diff --git a/App/App.pro b/App/App.pro
index cb11d808964c63c884cab7291e48e41fb758d260..e0ca2ee5acfa01092bb13a861038427c90702976 100644
--- a/App/App.pro
+++ b/App/App.pro
@@ -56,7 +56,6 @@ SOURCES += \
src/TestRoughness.cpp \
src/TestToyExperiment.cpp \
src/TreeEventStructure.cpp \
- src/ROOTMinimizerFunction.cpp \
src/ROOTGSLNLSMinimizer.cpp \
$${FUNCTIONAL_TESTS}/IsGISAXS01/IsGISAXS01.cpp
diff --git a/App/inc/ROOTMinimizer.h b/App/inc/ROOTMinimizer.h
index 3c7fe5c0492827d3b08b32f01a2f45ce0d90da5e..c610b715fe0780384cb61e612a93635318728c15 100644
--- a/App/inc/ROOTMinimizer.h
+++ b/App/inc/ROOTMinimizer.h
@@ -37,44 +37,33 @@ public:
ROOTMinimizer(const std::string &minimizer_name, const std::string &algo_type=std::string());
virtual ~ROOTMinimizer();
- virtual void setParameter(size_t index, const FitParameter *par);
- virtual void setParameters(const FitSuiteParameters ¶meters);
+ virtual void minimize();
+ virtual void setParameter(size_t index, const FitParameter *par);
- virtual void setFunction(function_chi2_t fun_chi2, size_t nparameters, function_gradient_t fun_gradient, size_t ndatasize);
+ virtual void setParameters(const FitSuiteParameters ¶meters);
- virtual void minimize();
+ virtual void setChiSquaredFunction(function_chi2_t fun_chi2, size_t nparameters);
- //! return created minimizer
- ROOT::Math::Minimizer *getROOTMinimizer() { return m_root_minimizer; }
+ virtual void setGradientFunction(function_gradient_t fun_gradient, size_t nparameters, size_t ndatasize);
- //! get number of variables to fit
virtual size_t getNumberOfVariables() const { return m_root_minimizer->NDim(); }
- //! return minimum function value
virtual double getMinValue() const { return m_root_minimizer->MinValue(); }
- //! return value of variable corresponding the minimum of the function
- virtual double getValueOfVariableAtMinimum(size_t i) const {
- if(i >= getNumberOfVariables() ) throw OutOfBoundsException("ROOTMinimizer::getValueOfVariableAtMinimum() -> Wrong number of the variable");
- return m_root_minimizer->X()[i];
- }
+ virtual double getValueOfVariableAtMinimum(size_t i) const {return m_root_minimizer->X()[check_index(i)]; }
- //! return value of variable corresponding the minimum of the function
- virtual double getErrorOfVariable(size_t i) const {
- if(i >= getNumberOfVariables() ) throw OutOfBoundsException("ROOTMinimizer::getErrorOfVariable() -> Wrong number of the variable");
- return (m_root_minimizer->Errors() == 0? 0 : m_root_minimizer->Errors()[i]);
- }
+ virtual double getErrorOfVariable(size_t i) const { return (m_root_minimizer->Errors() == 0? 0 : m_root_minimizer->Errors()[check_index(i)]); }
- //! printing results
virtual void printResults() const;
- //! clear resources (parameters) for consecutives minimizations
virtual void clear() { m_root_minimizer->Clear(); }
- //! printing minimizer description
virtual void printOptions() const;
+ //! return created minimizer
+ ROOT::Math::Minimizer *getROOTMinimizer() { return m_root_minimizer; }
+
//! checking validity of the combination minimizer_name and algo_type
bool isValidNames(const std::string &minimizer_name, const std::string &algo_type);
@@ -82,20 +71,16 @@ public:
bool isGradientBasedAgorithm();
private:
+ ROOTMinimizer(const ROOTMinimizer &);
+ ROOTMinimizer &operator=(const ROOTMinimizer &);
+
+ size_t check_index(size_t index) const { return index<getNumberOfVariables() ? index : throw OutOfBoundsException("ROOTMinimizer::getErrorOfVariable() -> Wrong number of the variable"); }
+
std::string m_minimizer_name;
std::string m_algo_type;
-
ROOT::Math::Minimizer *m_root_minimizer;
- ROOTMinimizerFunction * m_minfunc;
- ROOTMinimizerElementFunction * m_minfunc_element;
-
