/usr/include/trilinos/ROL_PrimalDualActiveSetStep.hpp is in libtrilinos-rol-dev 12.12.1-5.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 | // @HEADER
// ************************************************************************
//
// Rapid Optimization Library (ROL) Package
// Copyright (2014) Sandia Corporation
//
// Under terms of Contract DE-AC04-94AL85000, there is a non-exclusive
// license for use of this work by or on behalf of the U.S. Government.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// 3. Neither the name of the Corporation nor the names of the
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
// PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
// LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Questions? Contact lead developers:
// Drew Kouri (dpkouri@sandia.gov) and
// Denis Ridzal (dridzal@sandia.gov)
//
// ************************************************************************
// @HEADER
#ifndef ROL_PRIMALDUALACTIVESETSTEP_H
#define ROL_PRIMALDUALACTIVESETSTEP_H
#include "ROL_Step.hpp"
#include "ROL_Vector.hpp"
#include "ROL_Krylov.hpp"
#include "ROL_Objective.hpp"
#include "ROL_BoundConstraint.hpp"
#include "ROL_Types.hpp"
#include "ROL_Secant.hpp"
#include "Teuchos_ParameterList.hpp"
/** @ingroup step_group
\class ROL::PrimalDualActiveSetStep
\brief Implements the computation of optimization steps
with the Newton primal-dual active set method.
To describe primal-dual active set (PDAS), we consider the following
abstract setting. Suppose \f$\mathcal{X}\f$ is a Hilbert space of
functions mapping \f$\Xi\f$ to \f$\mathbb{R}\f$. For example,
\f$\Xi\subset\mathbb{R}^n\f$ and \f$\mathcal{X}=L^2(\Xi)\f$ or
\f$\Xi = \{1,\ldots,n\}\f$ and \f$\mathcal{X}=\mathbb{R}^n\f$. We
assume \f$ f:\mathcal{X}\to\mathbb{R}\f$ is twice-continuously Fréchet
differentiable and \f$a,\,b\in\mathcal{X}\f$ with \f$a\le b\f$ almost
everywhere in \f$\Xi\f$. Note that the PDAS algorithm will also work
with secant approximations of the Hessian.
Traditionally, PDAS is an algorithm for the minimizing quadratic objective
functions subject to bound constraints. ROL implements a Newton PDAS which
extends PDAS to general bound-constrained nonlinear programs, i.e.,
\f[
\min_x \quad f(x) \quad \text{s.t.} \quad a \le x \le b.
\f]
Given the \f$k\f$-th iterate \f$x_k\f$, the Newton PDAS algorithm computes
steps by applying PDAS to the quadratic subproblem
\f[
\min_s \quad \langle \nabla^2 f(x_k)s + \nabla f(x_k),s \rangle_{\mathcal{X}}
\quad \text{s.t.} \quad a \le x_k + s \le b.
\f]
For the \f$k\f$-th quadratic subproblem, PDAS builds an approximation of the
active set \f$\mathcal{A}_k\f$ using the dual variable \f$\lambda_k\f$ as
\f[
\mathcal{A}^+_k = \{\,\xi\in\Xi\,:\,(\lambda_k + c(x_k-b))(\xi) > 0\,\}, \quad
\mathcal{A}^-_k = \{\,\xi\in\Xi\,:\,(\lambda_k + c(x_k-a))(\xi) < 0\,\}, \quad\text{and}\quad
\mathcal{A}_k = \mathcal{A}^-_k\cup\mathcal{A}^+_k.
\f]
We define the inactive set \f$\mathcal{I}_k=\Xi\setminus\mathcal{A}_k\f$.
The solution to the quadratic subproblem is then computed iteratively by solving
\f[
\nabla^2 f(x_k) s_k + \lambda_{k+1} = -\nabla f(x_k), \quad
x_k+s_k = a \;\text{on}\;\mathcal{A}^-_k,\quad x_k+s_k = b\;\text{on}\;\mathcal{A}^+_k,
\quad\text{and}\quad
\lambda_{k+1} = 0\;\text{on}\;\mathcal{I}_k
\f]
and updating the active and inactive sets.
