ROL
poisson-control/example_02.cpp
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 // Burgers includes
11 #include "example_02.hpp"
12 // ROL includes
13 #include "ROL_Algorithm.hpp"
14 #include "ROL_TrustRegionStep.hpp"
15 #include "ROL_LineSearchStep.hpp"
16 #include "ROL_StdVector.hpp"
17 #include "ROL_StdTeuchosBatchManager.hpp"
18 #include "ROL_MonteCarloGenerator.hpp"
19 #include "ROL_Reduced_Objective_SimOpt.hpp"
20 #include "ROL_RiskNeutralObjective.hpp"
21 #include "ROL_Vector_SimOpt.hpp"
22 #include "ROL_Bounds.hpp"
23 #include "ROL_ParameterList.hpp"
24 
25 // Teuchos includes
26 #include "Teuchos_Time.hpp"
27 #include "ROL_Stream.hpp"
28 #include "Teuchos_GlobalMPISession.hpp"
29 #include "Teuchos_Comm.hpp"
30 #include "Teuchos_DefaultComm.hpp"
31 #include "Teuchos_CommHelpers.hpp"
32 
33 int main( int argc, char *argv[] ) {
34 
35  Teuchos::GlobalMPISession mpiSession(&argc, &argv);
36 
37  auto comm = ROL::toPtr( Teuchos::DefaultComm<int>::getComm() );
38 
39  // This little trick lets us print to std::cout only if a (dummy) command-line argument is provided.
40  int iprint = argc - 1;
41  ROL::Ptr<std::ostream> outStream;
42  ROL::nullstream bhs; // outputs nothing
43  if (iprint > 0)
44  outStream = ROL::makePtrFromRef(std::cout);
45  else
46  outStream = ROL::makePtrFromRef(bhs);
47 
48  int errorFlag = 0;
49 
50  // *** Example body.
51 
52  try {
53 
54  /***************************************************************************/
55  /***************** GRAB INPUTS *********************************************/
56  /***************************************************************************/
57  // Get finite element parameter list
58  std::string filename = "example_02.xml";
59  auto parlist = ROL::getParametersFromXmlFile( filename );
60 
61  if ( parlist->get("Display Option",0) && (comm->getRank() > 0) ) {
62  parlist->set("Display Option",0);
63  }
64  // Get ROL parameter list
65  filename = "input.xml";
66  auto ROL_parlist = ROL::getParametersFromXmlFile( filename );
67 
68  /***************************************************************************/
69  /***************** INITIALIZE SAMPLERS *************************************/
70  /***************************************************************************/
71  int dim = 2;
72  bool useSA = parlist->get("Use Stochastic Approximation",false);
73  int nSamp = 1;
74  if ( !useSA ) {
75  nSamp = parlist->get("Number of Monte Carlo Samples",1000);
76  }
77  std::vector<double> tmp(2); tmp[0] = -1.0; tmp[1] = 1.0;
78  std::vector<std::vector<double> > bounds(dim,tmp);
79  ROL::Ptr<ROL::BatchManager<double> > bman
80  = ROL::makePtr<ROL::StdTeuchosBatchManager<double,int>>(comm);
81  ROL::Ptr<ROL::SampleGenerator<double> > sampler
82  = ROL::makePtr<ROL::MonteCarloGenerator<double>>(nSamp,bounds,bman,useSA);
83 
84  /***************************************************************************/
85  /***************** INITIALIZE CONTROL VECTOR *******************************/
86  /***************************************************************************/
87  int nx = parlist->get("Number of Elements", 128);
88  ROL::Ptr<std::vector<double> > z_ptr = ROL::makePtr<std::vector<double>>(nx+1, 0.0);
89  ROL::Ptr<ROL::Vector<double> > z = ROL::makePtr<ROL::StdVector<double>>(z_ptr);
90  ROL::Ptr<std::vector<double> > u_ptr = ROL::makePtr<std::vector<double>>(nx-1, 0.0);
91  ROL::Ptr<ROL::Vector<double> > u = ROL::makePtr<ROL::StdVector<double>>(u_ptr);
92  ROL::Vector_SimOpt<double> x(u,z);
93  ROL::Ptr<std::vector<double> > p_ptr = ROL::makePtr<std::vector<double>>(nx-1, 0.0);
94  ROL::Ptr<ROL::Vector<double> > p = ROL::makePtr<ROL::StdVector<double>>(p_ptr);
95  ROL::Ptr<std::vector<double> > U_ptr = ROL::makePtr<std::vector<double>>(nx+1, 35.0);
96  ROL::Ptr<ROL::Vector<double> > U = ROL::makePtr<ROL::StdVector<double>>(U_ptr);
