ROL
example_05.hpp
Go to the documentation of this file.
1// @HEADER
2// *****************************************************************************
3// Rapid Optimization Library (ROL) Package
4//
5// Copyright 2014 NTESS and the ROL contributors.
6// SPDX-License-Identifier: BSD-3-Clause
7// *****************************************************************************
8// @HEADER
9
10#include "ROL_Stream.hpp"
11#include "Teuchos_LAPACK.hpp"
12#include "Teuchos_GlobalMPISession.hpp"
13#include "Teuchos_Comm.hpp"
14#include "Teuchos_DefaultComm.hpp"
15#include "Teuchos_CommHelpers.hpp"
16
17#include "ROL_ParameterList.hpp"
18// ROL_Types contains predefined constants and objects
19#include "ROL_Types.hpp"
20// ROL vectors
21#include "ROL_StdVector.hpp"
22// ROL objective functions and constraints
23#include "ROL_Objective_SimOpt.hpp"
24#include "ROL_Constraint_SimOpt.hpp"
25#include "ROL_Reduced_Objective_SimOpt.hpp"
26#include "ROL_OptimizationSolver.hpp"
27// ROL sample generators
28#include "ROL_MonteCarloGenerator.hpp"
29#include "ROL_StdTeuchosBatchManager.hpp"
30
31template<class Real>
32class Constraint_BurgersControl : public ROL::Constraint_SimOpt<Real> {
33private:
34 int nx_;
35 Real dx_;
36
37 Real compute_norm(const std::vector<Real> &r) {
38 return std::sqrt(dot(r,r));
39 }
40
41 Real dot(const std::vector<Real> &x, const std::vector<Real> &y) {
42 Real ip = 0.0;
43 Real c = (((int)x.size()==nx_) ? 4.0 : 2.0);
44 for (unsigned i=0; i<x.size(); i++) {
45 if ( i == 0 ) {
46 ip += dx_/6.0*(c*x[i] + x[i+1])*y[i];
47 }
48 else if ( i == x.size()-1 ) {
49 ip += dx_/6.0*(x[i-1] + c*x[i])*y[i];
50 }
51 else {
52 ip += dx_/6.0*(x[i-1] + 4.0*x[i] + x[i+1])*y[i];
53 }
54 }
55 return ip;
56 }
57
58 using ROL::Constraint_SimOpt<Real>::update;
59
60 void update(std::vector<Real> &u, const std::vector<Real> &s, const Real alpha=1.0) {
61 for (unsigned i=0; i<u.size(); i++) {
62 u[i] += alpha*s[i];
63 }
64 }
65
66 void scale(std::vector<Real> &u, const Real alpha=0.0) {
67 for (unsigned i=0; i<u.size(); i++) {
68 u[i] *= alpha;
69 }
70 }
71
72 void compute_residual(std::vector<Real> &r, const std::vector<Real> &u,
73 const std::vector<Real> &z) {
74 r.clear(); r.resize(nx_,0.0);
75 const std::vector<Real> param =
76 ROL::Constraint_SimOpt<Real>::getParameter();
77 Real nu = std::pow(10.0,param[0]-2.0);
78 Real f = param[1]/100.0;
79 Real u0 = 1.0+param[2]/1000.0;
80 Real u1 = param[3]/1000.0;
81 for (int i=0; i<nx_; i++) {
82 // Contribution from stiffness term
83 if ( i==0 ) {
84 r[i] = nu/dx_*(2.0*u[i]-u[i+1]);
85 }
86 else if (i==nx_-1) {
87 r[i] = nu/dx_*(2.0*u[i]-u[i-1]);
88 }
89 else {
90 r[i] = nu/dx_*(2.0*u[i]-u[i-1]-u[i+1]);
91 }
92 // Contribution from nonlinear term
93 if (i<nx_-1){
94 r[i] += u[i+1]*(u[i]+u[i+1])/6.0;
95 }
96 if (i>0) {
97 r[i] -= u[i-1]*(u[i-1]+u[i])/6.0;
98 }
99 // Contribution from control
100 r[i] -= dx_/6.0*(z[i]+4.0*z[i+1]+z[i+2]);
101 // Contribution from right-hand side
102 r[i] -= dx_*f;
103 }
104 // Contribution from Dirichlet boundary terms
105 r[ 0] -= u0*u[ 0]/6.0 + u0*u0/6.0 + nu*u0/dx_;
106 r[nx_-1] += u1*u[nx_-1]/6.0 + u1*u1/6.0 - nu*u1/dx_;
107 }
108