One can rewrite this system by consolidating active and inactive parts, i.e.,
\f[
\begin{pmatrix}
\nabla^2 f(x_k)_{\mathcal{A}_k,\mathcal{A}_k} & \nabla^2 f(x_k)_{\mathcal{A}_k,\mathcal{I}_k} \\
\nabla^2 f(x_k)_{\mathcal{I}_k,\mathcal{A}_k} & \nabla^2 f(x_k)_{\mathcal{I}_k,\mathcal{I}_k}
\end{pmatrix}
\begin{pmatrix}
(s_k)_{\mathcal{A}_k} \\
(s_k)_{\mathcal{I}_k}
\end{pmatrix}
+
\begin{pmatrix}
(\lambda_{k+1})_{\mathcal{A}_k} \\
0
\end{pmatrix}
= -
\begin{pmatrix}
\nabla f(x_k)_{\mathcal{A}_k}\\
\nabla f(x_k)_{\mathcal{I}_k}
\end{pmatrix}.
\f]
Here the subscripts \f$\mathcal{A}_k\f$ and \f$\mathcal{I}_k\f$ denote the active and inactive
components, respectively. Moreover, the active components of \f$s_k\f$ are
\f$s_k(\xi) = a(\xi)-x_k(\xi)\f$ if \f$\xi\in\mathcal{A}^-_k\f$ and \f$s_k(\xi) = b(\xi)-x_k(\xi)\f$
if \f$\xi\in\mathcal{A}^+_k\f$. Since \f$(s_k)_{\mathcal{A}_k}\f$ is fixed, we only need to solve
for the inactive components of \f$s_k\f$ which we can do this using conjugate residuals (CR) (i.e., the
Hessian operator corresponding to the inactive indices may not be positive definite). Once
\f$(s_k)_{\mathcal{I}_k}\f$ is computed, it is straight forward to update the dual variables.
*/
namespace ROL {
template <class Real>
class PrimalDualActiveSetStep : public Step<Real> {
private:
Teuchos::RCP<Krylov<Real> > krylov_;
// Krylov Parameters
int iterCR_; ///< CR iteration counter
int flagCR_; ///< CR termination flag
Real itol_; ///< Inexact CR tolerance
// PDAS Parameters
int maxit_; ///< Maximum number of PDAS iterations
int iter_; ///< PDAS iteration counter
int flag_; ///< PDAS termination flag
Real stol_; ///< PDAS minimum step size stopping tolerance
Real gtol_; ///< PDAS gradient stopping tolerance
Real scale_; ///< Scale for dual variables in the active set, \f$c\f$
Real neps_; ///< \f$\epsilon\f$-active set parameter
bool feasible_; ///< Flag whether the current iterate is feasible or not
// Dual Variable
Teuchos::RCP<Vector<Real> > lambda_; ///< Container for dual variables
Teuchos::RCP<Vector<Real> > xlam_; ///< Container for primal plus dual variables
Teuchos::RCP<Vector<Real> > x0_; ///< Container for initial priaml variables
Teuchos::RCP<Vector<Real> > xbnd_; ///< Container for primal variable bounds
Teuchos::RCP<Vector<Real> > As_; ///< Container for step projected onto active set
Teuchos::RCP<Vector<Real> > xtmp_; ///< Container for temporary primal storage
Teuchos::RCP<Vector<Real> > res_; ///< Container for optimality system residual for quadratic model
Teuchos::RCP<Vector<Real> > Ag_; ///< Container for gradient projected onto active set
Teuchos::RCP<Vector<Real> > rtmp_; ///< Container for temporary right hand side storage
Teuchos::RCP<Vector<Real> > gtmp_; ///< Container for temporary gradient storage
// Secant Information
ESecant esec_; ///< Enum for secant type
Teuchos::RCP<Secant<Real> > secant_; ///< Secant object
bool useSecantPrecond_;
bool useSecantHessVec_;
class HessianPD : public LinearOperator<Real> {
private:
const Teuchos::RCP<Objective<Real> > obj_;
const Teuchos::RCP<BoundConstraint<Real> > bnd_;
const Teuchos::RCP<Vector<Real> > x_;
const Teuchos::RCP<Vector<Real> > xlam_;
Teuchos::RCP<Vector<Real> > v_;
Real eps_;
const Teuchos::RCP<Secant<Real> > secant_;
bool useSecant_;
public:
HessianPD(const Teuchos::RCP<Objective<Real> > &obj,
const Teuchos::RCP<BoundConstraint<Real> > &bnd,
const Teuchos::RCP<Vector<Real> > &x,
const Teuchos::RCP<Vector<Real> > &xlam,
const Real eps = 0,
const Teuchos::RCP<Secant<Real> > &secant = Teuchos::null,
const bool useSecant = false )
: obj_(obj), bnd_(bnd), x_(x), xlam_(xlam),
eps_(eps), secant_(secant), useSecant_(useSecant) {
v_ = x_->clone();
if ( !useSecant || secant == Teuchos::null ) {
useSecant_ = false;
}
}
void apply( Vector<Real> &Hv, const Vector<Real> &v, Real &tol ) const {
v_->set(v);
bnd_->pruneActive(*v_,*xlam_,eps_);
if ( useSecant_ ) {
secant_->applyB(Hv,*v_);
}
else {
obj_->hessVec(Hv,*v_,*x_,tol);
}
bnd_->pruneActive(Hv,*xlam_,eps_);
}
};
class PrecondPD : public LinearOperator<Real> {
private:
const Teuchos::RCP<Objective<Real> > obj_;
const Teuchos::RCP<BoundConstraint<Real> > bnd_;
const Teuchos::RCP<Vector<Real> > x_;
const Teuchos::RCP<Vector<Real> > xlam_;
Teuchos::RCP<Vector<Real> > v_;
Real eps_;
const Teuchos::RCP<Secant<Real> > secant_;
bool useSecant_;
public:
PrecondPD(const Teuchos::RCP<Objective<Real> > &obj,
const Teuchos::RCP<BoundConstraint<Real> > &bnd,
const Teuchos::RCP<Vector<Real> > &x,
const Teuchos::RCP<Vector<Real> > &xlam,
const Real eps = 0,
const Teuchos::RCP<Secant<Real> > &secant = Teuchos::null,
const bool useSecant = false )
: obj_(obj), bnd_(bnd), x_(x), xlam_(xlam),
eps_(eps), secant_(secant), useSecant_(useSecant) {
v_ = x_->dual().clone();
if ( !useSecant || secant == Teuchos::null ) {
useSecant_ = false;
}
}
void apply( Vector<Real> &Hv, const Vector<Real> &v, Real &tol ) const {
Hv.set(v.dual());
}
void applyInverse( Vector<Real> &Hv, const Vector<Real> &v, Real &tol ) const {
v_->set(v);
bnd_->pruneActive(*v_,*xlam_,eps_);
if ( useSecant_ ) {
secant_->applyH(Hv,*v_);
}
else {
obj_->precond(Hv,*v_,*x_,tol);
}
bnd_->pruneActive(Hv,*xlam_,eps_);
}
};
/** \brief Compute the gradient-based criticality measure.
The criticality measure is
\f$\|x_k - P_{[a,b]}(x_k-\nabla f(x_k))\|_{\mathcal{X}}\f$.
Here, \f$P_{[a,b]}\f$ denotes the projection onto the
bound constraints.
@param[in] x is the current iteration
@param[in] obj is the objective function
@param[in] con are the bound constraints
@param[in] tol is a tolerance for inexact evaluations of the objective function
*/
Real computeCriticalityMeasure(Vector<Real> &x, Objective<Real> &obj, BoundConstraint<Real> &con, Real tol) {
Real one(1);
Teuchos::RCP<StepState<Real> > step_state = Step<Real>::getState();
obj.gradient(*(step_state->gradientVec),x,tol);
xtmp_->set(x);
xtmp_->axpy(-one,(step_state->gradientVec)->dual());
con.project(*xtmp_);
xtmp_->axpy(-one,x);
return xtmp_->norm();
}
public:
/** \brief Constructor.
