Adolc_debugtest.cpp

Go to the documentation of this file.

/*----------------------------------------------------------------------------
 ADOL-C -- Automatic Differentiation by Overloading in C++
 File:     ADOL-C_NLP.cpp
 Revision: $$
 Contents: class myADOLC_NPL for interfacing with Ipopt

 Copyright (c) Andrea Walther

 This file is part of ADOL-C. This software is provided as open source.
 Any use, reproduction, or distribution of the software constitutes
 recipient's acceptance of the terms of the accompanying license file.

 This code is based on the file  MyNLP.cpp contained in the Ipopt package
 with the authors:  Carl Laird, Andreas Waechter
----------------------------------------------------------------------------*/

/** C++ Example NLP for interfacing a problem with IPOPT and ADOL-C.
 *  MyADOL-C_NLP implements a C++ example showing how to interface
 *  with IPOPT and ADOL-C through the TNLP interface. This class
 *  implements the Example 5.1 from "Sparse and Parially Separable
 *  Test Problems for Unconstrained and Equality Constrained
 *  Optimization" by L. Luksan and J. Vlcek ignoring sparsity.
 *
 *  no exploitation of sparsity !!
 *
 */
#include <cassert>

#include "Adolc_debugtest.h"

using namespace Ipopt;

/* Constructor. */
MyADOLC_NLP::MyADOLC_NLP()
{}

MyADOLC_NLP::~MyADOLC_NLP(){}

bool MyADOLC_NLP::get_nlp_info(Index& n, Index& m, Index& nnz_jac_g,
                         Index& nnz_h_lag, IndexStyleEnum& index_style)
{
  n = 20;

  m = n-2;

  // in this example the jacobian is dense. Hence, it contains n*m nonzeros
  nnz_jac_g = n*m;

  // the hessian is also dense and has n*n total nonzeros, but we
  // only need the lower left corner (since it is symmetric)
  nnz_h_lag = n*(n-1)/2+n;

  generate_tapes(n, m);

  // use the C style indexing (0-based)
  index_style = C_STYLE;

  return true;
}

bool MyADOLC_NLP::get_bounds_info(Index n, Number* x_l, Number* x_u,
                            Index m, Number* g_l, Number* g_u)
{
  // none of the variables have bounds
  for (Index i=0; i<n; i++) {
    x_l[i] = -1e20;
    x_u[i] =  1e20;
  }

  // Set the bounds for the constraints
  for (Index i=0; i<m; i++) {
    g_l[i] = 0;
    g_u[i] = 0;
  }

  return true;
}

bool MyADOLC_NLP::get_starting_point(Index n, bool init_x, Number* x,
                               bool init_z, Number* z_L, Number* z_U,
                               Index m, bool init_lambda,
                               Number* lambda)
{
  // Here, we assume we only have starting values for x, if you code
  // your own NLP, you can provide starting values for the others if
  // you wish.
  assert(init_x == true);
  assert(init_z == false);
  assert(init_lambda == false);

  // set the starting point
  for (Index i=0; i<n/2; i++) {
    x[2*i] = -1.2;
    x[2*i+1] = 1.;
  }
  if (n != 2*(n/2)) {
    x[n-1] = -1.2;
  }

  return true;
}

template<class T> bool  MyADOLC_NLP::eval_obj(Index n, const T *x, T& obj_value)
{
  T a1, a2;
  obj_value = 0.;
  for (Index i=0; i<n-1; i++) {
    a1 = x[i]*x[i]-x[i+1];
    a2 = x[i] - 1.;
    obj_value += 100.*a1*a1 + a2*a2;
  }

  return true;
}

template<class T> bool  MyADOLC_NLP::eval_constraints(Index n, const T *x, Index m, T* g)
{
  for (Index i=0; i<m; i++) {
    g[i] = 3.*pow(x[i+1],3.) + 2.*x[i+2] - 5.
           + sin(x[i+1]-x[i+2])*sin(x[i+1]+x[i+2]) + 4.*x[i+1]
           - x[i]*exp(x[i]-x[i+1]) - 3.;
  }

  return true;
}

//*************************************************************************
//
//
//         Nothing has to be changed below this point !!
//
//
//*************************************************************************


bool MyADOLC_NLP::eval_f(Index n, const Number* x, bool new_x, Number& obj_value)
{
  eval_obj(n,x,obj_value);

  return true;
}

bool MyADOLC_NLP::eval_grad_f(Index n, const Number* x, bool new_x, Number* grad_f)
{

  gradient(tag_f,n,x,grad_f);

  return true;
}

bool MyADOLC_NLP::eval_g(Index n, const Number* x, bool new_x, Index m, Number* g)
{

  eval_constraints(n,x,m,g);

  return true;
}

bool MyADOLC_NLP::eval_jac_g(Index n, const Number* x, bool new_x,
                       Index m, Index nele_jac, Index* iRow, Index *jCol,
                       Number* values)
{
  if (values == NULL) {
    // return the structure of the jacobian,
    // assuming that the Jacobian is dense