109 void compute_pde_jacobian(std::vector<Real> &dl, std::vector<Real> &d, std::vector<Real> &du,
110 const std::vector<Real> &u) {
111 const std::vector<Real> param =
112 ROL::Constraint_SimOpt<Real>::getParameter();
113 Real nu = std::pow(10.0,param[0]-2.0);
114 Real u0 = 1.0+param[2]/1000.0;
115 Real u1 = param[3]/1000.0;
116 // Get Diagonal and Off-Diagonal Entries of linear PDE Jacobian
117 d.clear(); d.resize(nx_,nu*2.0/dx_);
118 dl.clear(); dl.resize(nx_-1,-nu/dx_);
119 du.clear(); du.resize(nx_-1,-nu/dx_);
120 // Contribution from nonlinearity
121 for (int i=0; i<nx_; i++) {
122 if (i<nx_-1) {
123 dl[i] += (-2.0*u[i]-u[i+1])/6.0;
124 d[i] += u[i+1]/6.0;
125 }
126 if (i>0) {
127 d[i] += -u[i-1]/6.0;
128 du[i-1] += (u[i-1]+2.0*u[i])/6.0;
129 }
130 }
131 // Contribution from Dirichlet boundary conditions
132 d[ 0] -= u0/6.0;
133 d[nx_-1] += u1/6.0;
134 }
135
136 void linear_solve(std::vector<Real> &u, std::vector<Real> &dl, std::vector<Real> &d, std::vector<Real> &du,
137 const std::vector<Real> &r, const bool transpose = false) {
138 u.assign(r.begin(),r.end());
139 // Perform LDL factorization
140 Teuchos::LAPACK<int,Real> lp;
141 std::vector<Real> du2(nx_-2,0.0);
142 std::vector<int> ipiv(nx_,0);
143 int info;
144 int ldb = nx_;
145 int nhrs = 1;
146 lp.GTTRF(nx_,&dl[0],&d[0],&du[0],&du2[0],&ipiv[0],&info);
147 char trans = 'N';
148 if ( transpose ) {
149 trans = 'T';
150 }
151 lp.GTTRS(trans,nx_,nhrs,&dl[0],&d[0],&du[0],&du2[0],&ipiv[0],&u[0],ldb,&info);
152 }
153
154public:
155
156 Constraint_BurgersControl(int nx = 128) : nx_(nx), dx_(1.0/((Real)nx+1.0)) {}
157
158 void value(ROL::Vector<Real> &c, const ROL::Vector<Real> &u,
159 const ROL::Vector<Real> &z, Real &tol) {
160 ROL::Ptr<std::vector<Real> > cp =
161 dynamic_cast<ROL::StdVector<Real>&>(c).getVector();
162 ROL::Ptr<const std::vector<Real> > up =
163 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
164 ROL::Ptr<const std::vector<Real> > zp =
165 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
166 compute_residual(*cp,*up,*zp);
167 }
168
169 void solve(ROL::Vector<Real> &c, ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol) {
170 ROL::Ptr<std::vector<Real> > up =
171 ROL::constPtrCast<std::vector<Real> >((dynamic_cast<ROL::StdVector<Real>&>(u)).getVector());
172 up->assign(up->size(),static_cast<Real>(1));
173 ROL::Constraint_SimOpt<Real>::solve(c,u,z,tol);
174 }
175
176 void applyJacobian_1(ROL::Vector<Real> &jv, const ROL::Vector<Real> &v, const ROL::Vector<Real> &u,
177 const ROL::Vector<Real> &z, Real &tol) {
178 ROL::Ptr<std::vector<Real> > jvp =
179 dynamic_cast<ROL::StdVector<Real>&>(jv).getVector();
180 ROL::Ptr<const std::vector<Real> > vp =
181 dynamic_cast<const ROL::StdVector<Real>&>(v).getVector();
182 ROL::Ptr<const std::vector<Real> > up =
183 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
184 ROL::Ptr<const std::vector<Real> > zp =
185 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
186 const std::vector<Real> param =
187 ROL::Constraint_SimOpt<Real>::getParameter();
188 Real nu = std::pow(10.0,param[0]-2.0);
189 Real u0 = 1.0+param[2]/1000.0;
190 Real u1 = param[3]/1000.0;
191 // Fill jvp
192 for (int i = 0; i < nx_; i++) {
193 (*jvp)[i] = nu/dx_*2.0*(*vp)[i];