@param[in] parlist is a parameter list containing relevent algorithmic information
@param[in] useSecant is a bool which determines whether or not the algorithm uses
a secant approximation of the Hessian
*/
PrimalDualActiveSetStep( Teuchos::ParameterList &parlist )
: Step<Real>::Step(), krylov_(Teuchos::null),
iterCR_(0), flagCR_(0), itol_(0),
maxit_(0), iter_(0), flag_(0), stol_(0), gtol_(0), scale_(0),
neps_(-ROL_EPSILON<Real>()), feasible_(false),
lambda_(Teuchos::null), xlam_(Teuchos::null), x0_(Teuchos::null),
xbnd_(Teuchos::null), As_(Teuchos::null), xtmp_(Teuchos::null),
res_(Teuchos::null), Ag_(Teuchos::null), rtmp_(Teuchos::null),
gtmp_(Teuchos::null),
esec_(SECANT_LBFGS), secant_(Teuchos::null), useSecantPrecond_(false),
useSecantHessVec_(false) {
Real one(1), oem6(1.e-6), oem8(1.e-8);
// Algorithmic parameters
maxit_ = parlist.sublist("Step").sublist("Primal Dual Active Set").get("Iteration Limit",10);
stol_ = parlist.sublist("Step").sublist("Primal Dual Active Set").get("Relative Step Tolerance",oem8);
gtol_ = parlist.sublist("Step").sublist("Primal Dual Active Set").get("Relative Gradient Tolerance",oem6);
scale_ = parlist.sublist("Step").sublist("Primal Dual Active Set").get("Dual Scaling", one);
// Build secant object
esec_ = StringToESecant(parlist.sublist("General").sublist("Secant").get("Type","Limited-Memory BFGS"));
useSecantHessVec_ = parlist.sublist("General").sublist("Secant").get("Use as Hessian", false);
useSecantPrecond_ = parlist.sublist("General").sublist("Secant").get("Use as Preconditioner", false);
if ( useSecantHessVec_ || useSecantPrecond_ ) {
secant_ = SecantFactory<Real>(parlist);
}
// Build Krylov object
krylov_ = KrylovFactory<Real>(parlist);
}
/** \brief Initialize step.
This includes projecting the initial guess onto the constraints,
computing the initial objective function value and gradient,
and initializing the dual variables.
@param[in,out] x is the initial guess
@param[in] obj is the objective function
@param[in] con are the bound constraints
@param[in] algo_state is the current state of the algorithm
*/
void initialize( Vector<Real> &x, const Vector<Real> &s, const Vector<Real> &g,
Objective<Real> &obj, BoundConstraint<Real> &con,
AlgorithmState<Real> &algo_state ) {
Teuchos::RCP<StepState<Real> > step_state = Step<Real>::getState();
Real zero(0), one(1);
// Initialize state descent direction and gradient storage
step_state->descentVec = s.clone();
step_state->gradientVec = g.clone();
step_state->searchSize = zero;
// Initialize additional storage
xlam_ = x.clone();
x0_ = x.clone();
xbnd_ = x.clone();
As_ = s.clone();
xtmp_ = x.clone();
res_ = g.clone();
Ag_ = g.clone();
rtmp_ = g.clone();
gtmp_ = g.clone();
// Project x onto constraint set
con.project(x);
// Update objective function, get value, and get gradient
Real tol = std::sqrt(ROL_EPSILON<Real>());
obj.update(x,true,algo_state.iter);
algo_state.value = obj.value(x,tol);
algo_state.nfval++;
algo_state.gnorm = computeCriticalityMeasure(x,obj,con,tol);
algo_state.ngrad++;
// Initialize dual variable
lambda_ = s.clone();
lambda_->set((step_state->gradientVec)->dual());
lambda_->scale(-one);
//con.setVectorToLowerBound(*lambda_);
}
/** \brief Compute step.
Given \f$x_k\f$, this function first builds the
primal-dual active sets
\f$\mathcal{A}_k^-\f$ and \f$\mathcal{A}_k^+\f$.
Next, it uses CR to compute the inactive
components of the step by solving
\f[
\nabla^2 f(x_k)_{\mathcal{I}_k,\mathcal{I}_k}(s_k)_{\mathcal{I}_k} =
-\nabla f(x_k)_{\mathcal{I}_k}
-\nabla^2 f(x_k)_{\mathcal{I}_k,\mathcal{A}_k} (s_k)_{\mathcal{A}_k}.
\f]
Finally, it updates the active components of the
dual variables as
\f[
\lambda_{k+1} = -\nabla f(x_k)_{\mathcal{A}_k}
-(\nabla^2 f(x_k) s_k)_{\mathcal{A}_k}.
