1 #include <../src/tao/leastsquares/impls/pounders/pounders.h> 2 3 static PetscErrorCode pounders_h(Tao subtao, Vec v, Mat H, Mat Hpre, void *ctx) 4 { 5 PetscFunctionBegin; 6 PetscFunctionReturn(0); 7 } 8 9 static PetscErrorCode pounders_fg(Tao subtao, Vec x, PetscReal *f, Vec g, void *ctx) 10 { 11 TAO_POUNDERS *mfqP = (TAO_POUNDERS*)ctx; 12 PetscReal d1,d2; 13 PetscErrorCode ierr; 14 15 PetscFunctionBegin; 16 /* g = A*x (add b later)*/ 17 ierr = MatMult(mfqP->subH,x,g);CHKERRQ(ierr); 18 19 /* f = 1/2 * x'*(Ax) + b'*x */ 20 ierr = VecDot(x,g,&d1);CHKERRQ(ierr); 21 ierr = VecDot(mfqP->subb,x,&d2);CHKERRQ(ierr); 22 *f = 0.5 *d1 + d2; 23 24 /* now g = g + b */ 25 ierr = VecAXPY(g, 1.0, mfqP->subb);CHKERRQ(ierr); 26 PetscFunctionReturn(0); 27 } 28 29 static PetscErrorCode pounders_feval(Tao tao, Vec x, Vec F, PetscReal *fsum) 30 { 31 PetscErrorCode ierr; 32 TAO_POUNDERS *mfqP = (TAO_POUNDERS*)tao->data; 33 PetscInt i,row,col; 34 PetscReal fr,fc; 35 36 PetscFunctionBegin; 37 ierr = TaoComputeResidual(tao,x,F);CHKERRQ(ierr); 38 if (tao->res_weights_v) { 39 ierr = VecPointwiseMult(mfqP->workfvec,tao->res_weights_v,F);CHKERRQ(ierr); 40 ierr = VecDot(mfqP->workfvec,mfqP->workfvec,fsum);CHKERRQ(ierr); 41 } else if (tao->res_weights_w) { 42 *fsum=0; 43 for (i=0;i<tao->res_weights_n;i++) { 44 row=tao->res_weights_rows[i]; 45 col=tao->res_weights_cols[i]; 46 ierr = VecGetValues(F,1,&row,&fr);CHKERRQ(ierr); 47 ierr = VecGetValues(F,1,&col,&fc);CHKERRQ(ierr); 48 *fsum += tao->res_weights_w[i]*fc*fr; 49 } 50 } else { 51 ierr = VecDot(F,F,fsum);CHKERRQ(ierr); 52 } 53 ierr = PetscInfo1(tao,"Least-squares residual norm: %20.19e\n",(double)*fsum);CHKERRQ(ierr); 54 if (PetscIsInfOrNanReal(*fsum)) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_USER, "User provided compute function generated Inf or NaN"); 55 PetscFunctionReturn(0); 56 } 57 58 static PetscErrorCode gqtwrap(Tao tao,PetscReal *gnorm, PetscReal *qmin) 59 { 60 PetscErrorCode ierr; 61 #if defined(PETSC_USE_REAL_SINGLE) 62 PetscReal atol=1.0e-5; 63 #else 64 PetscReal atol=1.0e-10; 65 #endif 66 PetscInt info,its; 67 TAO_POUNDERS *mfqP = (TAO_POUNDERS*)tao->data; 68 69 PetscFunctionBegin; 70 if (!mfqP->usegqt) { 71 PetscReal maxval; 72 PetscInt i,j; 73 74 ierr = VecSetValues(mfqP->subb,mfqP->n,mfqP->indices,mfqP->Gres,INSERT_VALUES);CHKERRQ(ierr); 75 ierr = VecAssemblyBegin(mfqP->subb);CHKERRQ(ierr); 76 ierr = VecAssemblyEnd(mfqP->subb);CHKERRQ(ierr); 77 78 ierr = VecSet(mfqP->subx,0.0);CHKERRQ(ierr); 79 80 ierr = VecSet(mfqP->subndel,-1.0);CHKERRQ(ierr); 81 ierr = VecSet(mfqP->subpdel,+1.0);CHKERRQ(ierr); 82 83 /* Complete the lower triangle of the Hessian matrix */ 84 for (i=0;i<mfqP->n;i++) { 85 for (j=i+1;j<mfqP->n;j++) { 86 mfqP->Hres[j+mfqP->n*i] = mfqP->Hres[mfqP->n*j+i]; 87 } 88 } 89 ierr = MatSetValues(mfqP->subH,mfqP->n,mfqP->indices,mfqP->n,mfqP->indices,mfqP->Hres,INSERT_VALUES);CHKERRQ(ierr); 90 ierr = MatAssemblyBegin(mfqP->subH,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 91 ierr = MatAssemblyEnd(mfqP->subH,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 92 93 ierr = TaoResetStatistics(mfqP->subtao);CHKERRQ(ierr); 94 /* ierr = TaoSetTolerances(mfqP->subtao,*gnorm,*gnorm,PETSC_DEFAULT);CHKERRQ(ierr); */ 95 /* enforce bound constraints -- experimental */ 96 if (tao->XU && tao->XL) { 97 ierr = VecCopy(tao->XU,mfqP->subxu);CHKERRQ(ierr); 98 ierr = VecAXPY(mfqP->subxu,-1.0,tao->solution);CHKERRQ(ierr); 99 ierr = VecScale(mfqP->subxu,1.0/mfqP->delta);CHKERRQ(ierr); 100 ierr = VecCopy(tao->XL,mfqP->subxl);CHKERRQ(ierr); 101 ierr = VecAXPY(mfqP->subxl,-1.0,tao->solution);CHKERRQ(ierr); 102 ierr = VecScale(mfqP->subxl,1.0/mfqP->delta);CHKERRQ(ierr); 103 104 ierr = VecPointwiseMin(mfqP->subxu,mfqP->subxu,mfqP->subpdel);CHKERRQ(ierr); 105 ierr = VecPointwiseMax(mfqP->subxl,mfqP->subxl,mfqP->subndel);CHKERRQ(ierr); 106 } else { 107 ierr = VecCopy(mfqP->subpdel,mfqP->subxu);CHKERRQ(ierr); 108 ierr = VecCopy(mfqP->subndel,mfqP->subxl);CHKERRQ(ierr); 109 } 110 /* Make sure xu > xl */ 111 ierr = VecCopy(mfqP->subxl,mfqP->subpdel);CHKERRQ(ierr); 112 ierr = VecAXPY(mfqP->subpdel,-1.0,mfqP->subxu);CHKERRQ(ierr); 113 ierr = VecMax(mfqP->subpdel,NULL,&maxval);CHKERRQ(ierr); 114 if (maxval > 1e-10) SETERRQ(PetscObjectComm((PetscObject)tao),PETSC_ERR_ARG_OUTOFRANGE,"upper bound < lower bound in subproblem"); 115 /* Make sure xu > tao->solution > xl */ 116 ierr = VecCopy(mfqP->subxl,mfqP->subpdel);CHKERRQ(ierr); 117 ierr = VecAXPY(mfqP->subpdel,-1.0,mfqP->subx);CHKERRQ(ierr); 118 ierr = VecMax(mfqP->subpdel,NULL,&maxval);CHKERRQ(ierr); 119 if (maxval > 1e-10) SETERRQ(PetscObjectComm((PetscObject)tao),PETSC_ERR_ARG_OUTOFRANGE,"initial guess < lower bound in subproblem"); 120 121 ierr = VecCopy(mfqP->subx,mfqP->subpdel);CHKERRQ(ierr); 122 ierr = VecAXPY(mfqP->subpdel,-1.0,mfqP->subxu);CHKERRQ(ierr); 123 ierr = VecMax(mfqP->subpdel,NULL,&maxval);CHKERRQ(ierr); 124 if (maxval > 1e-10) SETERRQ(PetscObjectComm((PetscObject)tao),PETSC_ERR_ARG_OUTOFRANGE,"initial guess > upper bound in subproblem"); 125 126 ierr = TaoSolve(mfqP->subtao);CHKERRQ(ierr); 127 ierr = TaoGetSolutionStatus(mfqP->subtao,NULL,qmin,NULL,NULL,NULL,NULL);CHKERRQ(ierr); 128 129 /* test bounds post-solution*/ 130 ierr = VecCopy(mfqP->subxl,mfqP->subpdel);CHKERRQ(ierr); 131 ierr = VecAXPY(mfqP->subpdel,-1.0,mfqP->subx);CHKERRQ(ierr); 132 ierr = VecMax(mfqP->subpdel,NULL,&maxval);CHKERRQ(ierr); 133 if (maxval > 1e-5) { 134 ierr = PetscInfo(tao,"subproblem solution < lower bound\n");CHKERRQ(ierr); 135 tao->reason = TAO_DIVERGED_TR_REDUCTION; 136 } 137 138 ierr = VecCopy(mfqP->subx,mfqP->subpdel);CHKERRQ(ierr); 139 ierr = VecAXPY(mfqP->subpdel,-1.0,mfqP->subxu);CHKERRQ(ierr); 140 ierr = VecMax(mfqP->subpdel,NULL,&maxval);CHKERRQ(ierr); 141 if (maxval > 1e-5) { 142 ierr = PetscInfo(tao,"subproblem solution > upper bound\n");CHKERRQ(ierr); 143 tao->reason = TAO_DIVERGED_TR_REDUCTION; 144 } 145 } else { 146 gqt(mfqP->n,mfqP->Hres,mfqP->n,mfqP->Gres,1.0,mfqP->gqt_rtol,atol,mfqP->gqt_maxits,gnorm,qmin,mfqP->Xsubproblem,&info,&its,mfqP->work,mfqP->work2, mfqP->work3); 147 } 148 *qmin *= -1; 149 PetscFunctionReturn(0); 150 } 151 152 static PetscErrorCode pounders_update_res(Tao tao) 153 { 154 TAO_POUNDERS *mfqP = (TAO_POUNDERS*)tao->data; 155 PetscInt i,row,col; 156 PetscBLASInt blasn=mfqP->n,blasn2=blasn*blasn,blasm=mfqP->m,ione=1; 157 PetscReal zero=0.0,one=1.0,wii,factor; 158 PetscErrorCode ierr; 159 160 PetscFunctionBegin; 161 for (i=0;i<mfqP->n;i++) { 162 mfqP->Gres[i]=0; 163 } 164 for (i=0;i<mfqP->n*mfqP->n;i++) { 165 mfqP->Hres[i]=0; 166 } 167 168 /* Compute Gres= sum_ij[wij * (cjgi + cigj)] */ 169 if (tao->res_weights_v) { 170 /* Vector(diagonal) weights: gres = sum_i(wii*ci*gi) */ 171 for (i=0;i<mfqP->m;i++) { 172 ierr = VecGetValues(tao->res_weights_v,1,&i,&factor);CHKERRQ(ierr); 173 factor=factor*mfqP->C[i]; 