194 if ( i > 0 ) {
195 (*jvp)[i] += -nu/dx_*(*vp)[i-1]
196 -(*up)[i-1]/6.0*(*vp)[i]
197 -((*up)[i]+2.0*(*up)[i-1])/6.0*(*vp)[i-1];
198 }
199 if ( i < nx_-1 ) {
200 (*jvp)[i] += -nu/dx_*(*vp)[i+1]
201 +(*up)[i+1]/6.0*(*vp)[i]
202 +((*up)[i]+2.0*(*up)[i+1])/6.0*(*vp)[i+1];
203 }
204 }
205 (*jvp)[ 0] -= u0/6.0*(*vp)[0];
206 (*jvp)[nx_-1] += u1/6.0*(*vp)[nx_-1];
207 }
208
209 void applyJacobian_2(ROL::Vector<Real> &jv, const ROL::Vector<Real> &v, const ROL::Vector<Real> &u,
210 const ROL::Vector<Real> &z, Real &tol) {
211 ROL::Ptr<std::vector<Real> > jvp =
212 dynamic_cast<ROL::StdVector<Real>&>(jv).getVector();
213 ROL::Ptr<const std::vector<Real>> vp =
214 dynamic_cast<const ROL::StdVector<Real>&>(v).getVector();
215 ROL::Ptr<const std::vector<Real>> up =
216 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
217 ROL::Ptr<const std::vector<Real>> zp =
218 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
219 for (int i=0; i<nx_; i++) {
220 // Contribution from control
221 (*jvp)[i] = -dx_/6.0*((*vp)[i]+4.0*(*vp)[i+1]+(*vp)[i+2]);
222 }
223 }
224
225 void applyInverseJacobian_1(ROL::Vector<Real> &ijv, const ROL::Vector<Real> &v, const ROL::Vector<Real> &u,
226 const ROL::Vector<Real> &z, Real &tol) {
227 ROL::Ptr<std::vector<Real> > ijvp =
228 dynamic_cast<ROL::StdVector<Real>&>(ijv).getVector();
229 ROL::Ptr<const std::vector<Real> > vp =
230 dynamic_cast<const ROL::StdVector<Real>&>(v).getVector();
231 ROL::Ptr<const std::vector<Real> > up =
232 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
233 ROL::Ptr<const std::vector<Real> > zp =
234 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
235 // Get PDE Jacobian
236 std::vector<Real> d(nx_,0.0);
237 std::vector<Real> dl(nx_-1,0.0);
238 std::vector<Real> du(nx_-1,0.0);
239 compute_pde_jacobian(dl,d,du,*up);
240 // Solve solve state sensitivity system at current time step
241 linear_solve(*ijvp,dl,d,du,*vp);
242 }
243
244 void applyAdjointJacobian_1(ROL::Vector<Real> &ajv, const ROL::Vector<Real> &v, const ROL::Vector<Real> &u,
245 const ROL::Vector<Real> &z, Real &tol) {
246 ROL::Ptr<std::vector<Real> > jvp =
247 dynamic_cast<ROL::StdVector<Real>&>(ajv).getVector();
248 ROL::Ptr<const std::vector<Real> > vp =
249 dynamic_cast<const ROL::StdVector<Real>&>(v).getVector();
250 ROL::Ptr<const std::vector<Real> > up =
251 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
252 ROL::Ptr<const std::vector<Real> > zp =
253 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
254 const std::vector<Real> param =
255 ROL::Constraint_SimOpt<Real>::getParameter();
256 Real nu = std::pow(10.0,param[0]-2.0);
257 Real u0 = 1.0+param[2]/1000.0;
258 Real u1 = param[3]/1000.0;
259 // Fill jvp
260 for (int i = 0; i < nx_; i++) {
261 (*jvp)[i] = nu/dx_*2.0*(*vp)[i];
262 if ( i > 0 ) {
263 (*jvp)[i] += -nu/dx_*(*vp)[i-1]
264 -(*up)[i-1]/6.0*(*vp)[i]
265 +((*up)[i-1]+2.0*(*up)[i])/6.0*(*vp)[i-1];
266 }
267 if ( i < nx_-1 ) {
268 (*jvp)[i] += -nu/dx_*(*vp)[i+1]
269 +(*up)[i+1]/6.0*(*vp)[i]
270 -((*up)[i+1]+2.0*(*up)[i])/6.0*(*vp)[i+1];
271 }
272 }
273 (*jvp)[ 0] -= u0/6.0*(*vp)[0];
274 (*jvp)[nx_-1] += u1/6.0*(*vp)[nx_-1];
275 }
276
277 void applyAdjointJacobian_2(ROL::Vector<Real> &jv, const ROL::Vector<Real> &v, const ROL::Vector<Real> &u,
