sparseLM.cpp File Reference

#include <iostream>
#include <fstream>
#include <iomanip>
#include "main.h"
#include <Eigen/LevenbergMarquardt>

Classes
struct	sparseGaussianTest< Scalar >

Functions
template<typename T >
void	test_sparseLM_T ()
	EIGEN_DECLARE_TEST (sparseLM)

Function Documentation

EIGEN_DECLARE_TEST()

EIGEN_DECLARE_TEST

(

sparseLM

)

                             {
  CALL_SUBTEST_1(test_sparseLM_T<double>());
 
  // CALL_SUBTEST_2(test_sparseLM_T<std::complex<double>());
}

References CALL_SUBTEST_1.

test_sparseLM_T()

template<typename T >

void test_sparseLM_T

(

)

                       {
  typedef Matrix<T, Dynamic, 1> VectorType;
 
  int inputs = 10;
  int values = 2000;
  sparseGaussianTest<T> sparse_gaussian(inputs, values);
  VectorType uv(inputs), uv_ref(inputs);
  VectorType x(values);
  // Generate the reference solution
  uv_ref << -2, 1, 4, 8, 6, 1.8, 1.2, 1.1, 1.9, 3;
  // Generate the reference data points
  x.setRandom();
  x = 10 * x;
  x.array() += 10;
  sparse_gaussian.initPoints(uv_ref, x);
 
  // Generate the initial parameters
  VectorBlock<VectorType> u(uv, 0, inputs / 2);
  VectorBlock<VectorType> v(uv, inputs / 2, inputs / 2);
  v.setOnes();
  // Generate u or Solve for u from v
  u.setOnes();
 
  // Solve the optimization problem
  LevenbergMarquardt<sparseGaussianTest<T> > lm(sparse_gaussian);
  int info;
  //   info = lm.minimize(uv);
 
  VERIFY_IS_EQUAL(info, 1);
  // Do a step by step solution and save the residual
  int maxiter = 200;
  int iter = 0;
  MatrixXd Err(values, maxiter);
  MatrixXd Mod(values, maxiter);
  LevenbergMarquardtSpace::Status status;
  status = lm.minimizeInit(uv);
  if (status == LevenbergMarquardtSpace::ImproperInputParameters) return;
}

References Eigen::LevenbergMarquardtSpace::ImproperInputParameters, info, sparseGaussianTest< Scalar >::initPoints(), Eigen::LevenbergMarquardt< FunctorType_ >::minimizeInit(), v, VERIFY_IS_EQUAL, and plotDoE::x.