\f]
@param[out] s is the step computed via PDAS
@param[in] x is the current iterate
@param[in] obj is the objective function
@param[in] con are the bound constraints
@param[in] algo_state is the current state of the algorithm
*/
void compute( Vector<Real> &s, const Vector<Real> &x, Objective<Real> &obj, BoundConstraint<Real> &con,
AlgorithmState<Real> &algo_state ) {
Teuchos::RCP<StepState<Real> > step_state = Step<Real>::getState();
Real zero(0), one(1);
s.zero();
x0_->set(x);
res_->set(*(step_state->gradientVec));
for ( iter_ = 0; iter_ < maxit_; iter_++ ) {
/********************************************************************/
// MODIFY ITERATE VECTOR TO CHECK ACTIVE SET
/********************************************************************/
xlam_->set(*x0_); // xlam = x0
xlam_->axpy(scale_,*(lambda_)); // xlam = x0 + c*lambda
/********************************************************************/
// PROJECT x ONTO PRIMAL DUAL FEASIBLE SET
/********************************************************************/
As_->zero(); // As = 0
con.setVectorToUpperBound(*xbnd_); // xbnd = u
xbnd_->axpy(-one,x); // xbnd = u - x
xtmp_->set(*xbnd_); // tmp = u - x
con.pruneUpperActive(*xtmp_,*xlam_,neps_); // tmp = I(u - x)
xbnd_->axpy(-one,*xtmp_); // xbnd = A(u - x)
As_->plus(*xbnd_); // As += A(u - x)
con.setVectorToLowerBound(*xbnd_); // xbnd = l
xbnd_->axpy(-one,x); // xbnd = l - x
xtmp_->set(*xbnd_); // tmp = l - x
con.pruneLowerActive(*xtmp_,*xlam_,neps_); // tmp = I(l - x)
xbnd_->axpy(-one,*xtmp_); // xbnd = A(l - x)
As_->plus(*xbnd_); // As += A(l - x)
/********************************************************************/
// APPLY HESSIAN TO ACTIVE COMPONENTS OF s AND REMOVE INACTIVE
/********************************************************************/
itol_ = std::sqrt(ROL_EPSILON<Real>());
if ( useSecantHessVec_ && secant_ != Teuchos::null ) { // IHAs = H*As
secant_->applyB(*gtmp_,*As_);
}
else {
obj.hessVec(*gtmp_,*As_,x,itol_);
}
con.pruneActive(*gtmp_,*xlam_,neps_); // IHAs = I(H*As)
/********************************************************************/
// SEPARATE ACTIVE AND INACTIVE COMPONENTS OF THE GRADIENT
/********************************************************************/
rtmp_->set(*(step_state->gradientVec)); // Inactive components
con.pruneActive(*rtmp_,*xlam_,neps_);
Ag_->set(*(step_state->gradientVec)); // Active components
Ag_->axpy(-one,*rtmp_);
/********************************************************************/
// SOLVE REDUCED NEWTON SYSTEM
/********************************************************************/
rtmp_->plus(*gtmp_);
rtmp_->scale(-one); // rhs = -Ig - I(H*As)
s.zero();
if ( rtmp_->norm() > zero ) {
// Initialize Hessian and preconditioner
Teuchos::RCP<Objective<Real> > obj_ptr = Teuchos::rcpFromRef(obj);
Teuchos::RCP<BoundConstraint<Real> > con_ptr = Teuchos::rcpFromRef(con);
Teuchos::RCP<LinearOperator<Real> > hessian
= Teuchos::rcp(new HessianPD(obj_ptr,con_ptr,
algo_state.iterateVec,xlam_,neps_,secant_,useSecantHessVec_));
Teuchos::RCP<LinearOperator<Real> > precond
= Teuchos::rcp(new PrecondPD(obj_ptr,con_ptr,
algo_state.iterateVec,xlam_,neps_,secant_,useSecantPrecond_));
//solve(s,*rtmp_,*xlam_,x,obj,con); // Call conjugate residuals
krylov_->run(s,*hessian,*rtmp_,*precond,iterCR_,flagCR_);
con.pruneActive(s,*xlam_,neps_); // s <- Is
}
s.plus(*As_); // s = Is + As
/********************************************************************/
// UPDATE MULTIPLIER
/********************************************************************/
if ( useSecantHessVec_ && secant_ != Teuchos::null ) {
secant_->applyB(*rtmp_,s);
}
else {
obj.hessVec(*rtmp_,s,x,itol_);
}
gtmp_->set(*rtmp_);
con.pruneActive(*gtmp_,*xlam_,neps_);
lambda_->set(*rtmp_);
lambda_->axpy(-one,*gtmp_);
lambda_->plus(*Ag_);
lambda_->scale(-one);
/********************************************************************/
// UPDATE STEP
/********************************************************************/
x0_->set(x);
x0_->plus(s);
res_->set(*(step_state->gradientVec));
res_->plus(*rtmp_);
// Compute criticality measure
xtmp_->set(*x0_);
xtmp_->axpy(-one,res_->dual());
con.project(*xtmp_);
xtmp_->axpy(-one,*x0_);
// std::cout << s.norm() << " "
// << tmp->norm() << " "
// << res_->norm() << " "
// << lambda_->norm() << " "
// << flagCR_ << " "
// << iterCR_ << "\n";
if ( xtmp_->norm() < gtol_*algo_state.gnorm ) {
flag_ = 0;
break;
}
if ( s.norm() < stol_*x.norm() ) {
flag_ = 2;
break;
}
}
if ( iter_ == maxit_ ) {
flag_ = 1;
}
else {
iter_++;
}
}
/** \brief Update step, if successful.