278 const ROL::Vector<Real> &z, Real &tol) {
279 ROL::Ptr<std::vector<Real> > jvp =
280 dynamic_cast<ROL::StdVector<Real>&>(jv).getVector();
281 ROL::Ptr<const std::vector<Real> > vp =
282 dynamic_cast<const ROL::StdVector<Real>&>(v).getVector();
283 ROL::Ptr<const std::vector<Real> > up =
284 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
285 ROL::Ptr<const std::vector<Real> > zp =
286 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
287 for (int i=0; i<nx_+2; i++) {
288 if ( i == 0 ) {
289 (*jvp)[i] = -dx_/6.0*(*vp)[i];
290 }
291 else if ( i == 1 ) {
292 (*jvp)[i] = -dx_/6.0*(4.0*(*vp)[i-1]+(*vp)[i]);
293 }
294 else if ( i == nx_ ) {
295 (*jvp)[i] = -dx_/6.0*(4.0*(*vp)[i-1]+(*vp)[i-2]);
296 }
297 else if ( i == nx_+1 ) {
298 (*jvp)[i] = -dx_/6.0*(*vp)[i-2];
299 }
300 else {
301 (*jvp)[i] = -dx_/6.0*((*vp)[i-2]+4.0*(*vp)[i-1]+(*vp)[i]);
302 }
303 }
304 }
305
306 void applyInverseAdjointJacobian_1(ROL::Vector<Real> &iajv, const ROL::Vector<Real> &v,
307 const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol) {
308 ROL::Ptr<std::vector<Real> > iajvp =
309 dynamic_cast<ROL::StdVector<Real>&>(iajv).getVector();
310 ROL::Ptr<const std::vector<Real> > vp =
311 dynamic_cast<const ROL::StdVector<Real>&>(v).getVector();
312 ROL::Ptr<const std::vector<Real>> up =
313 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
314 // Get PDE Jacobian
315 std::vector<Real> d(nx_,0.0);
316 std::vector<Real> du(nx_-1,0.0);
317 std::vector<Real> dl(nx_-1,0.0);
318 compute_pde_jacobian(dl,d,du,*up);
319 // Solve solve adjoint system at current time step
320 linear_solve(*iajvp,dl,d,du,*vp,true);
321 }
322
323 void applyAdjointHessian_11(ROL::Vector<Real> &ahwv, const ROL::Vector<Real> &w, const ROL::Vector<Real> &v,
324 const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol) {
325 ROL::Ptr<std::vector<Real> > ahwvp =
326 dynamic_cast<ROL::StdVector<Real>&>(ahwv).getVector();
327 ROL::Ptr<const std::vector<Real> > wp =
328 dynamic_cast<const ROL::StdVector<Real>&>(w).getVector();
329 ROL::Ptr<const std::vector<Real> > vp =
330 dynamic_cast<const ROL::StdVector<Real>&>(v).getVector();
331 ROL::Ptr<const std::vector<Real> > up =
332 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
333 ROL::Ptr<const std::vector<Real> > zp =
334 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
335 for (int i=0; i<nx_; i++) {
336 // Contribution from nonlinear term
337 (*ahwvp)[i] = 0.0;
338 if (i<nx_-1){
339 (*ahwvp)[i] += ((*wp)[i]*(*vp)[i+1] - (*wp)[i+1]*(2.0*(*vp)[i]+(*vp)[i+1]))/6.0;
340 }
341 if (i>0) {
342 (*ahwvp)[i] += ((*wp)[i-1]*((*vp)[i-1]+2.0*(*vp)[i]) - (*wp)[i]*(*vp)[i-1])/6.0;
343 }
344 }
345 }
346
347 void applyAdjointHessian_12(ROL::Vector<Real> &ahwv, const ROL::Vector<Real> &w, const ROL::Vector<Real> &v,
348 const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol) {
349 ahwv.zero();
350 }
351 void applyAdjointHessian_21(ROL::Vector<Real> &ahwv, const ROL::Vector<Real> &w, const ROL::Vector<Real> &v,
352 const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol) {
353 ahwv.zero();
354 }
355 void applyAdjointHessian_22(ROL::Vector<Real> &ahwv, const ROL::Vector<Real> &w, const ROL::Vector<Real> &v,
