 TaoConvergedReason reason = TAO_CONTINUE_ITERATING; 14 TaoLineSearchConvergedReason ls_status = TAOLINESEARCH_CONTINUE_ITERATING; 15 PetscReal f, fold, gdx, gnorm; 16 PetscReal stepsize = 1.0,delta; 17 18 PetscFunctionBegin; 19 /* Project initial point onto bounds */ 20 ierr = TaoComputeVariableBounds(tao);CHKERRQ(ierr); 21 ierr = VecMedian(tao->XL,tao->solution,tao->XU,tao->solution);CHKERRQ(ierr); 22 ierr = TaoLineSearchSetVariableBounds(tao->linesearch,tao->XL,tao->XU);CHKERRQ(ierr); 23 24 25 /* Check convergence criteria */ 26 ierr = TaoComputeObjectiveAndGradient(tao, tao->solution,&f,blmP->unprojected_gradient);CHKERRQ(ierr); 27 ierr = VecBoundGradientProjection(blmP->unprojected_gradient,tao->solution, tao->XL,tao->XU,tao->gradient);CHKERRQ(ierr); 28 29 ierr = TaoGradientNorm(tao, tao->gradient,NORM_2,&gnorm);CHKERRQ(ierr); 30 if (PetscIsInfOrNanReal(f) || PetscIsInfOrNanReal(gnorm)) SETERRQ(PETSC_COMM_SELF,1, "User provided compute function generated Inf pr NaN"); 31 32 ierr = TaoMonitor(tao, tao->niter, f, gnorm, 0.0, stepsize, &reason);CHKERRQ(ierr); 33 if (reason != TAO_CONTINUE_ITERATING) PetscFunctionReturn(0); 34 35 /* Set initial scaling for the function */ 36 if (f != 0.0) { 37 delta = 2.0*PetscAbsScalar(f) / (gnorm*gnorm); 38 } else { 39 delta = 2.0 / (gnorm*gnorm); 40 } 41 ierr = MatLMVMSetDelta(blmP->M,delta);CHKERRQ(ierr); 42 43 /* Set counter for gradient/reset steps */ 44 blmP->grad = 0; 45 blmP->reset = 0; 46 47 /* Have not converged; continue with Newton method */ 48 while (reason == TAO_CONTINUE_ITERATING) { 49 /* Compute direction */ 50 ierr = MatLMVMUpdate(blmP->M, tao->solution, tao->gradient);CHKERRQ(ierr); 51 ierr = MatLMVMSolve(blmP->M, blmP->unprojected_gradient, tao->stepdirection);CHKERRQ(ierr); 52 ierr = VecBoundGradientProjection(tao->stepdirection,tao->solution,tao->XL,tao->XU,tao->gradient);CHKERRQ(ierr); 53 54 /* Check for success (descent direction) */ 55 ierr = VecDot(blmP->unprojected_gradient, tao->gradient, &gdx);CHKERRQ(ierr); 56 if (gdx <= 0) { 57 /* Step is not descent or solve was not successful 58 Use steepest descent direction (scaled) */ 59 ++blmP->grad; 60 61 if (f != 0.0) { 62 delta = 2.0*PetscAbsScalar(f) / (gnorm*gnorm); 63 } else { 64 delta = 2.0 / (gnorm*gnorm); 65 } 66 ierr = MatLMVMSetDelta(blmP->M,delta);CHKERRQ(ierr); 67 ierr = MatLMVMReset(blmP->M);CHKERRQ(ierr); 68 ierr = MatLMVMUpdate(blmP->M, tao->solution, blmP->unprojected_gradient);CHKERRQ(ierr); 69 ierr = MatLMVMSolve(blmP->M,blmP->unprojected_gradient, tao->stepdirection);CHKERRQ(ierr); 70 } 71 ierr = VecScale(tao->stepdirection,-1.0);CHKERRQ(ierr); 72 73 /* Perform the linesearch */ 74 fold = f; 75 ierr = VecCopy(tao->solution, blmP->Xold);CHKERRQ(ierr); 76 ierr = VecCopy(blmP->unprojected_gradient, blmP->Gold);CHKERRQ(ierr); 77 ierr = TaoLineSearchSetInitialStepLength(tao->linesearch,1.0);CHKERRQ(ierr); 78 ierr = TaoLineSearchApply(tao->linesearch, tao->solution, &f, blmP->unprojected_gradient, tao->stepdirection, &stepsize, &ls_status);CHKERRQ(ierr); 