174 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&blasn,&factor,&mfqP->Fdiff[blasn*i],&ione,mfqP->Gres,&ione)); 175 } 176 177 /* compute Hres = sum_ij [wij * (*ci*Hj + cj*Hi + gi gj' + gj gi') ] */ 178 /* vector(diagonal weights) Hres = sum_i(wii*(ci*Hi + gi * gi' )*/ 179 for (i=0;i<mfqP->m;i++) { 180 ierr = VecGetValues(tao->res_weights_v,1,&i,&wii);CHKERRQ(ierr); 181 if (tao->niter>1) { 182 factor=wii*mfqP->C[i]; 183 /* add wii * ci * Hi */ 184 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&blasn2,&factor,&mfqP->H[i],&blasm,mfqP->Hres,&ione)); 185 } 186 /* add wii * gi * gi' */ 187 PetscStackCallBLAS("BLASgemm",BLASgemm_("N","T",&blasn,&blasn,&ione,&wii,&mfqP->Fdiff[blasn*i],&blasn,&mfqP->Fdiff[blasn*i],&blasn,&one,mfqP->Hres,&blasn)); 188 } 189 } else if (tao->res_weights_w) { 190 /* General case: .5 * Gres= sum_ij[wij * (cjgi + cigj)] */ 191 for (i=0;i<tao->res_weights_n;i++) { 192 row=tao->res_weights_rows[i]; 193 col=tao->res_weights_cols[i]; 194 195 factor = tao->res_weights_w[i]*mfqP->C[col]/2.0; 196 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&blasn,&factor,&mfqP->Fdiff[blasn*row],&ione,mfqP->Gres,&ione)); 197 factor = tao->res_weights_w[i]*mfqP->C[row]/2.0; 198 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&blasn,&factor,&mfqP->Fdiff[blasn*col],&ione,mfqP->Gres,&ione)); 199 } 200 201 /* compute Hres = sum_ij [wij * (*ci*Hj + cj*Hi + gi gj' + gj gi') ] */ 202 /* .5 * sum_ij [wij * (*ci*Hj + cj*Hi + gi gj' + gj gi') ] */ 203 for (i=0;i<tao->res_weights_n;i++) { 204 row=tao->res_weights_rows[i]; 205 col=tao->res_weights_cols[i]; 206 factor=tao->res_weights_w[i]/2.0; 207 /* add wij * gi gj' + wij * gj gi' */ 208 PetscStackCallBLAS("BLASgemm",BLASgemm_("N","T",&blasn,&blasn,&ione,&factor,&mfqP->Fdiff[blasn*row],&blasn,&mfqP->Fdiff[blasn*col],&blasn,&one,mfqP->Hres,&blasn)); 209 PetscStackCallBLAS("BLASgemm",BLASgemm_("N","T",&blasn,&blasn,&ione,&factor,&mfqP->Fdiff[blasn*col],&blasn,&mfqP->Fdiff[blasn*row],&blasn,&one,mfqP->Hres,&blasn)); 210 } 211 if (tao->niter > 1) { 212 for (i=0;i<tao->res_weights_n;i++) { 213 row=tao->res_weights_rows[i]; 214 col=tao->res_weights_cols[i]; 215 216 /* add wij*cj*Hi */ 217 factor = tao->res_weights_w[i]*mfqP->C[col]/2.0; 218 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&blasn2,&factor,&mfqP->H[row],&blasm,mfqP->Hres,&ione)); 219 220 /* add wij*ci*Hj */ 221 factor = tao->res_weights_w[i]*mfqP->C[row]/2.0; 222 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&blasn2,&factor,&mfqP->H[col],&blasm,mfqP->Hres,&ione)); 223 } 224 } 225 } else { 226 /* Default: Gres= sum_i[cigi] = G*c' */ 227 ierr = PetscInfo(tao,"Identity weights\n");CHKERRQ(ierr); 228 PetscStackCallBLAS("BLASgemv",BLASgemv_("N",&blasn,&blasm,&one,mfqP->Fdiff,&blasn,mfqP->C,&ione,&zero,mfqP->Gres,&ione)); 229 230 /* compute Hres = sum_ij [wij * (*ci*Hj + cj*Hi + gi gj' + gj gi') ] */ 231 /* Hres = G*G' + 0.5 sum {F(xkin,i)*H(:,:,i)} */ 232 PetscStackCallBLAS("BLASgemm",BLASgemm_("N","T",&blasn,&blasn,&blasm,&one,mfqP->Fdiff, &blasn,mfqP->Fdiff, &blasn,&zero,mfqP->Hres,&blasn)); 233 234 /* sum(F(xkin,i)*H(:,:,i)) */ 235 if (tao->niter>1) { 236 for (i=0;i<mfqP->m;i++) { 237 factor = mfqP->C[i]; 238 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&blasn2,&factor,&mfqP->H[i],&blasm,mfqP->Hres,&ione)); 239 } 240 } 241 } 242 PetscFunctionReturn(0); 243 } 244 245 static PetscErrorCode phi2eval(PetscReal *x, PetscInt n, PetscReal *phi) 246 { 247 /* Phi = .5*[x(1)^2 sqrt(2)*x(1)*x(2) ... sqrt(2)*x(1)*x(n) ... x(2)^2 sqrt(2)*x(2)*x(3) .. x(n)^2] */ 248 PetscInt i,j,k; 249 PetscReal sqrt2 = PetscSqrtReal(2.0); 250 251 PetscFunctionBegin; 252 j=0; 253 for (i=0;i<n;i++) { 254 phi[j] = 0.5 * x[i]*x[i]; 255 j++; 256 for (k=i+1;k<n;k++) { 257 phi[j] = x[i]*x[k]/sqrt2; 258 j++; 259 } 260 } 261 PetscFunctionReturn(0); 262 } 263 264 static PetscErrorCode getquadpounders(TAO_POUNDERS *mfqP) 265 { 266 /* Computes the parameters of the quadratic Q(x) = c + g'*x + 0.5*x*G*x' 267 that satisfies the interpolation conditions Q(X[:,j]) = f(j) 268 for j=1,...,m and with a Hessian matrix of least Frobenius norm */ 269 270 /* NB --we are ignoring c */ 271 PetscInt i,j,k,num,np = mfqP->nmodelpoints; 272 PetscReal one = 1.0,zero=0.0,negone=-1.0; 273 PetscBLASInt blasnpmax = mfqP->npmax; 274 PetscBLASInt blasnplus1 = mfqP->n+1; 275 PetscBLASInt blasnp = np; 276 PetscBLASInt blasint = mfqP->n*(mfqP->n+1) / 2; 277 PetscBLASInt blasint2 = np - mfqP->n-1; 278 PetscBLASInt info,ione=1; 279 PetscReal sqrt2 = PetscSqrtReal(2.0); 280 281 PetscFunctionBegin; 282 for (i=0;i<mfqP->n*mfqP->m;i++) { 283 mfqP->Gdel[i] = 0; 284 } 285 for (i=0;i<mfqP->n*mfqP->n*mfqP->m;i++) { 286 mfqP->Hdel[i] = 0; 287 } 288 289 /* factor M */ 290 PetscStackCallBLAS("LAPACKgetrf",LAPACKgetrf_(&blasnplus1,&blasnp,mfqP->M,&blasnplus1,mfqP->npmaxiwork,&info)); 291 if (info != 0) SETERRQ1(PETSC_COMM_SELF,1,"LAPACK routine getrf returned with value %d\n",info); 292 293 if (np == mfqP->n+1) { 294 for (i=0;i<mfqP->npmax-mfqP->n-1;i++) { 295 mfqP->omega[i]=0.0; 296 } 297 for (i=0;i<mfqP->n*(mfqP->n+1)/2;i++) { 298 mfqP->beta[i]=0.0; 299 } 300 } else { 301 /* Let Ltmp = (L'*L) */ 302 PetscStackCallBLAS("BLASgemm",BLASgemm_("T","N",&blasint2,&blasint2,&blasint,&one,&mfqP->L[(mfqP->n+1)*blasint],&blasint,&mfqP->L[(mfqP->n+1)*blasint],&blasint,&zero,mfqP->L_tmp,&blasint)); 303 304 /* factor Ltmp */ 305 PetscStackCallBLAS("LAPACKpotrf",LAPACKpotrf_("L",&blasint2,mfqP->L_tmp,&blasint,&info)); 306 if (info != 0) SETERRQ1(PETSC_COMM_SELF,1,"LAPACK routine potrf returned with value %d\n",info); 307 } 308 309 for (k=0;k<mfqP->m;k++) { 310 if (np != mfqP->n+1) { 311 /* Solve L'*L*Omega = Z' * RESk*/ 312 PetscStackCallBLAS("BLASgemv",BLASgemv_("T",&blasnp,&blasint2,&one,mfqP->Z,&blasnpmax,&mfqP->RES[mfqP->npmax*k],&ione,&zero,mfqP->omega,&ione)); 313 PetscStackCallBLAS("LAPACKpotrs",LAPACKpotrs_("L",&blasint2,&ione,mfqP->L_tmp,&blasint,mfqP->omega,&blasint2,&info)); 314 if (info != 0) SETERRQ1(PETSC_COMM_SELF,1,"LAPACK routine potrs returned with value %d\n",info); 315 316 /* Beta = L*Omega */ 317 PetscStackCallBLAS("BLASgemv",BLASgemv_("N",&blasint,&blasint2,&one,&mfqP->L[(mfqP->n+1)*blasint],&blasint,mfqP->omega,&ione,&zero,mfqP->beta,&ione)); 318 } 319 320 /* solve M'*Alpha = RESk - N'*Beta */ 321 PetscStackCallBLAS("BLASgemv",BLASgemv_("T",&blasint,&blasnp,&negone,mfqP->N,&blasint,mfqP->beta,&ione,&one,&mfqP->RES[mfqP->npmax*k],&ione)); 322 PetscStackCallBLAS("LAPACKgetrs",LAPACKgetrs_("T",&blasnplus1,&ione,mfqP->M,&blasnplus1,mfqP->npmaxiwork,&mfqP->RES[mfqP->npmax*k],&blasnplus1,&info)); 323 if (info != 0) SETERRQ1(PETSC_COMM_SELF,1,"LAPACK routine getrs returned with value %d\n",info); 324 325 /* Gdel(:,k) = Alpha(2:n+1) */ 326 for (i=0;i<mfqP->n;i++) { 327 mfqP->Gdel[i + mfqP->n*k] = mfqP->RES[mfqP->npmax*k + i+1]; 328 } 329 330 /* Set Hdels */ 