356 const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol) {
357 ahwv.zero();
358 }
359};
360
361template<class Real>
362class Objective_BurgersControl : public ROL::Objective_SimOpt<Real> {
363private:
364 Real alpha_; // Penalty Parameter
365
366 int nx_; // Number of interior nodes
367 Real dx_; // Mesh spacing (i.e. 1/(nx+1))
368
369/***************************************************************/
370/********** BEGIN PRIVATE MEMBER FUNCTION DECLARATION **********/
371/***************************************************************/
372 Real evaluate_target(Real x) {
373 Real val = 0.0;
374 int example = 2;
375 switch (example) {
376 case 1: val = ((x<0.5) ? 1.0 : 0.0); break;
377 case 2: val = 1.0; break;
378 case 3: val = std::abs(std::sin(8.0*M_PI*x)); break;
379 case 4: val = std::exp(-0.5*(x-0.5)*(x-0.5)); break;
380 }
381 return val;
382 }
383
384 Real dot(const std::vector<Real> &x, const std::vector<Real> &y) {
385 Real ip = 0.0;
386 Real c = (((int)x.size()==nx_) ? 4.0 : 2.0);
387 for (unsigned i=0; i<x.size(); i++) {
388 if ( i == 0 ) {
389 ip += dx_/6.0*(c*x[i] + x[i+1])*y[i];
390 }
391 else if ( i == x.size()-1 ) {
392 ip += dx_/6.0*(x[i-1] + c*x[i])*y[i];
393 }
394 else {
395 ip += dx_/6.0*(x[i-1] + 4.0*x[i] + x[i+1])*y[i];
396 }
397 }
398 return ip;
399 }
400
401 void apply_mass(std::vector<Real> &Mu, const std::vector<Real> &u ) {
402 Mu.resize(u.size(),0.0);
403 Real c = (((int)u.size()==nx_) ? 4.0 : 2.0);
404 for (unsigned i=0; i<u.size(); i++) {
405 if ( i == 0 ) {
406 Mu[i] = dx_/6.0*(c*u[i] + u[i+1]);
407 }
408 else if ( i == u.size()-1 ) {
409 Mu[i] = dx_/6.0*(u[i-1] + c*u[i]);
410 }
411 else {
412 Mu[i] = dx_/6.0*(u[i-1] + 4.0*u[i] + u[i+1]);
413 }
414 }
415 }
416/*************************************************************/
417/********** END PRIVATE MEMBER FUNCTION DECLARATION **********/
418/*************************************************************/
419
420public:
421
422 Objective_BurgersControl(Real alpha = 1.e-4, int nx = 128) : alpha_(alpha), nx_(nx) {
423 dx_ = 1.0/((Real)nx+1.0);
424 }
425
426 using ROL::Objective_SimOpt<Real>::value;
427
428 Real value( const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol ) {
429 ROL::Ptr<const std::vector<Real> > up =
430 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
431 ROL::Ptr<const std::vector<Real> > zp =
432 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
433 // COMPUTE RESIDUAL
434 Real res1 = 0.0, res2 = 0.0, res3 = 0.0;
435 Real valu = 0.0, valz = dot(*zp,*zp);
436 for (int i=0; i<nx_; i++) {
437 if ( i == 0 ) {
438 res1 = (*up)[i]-evaluate_target((Real)(i+1)*dx_);
439 res2 = (*up)[i+1]-evaluate_target((Real)(i+2)*dx_);
440 valu += dx_/6.0*(4.0*res1 + res2)*res1;
441 }
442 else if ( i == nx_-1 ) {
443 res1 = (*up)[i-1]-evaluate_target((Real)i*dx_);
444 res2 = (*up)[i]-evaluate_target((Real)(i+1)*dx_);
445 valu += dx_/6.0*(res1 + 4.0*res2)*res2;
446 }
447 else {
448 res1 = (*up)[i-1]-evaluate_target((Real)i*dx_);
449 res2 = (*up)[i]-evaluate_target((Real)(i+1)*dx_);
450 res3 = (*up)[i+1]-evaluate_target((Real)(i+2)*dx_);
451 valu += dx_/6.0*(res1 + 4.0*res2 + res3)*res2;
452 }
453 }
454 return 0.5*(valu + alpha_*valz);