This function returns \f$x_{k+1} = x_k + s_k\f$.
It also updates secant information if being used.
@param[in] x is the new iterate
@param[out] s is the step computed via PDAS
@param[in] obj is the objective function
@param[in] con are the bound constraints
@param[in] algo_state is the current state of the algorithm
*/
void update( Vector<Real> &x, const Vector<Real> &s, Objective<Real> &obj, BoundConstraint<Real> &con,
AlgorithmState<Real> &algo_state ) {
Teuchos::RCP<StepState<Real> > step_state = Step<Real>::getState();
x.plus(s);
feasible_ = con.isFeasible(x);
algo_state.snorm = s.norm();
algo_state.iter++;
Real tol = std::sqrt(ROL_EPSILON<Real>());
obj.update(x,true,algo_state.iter);
algo_state.value = obj.value(x,tol);
algo_state.nfval++;
if ( secant_ != Teuchos::null ) {
gtmp_->set(*(step_state->gradientVec));
}
algo_state.gnorm = computeCriticalityMeasure(x,obj,con,tol);
algo_state.ngrad++;
if ( secant_ != Teuchos::null ) {
secant_->updateStorage(x,*(step_state->gradientVec),*gtmp_,s,algo_state.snorm,algo_state.iter+1);
}
(algo_state.iterateVec)->set(x);
}
/** \brief Print iterate header.
This function produces a string containing
header information.
*/
std::string printHeader( void ) const {
std::stringstream hist;
hist << " ";
hist << std::setw(6) << std::left << "iter";
hist << std::setw(15) << std::left << "value";
hist << std::setw(15) << std::left << "gnorm";
hist << std::setw(15) << std::left << "snorm";
hist << std::setw(10) << std::left << "#fval";
hist << std::setw(10) << std::left << "#grad";
if ( maxit_ > 1 ) {
hist << std::setw(10) << std::left << "iterPDAS";
hist << std::setw(10) << std::left << "flagPDAS";
}
else {
hist << std::setw(10) << std::left << "iterCR";
hist << std::setw(10) << std::left << "flagCR";
}
hist << std::setw(10) << std::left << "feasible";
hist << "\n";
return hist.str();
}
/** \brief Print step name.
This function produces a string containing
the algorithmic step information.
*/
std::string printName( void ) const {
std::stringstream hist;
hist << "\nPrimal Dual Active Set Newton's Method\n";
return hist.str();
}
/** \brief Print iterate status.
This function prints the iteration status.