331 num=0; 332 for (i=0;i<mfqP->n;i++) { 333 /* H[i,i,k] = Beta(num) */ 334 mfqP->Hdel[(i*mfqP->n + i)*mfqP->m + k] = mfqP->beta[num]; 335 num++; 336 for (j=i+1;j<mfqP->n;j++) { 337 /* H[i,j,k] = H[j,i,k] = Beta(num)/sqrt(2) */ 338 mfqP->Hdel[(j*mfqP->n + i)*mfqP->m + k] = mfqP->beta[num]/sqrt2; 339 mfqP->Hdel[(i*mfqP->n + j)*mfqP->m + k] = mfqP->beta[num]/sqrt2; 340 num++; 341 } 342 } 343 } 344 PetscFunctionReturn(0); 345 } 346 347 static PetscErrorCode morepoints(TAO_POUNDERS *mfqP) 348 { 349 /* Assumes mfqP->model_indices[0] is minimum index 350 Finishes adding points to mfqP->model_indices (up to npmax) 351 Computes L,Z,M,N 352 np is actual number of points in model (should equal npmax?) */ 353 PetscInt point,i,j,offset; 354 PetscInt reject; 355 PetscBLASInt blasn=mfqP->n,blasnpmax=mfqP->npmax,blasnplus1=mfqP->n+1,info,blasnmax=mfqP->nmax,blasint,blasint2,blasnp,blasmaxmn; 356 const PetscReal *x; 357 PetscReal normd; 358 PetscErrorCode ierr; 359 360 PetscFunctionBegin; 361 /* Initialize M,N */ 362 for (i=0;i<mfqP->n+1;i++) { 363 ierr = VecGetArrayRead(mfqP->Xhist[mfqP->model_indices[i]],&x);CHKERRQ(ierr); 364 mfqP->M[(mfqP->n+1)*i] = 1.0; 365 for (j=0;j<mfqP->n;j++) { 366 mfqP->M[j+1+((mfqP->n+1)*i)] = (x[j] - mfqP->xmin[j]) / mfqP->delta; 367 } 368 ierr = VecRestoreArrayRead(mfqP->Xhist[mfqP->model_indices[i]],&x);CHKERRQ(ierr); 369 ierr = phi2eval(&mfqP->M[1+((mfqP->n+1)*i)],mfqP->n,&mfqP->N[mfqP->n*(mfqP->n+1)/2 * i]);CHKERRQ(ierr); 370 } 371 372 /* Now we add points until we have npmax starting with the most recent ones */ 373 point = mfqP->nHist-1; 374 mfqP->nmodelpoints = mfqP->n+1; 375 while (mfqP->nmodelpoints < mfqP->npmax && point>=0) { 376 /* Reject any points already in the model */ 377 reject = 0; 378 for (j=0;j<mfqP->n+1;j++) { 379 if (point == mfqP->model_indices[j]) { 380 reject = 1; 381 break; 382 } 383 } 384 385 /* Reject if norm(d) >c2 */ 386 if (!reject) { 387 ierr = VecCopy(mfqP->Xhist[point],mfqP->workxvec);CHKERRQ(ierr); 388 ierr = VecAXPY(mfqP->workxvec,-1.0,mfqP->Xhist[mfqP->minindex]);CHKERRQ(ierr); 389 ierr = VecNorm(mfqP->workxvec,NORM_2,&normd);CHKERRQ(ierr); 390 normd /= mfqP->delta; 391 if (normd > mfqP->c2) { 392 reject =1; 393 } 394 } 395 if (reject){ 396 point--; 397 continue; 398 } 399 400 ierr = VecGetArrayRead(mfqP->Xhist[point],&x);CHKERRQ(ierr); 401 mfqP->M[(mfqP->n+1)*mfqP->nmodelpoints] = 1.0; 402 for (j=0;j<mfqP->n;j++) { 403 mfqP->M[j+1+((mfqP->n+1)*mfqP->nmodelpoints)] = (x[j] - mfqP->xmin[j]) / mfqP->delta; 404 } 405 ierr = VecRestoreArrayRead(mfqP->Xhist[point],&x);CHKERRQ(ierr); 406 ierr = phi2eval(&mfqP->M[1+(mfqP->n+1)*mfqP->nmodelpoints],mfqP->n,&mfqP->N[mfqP->n*(mfqP->n+1)/2 * (mfqP->nmodelpoints)]);CHKERRQ(ierr); 407 408 /* Update QR factorization */ 409 /* Copy M' to Q_tmp */ 410 for (i=0;i<mfqP->n+1;i++) { 411 for (j=0;j<mfqP->npmax;j++) { 412 mfqP->Q_tmp[j+mfqP->npmax*i] = mfqP->M[i+(mfqP->n+1)*j]; 413 } 414 } 415 blasnp = mfqP->nmodelpoints+1; 416 /* Q_tmp,R = qr(M') */ 417 blasmaxmn=PetscMax(mfqP->m,mfqP->n+1); 418 PetscStackCallBLAS("LAPACKgeqrf",LAPACKgeqrf_(&blasnp,&blasnplus1,mfqP->Q_tmp,&blasnpmax,mfqP->tau_tmp,mfqP->mwork,&blasmaxmn,&info)); 419 if (info != 0) SETERRQ1(PETSC_COMM_SELF,1,"LAPACK routine geqrf returned with value %d\n",info); 420 421 /* Reject if min(svd(N*Q(:,n+2:np+1)) <= theta2 */ 422 /* L = N*Qtmp */ 423 blasint2 = mfqP->n * (mfqP->n+1) / 2; 424 /* Copy N to L_tmp */ 425 for (i=0;i<mfqP->n*(mfqP->n+1)/2 * mfqP->npmax;i++) { 426 mfqP->L_tmp[i]= mfqP->N[i]; 427 } 428 /* Copy L_save to L_tmp */ 429 430 /* L_tmp = N*Qtmp' */ 431 PetscStackCallBLAS("LAPACKormqr",LAPACKormqr_("R","N",&blasint2,&blasnp,&blasnplus1,mfqP->Q_tmp,&blasnpmax,mfqP->tau_tmp,mfqP->L_tmp,&blasint2,mfqP->npmaxwork,&blasnmax,&info)); 432 if (info != 0) SETERRQ1(PETSC_COMM_SELF,1,"LAPACK routine ormqr returned with value %d\n",info); 433 434 /* Copy L_tmp to L_save */ 435 for (i=0;i<mfqP->npmax * mfqP->n*(mfqP->n+1)/2;i++) { 436 mfqP->L_save[i] = mfqP->L_tmp[i]; 437 } 438 439 /* Get svd for L_tmp(:,n+2:np+1) (L_tmp is modified in process) */ 440 blasint = mfqP->nmodelpoints - mfqP->n; 441 ierr = PetscFPTrapPush(PETSC_FP_TRAP_OFF);CHKERRQ(ierr); 442 PetscStackCallBLAS("LAPACKgesvd",LAPACKgesvd_("N","N",&blasint2,&blasint,&mfqP->L_tmp[(mfqP->n+1)*blasint2],&blasint2,mfqP->beta,mfqP->work,&blasn,mfqP->work,&blasn,mfqP->npmaxwork,&blasnmax,&info)); 443 ierr = PetscFPTrapPop();CHKERRQ(ierr); 444 if (info != 0) SETERRQ1(PETSC_COMM_SELF,1,"LAPACK routine gesvd returned with value %d\n",info); 445 446 if (mfqP->beta[PetscMin(blasint,blasint2)-1] > mfqP->theta2) { 447 /* accept point */ 448 mfqP->model_indices[mfqP->nmodelpoints] = point; 449 /* Copy Q_tmp to Q */ 450 for (i=0;i<mfqP->npmax* mfqP->npmax;i++) { 451 mfqP->Q[i] = mfqP->Q_tmp[i]; 452 } 453 for (i=0;i<mfqP->npmax;i++){ 454 mfqP->tau[i] = mfqP->tau_tmp[i]; 455 } 456 mfqP->nmodelpoints++; 457 blasnp = mfqP->nmodelpoints; 458 459 /* Copy L_save to L */ 460 for (i=0;i<mfqP->npmax * mfqP->n*(mfqP->n+1)/2;i++) { 461 mfqP->L[i] = mfqP->L_save[i]; 462 } 463 } 464 point--; 465 } 466 467 blasnp = mfqP->nmodelpoints; 468 /* Copy Q(:,n+2:np) to Z */ 469 /* First set Q_tmp to I */ 470 for (i=0;i<mfqP->npmax*mfqP->npmax;i++) { 471 mfqP->Q_tmp[i] = 0.0; 472 } 473 for (i=0;i<mfqP->npmax;i++) { 474 mfqP->Q_tmp[i + mfqP->npmax*i] = 1.0; 475 } 476 477 /* Q_tmp = I * Q */ 478 PetscStackCallBLAS("LAPACKormqr",LAPACKormqr_("R","N",&blasnp,&blasnp,&blasnplus1,mfqP->Q,&blasnpmax,mfqP->tau,mfqP->Q_tmp,&blasnpmax,mfqP->npmaxwork,&blasnmax,&info)); 479 if (info != 0) SETERRQ1(PETSC_COMM_SELF,1,"LAPACK routine ormqr returned with value %d\n",info); 480 481 /* Copy Q_tmp(:,n+2:np) to Z) */ 482 offset = mfqP->npmax * (mfqP->n+1); 483 for (i=offset;i<mfqP->npmax*mfqP->npmax;i++) { 484 mfqP->Z[i-offset] = mfqP->Q_tmp[i]; 485 } 486 487 if (mfqP->nmodelpoints == mfqP->n + 1) { 488 /* Set L to I_{n+1} */ 489 for (i=0;i<mfqP->npmax * mfqP->n*(mfqP->n+1)/2;i++) { 490 mfqP->L[i] = 0.0; 491 } 492 for (i=0;i<mfqP->n;i++) { 493 mfqP->L[(mfqP->n*(mfqP->n+1)/2)*i + i] = 1.0; 494 } 495 } 496 PetscFunctionReturn(0); 497 } 498 499 /* Only call from modelimprove, addpoint() needs ->Q_tmp and ->work to be set */ 500 static PetscErrorCode addpoint(Tao tao, TAO_POUNDERS *mfqP, PetscInt index) 501 { 502 PetscErrorCode ierr; 503 504 PetscFunctionBegin; 505 /* Create new vector in history: X[newidx] = X[mfqP->index] + delta*X[index]*/ 506 ierr = VecDuplicate(mfqP->Xhist[0],&mfqP->Xhist[mfqP->nHist]);CHKERRQ(ierr); 507 ierr = VecSetValues(mfqP->Xhist[mfqP->nHist],mfqP->n,mfqP->indices,&mfqP->Q_tmp[index*mfqP->npmax],INSERT_VALUES);CHKERRQ(ierr); 508 ierr = VecAssemblyBegin(mfqP->Xhist[mfqP->nHist]);CHKERRQ(ierr); 509 ierr = VecAssemblyEnd(mfqP->Xhist[mfqP->nHist]);CHKERRQ(ierr); 510 ierr = VecAYPX(mfqP->Xhist[mfqP->nHist],mfqP->delta,mfqP->Xhist[mfqP->minindex]);CHKERRQ(ierr); 