455 }
456
457 void gradient_1( ROL::Vector<Real> &g, const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol ) {
458 // Unwrap g
459 ROL::Ptr<std::vector<Real> > gup =
460 dynamic_cast<ROL::StdVector<Real>&>(g).getVector();
461 // Unwrap x
462 ROL::Ptr<const std::vector<Real> > up =
463 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
464 ROL::Ptr<const std::vector<Real> > zp =
465 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
466 // COMPUTE GRADIENT WRT U
467 std::vector<Real> diff(nx_,0.0);
468 for (int i=0; i<nx_; i++) {
469 diff[i] = ((*up)[i]-evaluate_target((Real)(i+1)*dx_));
470 }
471 apply_mass(*gup,diff);
472 }
473
474 void gradient_2( ROL::Vector<Real> &g, const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol ) {
475 // Unwrap g
476 ROL::Ptr<std::vector<Real> > gzp =
477 dynamic_cast<ROL::StdVector<Real>&>(g).getVector();
478 // Unwrap x
479 ROL::Ptr<const std::vector<Real> > up =
480 dynamic_cast<const ROL::StdVector<Real>&>(u).getVector();
481 ROL::Ptr<const std::vector<Real> > zp =
482 dynamic_cast<const ROL::StdVector<Real>&>(z).getVector();
483 // COMPUTE GRADIENT WRT Z
484 for (int i=0; i<nx_+2; i++) {
485 if (i==0) {
486 (*gzp)[i] = alpha_*dx_/6.0*(2.0*(*zp)[i]+(*zp)[i+1]);
487 }
488 else if (i==nx_+1) {
489 (*gzp)[i] = alpha_*dx_/6.0*(2.0*(*zp)[i]+(*zp)[i-1]);
490 }
491 else {
492 (*gzp)[i] = alpha_*dx_/6.0*((*zp)[i-1]+4.0*(*zp)[i]+(*zp)[i+1]);
493 }
494 }
495 }
496
497 void hessVec_11( ROL::Vector<Real> &hv, const ROL::Vector<Real> &v,
498 const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol ) {
499 ROL::Ptr<std::vector<Real> > hvup =
500 dynamic_cast<ROL::StdVector<Real>&>(hv).getVector();
501 // Unwrap v
502 ROL::Ptr<const std::vector<Real> > vup =
503 dynamic_cast<const ROL::StdVector<Real>&>(v).getVector();
504 // COMPUTE GRADIENT WRT U
505 apply_mass(*hvup,*vup);
506 }
507
508 void hessVec_12( ROL::Vector<Real> &hv, const ROL::Vector<Real> &v,
509 const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol ) {
510 hv.zero();
511 }
512
513 void hessVec_21( ROL::Vector<Real> &hv, const ROL::Vector<Real> &v,
514 const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol ) {
515 hv.zero();
516 }
517
518 void hessVec_22( ROL::Vector<Real> &hv, const ROL::Vector<Real> &v,
519 const ROL::Vector<Real> &u, const ROL::Vector<Real> &z, Real &tol ) {
520 ROL::Ptr<std::vector<Real> > hvzp =
521 dynamic_cast<ROL::StdVector<Real>&>(hv).getVector();
522 // Unwrap v
523 ROL::Ptr<const std::vector<Real> > vzp =
524 dynamic_cast<const ROL::StdVector<Real>&>(v).getVector();
525 // COMPUTE GRADIENT WRT Z
526 for (int i=0; i<nx_+2; i++) {
527 if (i==0) {
528 (*hvzp)[i] = alpha_*dx_/6.0*(2.0*(*vzp)[i]+(*vzp)[i+1]);
529 }
530 else if (i==nx_+1) {
531 (*hvzp)[i] = alpha_*dx_/6.0*(2.0*(*vzp)[i]+(*vzp)[i-1]);
532 }
533 else {
534 (*hvzp)[i] = alpha_*dx_/6.0*((*vzp)[i-1]+4.0*(*vzp)[i]+(*vzp)[i+1]);
535 }
536 }
537 }
538};
ROL_Constraint_SimOpt.hpp
ROL_MonteCarloGenerator.hpp
ROL_Objective_SimOpt.hpp
ROL_OptimizationSolver.hpp
ROL_Reduced_Objective_SimOpt.hpp
ROL_StdVector.hpp
ROL_Stream.hpp
Defines a no-output stream class ROL::NullStream and a function makeStreamPtr which either wraps a re...