@param[in] algo_state is the current state of the algorithm
@param[in] printHeader if set to true will print the header at each iteration
*/
virtual std::string print( AlgorithmState<Real> &algo_state, bool print_header = false ) const {
std::stringstream hist;
hist << std::scientific << std::setprecision(6);
if ( algo_state.iter == 0 ) {
hist << printName();
}
if ( print_header ) {
hist << printHeader();
}
if ( algo_state.iter == 0 ) {
hist << " ";
hist << std::setw(6) << std::left << algo_state.iter;
hist << std::setw(15) << std::left << algo_state.value;
hist << std::setw(15) << std::left << algo_state.gnorm;
hist << "\n";
}
else {
hist << " ";
hist << std::setw(6) << std::left << algo_state.iter;
hist << std::setw(15) << std::left << algo_state.value;
hist << std::setw(15) << std::left << algo_state.gnorm;
hist << std::setw(15) << std::left << algo_state.snorm;
hist << std::setw(10) << std::left << algo_state.nfval;
hist << std::setw(10) << std::left << algo_state.ngrad;
if ( maxit_ > 1 ) {
hist << std::setw(10) << std::left << iter_;
hist << std::setw(10) << std::left << flag_;
}
else {
hist << std::setw(10) << std::left << iterCR_;
hist << std::setw(10) << std::left << flagCR_;
}
if ( feasible_ ) {
hist << std::setw(10) << std::left << "YES";
}
else {
hist << std::setw(10) << std::left << "NO";
}
hist << "\n";
}
return hist.str();
}
}; // class PrimalDualActiveSetStep
} // namespace ROL
#endif
// void solve(Vector<Real> &sol, const Vector<Real> &rhs, const Vector<Real> &xlam, const Vector<Real> &x,
// Objective<Real> &obj, BoundConstraint<Real> &con) {
// Real rnorm = rhs.norm();
// Real rtol = std::min(tol1_,tol2_*rnorm);
// itol_ = std::sqrt(ROL_EPSILON<Real>());
// sol.zero();
//
// Teuchos::RCP<Vector<Real> > res = rhs.clone();
// res->set(rhs);
//
// Teuchos::RCP<Vector<Real> > v = x.clone();
// con.pruneActive(*res,xlam,neps_);
// obj.precond(*v,*res,x,itol_);
// con.pruneActive(*v,xlam,neps_);
//
// Teuchos::RCP<Vector<Real> > p = x.clone();
// p->set(*v);
//
// Teuchos::RCP<Vector<Real> > Hp = x.clone();
//
// iterCR_ = 0;
// flagCR_ = 0;
//
// Real kappa = 0.0, beta = 0.0, alpha = 0.0, tmp = 0.0, rv = v->dot(*res);
//
// for (iterCR_ = 0; iterCR_ < maxitCR_; iterCR_++) {
// if ( false ) {
// itol_ = rtol/(maxitCR_*rnorm);
// }
// con.pruneActive(*p,xlam,neps_);
// if ( secant_ == Teuchos::null ) {
// obj.hessVec(*Hp, *p, x, itol_);
// }
// else {
// secant_->applyB( *Hp, *p, x );
// }
// con.pruneActive(*Hp,xlam,neps_);
//
// kappa = p->dot(*Hp);
// if ( kappa <= 0.0 ) { flagCR_ = 2; break; }
// alpha = rv/kappa;
// sol.axpy(alpha,*p);
//
// res->axpy(-alpha,*Hp);
// rnorm = res->norm();
// if ( rnorm < rtol ) { break; }
//
// con.pruneActive(*res,xlam,neps_);
// obj.precond(*v,*res,x,itol_);
// con.pruneActive(*v,xlam,neps_);
// tmp = rv;
// rv = v->dot(*res);
// beta = rv/tmp;
//
// p->scale(beta);
// p->axpy(1.0,*v);
// }
// if ( iterCR_ == maxitCR_ ) {
// flagCR_ = 1;
// }
// else {
// iterCR_++;
// }
// }
// /** \brief Apply the inactive components of the Hessian operator.
//
// I.e., the components corresponding to \f$\mathcal{I}_k\f$.
//
// @param[out] hv is the result of applying the Hessian at @b x to
// @b v
// @param[in] v is the direction in which we apply the Hessian
// @param[in] x is the current iteration vector \f$x_k\f$
// @param[in] xlam is the vector \f$x_k + c\lambda_k\f$
// @param[in] obj is the objective function
// @param[in] con are the bound constraints
// */
// void applyInactiveHessian(Vector<Real> &hv, const Vector<Real> &v, const Vector<Real> &x,
// const Vector<Real> &xlam, Objective<Real> &obj, BoundConstraint<Real> &con) {
// Teuchos::RCP<Vector<Real> > tmp = v.clone();
// tmp->set(v);
// con.pruneActive(*tmp,xlam,neps_);
// if ( secant_ == Teuchos::null ) {
// obj.hessVec(hv,*tmp,x,itol_);
// }
// else {
// secant_->applyB(hv,*tmp,x);
// }
// con.pruneActive(hv,xlam,neps_);
// }
//
// /** \brief Apply the inactive components of the preconditioner operator.
//
// I.e., the components corresponding to \f$\mathcal{I}_k\f$.