-// IMinimizerFunction *m_minimizer_function;
-// ROOT::Math::Functor *m_fcn; //! function to minimize
-// ROOT::Math::GradFunctor *m_fcn_grad; //! gradient of function to minimize
-// function_t m_fcn;
-// element_function_t m_element_fcn;
-// int m_ndims;
-// int m_nelements;
+ ROOTMinimizerChiSquaredFunction *m_chi2_func;
+ ROOTMinimizerGradientFunction *m_gradient_func;
};
#endif // ROOTMINIMIZER_H
diff --git a/App/inc/ROOTMinimizerFunction.h b/App/inc/ROOTMinimizerFunction.h
index 6e242c2ced6d29819145ffcb8bbd406d874a663c..407c7bbc5847a974990380306b7f959ac6aeb99e 100644
--- a/App/inc/ROOTMinimizerFunction.h
+++ b/App/inc/ROOTMinimizerFunction.h
@@ -14,72 +14,61 @@
//! @author Scientific Computing Group at FRM II
//! @date Dec 10, 2012
-
-#include "Numeric.h"
#include "IMinimizer.h"
#include "Math/Functor.h"
#include "Math/FitMethodFunction.h"
-#include <vector>
-#include <cmath>
+
//- -------------------------------------------------------------------
-//! @class ROOTMinimizerFunction
-//! @brief Basic minimizer function
+//! @class ROOTMinimizerChiSquaredFunction
+//! @brief minimizer chi2 function
//- -------------------------------------------------------------------
-class ROOTMinimizerFunction : public ROOT::Math::Functor
+class ROOTMinimizerChiSquaredFunction : public ROOT::Math::Functor
{
public:
- ROOTMinimizerFunction(IMinimizer::function_chi2_t fcn, int ndims ) : ROOT::Math::Functor(fcn, ndims), m_fcn(fcn) {}
- virtual ~ROOTMinimizerFunction(){}
+ ROOTMinimizerChiSquaredFunction(IMinimizer::function_chi2_t fcn, int ndims ) : ROOT::Math::Functor(fcn, ndims), m_fcn(fcn) {}
+ virtual ~ROOTMinimizerChiSquaredFunction(){}
IMinimizer::function_chi2_t m_fcn;
};
//- -------------------------------------------------------------------
-//! @class ROOTMinimizerElementFunction
+//! @class ROOTMinimizerGradientFunction
//! @brief Minimizer function with access to single data element residuals.
//! Required by Fumili, Fumili2 and GSLMultiMin minimizers
//- -------------------------------------------------------------------
-class ROOTMinimizerElementFunction : public ROOT::Math::FitMethodFunction
+class ROOTMinimizerGradientFunction : public ROOT::Math::FitMethodFunction
{
public:
typedef ROOT::Math::BasicFitMethodFunction<ROOT::Math::IMultiGenFunction>::Type_t Type_t;
- ROOTMinimizerElementFunction(IMinimizer::function_gradient_t fun_gradient, size_t npars, size_t ndatasize)
+ ROOTMinimizerGradientFunction(IMinimizer::function_gradient_t fun_gradient, size_t npars, size_t ndatasize)
: ROOT::Math::FitMethodFunction(npars, ndatasize)
, m_fun_gradient(fun_gradient) { }
- virtual ~ROOTMinimizerElementFunction(){}
+ virtual ~ROOTMinimizerGradientFunction(){}
Type_t Type() const { return ROOT::Math::FitMethodFunction::kLeastSquare; }
- ROOT::Math::IMultiGenFunction * Clone() const { return new ROOTMinimizerElementFunction(m_fun_gradient, NDim(), NPoints()); }
+ ROOT::Math::IMultiGenFunction * Clone() const { return new ROOTMinimizerGradientFunction(m_fun_gradient, NDim(), NPoints()); }
//! evaluation of single data element residual
- double DataElement(const double *pars, unsigned int i_data, double *gradient = 0) const
- {
+ double DataElement(const double *pars, unsigned int i_data, double *gradient = 0) const {
return m_fun_gradient(pars, i_data, gradient);
}
private:
//! evaluation of chi2
- double DoEval(const double * pars) const
- {
- //std::cout << "ROOTMinimizerFuncion::DoEval() -> 1.1 ndim:" <<NDim() <<" npoint:" << NPoints() << std::endl;
+ double DoEval(const double * pars) const {
double chi2 = 0.0;
for(size_t i_data=0; i_data<NPoints(); ++i_data) {
double res = DataElement(pars, i_data);
chi2 += res*res;
}