79 ierr = TaoAddLineSearchCounts(tao);CHKERRQ(ierr); 80 81 if (ls_status != TAOLINESEARCH_SUCCESS && ls_status != TAOLINESEARCH_SUCCESS_USER) { 82 /* Linesearch failed 83 Reset factors and use scaled (projected) gradient step */ 84 ++blmP->reset; 85 86 f = fold; 87 ierr = VecCopy(blmP->Xold, tao->solution);CHKERRQ(ierr); 88 ierr = VecCopy(blmP->Gold, blmP->unprojected_gradient);CHKERRQ(ierr); 89 90 if (f != 0.0) { 91 delta = 2.0* PetscAbsScalar(f) / (gnorm*gnorm); 92 } else { 93 delta = 2.0/ (gnorm*gnorm); 94 } 95 ierr = MatLMVMSetDelta(blmP->M,delta);CHKERRQ(ierr); 96 ierr = MatLMVMReset(blmP->M);CHKERRQ(ierr); 97 ierr = MatLMVMUpdate(blmP->M, tao->solution, blmP->unprojected_gradient);CHKERRQ(ierr); 98 ierr = MatLMVMSolve(blmP->M, blmP->unprojected_gradient, tao->stepdirection);CHKERRQ(ierr); 99 ierr = VecScale(tao->stepdirection, -1.0);CHKERRQ(ierr); 100 101 /* This may be incorrect; linesearch has values fo stepmax and stepmin 102 that should be reset. */ 103 ierr = TaoLineSearchSetInitialStepLength(tao->linesearch,1.0);CHKERRQ(ierr); 104 ierr = TaoLineSearchApply(tao->linesearch,tao->solution,&f, blmP->unprojected_gradient, tao->stepdirection, &stepsize, &ls_status);CHKERRQ(ierr); 105 ierr = TaoAddLineSearchCounts(tao);CHKERRQ(ierr); 106 107 if (ls_status != TAOLINESEARCH_SUCCESS && ls_status != TAOLINESEARCH_SUCCESS_USER) { 108 tao->reason = TAO_DIVERGED_LS_FAILURE; 109 break; 110 } 111 } 112 113 /* Check for converged */ 114 ierr = VecBoundGradientProjection(blmP->unprojected_gradient, tao->solution, tao->XL, tao->XU, tao->gradient);CHKERRQ(ierr); 115 ierr = TaoGradientNorm(tao, tao->gradient, NORM_2, &gnorm);CHKERRQ(ierr); 116 117 118 if (PetscIsInfOrNanReal(f) || PetscIsInfOrNanReal(gnorm)) SETERRQ(PETSC_COMM_SELF,1, "User provided compute function generated Not-a-Number"); 119 tao->niter++; 120 ierr = TaoMonitor(tao, tao->niter, f, gnorm, 0.0, stepsize, &reason);CHKERRQ(ierr); 121 } 122 PetscFunctionReturn(0); 123 } 124 125 #undef __FUNCT__ 126 #define __FUNCT__ "TaoSetup_BLMVM" 127 static PetscErrorCode TaoSetup_BLMVM(Tao tao) 128 { 129 TAO_BLMVM *blmP = (TAO_BLMVM *)tao->data; 130 PetscInt n,N; 131 PetscErrorCode ierr; 132 KSP H0ksp; 133 134 PetscFunctionBegin; 135 /* Existence of tao->solution checked in TaoSetup() */ 136 ierr = VecDuplicate(tao->solution,&blmP->Xold);CHKERRQ(ierr); 137 ierr = VecDuplicate(tao->solution,&blmP->Gold);CHKERRQ(ierr); 138 ierr = VecDuplicate(tao->solution, &blmP->unprojected_gradient);CHKERRQ(ierr); 139 140 if (!tao->stepdirection) { 141 ierr = VecDuplicate(tao->solution, &tao->stepdirection);CHKERRQ(ierr); 142 } 143 if (!tao->gradient) { 144 ierr = VecDuplicate(tao->solution,&tao->gradient);CHKERRQ(ierr); 145 } 146 if (!tao->XL) { 147 ierr = VecDuplicate(tao->solution,&tao->XL);CHKERRQ(ierr); 148 ierr = VecSet(tao->XL,PETSC_NINFINITY);CHKERRQ(ierr); 149 } 150 if (!tao->XU) { 151 ierr = VecDuplicate(tao->solution,&tao->XU);CHKERRQ(ierr); 152 ierr = VecSet(tao->XU,PETSC_INFINITY);CHKERRQ(ierr); 153 } 154 /* Create matrix for the limited memory approximation */ 155 ierr = VecGetLocalSize(tao->solution,&n);CHKERRQ(ierr); 156 ierr = VecGetSize(tao->solution,&N);CHKERRQ(ierr); 157 ierr = MatCreateLMVM(((PetscObject)tao)->comm,n,N,&blmP->M);CHKERRQ(ierr); 158 ierr = MatLMVMAllocateVectors(blmP->M,tao->solution);CHKERRQ(ierr); 159 160 /* If the user has set a matrix to solve as the initial H0, set the options prefix here, and set up the KSP */ 161 if (blmP->H0) { 162 const char *prefix; 163 PC H0pc; 164 165 ierr = MatLMVMSetH0(blmP->M, blmP->H0);CHKERRQ(ierr); 166 ierr = MatLMVMGetH0KSP(blmP->M, &H0ksp);CHKERRQ(ierr); 167 168 ierr = TaoGetOptionsPrefix(tao, &prefix);CHKERRQ(ierr); 169 ierr = KSPSetOptionsPrefix(H0ksp, prefix);CHKERRQ(ierr); 170 ierr = PetscObjectAppendOptionsPrefix((PetscObject)H0ksp, "tao_h0_");CHKERRQ(ierr); 171 ierr = KSPGetPC(H0ksp, &H0pc);CHKERRQ(ierr); 172 ierr = PetscObjectAppendOptionsPrefix((PetscObject)H0pc, "tao_h0_");CHKERRQ(ierr); 173 174 ierr = KSPSetFromOptions(H0ksp);CHKERRQ(ierr); 175 ierr = KSPSetUp(H0ksp);CHKERRQ(ierr); 176 } 177 178 PetscFunctionReturn(0); 179 } 180 181 /* ---------------------------------------------------------- */ 182 #undef __FUNCT__ 183 #define __FUNCT__ "TaoDestroy_BLMVM" 184 static PetscErrorCode TaoDestroy_BLMVM(Tao tao) 185 { 186 TAO_BLMVM *blmP = (TAO_BLMVM *)tao->data; 187 PetscErrorCode ierr; 188 189 PetscFunctionBegin; 190 if (tao->setupcalled) { 191 ierr = MatDestroy(&blmP->M);CHKERRQ(ierr); 192 ierr = VecDestroy(&blmP->unprojected_gradient);CHKERRQ(ierr); 193 ierr = VecDestroy(&blmP->Xold);CHKERRQ(ierr); 194 ierr = VecDestroy(&blmP->Gold);CHKERRQ(ierr); 195 } 196 197 if (blmP->H0) { 198 PetscObjectDereference((PetscObject)blmP->H0); 199 } 200 201 ierr = PetscFree(tao->data);CHKERRQ(ierr); 202 PetscFunctionReturn(0); 203 } 204 205 /*------------------------------------------------------------*/ 206 #undef __FUNCT__ 207 #define __FUNCT__ "TaoSetFromOptions_BLMVM" 208 static PetscErrorCode TaoSetFromOptions_BLMVM(PetscOptionItems* PetscOptionsObject,Tao tao) 209 { 210 PetscErrorCode ierr; 211 212 PetscFunctionBegin; 213 ierr = PetscOptionsHead(PetscOptionsObject,"Limited-memory variable-metric method for bound constrained optimization");CHKERRQ(ierr); 214 ierr = TaoLineSearchSetFromOptions(tao->linesearch);CHKERRQ(ierr); 215 ierr = PetscOptionsTail();CHKERRQ(ierr); 216 PetscFunctionReturn(0); 217 } 218 219 220 /*------------------------------------------------------------*/ 221 #undef __FUNCT__ 222 #define __FUNCT__ "TaoView_BLMVM" 223 static int TaoView_BLMVM(Tao tao, PetscViewer viewer) 224 { 225 TAO_BLMVM *lmP = (TAO_BLMVM *)tao->data; 226 PetscBool isascii; 227 PetscErrorCode ierr; 228 229 PetscFunctionBegin; 230 ierr = PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &isascii);CHKERRQ(ierr); 231 if (isascii) { 232 ierr = PetscViewerASCIIPushTab(viewer);CHKERRQ(ierr); 233 ierr = PetscViewerASCIIPrintf(viewer, "Gradient steps: %D\n", lmP->grad);CHKERRQ(ierr); 234 ierr = PetscViewerASCIIPopTab(viewer);CHKERRQ(ierr); 235 } 236 PetscFunctionReturn(0); 237 } 238 239 #undef __FUNCT__ 240 #define __FUNCT__ "TaoComputeDual_BLMVM" 241 static PetscErrorCode TaoComputeDual_BLMVM(Tao tao, Vec DXL, Vec DXU) 242 { 243 TAO_BLMVM *blm = (TAO_BLMVM *) tao->data; 244 PetscErrorCode ierr; 245 246 PetscFunctionBegin; 247 PetscValidHeaderSpecific(tao,TAO_CLASSID,1); 248 PetscValidHeaderSpecific(DXL,VEC_CLASSID,2); 249 PetscValidHeaderSpecific(DXU,VEC_CLASSID,3); 250 if (!tao->gradient || !blm->unprojected_gradient) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Dual variables don't exist yet or no longer exist.\n"); 251 252 ierr = VecCopy(tao->gradient,DXL);CHKERRQ(ierr); 253 ierr = VecAXPY(DXL,-1.0,blm->unprojected_gradient);CHKERRQ(ierr); 254 ierr = VecSet(DXU,0.0);CHKERRQ(ierr); 255 ierr = VecPointwiseMax(DXL,DXL,DXU);CHKERRQ(ierr); 256 257 ierr = VecCopy(blm->unprojected_gradient,DXU);CHKERRQ(ierr); 258 ierr = VecAXPY(DXU,-1.0,tao->gradient);CHKERRQ(ierr); 259 ierr = VecAXPY(DXU,1.0,DXL);CHKERRQ(ierr); 260 PetscFunctionReturn(0); 261 } 262 263 /* ---------------------------------------------------------- */ 264 /*MC 265 TAOBLMVM - Bounded limited memory variable metric is a quasi-Newton method 266 for nonlinear minimization with bound constraints. It is an extension 267 of TAOLMVM 268 269 Options Database Keys: 270 + -tao_lmm_vectors - number of vectors to use for approximation 271 . -tao_lmm_scale_type - "none","scalar","broyden" 272 . -tao_lmm_limit_type - "none","average","relative","absolute" 273 . -tao_lmm_rescale_type - "none","scalar","gl" 274 . -tao_lmm_limit_mu - mu limiting factor 275 . -tao_lmm_limit_nu - nu limiting factor 276 . -tao_lmm_delta_min - minimum delta value 277 . -tao_lmm_delta_max - maximum delta value 278 . -tao_lmm_broyden_phi - phi factor for Broyden scaling 279 . -tao_lmm_scalar_alpha - alpha factor for scalar scaling 280 . -tao_lmm_rescale_alpha - alpha factor for rescaling diagonal 281 . -tao_lmm_rescale_beta - beta factor for rescaling diagonal 282 . -tao_lmm_scalar_history - amount of history for scalar scaling 283 . -tao_lmm_rescale_history - amount of history for rescaling diagonal 284 - -tao_lmm_eps - rejection tolerance 285 286 Level: beginner 287 M*/ 288 #undef __FUNCT__ 289 #define __FUNCT__ "TaoCreate_BLMVM" 290 PETSC_EXTERN PetscErrorCode TaoCreate_BLMVM(Tao tao) 291 { 292 TAO_BLMVM *blmP; 293 const char *morethuente_type = TAOLINESEARCHMT; 294 PetscErrorCode ierr; 295 296 PetscFunctionBegin; 297 tao->ops->setup = TaoSetup_BLMVM; 298 tao->ops->solve = TaoSolve_BLMVM; 299 tao->ops->view = TaoView_BLMVM; 300 tao->ops->setfromoptions = TaoSetFromOptions_BLMVM; 301 tao->ops->destroy = TaoDestroy_BLMVM; 302 tao->ops->computedual = TaoComputeDual_BLMVM; 303 304 ierr = PetscNewLog(tao,&blmP);CHKERRQ(ierr); 305 blmP->H0 = NULL; 306 tao->data = (void*)blmP; 307 308 /* Override default settings (unless already changed) */ 309 if (!tao->max_it_changed) tao->max_it = 2000; 310 if (!tao->max_funcs_changed) tao->max_funcs = 4000; 311 312 ierr = TaoLineSearchCreate(((PetscObject)tao)->comm, &tao->linesearch);CHKERRQ(ierr); 313 ierr = TaoLineSearchSetType(tao->linesearch, morethuente_type);CHKERRQ(ierr); 314 ierr = TaoLineSearchUseTaoRoutines(tao->linesearch,tao);CHKERRQ(ierr); 315 ierr = TaoLineSearchSetOptionsPrefix(tao->linesearch,tao->hdr.prefix);CHKERRQ(ierr); 316 PetscFunctionReturn(0); 317 } 318 319 #undef __FUNCT__ 320 #define __FUNCT__ "TaoLMVMSetH0" 321 PETSC_EXTERN PetscErrorCode TaoLMVMSetH0(Tao tao, Mat H0) 322 { 323 TAO_LMVM *lmP; 324 TAO_BLMVM *blmP; 325 const TaoType type; 326 PetscBool is_lmvm, is_blmvm; 327 328 PetscErrorCode ierr; 329 330 ierr = TaoGetType(tao, &type);CHKERRQ(ierr); 331 ierr = PetscStrcmp(type, TAOLMVM, &is_lmvm);CHKERRQ(ierr); 332 ierr = PetscStrcmp(type, TAOBLMVM, &is_blmvm);CHKERRQ(ierr); 333 334 if (is_lmvm) { 335 lmP = (TAO_LMVM *)tao->data; 336 ierr = PetscObjectReference((PetscObject)H0);CHKERRQ(ierr); 337 lmP->H0 = H0; 338 } else if (is_blmvm) { 339 blmP = (TAO_BLMVM *)tao->data; 340 ierr = PetscObjectReference((PetscObject)H0);CHKERRQ(ierr); 341 blmP->H0 = H0; 342 } else SETERRQ(PetscObjectComm((PetscObject)tao), PETSC_ERR_ARG_WRONGSTATE, "This routine applies to TAO_LMVM and TAO_BLMVM."); 343 344 PetscFunctionReturn(0); 345 } 346 347 #undef __FUNCT__ 348 #define __FUNCT__ "TaoLMVMGetH0" 349 PETSC_EXTERN PetscErrorCode TaoLMVMGetH0(Tao tao, Mat *H0) 350 { 351 TAO_LMVM *lmP; 352 TAO_BLMVM *blmP; 353 const TaoType type; 354 PetscBool is_lmvm, is_blmvm; 355 Mat M; 356 357 PetscErrorCode ierr; 358 359 ierr = TaoGetType(tao, &type);CHKERRQ(ierr); 360 ierr = PetscStrcmp(type, TAOLMVM, &is_lmvm);CHKERRQ(ierr); 361 ierr = PetscStrcmp(type, TAOBLMVM, &is_blmvm);CHKERRQ(ierr); 362 363 if (is_lmvm) { 364 lmP = (TAO_LMVM *)tao->data; 365 M = lmP->M; 366 } else if (is_blmvm) { 367 blmP = (TAO_BLMVM *)tao->data; 368 M = blmP->M; 369 } else SETERRQ(PetscObjectComm((PetscObject)tao), PETSC_ERR_ARG_WRONGSTATE, "This routine applies to TAO_LMVM and TAO_BLMVM."); 370 371 ierr = MatLMVMGetH0(M, H0);CHKERRQ(ierr); 372 PetscFunctionReturn(0); 373 } 374 375 #undef __FUNCT__ 376 #define __FUNCT__ "TaoLMVMGetH0KSP" 377 PETSC_EXTERN PetscErrorCode TaoLMVMGetH0KSP(Tao tao, KSP *ksp) 378 { 379 TAO_LMVM *lmP; 380 TAO_BLMVM *blmP; 381 const TaoType type; 382 PetscBool is_lmvm, is_blmvm; 383 Mat M; 384 PetscErrorCode ierr; 385 386 ierr = TaoGetType(tao, &type);CHKERRQ(ierr); 387 ierr = PetscStrcmp(type, TAOLMVM, &is_lmvm);CHKERRQ(ierr); 388 ierr = PetscStrcmp(type, TAOBLMVM, &is_blmvm);CHKERRQ(ierr); 389 390 if (is_lmvm) { 391 lmP = (TAO_LMVM *)tao->data; 392 M = lmP->M; 393 } else if (is_blmvm) { 394 blmP = (TAO_BLMVM *)tao->data; 395 M = blmP->M; 396 } else SETERRQ(PetscObjectComm((PetscObject)tao), PETSC_ERR_ARG_WRONGSTATE, "This routine applies to TAO_LMVM and TAO_BLMVM."); 397 398 ierr = MatLMVMGetH0KSP(M, ksp);CHKERRQ(ierr); 399 PetscFunctionReturn(0); 400 } 401