    Index idx = 0;
    for(Index i=0; i<m; i++)
      for(Index j=0; j<n; j++)
    {
      iRow[idx] = i;
      jCol[idx++] = j;
    }
 }
  else {
    // return the values of the jacobian of the constraints

    jacobian(tag_g,m,n,x,Jac);

    Index idx = 0;
    for(Index i=0; i<m; i++)
      for(Index j=0; j<n; j++)
      values[idx++] = Jac[i][j];

  }

  return true;
}

bool MyADOLC_NLP::eval_h(Index n, const Number* x, bool new_x,
                   Number obj_factor, Index m, const Number* lambda,
                   bool new_lambda, Index nele_hess, Index* iRow,
                   Index* jCol, Number* values)
{
  if (values == NULL) {
    // return the structure. This is a symmetric matrix, fill the lower left
    // triangle only.

    // the hessian for this problem is actually dense
    Index idx=0;
    for (Index row = 0; row < n; row++) {
      for (Index col = 0; col <= row; col++) {
        iRow[idx] = row;
        jCol[idx] = col;
        idx++;
      }
    }

    assert(idx == nele_hess);
  }
  else {
    // return the values. This is a symmetric matrix, fill the lower left
    // triangle only

    for(Index i = 0; i<n ; i++)
      x_lam[i] = x[i];
    for(Index i = 0; i<m ; i++)
      x_lam[n+i] = lambda[i];
    x_lam[n+m] = obj_factor;

    hessian(tag_L,n+m+1,x_lam,Hess);

    Index idx = 0;

    for(Index i = 0; i<n ; i++)
      {
    for(Index j = 0; j<=i ; j++)
      {
        values[idx++] = Hess[i][j];
      }
      }
  }

  return true;
}

void MyADOLC_NLP::finalize_solution(SolverReturn status,
                              Index n, const Number* x, const Number* z_L, const Number* z_U,
                              Index m, const Number* g, const Number* lambda,
                              Number obj_value,
                  const IpoptData* ip_data,
                  IpoptCalculatedQuantities* ip_cq)
{

  printf("\n\nObjective value\n");
  printf("f(x*) = %e\n", obj_value);

// Memory deallocation for ADOL-C variables

  delete[] x_lam;

  for(Index i=0;i<m;i++)
    delete[] Jac[i];
  delete[] Jac;

  for(Index i=0;i<n+m+1;i++)
    delete[] Hess[i];
  delete[] Hess;
}


//***************    ADOL-C part ***********************************

void MyADOLC_NLP::generate_tapes(Index n, Index m)
{
  Number *xp    = new double[n];
  Number *lamp  = new double[m];
  Number *zl    = new double[m];
  Number *zu    = new double[m];

  adouble *xa   = new adouble[n];
  adouble *g    = new adouble[m];
  adouble *lam  = new adouble[m];
  adouble sig;
  adouble obj_value;

  double dummy;

  Jac = new double*[m];
  for(Index i=0;i<m;i++)
    Jac[i] = new double[n];

  x_lam   = new double[n+m+1];

  Hess = new double*[n+m+1];
  for(Index i=0;i<n+m+1;i++)
    Hess[i] = new double[i+1];

  get_starting_point(n, 1, xp, 0, zl, zu, m, 0, lamp);

  trace_on(tag_f);

    for(Index i=0;i<n;i++)
      xa[i] <<= xp[i];

    eval_obj(n,xa,obj_value);

    obj_value >>= dummy;

  trace_off();

  trace_on(tag_g);

    for(Index i=0;i<n;i++)
      xa[i] <<= xp[i];

    eval_constraints(n,xa,m,g);


    for(Index i=0;i<m;i++)
      g[i] >>= dummy;

  trace_off();

   trace_on(tag_L);

    for(Index i=0;i<n;i++)
      xa[i] <<= xp[i];
    for(Index i=0;i<m;i++)
      lam[i] <<= 1.0;
    sig <<= 1.0;

    eval_obj(n,xa,obj_value);

    obj_value *= sig;
    eval_constraints(n,xa,m,g);

    for(Index i=0;i<m;i++)
      obj_value += g[i]*lam[i];

    obj_value >>= dummy;

  trace_off();

  delete[] xa;
  delete[] xp;
  delete[] g;
  delete[] lam;
  delete[] lamp;
  delete[] zu;
  delete[] zl;

}