511 512 /* Project into feasible region */ 513 if (tao->XU && tao->XL) { 514 ierr = VecMedian(mfqP->Xhist[mfqP->nHist], tao->XL, tao->XU, mfqP->Xhist[mfqP->nHist]);CHKERRQ(ierr); 515 } 516 517 /* Compute value of new vector */ 518 ierr = VecDuplicate(mfqP->Fhist[0],&mfqP->Fhist[mfqP->nHist]);CHKERRQ(ierr); 519 CHKMEMQ; 520 ierr = pounders_feval(tao,mfqP->Xhist[mfqP->nHist],mfqP->Fhist[mfqP->nHist],&mfqP->Fres[mfqP->nHist]);CHKERRQ(ierr); 521 522 /* Add new vector to model */ 523 mfqP->model_indices[mfqP->nmodelpoints] = mfqP->nHist; 524 mfqP->nmodelpoints++; 525 mfqP->nHist++; 526 PetscFunctionReturn(0); 527 } 528 529 static PetscErrorCode modelimprove(Tao tao, TAO_POUNDERS *mfqP, PetscInt addallpoints) 530 { 531 /* modeld = Q(:,np+1:n)' */ 532 PetscErrorCode ierr; 533 PetscInt i,j,minindex=0; 534 PetscReal dp,half=0.5,one=1.0,minvalue=PETSC_INFINITY; 535 PetscBLASInt blasn=mfqP->n, blasnpmax = mfqP->npmax, blask,info; 536 PetscBLASInt blas1=1,blasnmax = mfqP->nmax; 537 538 blask = mfqP->nmodelpoints; 539 /* Qtmp = I(n x n) */ 540 for (i=0;i<mfqP->n;i++) { 541 for (j=0;j<mfqP->n;j++) { 542 mfqP->Q_tmp[i + mfqP->npmax*j] = 0.0; 543 } 544 } 545 for (j=0;j<mfqP->n;j++) { 546 mfqP->Q_tmp[j + mfqP->npmax*j] = 1.0; 547 } 548 549 /* Qtmp = Q * I */ 550 PetscStackCallBLAS("LAPACKormqr",LAPACKormqr_("R","N",&blasn,&blasn,&blask,mfqP->Q,&blasnpmax,mfqP->tau, mfqP->Q_tmp, &blasnpmax, mfqP->npmaxwork,&blasnmax, &info)); 551 552 for (i=mfqP->nmodelpoints;i<mfqP->n;i++) { 553 dp = BLASdot_(&blasn,&mfqP->Q_tmp[i*mfqP->npmax],&blas1,mfqP->Gres,&blas1); 554 if (dp>0.0) { /* Model says use the other direction! */ 555 for (j=0;j<mfqP->n;j++) { 556 mfqP->Q_tmp[i*mfqP->npmax+j] *= -1; 557 } 558 } 559 /* mfqP->work[i] = Cres+Modeld(i,:)*(Gres+.5*Hres*Modeld(i,:)') */ 560 for (j=0;j<mfqP->n;j++) { 561 mfqP->work2[j] = mfqP->Gres[j]; 562 } 563 PetscStackCallBLAS("BLASgemv",BLASgemv_("N",&blasn,&blasn,&half,mfqP->Hres,&blasn,&mfqP->Q_tmp[i*mfqP->npmax], &blas1, &one, mfqP->work2,&blas1)); 564 mfqP->work[i] = BLASdot_(&blasn,&mfqP->Q_tmp[i*mfqP->npmax],&blas1,mfqP->work2,&blas1); 565 if (i==mfqP->nmodelpoints || mfqP->work[i] < minvalue) { 566 minindex=i; 567 minvalue = mfqP->work[i]; 568 } 569 if (addallpoints != 0) { 570 ierr = addpoint(tao,mfqP,i);CHKERRQ(ierr); 571 } 572 } 573 if (!addallpoints) { 574 ierr = addpoint(tao,mfqP,minindex);CHKERRQ(ierr); 575 } 576 PetscFunctionReturn(0); 577 } 578 579 static PetscErrorCode affpoints(TAO_POUNDERS *mfqP, PetscReal *xmin,PetscReal c) 580 { 581 PetscInt i,j; 582 PetscBLASInt blasm=mfqP->m,blasj,blask,blasn=mfqP->n,ione=1,info; 583 PetscBLASInt blasnpmax = mfqP->npmax,blasmaxmn; 584 PetscReal proj,normd; 585 const PetscReal *x; 586 PetscErrorCode ierr; 587 588 PetscFunctionBegin; 589 for (i=mfqP->nHist-1;i>=0;i--) { 590 ierr = VecGetArrayRead(mfqP->Xhist[i],&x);CHKERRQ(ierr); 591 for (j=0;j<mfqP->n;j++) { 592 mfqP->work[j] = (x[j] - xmin[j])/mfqP->delta; 593 } 594 ierr = VecRestoreArrayRead(mfqP->Xhist[i],&x);CHKERRQ(ierr); 595 PetscStackCallBLAS("BLAScopy",BLAScopy_(&blasn,mfqP->work,&ione,mfqP->work2,&ione)); 596 normd = BLASnrm2_(&blasn,mfqP->work,&ione); 597 if (normd <= c) { 598 blasj=PetscMax((mfqP->n - mfqP->nmodelpoints),0); 599 if (!mfqP->q_is_I) { 600 /* project D onto null */ 601 blask=(mfqP->nmodelpoints); 602 PetscStackCallBLAS("LAPACKormqr",LAPACKormqr_("R","N",&ione,&blasn,&blask,mfqP->Q,&blasnpmax,mfqP->tau,mfqP->work2,&ione,mfqP->mwork,&blasm,&info)); 603 if (info < 0) SETERRQ1(PETSC_COMM_SELF,1,"ormqr returned value %d\n",info); 604 } 605 proj = BLASnrm2_(&blasj,&mfqP->work2[mfqP->nmodelpoints],&ione); 606 607 if (proj >= mfqP->theta1) { /* add this index to model */ 608 mfqP->model_indices[mfqP->nmodelpoints]=i; 609 mfqP->nmodelpoints++; 610 PetscStackCallBLAS("BLAScopy",BLAScopy_(&blasn,mfqP->work,&ione,&mfqP->Q_tmp[mfqP->npmax*(mfqP->nmodelpoints-1)],&ione)); 611 blask=mfqP->npmax*(mfqP->nmodelpoints); 612 PetscStackCallBLAS("BLAScopy",BLAScopy_(&blask,mfqP->Q_tmp,&ione,mfqP->Q,&ione)); 613 blask = mfqP->nmodelpoints; 614 blasmaxmn = PetscMax(mfqP->m,mfqP->n); 615 PetscStackCallBLAS("LAPACKgeqrf",LAPACKgeqrf_(&blasn,&blask,mfqP->Q,&blasnpmax,mfqP->tau,mfqP->mwork,&blasmaxmn,&info)); 616 if (info < 0) SETERRQ1(PETSC_COMM_SELF,1,"geqrf returned value %d\n",info); 617 mfqP->q_is_I = 0; 618 } 619 if (mfqP->nmodelpoints == mfqP->n) { 620 break; 621 } 622 } 623 } 624 625 PetscFunctionReturn(0); 626 } 627 628 static PetscErrorCode TaoSolve_POUNDERS(Tao tao) 629 { 630 TAO_POUNDERS *mfqP = (TAO_POUNDERS *)tao->data; 631 PetscInt i,ii,j,k,l; 632 PetscReal step=1.0; 633 PetscInt low,high; 634 PetscReal minnorm; 635 PetscReal *x,*f; 636 const PetscReal *xmint,*fmin; 637 PetscReal cres,deltaold; 638 PetscReal gnorm; 639 PetscBLASInt info,ione=1,iblas; 640 PetscBool valid,same; 641 PetscReal mdec, rho, normxsp; 642 PetscReal one=1.0,zero=0.0,ratio; 643 PetscBLASInt blasm,blasn,blasncopy,blasnpmax; 644 PetscErrorCode ierr; 645 static PetscBool set = PETSC_FALSE; 646 647 /* n = # of parameters 648 m = dimension (components) of function */ 649 PetscFunctionBegin; 650 ierr = PetscCitationsRegister("@article{UNEDF0,\n" 651 "title = {Nuclear energy density optimization},\n" 652 "author = {Kortelainen, M. and Lesinski, T. and Mor\'e, J. and Nazarewicz, W.\n" 653 " and Sarich, J. and Schunck, N. and Stoitsov, M. V. and Wild, S. },\n" 654 "journal = {Phys. Rev. C},\n" 655 "volume = {82},\n" 656 "number = {2},\n" 657 "pages = {024313},\n" 658 "numpages = {18},\n" 659 "year = {2010},\n" 660 "month = {Aug},\n" 661 "doi = {10.1103/PhysRevC.82.024313}\n}\n",&set);CHKERRQ(ierr); 662 tao->niter=0; 663 if (tao->XL && tao->XU) { 664 /* Check x0 <= XU */ 665 PetscReal val; 666 667 ierr = VecCopy(tao->solution,mfqP->Xhist[0]);CHKERRQ(ierr); 668 ierr = VecAXPY(mfqP->Xhist[0],-1.0,tao->XU);CHKERRQ(ierr); 669 ierr = VecMax(mfqP->Xhist[0],NULL,&val);CHKERRQ(ierr); 670 if (val > 1e-10) SETERRQ(PetscObjectComm((PetscObject)tao),PETSC_ERR_ARG_OUTOFRANGE,"X0 > upper bound"); 671 672 /* Check x0 >= xl */ 673 ierr = VecCopy(tao->XL,mfqP->Xhist[0]);CHKERRQ(ierr); 674 ierr = VecAXPY(mfqP->Xhist[0],-1.0,tao->solution);CHKERRQ(ierr); 675 ierr = VecMax(mfqP->Xhist[0],NULL,&val);CHKERRQ(ierr); 676 if (val > 1e-10) SETERRQ(PetscObjectComm((PetscObject)tao),PETSC_ERR_ARG_OUTOFRANGE,"X0 < lower bound"); 677 678 /* Check x0 + delta < XU -- should be able to get around this eventually */ 679 680 ierr = VecSet(mfqP->Xhist[0],mfqP->delta);CHKERRQ(ierr); 681 ierr = VecAXPY(mfqP->Xhist[0],1.0,tao->solution);CHKERRQ(ierr); 682 ierr = VecAXPY(mfqP->Xhist[0],-1.0,tao->XU);CHKERRQ(ierr); 683 ierr = VecMax(mfqP->Xhist[0],NULL,&val);CHKERRQ(ierr); 684 if (val > 1e-10) SETERRQ(PetscObjectComm((PetscObject)tao),PETSC_ERR_ARG_OUTOFRANGE,"X0 + delta > upper bound"); 685 } 686 687 blasm = mfqP->m; blasn=mfqP->n; blasnpmax = mfqP->npmax; 