97  ROL::Ptr<std::vector<double> > L_ptr = ROL::makePtr<std::vector<double>>(nx+1, -5.0);
98  ROL::Ptr<ROL::Vector<double> > L = ROL::makePtr<ROL::StdVector<double>>(L_ptr);
99  ROL::Bounds<double> bnd(L,U);
100 
101  /***************************************************************************/
102  /***************** INITIALIZE OBJECTIVE FUNCTION ***************************/
103  /***************************************************************************/
104  double alpha = parlist->get("Penalty Parameter", 1.e-4);
105  ROL::Ptr<FEM<double> > fem = ROL::makePtr<FEM<double>>(nx);
106  ROL::Ptr<ROL::Objective_SimOpt<double> > pObj
107  = ROL::makePtr<DiffusionObjective<double>>(fem, alpha);
108  ROL::Ptr<ROL::Constraint_SimOpt<double> > pCon
109  = ROL::makePtr<DiffusionConstraint<double>>(fem);
110  ROL::Ptr<ROL::Objective<double> > robj
111  = ROL::makePtr<ROL::Reduced_Objective_SimOpt<double>>(pObj,pCon,u,z,p);
112  ROL::RiskNeutralObjective<double> obj(robj,sampler);
113 
114  /***************************************************************************/
115  /***************** RUN DERIVATIVE CHECK ************************************/
116  /***************************************************************************/
117  if (parlist->get("Run Derivative Check",false)) {
118  // Direction to test finite differences
119  ROL::Ptr<std::vector<double> > dz_ptr = ROL::makePtr<std::vector<double>>(nx+1, 0.0);
120  ROL::Ptr<ROL::Vector<double> > dz = ROL::makePtr<ROL::StdVector<double>>(dz_ptr);
121  ROL::Ptr<std::vector<double> > du_ptr = ROL::makePtr<std::vector<double>>(nx-1, 0.0);
122  ROL::Ptr<ROL::Vector<double> > du = ROL::makePtr<ROL::StdVector<double>>(du_ptr);
123  ROL::Vector_SimOpt<double> d(du,dz);
124  // Set to random vectors
125  srand(12345);
126  for (int i=0; i<nx+1; i++) {
127  (*dz_ptr)[i] = 2.0*(double)rand()/(double)RAND_MAX - 1.0;
128  (*z_ptr)[i] = 2.0*(double)rand()/(double)RAND_MAX - 1.0;
129  }
130  for (int i=0; i<nx-1; i++) {
131  (*du_ptr)[i] = 2.0*(double)rand()/(double)RAND_MAX - 1.0;
132  (*u_ptr)[i] = 2.0*(double)rand()/(double)RAND_MAX - 1.0;
133  }
134  // Run derivative checks
135  std::vector<double> param(dim,0.0);
136  robj->setParameter(param);
137  if ( comm->getRank() == 0 ) {
138  std::cout << "\nRUN DERIVATIVE CHECK FOR PARAMETRIZED OBJECTIVE FUNCTION SIMOPT\n";
139  }
140  pObj->checkGradient(x,d,(comm->getRank()==0));
141  pObj->checkHessVec(x,d,(comm->getRank()==0));
142  if ( comm->getRank() == 0 ) {
143  std::cout << "\nRUN DERIVATIVE CHECK FOR PARAMETRIZED EQUALITY CONSTRAINT SIMOPT\n";
144  }
145  pCon->checkApplyJacobian(x,d,*p,(comm->getRank()==0));
146  pCon->checkApplyAdjointJacobian(x,*du,*p,x,(comm->getRank()==0));
147  pCon->checkApplyAdjointHessian(x,*du,d,x,(comm->getRank()==0));
148  if ( comm->getRank() == 0 ) {
149  std::cout << "\nRUN DERIVATIVE CHECK FOR PARAMETRIZED OBJECTIVE FUNCTION\n";
150  }
151  robj->checkGradient(*z,*dz,(comm->getRank()==0));
152  robj->checkHessVec(*z,*dz,(comm->getRank()==0));
153  // Run derivative checks
154  if ( comm->getRank() == 0 ) {
155  std::cout << "\nRUN DERIVATIVE CHECK FOR RISK-NEUTRAL OBJECTIVE FUNCTION\n";
156  }
157  obj.checkGradient(*z,*dz,(comm->getRank()==0));
158  obj.checkHessVec(*z,*dz,(comm->getRank()==0));
159  }
160 
161  /***************************************************************************/
162  /***************** INITIALIZE ROL ALGORITHM ********************************/
163  /***************************************************************************/
164  ROL::Ptr<ROL::Algorithm<double>> algo;
165  ROL::Ptr<ROL::Step<double>> step;
166  ROL::Ptr<ROL::StatusTest<double>> status;
167  if ( useSA ) {
168  ROL_parlist->sublist("General").set("Recompute Objective Function",false);
169  ROL_parlist->sublist("Step").sublist("Line Search").set("Initial Step Size",0.1/alpha);