//
// @param[out] hv is the result of applying the preconditioner at @b x to
// @b v
// @param[in] v is the direction in which we apply the preconditioner
// @param[in] x is the current iteration vector \f$x_k\f$
// @param[in] xlam is the vector \f$x_k + c\lambda_k\f$
// @param[in] obj is the objective function
// @param[in] con are the bound constraints
// */
// void applyInactivePrecond(Vector<Real> &pv, const Vector<Real> &v, const Vector<Real> &x,
// const Vector<Real> &xlam, Objective<Real> &obj, BoundConstraint<Real> &con) {
// Teuchos::RCP<Vector<Real> > tmp = v.clone();
// tmp->set(v);
// con.pruneActive(*tmp,xlam,neps_);
// obj.precond(pv,*tmp,x,itol_);
// con.pruneActive(pv,xlam,neps_);
// }
//
// /** \brief Solve the inactive part of the PDAS optimality system.
//
// The inactive PDAS optimality system is
// \f[
// \nabla^2 f(x_k)_{\mathcal{I}_k,\mathcal{I}_k}s =
// -\nabla f(x_k)_{\mathcal{I}_k}
// -\nabla^2 f(x_k)_{\mathcal{I}_k,\mathcal{A}_k} (s_k)_{\mathcal{A}_k}.
// \f]
// Since the inactive part of the Hessian may not be positive definite, we solve
// using CR.
//
// @param[out] sol is the vector containing the solution
// @param[in] rhs is the right-hand side vector
// @param[in] xlam is the vector \f$x_k + c\lambda_k\f$
// @param[in] x is the current iteration vector \f$x_k\f$
// @param[in] obj is the objective function
// @param[in] con are the bound constraints
// */
// // Solve the inactive part of the optimality system using conjugate residuals
// void solve(Vector<Real> &sol, const Vector<Real> &rhs, const Vector<Real> &xlam, const Vector<Real> &x,
// Objective<Real> &obj, BoundConstraint<Real> &con) {
// // Initialize Residual
// Teuchos::RCP<Vector<Real> > res = rhs.clone();
// res->set(rhs);
// Real rnorm = res->norm();
// Real rtol = std::min(tol1_,tol2_*rnorm);
// if ( false ) { itol_ = rtol/(maxitCR_*rnorm); }
// sol.zero();
//
// // Apply preconditioner to residual r = Mres
// Teuchos::RCP<Vector<Real> > r = x.clone();
// applyInactivePrecond(*r,*res,x,xlam,obj,con);
//
// // Initialize direction p = v
// Teuchos::RCP<Vector<Real> > p = x.clone();
// p->set(*r);
//
// // Apply Hessian to v
// Teuchos::RCP<Vector<Real> > Hr = x.clone();
// applyInactiveHessian(*Hr,*r,x,xlam,obj,con);
//
// // Apply Hessian to p
// Teuchos::RCP<Vector<Real> > Hp = x.clone();
// Teuchos::RCP<Vector<Real> > MHp = x.clone();
// Hp->set(*Hr);
//
// iterCR_ = 0;
// flagCR_ = 0;
//
// Real kappa = 0.0, beta = 0.0, alpha = 0.0, tmp = 0.0, rHr = Hr->dot(*r);
//
// for (iterCR_ = 0; iterCR_ < maxitCR_; iterCR_++) {
// // Precondition Hp
// applyInactivePrecond(*MHp,*Hp,x,xlam,obj,con);
//
// kappa = Hp->dot(*MHp); // p' H M H p
// alpha = rHr/kappa; // r' M H M r
// sol.axpy(alpha,*p); // update step
// res->axpy(-alpha,*Hp); // residual
// r->axpy(-alpha,*MHp); // preconditioned residual
//
// // recompute rnorm and decide whether or not to exit
// rnorm = res->norm();
// if ( rnorm < rtol ) { break; }
//
// // Apply Hessian to v
// itol_ = rtol/(maxitCR_*rnorm);
// applyInactiveHessian(*Hr,*r,x,xlam,obj,con);
//
// tmp = rHr;
// rHr = Hr->dot(*r);
// beta = rHr/tmp;
// p->scale(beta);
// p->axpy(1.0,*r);
// Hp->scale(beta);
// Hp->axpy(1.0,*Hr);
// }
// if ( iterCR_ == maxitCR_ ) {
// flagCR_ = 1;
// }
// else {
// iterCR_++;
// }
// }
|