688 for (i=0;i<mfqP->n*mfqP->n*mfqP->m;++i) mfqP->H[i]=0; 689 690 ierr = VecCopy(tao->solution,mfqP->Xhist[0]);CHKERRQ(ierr); 691 692 /* This provides enough information to approximate the gradient of the objective */ 693 /* using a forward difference scheme. */ 694 695 ierr = PetscInfo1(tao,"Initialize simplex; delta = %10.9e\n",(double)mfqP->delta);CHKERRQ(ierr); 696 ierr = pounders_feval(tao,mfqP->Xhist[0],mfqP->Fhist[0],&mfqP->Fres[0]);CHKERRQ(ierr); 697 mfqP->minindex = 0; 698 minnorm = mfqP->Fres[0]; 699 700 ierr = VecGetOwnershipRange(mfqP->Xhist[0],&low,&high);CHKERRQ(ierr); 701 for (i=1;i<mfqP->n+1;++i) { 702 ierr = VecCopy(mfqP->Xhist[0],mfqP->Xhist[i]);CHKERRQ(ierr); 703 704 if (i-1 >= low && i-1 < high) { 705 ierr = VecGetArray(mfqP->Xhist[i],&x);CHKERRQ(ierr); 706 x[i-1-low] += mfqP->delta; 707 ierr = VecRestoreArray(mfqP->Xhist[i],&x);CHKERRQ(ierr); 708 } 709 CHKMEMQ; 710 ierr = pounders_feval(tao,mfqP->Xhist[i],mfqP->Fhist[i],&mfqP->Fres[i]);CHKERRQ(ierr); 711 if (mfqP->Fres[i] < minnorm) { 712 mfqP->minindex = i; 713 minnorm = mfqP->Fres[i]; 714 } 715 } 716 ierr = VecCopy(mfqP->Xhist[mfqP->minindex],tao->solution);CHKERRQ(ierr); 717 ierr = VecCopy(mfqP->Fhist[mfqP->minindex],tao->ls_res);CHKERRQ(ierr); 718 ierr = PetscInfo1(tao,"Finalize simplex; minnorm = %10.9e\n",(double)minnorm);CHKERRQ(ierr); 719 720 /* Gather mpi vecs to one big local vec */ 721 722 /* Begin serial code */ 723 724 /* Disp[i] = Xi-xmin, i=1,..,mfqP->minindex-1,mfqP->minindex+1,..,n */ 725 /* Fdiff[i] = (Fi-Fmin)', i=1,..,mfqP->minindex-1,mfqP->minindex+1,..,n */ 726 /* (Column oriented for blas calls) */ 727 ii=0; 728 729 ierr = PetscInfo1(tao,"Build matrix: %D\n",(PetscInt)mfqP->size);CHKERRQ(ierr); 730 if (1 == mfqP->size) { 731 ierr = VecGetArrayRead(mfqP->Xhist[mfqP->minindex],&xmint);CHKERRQ(ierr); 732 for (i=0;i<mfqP->n;i++) mfqP->xmin[i] = xmint[i]; 733 ierr = VecRestoreArrayRead(mfqP->Xhist[mfqP->minindex],&xmint);CHKERRQ(ierr); 734 ierr = VecGetArrayRead(mfqP->Fhist[mfqP->minindex],&fmin);CHKERRQ(ierr); 735 for (i=0;i<mfqP->n+1;i++) { 736 if (i == mfqP->minindex) continue; 737 738 ierr = VecGetArray(mfqP->Xhist[i],&x);CHKERRQ(ierr); 739 for (j=0;j<mfqP->n;j++) { 740 mfqP->Disp[ii+mfqP->npmax*j] = (x[j] - mfqP->xmin[j])/mfqP->delta; 741 } 742 ierr = VecRestoreArray(mfqP->Xhist[i],&x);CHKERRQ(ierr); 743 744 ierr = VecGetArray(mfqP->Fhist[i],&f);CHKERRQ(ierr); 745 for (j=0;j<mfqP->m;j++) { 746 mfqP->Fdiff[ii+mfqP->n*j] = f[j] - fmin[j]; 747 } 748 ierr = VecRestoreArray(mfqP->Fhist[i],&f);CHKERRQ(ierr); 749 750 mfqP->model_indices[ii++] = i; 751 } 752 for (j=0;j<mfqP->m;j++) { 753 mfqP->C[j] = fmin[j]; 754 } 755 ierr = VecRestoreArrayRead(mfqP->Fhist[mfqP->minindex],&fmin);CHKERRQ(ierr); 756 } else { 757 ierr = VecSet(mfqP->localxmin,0);CHKERRQ(ierr); 758 ierr = VecScatterBegin(mfqP->scatterx,mfqP->Xhist[mfqP->minindex],mfqP->localxmin,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 759 ierr = VecScatterEnd(mfqP->scatterx,mfqP->Xhist[mfqP->minindex],mfqP->localxmin,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 760 761 ierr = VecGetArrayRead(mfqP->localxmin,&xmint);CHKERRQ(ierr); 762 for (i=0;i<mfqP->n;i++) mfqP->xmin[i] = xmint[i]; 763 ierr = VecRestoreArrayRead(mfqP->localxmin,&xmint);CHKERRQ(ierr); 764 765 ierr = VecScatterBegin(mfqP->scatterf,mfqP->Fhist[mfqP->minindex],mfqP->localfmin,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 766 ierr = VecScatterEnd(mfqP->scatterf,mfqP->Fhist[mfqP->minindex],mfqP->localfmin,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 767 ierr = VecGetArrayRead(mfqP->localfmin,&fmin);CHKERRQ(ierr); 768 for (i=0;i<mfqP->n+1;i++) { 769 if (i == mfqP->minindex) continue; 770 771 ierr = VecScatterBegin(mfqP->scatterx,mfqP->Xhist[ii],mfqP->localx,INSERT_VALUES, SCATTER_FORWARD);CHKERRQ(ierr); 772 ierr = VecScatterEnd(mfqP->scatterx,mfqP->Xhist[ii],mfqP->localx,INSERT_VALUES, SCATTER_FORWARD);CHKERRQ(ierr); 773 ierr = VecGetArray(mfqP->localx,&x);CHKERRQ(ierr); 774 for (j=0;j<mfqP->n;j++) { 775 mfqP->Disp[ii+mfqP->npmax*j] = (x[j] - mfqP->xmin[j])/mfqP->delta; 776 } 777 ierr = VecRestoreArray(mfqP->localx,&x);CHKERRQ(ierr); 778 779 ierr = VecScatterBegin(mfqP->scatterf,mfqP->Fhist[ii],mfqP->localf,INSERT_VALUES, SCATTER_FORWARD);CHKERRQ(ierr); 780 ierr = VecScatterEnd(mfqP->scatterf,mfqP->Fhist[ii],mfqP->localf,INSERT_VALUES, SCATTER_FORWARD);CHKERRQ(ierr); 781 ierr = VecGetArray(mfqP->localf,&f);CHKERRQ(ierr); 782 for (j=0;j<mfqP->m;j++) { 783 mfqP->Fdiff[ii+mfqP->n*j] = f[j] - fmin[j]; 784 } 785 ierr = VecRestoreArray(mfqP->localf,&f);CHKERRQ(ierr); 786 787 mfqP->model_indices[ii++] = i; 788 } 789 for (j=0;j<mfqP->m;j++) { 790 mfqP->C[j] = fmin[j]; 791 } 792 ierr = VecRestoreArrayRead(mfqP->localfmin,&fmin);CHKERRQ(ierr); 793 } 794 795 /* Determine the initial quadratic models */ 796 /* G = D(ModelIn,:) \ (F(ModelIn,1:m)-repmat(F(xkin,1:m),n,1)); */ 797 /* D (nxn) Fdiff (nxm) => G (nxm) */ 798 blasncopy = blasn; 799 PetscStackCallBLAS("LAPACKgesv",LAPACKgesv_(&blasn,&blasm,mfqP->Disp,&blasnpmax,mfqP->iwork,mfqP->Fdiff,&blasncopy,&info)); 800 ierr = PetscInfo1(tao,"Linear solve return: %D\n",(PetscInt)info);CHKERRQ(ierr); 801 802 cres = minnorm; 803 ierr = pounders_update_res(tao);CHKERRQ(ierr); 804 805 valid = PETSC_TRUE; 806 807 ierr = VecSetValues(tao->gradient,mfqP->n,mfqP->indices,mfqP->Gres,INSERT_VALUES);CHKERRQ(ierr); 808 ierr = VecAssemblyBegin(tao->gradient);CHKERRQ(ierr); 809 ierr = VecAssemblyEnd(tao->gradient);CHKERRQ(ierr); 810 ierr = VecNorm(tao->gradient,NORM_2,&gnorm);CHKERRQ(ierr); 811 gnorm *= mfqP->delta; 812 ierr = VecCopy(mfqP->Xhist[mfqP->minindex],tao->solution);CHKERRQ(ierr); 813 814 tao->reason = TAO_CONTINUE_ITERATING; 815 ierr = TaoLogConvergenceHistory(tao,minnorm,gnorm,0.0,tao->ksp_its);CHKERRQ(ierr); 816 ierr = TaoMonitor(tao,tao->niter,minnorm,gnorm,0.0,step);CHKERRQ(ierr); 817 ierr = (*tao->ops->convergencetest)(tao,tao->cnvP);CHKERRQ(ierr); 818 819 mfqP->nHist = mfqP->n+1; 820 mfqP->nmodelpoints = mfqP->n+1; 821 ierr = PetscInfo1(tao,"Initial gradient: %20.19e\n",(double)gnorm);CHKERRQ(ierr); 822 823 while (tao->reason == TAO_CONTINUE_ITERATING) { 824 PetscReal gnm = 1e-4; 825 /* Call general purpose update function */ 826 if (tao->ops->update) { 827 ierr = (*tao->ops->update)(tao, tao->niter, tao->user_update);CHKERRQ(ierr); 828 } 829 tao->niter++; 830 /* Solve the subproblem min{Q(s): ||s|| <= 1.0} */ 831 ierr = gqtwrap(tao,&gnm,&mdec);CHKERRQ(ierr); 832 /* Evaluate the function at the new point */ 833 834 for (i=0;i<mfqP->n;i++) { 835 mfqP->work[i] = mfqP->Xsubproblem[i]*mfqP->delta + mfqP->xmin[i]; 836 } 837 ierr = VecDuplicate(tao->solution,&mfqP->Xhist[mfqP->nHist]);CHKERRQ(ierr); 838 ierr = VecDuplicate(tao->ls_res,&mfqP->Fhist[mfqP->nHist]);CHKERRQ(ierr); 839 ierr = VecSetValues(mfqP->Xhist[mfqP->nHist],mfqP->n,mfqP->indices,mfqP->work,INSERT_VALUES);CHKERRQ(ierr); 