170  ROL_parlist->sublist("Step").sublist("Line Search").set("User Defined Initial Step Size",true);
171  ROL_parlist->sublist("Step").sublist("Line Search").sublist("Line-Search Method").set("Type","Iteration Scaling");
172  ROL_parlist->sublist("Step").sublist("Line Search").sublist("Descent Method").set("Type","Steepest Descent");
173  ROL_parlist->sublist("Step").sublist("Line Search").sublist("Curvature Condition").set("Type","Null Curvature Condition");
174  status = ROL::makePtr<ROL::StatusTest<double>>(*ROL_parlist);
175  step = ROL::makePtr<ROL::LineSearchStep<double>>(*ROL_parlist);
176  algo = ROL::makePtr<ROL::Algorithm<double>>(step,status,false);
177  }
178  else {
179  status = ROL::makePtr<ROL::StatusTest<double>>(*ROL_parlist);
180  step = ROL::makePtr<ROL::TrustRegionStep<double>>(*ROL_parlist);
181  algo = ROL::makePtr<ROL::Algorithm<double>>(step,status,false);
182  }
183 
184  /***************************************************************************/
185  /***************** PERFORM OPTIMIZATION ************************************/
186  /***************************************************************************/
187  Teuchos::Time timer("Optimization Time",true);
188  z->zero();
189  algo->run(*z,obj,bnd,(comm->getRank()==0));
190  double optTime = timer.stop();
191 
192  /***************************************************************************/
193  /***************** PRINT RESULTS *******************************************/
194  /***************************************************************************/
195  int my_number_samples = sampler->numMySamples(), number_samples = 0;
196  Teuchos::reduceAll<int,int>(*comm,Teuchos::REDUCE_SUM,1,&my_number_samples,&number_samples);
197  int my_number_solves = ROL::dynamicPtrCast<DiffusionConstraint<double> >(pCon)->getNumSolves(), number_solves = 0;
198  Teuchos::reduceAll<int,int>(*comm,Teuchos::REDUCE_SUM,1,&my_number_solves,&number_solves);
199  if (comm->getRank() == 0) {
200  std::cout << "Number of Samples = " << number_samples << "\n";
201  std::cout << "Number of Solves = " << number_solves << "\n";
202  std::cout << "Optimization Time = " << optTime << "\n\n";
203  }
204 
205  if ( comm->getRank() == 0 ) {
206  std::ofstream file;
207  if (useSA) {
208  file.open("control_SA.txt");
209  }
210  else {
211  file.open("control_SAA.txt");
212  }
213  std::vector<double> xmesh(fem->nz(),0.0);
214  fem->build_mesh(xmesh);
215  for (int i = 0; i < fem->nz(); i++ ) {
216  file << std::setprecision(std::numeric_limits<double>::digits10) << std::scientific << xmesh[i] << " "
217  << std::setprecision(std::numeric_limits<double>::digits10) << std::scientific << (*z_ptr)[i]
218  << "\n";
219  }
220  file.close();
221  }
222  }
223  catch (std::logic_error& err) {
224  *outStream << err.what() << "\n";
225  errorFlag = -1000;
226  }; // end try
227 
228  if (errorFlag != 0)
229  std::cout << "End Result: TEST FAILED\n";
230  else
231  std::cout << "End Result: TEST PASSED\n";
232 
233  return 0;
234 }
235 
236 
237 
238 
ROL_Algorithm.hpp
ROL_Reduced_Objective_SimOpt.hpp
ROL::Vector_SimOpt
Defines the linear algebra or vector space interface for simulation-based optimization.
Definition: ROL_Vector_SimOpt.hpp:23
ROL_LineSearchStep.hpp
ROL_RiskNeutralObjective.hpp
ROL_Stream.hpp
Defines a no-output stream class ROL::NullStream and a function makeStreamPtr which either wraps a re...
ROL_Bounds.hpp
ROL::details::nullstream
basic_nullstream< char, std::char_traits< char >> nullstream
Definition: ROL_Stream.hpp:36
ROL_StdVector.hpp
ROL_TrustRegionStep.hpp
ROL::Bounds
Provides the elementwise interface to apply upper and lower bound constraints.
Definition: ROL_Bounds.hpp:25
main
int main(int argc, char *argv[])
Definition: function/constraint/test_01.cpp:31
example_02.hpp
ROL_MonteCarloGenerator.hpp
dim
constexpr auto dim
Definition: vector/test_11.cpp:23
ROL_Vector_SimOpt.hpp
Generated by   doxygen 1.8.5