ROL_Types.hpp
Contains definitions of custom data types in ROL.
Constraint_BurgersControl
Definition test_04.hpp:670
Constraint_BurgersControl::scale
void scale(std::vector< Real > &u, const Real alpha=0.0)
Definition example_05.hpp:66
Constraint_BurgersControl::applyAdjointHessian_11
void applyAdjointHessian_11(ROL::Vector< Real > &ahwv, const ROL::Vector< Real > &w, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the simulation-space derivative of the adjoint of the constraint simulation-space Jacobian at ...
Definition example_05.hpp:323
Constraint_BurgersControl::applyAdjointHessian_22
void applyAdjointHessian_22(ROL::Vector< Real > &ahwv, const ROL::Vector< Real > &w, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the optimization-space derivative of the adjoint of the constraint optimization-space Jacobian ...
Definition example_05.hpp:355
Constraint_BurgersControl::nx_
int nx_
Definition burgers-control/example_02.hpp:22
Constraint_BurgersControl::compute_residual
void compute_residual(std::vector< Real > &r, const std::vector< Real > &u, const std::vector< Real > &z)
Definition example_05.hpp:72
Constraint_BurgersControl::Constraint_BurgersControl
Constraint_BurgersControl(int nx=128)
Definition example_05.hpp:156
Constraint_BurgersControl::applyAdjointJacobian_2
void applyAdjointJacobian_2(ROL::Vector< Real > &jv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the adjoint of the partial constraint Jacobian at , , to vector . This is the primary interface...
Definition example_05.hpp:277
Constraint_BurgersControl::applyAdjointJacobian_1
void applyAdjointJacobian_1(ROL::Vector< Real > &ajv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the adjoint of the partial constraint Jacobian at , , to the vector . This is the primary inter...
Definition example_05.hpp:244
Constraint_BurgersControl::compute_pde_jacobian
void compute_pde_jacobian(std::vector< Real > &dl, std::vector< Real > &d, std::vector< Real > &du, const std::vector< Real > &u)
Definition example_05.hpp:109
Constraint_BurgersControl::applyJacobian_2
void applyJacobian_2(ROL::Vector< Real > &jv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the partial constraint Jacobian at , , to the vector .
Definition example_05.hpp:209
Constraint_BurgersControl::compute_norm
Real compute_norm(const std::vector< Real > &r)
Definition example_05.hpp:37
Constraint_BurgersControl::applyAdjointHessian_21
void applyAdjointHessian_21(ROL::Vector< Real > &ahwv, const ROL::Vector< Real > &w, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the simulation-space derivative of the adjoint of the constraint optimization-space Jacobian at...
Definition example_05.hpp:351
Constraint_BurgersControl::update
void update(std::vector< Real > &u, const std::vector< Real > &s, const Real alpha=1.0)
Definition burgers-control/example_02.hpp:53
Constraint_BurgersControl::applyInverseJacobian_1
void applyInverseJacobian_1(ROL::Vector< Real > &ijv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the inverse partial constraint Jacobian at , , to the vector .
Definition example_05.hpp:225
Constraint_BurgersControl::dx_
Real dx_
Definition burgers-control/example_02.hpp:23
Constraint_BurgersControl::value
void value(ROL::Vector< Real > &c, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Evaluate the constraint operator at .
Definition example_05.hpp:158
Constraint_BurgersControl::dot
Real dot(const std::vector< Real > &x, const std::vector< Real > &y)
Definition burgers-control/example_02.hpp:34
Constraint_BurgersControl::applyAdjointHessian_12
void applyAdjointHessian_12(ROL::Vector< Real > &ahwv, const ROL::Vector< Real > &w, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the optimization-space derivative of the adjoint of the constraint simulation-space Jacobian at...