840 ierr = VecAssemblyBegin(mfqP->Xhist[mfqP->nHist]);CHKERRQ(ierr); 841 ierr = VecAssemblyEnd(mfqP->Xhist[mfqP->nHist]);CHKERRQ(ierr); 842 843 ierr = pounders_feval(tao,mfqP->Xhist[mfqP->nHist],mfqP->Fhist[mfqP->nHist],&mfqP->Fres[mfqP->nHist]);CHKERRQ(ierr); 844 845 rho = (mfqP->Fres[mfqP->minindex] - mfqP->Fres[mfqP->nHist]) / mdec; 846 mfqP->nHist++; 847 848 /* Update the center */ 849 if ((rho >= mfqP->eta1) || (rho > mfqP->eta0 && valid==PETSC_TRUE)) { 850 /* Update model to reflect new base point */ 851 for (i=0;i<mfqP->n;i++) { 852 mfqP->work[i] = (mfqP->work[i] - mfqP->xmin[i])/mfqP->delta; 853 } 854 for (j=0;j<mfqP->m;j++) { 855 /* C(j) = C(j) + work*G(:,j) + .5*work*H(:,:,j)*work'; 856 G(:,j) = G(:,j) + H(:,:,j)*work' */ 857 for (k=0;k<mfqP->n;k++) { 858 mfqP->work2[k]=0.0; 859 for (l=0;l<mfqP->n;l++) { 860 mfqP->work2[k]+=mfqP->H[j + mfqP->m*(k + l*mfqP->n)]*mfqP->work[l]; 861 } 862 } 863 for (i=0;i<mfqP->n;i++) { 864 mfqP->C[j]+=mfqP->work[i]*(mfqP->Fdiff[i + mfqP->n* j] + 0.5*mfqP->work2[i]); 865 mfqP->Fdiff[i+mfqP->n*j] +=mfqP-> work2[i]; 866 } 867 } 868 /* Cres += work*Gres + .5*work*Hres*work'; 869 Gres += Hres*work'; */ 870 871 PetscStackCallBLAS("BLASgemv",BLASgemv_("N",&blasn,&blasn,&one,mfqP->Hres,&blasn,mfqP->work,&ione,&zero,mfqP->work2,&ione)); 872 for (i=0;i<mfqP->n;i++) { 873 cres += mfqP->work[i]*(mfqP->Gres[i] + 0.5*mfqP->work2[i]); 874 mfqP->Gres[i] += mfqP->work2[i]; 875 } 876 mfqP->minindex = mfqP->nHist-1; 877 minnorm = mfqP->Fres[mfqP->minindex]; 878 ierr = VecCopy(mfqP->Fhist[mfqP->minindex],tao->ls_res);CHKERRQ(ierr); 879 /* Change current center */ 880 ierr = VecGetArrayRead(mfqP->Xhist[mfqP->minindex],&xmint);CHKERRQ(ierr); 881 for (i=0;i<mfqP->n;i++) { 882 mfqP->xmin[i] = xmint[i]; 883 } 884 ierr = VecRestoreArrayRead(mfqP->Xhist[mfqP->minindex],&xmint);CHKERRQ(ierr); 885 } 886 887 /* Evaluate at a model-improving point if necessary */ 888 if (valid == PETSC_FALSE) { 889 mfqP->q_is_I = 1; 890 mfqP->nmodelpoints = 0; 891 ierr = affpoints(mfqP,mfqP->xmin,mfqP->c1);CHKERRQ(ierr); 892 if (mfqP->nmodelpoints < mfqP->n) { 893 ierr = PetscInfo(tao,"Model not valid -- model-improving\n");CHKERRQ(ierr); 894 ierr = modelimprove(tao,mfqP,1);CHKERRQ(ierr); 895 } 896 } 897 898 /* Update the trust region radius */ 899 deltaold = mfqP->delta; 900 normxsp = 0; 901 for (i=0;i<mfqP->n;i++) { 902 normxsp += mfqP->Xsubproblem[i]*mfqP->Xsubproblem[i]; 903 } 904 normxsp = PetscSqrtReal(normxsp); 905 if (rho >= mfqP->eta1 && normxsp > 0.5*mfqP->delta) { 906 mfqP->delta = PetscMin(mfqP->delta*mfqP->gamma1,mfqP->deltamax); 907 } else if (valid == PETSC_TRUE) { 908 mfqP->delta = PetscMax(mfqP->delta*mfqP->gamma0,mfqP->deltamin); 909 } 910 911 /* Compute the next interpolation set */ 912 mfqP->q_is_I = 1; 913 mfqP->nmodelpoints=0; 914 ierr = PetscInfo2(tao,"Affine Points: xmin = %20.19e, c1 = %20.19e\n",(double)*mfqP->xmin,(double)mfqP->c1);CHKERRQ(ierr); 915 ierr = affpoints(mfqP,mfqP->xmin,mfqP->c1);CHKERRQ(ierr); 916 if (mfqP->nmodelpoints == mfqP->n) { 917 valid = PETSC_TRUE; 918 } else { 919 valid = PETSC_FALSE; 920 ierr = PetscInfo2(tao,"Affine Points: xmin = %20.19e, c2 = %20.19e\n",(double)*mfqP->xmin,(double)mfqP->c2);CHKERRQ(ierr); 921 ierr = affpoints(mfqP,mfqP->xmin,mfqP->c2);CHKERRQ(ierr); 922 if (mfqP->n > mfqP->nmodelpoints) { 923 ierr = PetscInfo(tao,"Model not valid -- adding geometry points\n");CHKERRQ(ierr); 924 ierr = modelimprove(tao,mfqP,mfqP->n - mfqP->nmodelpoints);CHKERRQ(ierr); 925 } 926 } 927 for (i=mfqP->nmodelpoints;i>0;i--) { 928 mfqP->model_indices[i] = mfqP->model_indices[i-1]; 929 } 930 mfqP->nmodelpoints++; 931 mfqP->model_indices[0] = mfqP->minindex; 932 ierr = morepoints(mfqP);CHKERRQ(ierr); 933 for (i=0;i<mfqP->nmodelpoints;i++) { 934 ierr = VecGetArray(mfqP->Xhist[mfqP->model_indices[i]],&x);CHKERRQ(ierr); 935 for (j=0;j<mfqP->n;j++) { 936 mfqP->Disp[i + mfqP->npmax*j] = (x[j] - mfqP->xmin[j]) / deltaold; 937 } 938 ierr = VecRestoreArray(mfqP->Xhist[mfqP->model_indices[i]],&x);CHKERRQ(ierr); 939 ierr = VecGetArray(mfqP->Fhist[mfqP->model_indices[i]],&f);CHKERRQ(ierr); 940 for (j=0;j<mfqP->m;j++) { 941 for (k=0;k<mfqP->n;k++) { 942 mfqP->work[k]=0.0; 943 for (l=0;l<mfqP->n;l++) { 944 mfqP->work[k] += mfqP->H[j + mfqP->m*(k + mfqP->n*l)] * mfqP->Disp[i + mfqP->npmax*l]; 945 } 946 } 947 mfqP->RES[j*mfqP->npmax + i] = -mfqP->C[j] - BLASdot_(&blasn,&mfqP->Fdiff[j*mfqP->n],&ione,&mfqP->Disp[i],&blasnpmax) - 0.5*BLASdot_(&blasn,mfqP->work,&ione,&mfqP->Disp[i],&blasnpmax) + f[j]; 948 } 949 ierr = VecRestoreArray(mfqP->Fhist[mfqP->model_indices[i]],&f);CHKERRQ(ierr); 950 } 951 952 /* Update the quadratic model */ 953 ierr = PetscInfo2(tao,"Get Quad, size: %D, points: %D\n",mfqP->n,mfqP->nmodelpoints);CHKERRQ(ierr); 954 ierr = getquadpounders(mfqP);CHKERRQ(ierr); 955 ierr = VecGetArrayRead(mfqP->Fhist[mfqP->minindex],&fmin);CHKERRQ(ierr); 956 PetscStackCallBLAS("BLAScopy",BLAScopy_(&blasm,fmin,&ione,mfqP->C,&ione)); 957 /* G = G*(delta/deltaold) + Gdel */ 958 ratio = mfqP->delta/deltaold; 959 iblas = blasm*blasn; 960 PetscStackCallBLAS("BLASscal",BLASscal_(&iblas,&ratio,mfqP->Fdiff,&ione)); 961 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&iblas,&one,mfqP->Gdel,&ione,mfqP->Fdiff,&ione)); 962 /* H = H*(delta/deltaold)^2 + Hdel */ 963 iblas = blasm*blasn*blasn; 964 ratio *= ratio; 965 PetscStackCallBLAS("BLASscal",BLASscal_(&iblas,&ratio,mfqP->H,&ione)); 966 PetscStackCallBLAS("BLASaxpy",BLASaxpy_(&iblas,&one,mfqP->Hdel,&ione,mfqP->H,&ione)); 967 968 /* Get residuals */ 969 cres = mfqP->Fres[mfqP->minindex]; 970 ierr = pounders_update_res(tao);CHKERRQ(ierr); 971 972 /* Export solution and gradient residual to TAO */ 973 ierr = VecCopy(mfqP->Xhist[mfqP->minindex],tao->solution);CHKERRQ(ierr); 974 ierr = VecSetValues(tao->gradient,mfqP->n,mfqP->indices,mfqP->Gres,INSERT_VALUES);CHKERRQ(ierr); 975 ierr = VecAssemblyBegin(tao->gradient);CHKERRQ(ierr); 976 ierr = VecAssemblyEnd(tao->gradient);CHKERRQ(ierr); 977 ierr = VecNorm(tao->gradient,NORM_2,&gnorm);CHKERRQ(ierr); 978 gnorm *= mfqP->delta; 979 /* final criticality test */ 980 ierr = TaoLogConvergenceHistory(tao,minnorm,gnorm,0.0,tao->ksp_its);CHKERRQ(ierr); 981 ierr = TaoMonitor(tao,tao->niter,minnorm,gnorm,0.0,step);CHKERRQ(ierr); 982 ierr = (*tao->ops->convergencetest)(tao,tao->cnvP);CHKERRQ(ierr); 983 /* test for repeated model */ 984 if (mfqP->nmodelpoints==mfqP->last_nmodelpoints) { 985 same = PETSC_TRUE; 986 } else { 987 same = PETSC_FALSE; 988 } 989 for (i=0;i<mfqP->nmodelpoints;i++) { 990 if (same) { 991 if (mfqP->model_indices[i] == mfqP->last_model_indices[i]) { 992 same = PETSC_TRUE; 993 } else { 994 same = PETSC_FALSE; 995 } 996 } 997 mfqP->last_model_indices[i] = mfqP->model_indices[i]; 998 } 999 mfqP->last_nmodelpoints = mfqP->nmodelpoints; 1000 if (same && mfqP->delta == deltaold) { 1001 ierr = PetscInfo(tao,"Identical model used in successive iterations\n");CHKERRQ(ierr); 1002 tao->reason = TAO_CONVERGED_STEPTOL; 1003 } 1004 } 1005 PetscFunctionReturn(0); 1006 } 1007 1008 static PetscErrorCode TaoSetUp_POUNDERS(Tao tao) 1009 { 1010 TAO_POUNDERS *mfqP = (TAO_POUNDERS*)tao->data; 1011 PetscInt i,j; 1012 IS isfloc,isfglob,isxloc,isxglob; 1013 PetscErrorCode ierr; 1014 1015 PetscFunctionBegin; 1016 if (!tao->gradient) {ierr = VecDuplicate(tao->solution,&tao->gradient);CHKERRQ(ierr); } 1017 if (!tao->stepdirection) {ierr = VecDuplicate(tao->solution,&tao->stepdirection);CHKERRQ(ierr); } 1018 ierr = VecGetSize(tao->solution,&mfqP->n);CHKERRQ(ierr); 1019 ierr = VecGetSize(tao->ls_res,&mfqP->m);CHKERRQ(ierr); 1020 mfqP->c1 = PetscSqrtReal((PetscReal)mfqP->n); 1021 if (mfqP->npmax == PETSC_DEFAULT) { 1022 mfqP->npmax = 2*mfqP->n + 1; 1023 } 1024 mfqP->npmax = PetscMin((mfqP->n+1)*(mfqP->n+2)/2,mfqP->npmax); 1025 mfqP->npmax = PetscMax(mfqP->npmax, mfqP->n+2); 1026 1027 ierr = PetscMalloc1(tao->max_funcs+100,&mfqP->Xhist);CHKERRQ(ierr); 1028 ierr = PetscMalloc1(tao->max_funcs+100,&mfqP->Fhist);CHKERRQ(ierr); 1029 for (i=0;i<mfqP->n+1;i++) { 1030 ierr = VecDuplicate(tao->solution,&mfqP->Xhist[i]);CHKERRQ(ierr); 1031 ierr = VecDuplicate(tao->ls_res,&mfqP->Fhist[i]);CHKERRQ(ierr); 1032 } 1033 ierr = VecDuplicate(tao->solution,&mfqP->workxvec);CHKERRQ(ierr); 1034 ierr = VecDuplicate(tao->ls_res,&mfqP->workfvec);CHKERRQ(ierr); 1035 mfqP->nHist = 0; 1036 1037 ierr = PetscMalloc1(tao->max_funcs+100,&mfqP->Fres);CHKERRQ(ierr); 1038 ierr = PetscMalloc1(mfqP->npmax*mfqP->m,&mfqP->RES);CHKERRQ(ierr); 1039 ierr = PetscMalloc1(mfqP->n,&mfqP->work);CHKERRQ(ierr); 1040 ierr = PetscMalloc1(mfqP->n,&mfqP->work2);CHKERRQ(ierr); 1041 ierr = PetscMalloc1(mfqP->n,&mfqP->work3);CHKERRQ(ierr); 1042 ierr = PetscMalloc1(PetscMax(mfqP->m,mfqP->n+1),&mfqP->mwork);CHKERRQ(ierr); 1043 ierr = PetscMalloc1(mfqP->npmax - mfqP->n - 1,&mfqP->omega);CHKERRQ(ierr); 1044 ierr = PetscMalloc1(mfqP->n * (mfqP->n+1) / 2,&mfqP->beta);CHKERRQ(ierr); 1045 ierr = PetscMalloc1(mfqP->n + 1 ,&mfqP->alpha);CHKERRQ(ierr); 1046 1047 ierr = PetscMalloc1(mfqP->n*mfqP->n*mfqP->m,&mfqP->H);CHKERRQ(ierr); 1048 ierr = PetscMalloc1(mfqP->npmax*mfqP->npmax,&mfqP->Q);CHKERRQ(ierr); 1049 ierr = PetscMalloc1(mfqP->npmax*mfqP->npmax,&mfqP->Q_tmp);CHKERRQ(ierr); 1050 ierr = PetscMalloc1(mfqP->n*(mfqP->n+1)/2*(mfqP->npmax),&mfqP->L);CHKERRQ(ierr); 1051 ierr = PetscMalloc1(mfqP->n*(mfqP->n+1)/2*(mfqP->npmax),&mfqP->L_tmp);CHKERRQ(ierr); 1052 ierr = PetscMalloc1(mfqP->n*(mfqP->n+1)/2*(mfqP->npmax),&mfqP->L_save);CHKERRQ(ierr); 1053 ierr = PetscMalloc1(mfqP->n*(mfqP->n+1)/2*(mfqP->npmax),&mfqP->N);CHKERRQ(ierr); 1054 ierr = PetscMalloc1(mfqP->npmax * (mfqP->n+1) ,&mfqP->M);CHKERRQ(ierr); 1055 ierr = PetscMalloc1(mfqP->npmax * (mfqP->npmax - mfqP->n - 1) , &mfqP->Z);CHKERRQ(ierr); 1056 ierr = PetscMalloc1(mfqP->npmax,&mfqP->tau);CHKERRQ(ierr); 1057 ierr = PetscMalloc1(mfqP->npmax,&mfqP->tau_tmp);CHKERRQ(ierr); 1058 mfqP->nmax = PetscMax(5*mfqP->npmax,mfqP->n*(mfqP->n+1)/2); 1059 ierr = PetscMalloc1(mfqP->nmax,&mfqP->npmaxwork);CHKERRQ(ierr); 1060 ierr = PetscMalloc1(mfqP->nmax,&mfqP->npmaxiwork);CHKERRQ(ierr); 1061 ierr = PetscMalloc1(mfqP->n,&mfqP->xmin);CHKERRQ(ierr); 1062 ierr = PetscMalloc1(mfqP->m,&mfqP->C);CHKERRQ(ierr); 1063 ierr = PetscMalloc1(mfqP->m*mfqP->n,&mfqP->Fdiff);CHKERRQ(ierr); 1064 ierr = PetscMalloc1(mfqP->npmax*mfqP->n,&mfqP->Disp);CHKERRQ(ierr); 1065 ierr = PetscMalloc1(mfqP->n,&mfqP->Gres);CHKERRQ(ierr); 1066 ierr = PetscMalloc1(mfqP->n*mfqP->n,&mfqP->Hres);CHKERRQ(ierr); 1067 ierr = PetscMalloc1(mfqP->n*mfqP->n,&mfqP->Gpoints);CHKERRQ(ierr); 1068 ierr = PetscMalloc1(mfqP->npmax,&mfqP->model_indices);CHKERRQ(ierr); 1069 ierr = PetscMalloc1(mfqP->npmax,&mfqP->last_model_indices);CHKERRQ(ierr); 1070 ierr = PetscMalloc1(mfqP->n,&mfqP->Xsubproblem);CHKERRQ(ierr); 1071 ierr = PetscMalloc1(mfqP->m*mfqP->n,&mfqP->Gdel);CHKERRQ(ierr); 1072 ierr = PetscMalloc1(mfqP->n*mfqP->n*mfqP->m, &mfqP->Hdel);CHKERRQ(ierr); 1073 ierr = PetscMalloc1(PetscMax(mfqP->m,mfqP->n),&mfqP->indices);CHKERRQ(ierr); 1074 ierr = PetscMalloc1(mfqP->n,&mfqP->iwork);CHKERRQ(ierr); 1075 ierr = PetscMalloc1(mfqP->m*mfqP->m,&mfqP->w);CHKERRQ(ierr); 1076 for (i=0;i<mfqP->m;i++) { 1077 for (j=0;j<mfqP->m;j++) { 1078 if (i==j) { 1079 mfqP->w[i+mfqP->m*j]=1.0; 1080 } else { 1081 mfqP->w[i+mfqP->m*j]=0.0; 1082 } 1083 } 1084 } 1085 for (i=0;i<PetscMax(mfqP->m,mfqP->n);i++) { 1086 mfqP->indices[i] = i; 1087 } 1088 ierr = MPI_Comm_size(((PetscObject)tao)->comm,&mfqP->size);CHKERRQ(ierr); 1089 if (mfqP->size > 1) { 1090 ierr = VecCreateSeq(PETSC_COMM_SELF,mfqP->n,&mfqP->localx);CHKERRQ(ierr); 1091 ierr = VecCreateSeq(PETSC_COMM_SELF,mfqP->n,&mfqP->localxmin);CHKERRQ(ierr); 1092 ierr = VecCreateSeq(PETSC_COMM_SELF,mfqP->m,&mfqP->localf);CHKERRQ(ierr); 1093 ierr = VecCreateSeq(PETSC_COMM_SELF,mfqP->m,&mfqP->localfmin);CHKERRQ(ierr); 1094 ierr = ISCreateStride(MPI_COMM_SELF,mfqP->n,0,1,&isxloc);CHKERRQ(ierr); 1095 ierr = ISCreateStride(MPI_COMM_SELF,mfqP->n,0,1,&isxglob);CHKERRQ(ierr); 1096 ierr = ISCreateStride(MPI_COMM_SELF,mfqP->m,0,1,&isfloc);CHKERRQ(ierr); 1097 ierr = ISCreateStride(MPI_COMM_SELF,mfqP->m,0,1,&isfglob);CHKERRQ(ierr); 1098 1099 1100 ierr = VecScatterCreate(tao->solution,isxglob,mfqP->localx,isxloc,&mfqP->scatterx);CHKERRQ(ierr); 1101 ierr = VecScatterCreate(tao->ls_res,isfglob,mfqP->localf,isfloc,&mfqP->scatterf);CHKERRQ(ierr); 1102 1103 ierr = ISDestroy(&isxloc);CHKERRQ(ierr); 1104 ierr = ISDestroy(&isxglob);CHKERRQ(ierr); 1105 ierr = ISDestroy(&isfloc);CHKERRQ(ierr); 1106 ierr = ISDestroy(&isfglob);CHKERRQ(ierr); 1107 } 1108 1109 if (!mfqP->usegqt) { 1110 KSP ksp; 1111 PC pc; 1112 ierr = VecCreateSeqWithArray(PETSC_COMM_SELF,mfqP->n,mfqP->n,mfqP->Xsubproblem,&mfqP->subx);CHKERRQ(ierr); 1113 ierr = VecCreateSeq(PETSC_COMM_SELF,mfqP->n,&mfqP->subxl);CHKERRQ(ierr); 1114 ierr = VecDuplicate(mfqP->subxl,&mfqP->subb);CHKERRQ(ierr); 1115 ierr = VecDuplicate(mfqP->subxl,&mfqP->subxu);CHKERRQ(ierr); 1116 ierr = VecDuplicate(mfqP->subxl,&mfqP->subpdel);CHKERRQ(ierr); 1117 ierr = VecDuplicate(mfqP->subxl,&mfqP->subndel);CHKERRQ(ierr); 1118 ierr = TaoCreate(PETSC_COMM_SELF,&mfqP->subtao);CHKERRQ(ierr); 1119 ierr = PetscObjectIncrementTabLevel((PetscObject)mfqP->subtao, (PetscObject)tao, 1);CHKERRQ(ierr); 1120 ierr = TaoSetType(mfqP->subtao,TAOBNTR);CHKERRQ(ierr); 1121 ierr = TaoSetOptionsPrefix(mfqP->subtao,"pounders_subsolver_");CHKERRQ(ierr); 1122 ierr = TaoSetInitialVector(mfqP->subtao,mfqP->subx);CHKERRQ(ierr); 1123 ierr = TaoSetObjectiveAndGradientRoutine(mfqP->subtao,pounders_fg,(void*)mfqP);CHKERRQ(ierr); 1124 ierr = TaoSetMaximumIterations(mfqP->subtao,mfqP->gqt_maxits);CHKERRQ(ierr); 1125 ierr = TaoSetFromOptions(mfqP->subtao);CHKERRQ(ierr); 1126 ierr = TaoGetKSP(mfqP->subtao,&ksp);CHKERRQ(ierr); 1127 if (ksp) { 1128 ierr = KSPGetPC(ksp,&pc);CHKERRQ(ierr); 1129 ierr = PCSetType(pc,PCNONE);CHKERRQ(ierr); 1130 } 1131 ierr = TaoSetVariableBounds(mfqP->subtao,mfqP->subxl,mfqP->subxu);CHKERRQ(ierr); 1132 ierr = MatCreateSeqDense(PETSC_COMM_SELF,mfqP->n,mfqP->n,mfqP->Hres,&mfqP->subH);CHKERRQ(ierr); 1133 ierr = TaoSetHessianRoutine(mfqP->subtao,mfqP->subH,mfqP->subH,pounders_h,(void*)mfqP);CHKERRQ(ierr); 1134 } 1135 PetscFunctionReturn(0); 1136 } 1137 1138 static PetscErrorCode TaoDestroy_POUNDERS(Tao tao) 1139 { 1140 TAO_POUNDERS *mfqP = (TAO_POUNDERS*)tao->data; 1141 PetscInt i; 1142 PetscErrorCode ierr; 1143 1144 PetscFunctionBegin; 1145 if (!mfqP->usegqt) { 1146 ierr = TaoDestroy(&mfqP->subtao);CHKERRQ(ierr); 1147 ierr = VecDestroy(&mfqP->subx);CHKERRQ(ierr); 1148 ierr = VecDestroy(&mfqP->subxl);CHKERRQ(ierr); 1149 ierr = VecDestroy(&mfqP->subxu);CHKERRQ(ierr); 1150 ierr = VecDestroy(&mfqP->subb);CHKERRQ(ierr); 1151 ierr = VecDestroy(&mfqP->subpdel);CHKERRQ(ierr); 1152 ierr = VecDestroy(&mfqP->subndel);CHKERRQ(ierr); 1153 ierr = MatDestroy(&mfqP->subH);CHKERRQ(ierr); 1154 } 1155 ierr = PetscFree(mfqP->Fres);CHKERRQ(ierr); 1156 ierr = PetscFree(mfqP->RES);CHKERRQ(ierr); 1157 ierr = PetscFree(mfqP->work);CHKERRQ(ierr); 1158 ierr = PetscFree(mfqP->work2);CHKERRQ(ierr); 1159 ierr = PetscFree(mfqP->work3);CHKERRQ(ierr); 1160 ierr = PetscFree(mfqP->mwork);CHKERRQ(ierr); 1161 ierr = PetscFree(mfqP->omega);CHKERRQ(ierr); 1162 ierr = PetscFree(mfqP->beta);CHKERRQ(ierr); 1163 ierr = PetscFree(mfqP->alpha);CHKERRQ(ierr); 1164 ierr = PetscFree(mfqP->H);CHKERRQ(ierr); 1165 ierr = PetscFree(mfqP->Q);CHKERRQ(ierr); 1166 ierr = PetscFree(mfqP->Q_tmp);CHKERRQ(ierr); 1167 ierr = PetscFree(mfqP->L);CHKERRQ(ierr); 1168 ierr = PetscFree(mfqP->L_tmp);CHKERRQ(ierr); 1169 ierr = PetscFree(mfqP->L_save);CHKERRQ(ierr); 1170 ierr = PetscFree(mfqP->N);CHKERRQ(ierr); 1171 ierr = PetscFree(mfqP->M);CHKERRQ(ierr); 1172 ierr = PetscFree(mfqP->Z);CHKERRQ(ierr); 1173 ierr = PetscFree(mfqP->tau);CHKERRQ(ierr); 1174 ierr = PetscFree(mfqP->tau_tmp);CHKERRQ(ierr); 1175 ierr = PetscFree(mfqP->npmaxwork);CHKERRQ(ierr); 1176 ierr = PetscFree(mfqP->npmaxiwork);CHKERRQ(ierr); 1177 ierr = PetscFree(mfqP->xmin);CHKERRQ(ierr); 1178 ierr = PetscFree(mfqP->C);CHKERRQ(ierr); 1179 ierr = PetscFree(mfqP->Fdiff);CHKERRQ(ierr); 1180 ierr = PetscFree(mfqP->Disp);CHKERRQ(ierr); 1181 ierr = PetscFree(mfqP->Gres);CHKERRQ(ierr); 1182 ierr = PetscFree(mfqP->Hres);CHKERRQ(ierr); 1183 ierr = PetscFree(mfqP->Gpoints);CHKERRQ(ierr); 1184 ierr = PetscFree(mfqP->model_indices);CHKERRQ(ierr); 1185 ierr = PetscFree(mfqP->last_model_indices);CHKERRQ(ierr); 1186 ierr = PetscFree(mfqP->Xsubproblem);CHKERRQ(ierr); 1187 ierr = PetscFree(mfqP->Gdel);CHKERRQ(ierr); 1188 ierr = PetscFree(mfqP->Hdel);CHKERRQ(ierr); 1189 ierr = PetscFree(mfqP->indices);CHKERRQ(ierr); 1190 ierr = PetscFree(mfqP->iwork);CHKERRQ(ierr); 1191 ierr = PetscFree(mfqP->w);CHKERRQ(ierr); 1192 for (i=0;i<mfqP->nHist;i++) { 1193 ierr = VecDestroy(&mfqP->Xhist[i]);CHKERRQ(ierr); 1194 ierr = VecDestroy(&mfqP->Fhist[i]);CHKERRQ(ierr); 1195 } 1196 ierr = VecDestroy(&mfqP->workxvec);CHKERRQ(ierr); 1197 ierr = VecDestroy(&mfqP->workfvec);CHKERRQ(ierr); 1198 ierr = PetscFree(mfqP->Xhist);CHKERRQ(ierr); 1199 ierr = PetscFree(mfqP->Fhist);CHKERRQ(ierr); 1200 1201 if (mfqP->size > 1) { 1202 ierr = VecDestroy(&mfqP->localx);CHKERRQ(ierr); 1203 ierr = VecDestroy(&mfqP->localxmin);CHKERRQ(ierr); 1204 ierr = VecDestroy(&mfqP->localf);CHKERRQ(ierr); 1205 ierr = VecDestroy(&mfqP->localfmin);CHKERRQ(ierr); 1206 } 1207 ierr = PetscFree(tao->data);CHKERRQ(ierr); 1208 PetscFunctionReturn(0); 1209 } 1210 1211 static PetscErrorCode TaoSetFromOptions_POUNDERS(PetscOptionItems *PetscOptionsObject,Tao tao) 1212 { 1213 TAO_POUNDERS *mfqP = (TAO_POUNDERS*)tao->data; 1214 PetscErrorCode ierr; 1215 1216 PetscFunctionBegin; 1217 ierr = PetscOptionsHead(PetscOptionsObject,"POUNDERS method for least-squares optimization");CHKERRQ(ierr); 1218 ierr = PetscOptionsReal("-tao_pounders_delta","initial delta","",mfqP->delta,&mfqP->delta0,NULL);CHKERRQ(ierr); 1219 mfqP->delta = mfqP->delta0; 1220 ierr = PetscOptionsInt("-tao_pounders_npmax","max number of points in model","",mfqP->npmax,&mfqP->npmax,NULL);CHKERRQ(ierr); 1221 ierr = PetscOptionsBool("-tao_pounders_gqt","use gqt algorithm for subproblem","",mfqP->usegqt,&mfqP->usegqt,NULL);CHKERRQ(ierr); 1222 ierr = PetscOptionsTail();CHKERRQ(ierr); 1223 PetscFunctionReturn(0); 1224 } 1225 1226 static PetscErrorCode TaoView_POUNDERS(Tao tao, PetscViewer viewer) 1227 { 1228 TAO_POUNDERS *mfqP = (TAO_POUNDERS *)tao->data; 1229 PetscBool isascii; 1230 PetscErrorCode ierr; 1231 1232 PetscFunctionBegin; 1233 ierr = PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&isascii);CHKERRQ(ierr); 1234 if (isascii) { 1235 ierr = PetscViewerASCIIPrintf(viewer, "initial delta: %g\n",(double)mfqP->delta0);CHKERRQ(ierr); 1236 ierr = PetscViewerASCIIPrintf(viewer, "final delta: %g\n",(double)mfqP->delta);CHKERRQ(ierr); 1237 ierr = PetscViewerASCIIPrintf(viewer, "model points: %D\n",mfqP->nmodelpoints);CHKERRQ(ierr); 1238 if (mfqP->usegqt) { 1239 ierr = PetscViewerASCIIPrintf(viewer, "subproblem solver: gqt\n");CHKERRQ(ierr); 1240 } else { 1241 ierr = TaoView(mfqP->subtao, viewer);CHKERRQ(ierr); 1242 } 1243 } 1244 PetscFunctionReturn(0); 1245 } 1246 /*MC 1247 TAOPOUNDERS - POUNDERS derivate-free model-based algorithm for nonlinear least squares 1248 1249 Options Database Keys: 1250 + -tao_pounders_delta - initial step length 1251 . -tao_pounders_npmax - maximum number of points in model 1252 - -tao_pounders_gqt - use gqt algorithm for subproblem instead of TRON 1253 1254 Level: beginner 1255 1256 M*/ 1257 1258 PETSC_EXTERN PetscErrorCode TaoCreate_POUNDERS(Tao tao) 1259 { 1260 TAO_POUNDERS *mfqP = (TAO_POUNDERS*)tao->data; 1261 PetscErrorCode ierr; 1262 1263 PetscFunctionBegin; 1264 tao->ops->setup = TaoSetUp_POUNDERS; 1265 tao->ops->solve = TaoSolve_POUNDERS; 1266 tao->ops->view = TaoView_POUNDERS; 1267 tao->ops->setfromoptions = TaoSetFromOptions_POUNDERS; 1268 tao->ops->destroy = TaoDestroy_POUNDERS; 1269 1270 ierr = PetscNewLog(tao,&mfqP);CHKERRQ(ierr); 1271 tao->data = (void*)mfqP; 1272 /* Override default settings (unless already changed) */ 1273 if (!tao->max_it_changed) tao->max_it = 2000; 1274 if (!tao->max_funcs_changed) tao->max_funcs = 4000; 1275 mfqP->npmax = PETSC_DEFAULT; 1276 mfqP->delta0 = 0.1; 1277 mfqP->delta = 0.1; 1278 mfqP->deltamax=1e3; 1279 mfqP->deltamin=1e-6; 1280 mfqP->c2 = 10.0; 1281 mfqP->theta1=1e-5; 1282 mfqP->theta2=1e-4; 1283 mfqP->gamma0=0.5; 1284 mfqP->gamma1=2.0; 1285 mfqP->eta0 = 0.0; 1286 mfqP->eta1 = 0.1; 1287 mfqP->usegqt = PETSC_FALSE; 1288 mfqP->gqt_rtol = 0.001; 1289 mfqP->gqt_maxits = 50; 1290 mfqP->workxvec = NULL; 1291 PetscFunctionReturn(0); 1292 } 1293