Definition example_05.hpp:347
Constraint_BurgersControl::linear_solve
void linear_solve(std::vector< Real > &u, std::vector< Real > &dl, std::vector< Real > &d, std::vector< Real > &du, const std::vector< Real > &r, const bool transpose=false)
Definition example_05.hpp:136
Constraint_BurgersControl::applyJacobian_1
void applyJacobian_1(ROL::Vector< Real > &jv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the partial constraint Jacobian at , , to the vector .
Definition example_05.hpp:176
Constraint_BurgersControl::applyInverseAdjointJacobian_1
void applyInverseAdjointJacobian_1(ROL::Vector< Real > &iajv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply the inverse of the adjoint of the partial constraint Jacobian at , , to the vector .
Definition example_05.hpp:306
Constraint_BurgersControl::solve
void solve(ROL::Vector< Real > &c, ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Given , solve for .
Definition example_05.hpp:169
Objective_BurgersControl
Definition burgers-control/example_01.hpp:20
Objective_BurgersControl::value
Real value(const ROL::Vector< Real > &z, Real &tol)
Compute value.
Definition burgers-control/example_01.hpp:306
Objective_BurgersControl::dot
Real dot(const std::vector< Real > &x, const std::vector< Real > &y)
Definition example_05.hpp:384
Objective_BurgersControl::apply_mass
void apply_mass(std::vector< Real > &Mu, const std::vector< Real > &u)
Definition example_05.hpp:401
Objective_BurgersControl::hessVec_21
void hessVec_21(ROL::Vector< Real > &hv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Definition example_05.hpp:513
Objective_BurgersControl::gradient_1
void gradient_1(ROL::Vector< Real > &g, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Compute gradient with respect to first component.
Definition example_05.hpp:457
Objective_BurgersControl::evaluate_target
Real evaluate_target(Real x)
Definition example_05.hpp:372
Objective_BurgersControl::Objective_BurgersControl
Objective_BurgersControl(Real alpha=1.e-4, int nx=128)
Definition example_05.hpp:422
Objective_BurgersControl::hessVec_11
void hessVec_11(ROL::Vector< Real > &hv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Apply Hessian approximation to vector.
Definition example_05.hpp:497
Objective_BurgersControl::hessVec_22
void hessVec_22(ROL::Vector< Real > &hv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Definition example_05.hpp:518
Objective_BurgersControl::gradient_2
void gradient_2(ROL::Vector< Real > &g, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Compute gradient with respect to second component.
Definition example_05.hpp:474
Objective_BurgersControl::nx_
int nx_
Definition burgers-control/example_01.hpp:24
Objective_BurgersControl::dx_
Real dx_
Definition burgers-control/example_01.hpp:25
Objective_BurgersControl::hessVec_12
void hessVec_12(ROL::Vector< Real > &hv, const ROL::Vector< Real > &v, const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Definition example_05.hpp:508
Objective_BurgersControl::alpha_
Real alpha_
Definition burgers-control/example_01.hpp:22
Objective_BurgersControl::value
Real value(const ROL::Vector< Real > &u, const ROL::Vector< Real > &z, Real &tol)
Compute value.
Definition example_05.hpp:428
ROL::Constraint_SimOpt
Defines the constraint operator interface for simulation-based optimization.
Definition ROL_Constraint_SimOpt.hpp:74
ROL::Constraint_SimOpt::solve
virtual void solve(Vector< Real > &c, Vector< Real > &u, const Vector< Real > &z, Real &tol)
Given , solve for .
Definition ROL_Constraint_SimOpt.hpp:191
ROL::Constraint::getParameter
const std::vector< Real > getParameter(void) const
Definition ROL_Constraint.hpp:369
ROL::Objective_SimOpt
Provides the interface to evaluate simulation-based objective functions.
Definition ROL_Objective_SimOpt.hpp:25
ROL::StdVector
Provides the ROL::Vector interface for scalar values, to be used, for example, with scalar constraint...
Definition ROL_StdVector.hpp:27
ROL::Vector
Defines the linear algebra or vector space interface.
Definition ROL_Vector.hpp:50
ROL::Vector::zero
virtual void zero()
Set to zero vector.
Definition ROL_Vector.hpp:133
Generated by doxygen 1.9.8