- //std::cout << "ROOTMinimizerFuncion::DoEval() -> 1.2 chi:" <<chi2 << " ndim:" << NDim() << " np:" << NPoints() << std::endl;
return chi2/double(NPoints());
- //return chi2;
}
-
IMinimizer::function_gradient_t m_fun_gradient;
};
-
-
#endif // ROOTMINIMIZERFUNCTION_H
diff --git a/App/src/IsGISAXSData.cpp b/App/src/IsGISAXSData.cpp
index 419aec4471854562d72b6f8cc2976152bb669fb0..d209a2f9c5f966c61cc84cdce01bfe9180a59c44 100644
--- a/App/src/IsGISAXSData.cpp
+++ b/App/src/IsGISAXSData.cpp
@@ -162,7 +162,6 @@ OutputData<double> *IsGISAXSData::convert_isgi_scan(std::vector<IsgiData > &isgi
for(size_t i_point=0; i_point<isgi_data.size(); ++i_point) {
phi_axis.push_back(isgi_data[i_point].phif);
}
-
}
OutputData<double > * data = new OutputData<double >;
data->addAxis(phi_axis);
diff --git a/App/src/IsGISAXSTools.cpp b/App/src/IsGISAXSTools.cpp
index 95ea8cc788eb4104174ecc923494ce0a65ae5012..1fe11f34fac822aace4a996da59c62ae8dc0453d 100644
--- a/App/src/IsGISAXSTools.cpp
+++ b/App/src/IsGISAXSTools.cpp
@@ -20,6 +20,7 @@
#include <iomanip>
#include <cmath>
#include <algorithm>
+#include <cassert>
double IsGISAXSTools::m_hist_min = 0;
double IsGISAXSTools::m_hist_max = 0;
@@ -34,6 +35,7 @@ void IsGISAXSTools::drawLogOutputData(const OutputData<double>& output,
const std::string& canvas_name, const std::string& canvas_title,
const std::string& draw_options, const std::string &histogram_title)
{
+ assert(&output);
TCanvas *c1 = new TCanvas(canvas_name.c_str(), canvas_title.c_str(), 0, 0, 1024, 768);
IsGISAXSTools::setMinimum(1.);
c1->cd();
@@ -48,6 +50,7 @@ void IsGISAXSTools::drawOutputData(const OutputData<double>& output,
const std::string& canvas_name, const std::string& canvas_title,
const std::string& draw_options, const std::string &histogram_title)
{
+ assert(&output);
TCanvas *c1 = new TCanvas(canvas_name.c_str(), canvas_title.c_str(), 0, 0, 1024, 768);
c1->cd();
drawOutputDataInPad(output, draw_options, histogram_title);
@@ -59,6 +62,7 @@ void IsGISAXSTools::drawOutputData(const OutputData<double>& output,
void IsGISAXSTools::drawOutputDataInPad(const OutputData<double>& output,
const std::string& draw_options, const std::string &histogram_title)
{
+ assert(&output);
if(!gPad) {
throw NullPointerException("IsGISAXSTools::drawOutputDataInPad() -> Error! No canvas exists.");
}
@@ -85,6 +89,7 @@ void IsGISAXSTools::drawOutputDataInPad(const OutputData<double>& output,
/* ************************************************************************* */
TH2D *IsGISAXSTools::getOutputDataTH2D(const OutputData<double>& output, const std::string &histo_name)
{
+ assert(&output);
if (output.getNdimensions() !=2) throw( "IsGISAXSTools::getOutputDataTH2D() -> Warning! Expected number of dimensiobs is 2.");
std::vector<AxisStructure > haxises;
@@ -146,6 +151,7 @@ TH2D *IsGISAXSTools::getOutputDataTH2D(const OutputData<double>& output, const s
/* ************************************************************************* */
TH1 *IsGISAXSTools::getOutputDataTH123D(const OutputData<double>& output, const std::string &histo_name)
{
+ assert(&output);
if (output.getNdimensions() >3) {
std::cout << "IsGISAXSTools::getOutputDataTH123D() -> Warning! Expected number of dimensions should be not more than 3" << std::endl;
return 0;
@@ -270,6 +276,8 @@ void IsGISAXSTools::drawOutputDataDistribution1D(const OutputData<double> &outpu
/* ************************************************************************* */
void IsGISAXSTools::drawOutputDataDifference1D(const OutputData<double> &left, const OutputData<double> &right, const std::string &draw_options, const std::string &histogram_title)
{
+ assert(&left);
+ assert(&right);
if(!gPad) {
throw NullPointerException("IsGISAXSTools::drawOutputDataDifference1D() -> Error! No canvas exists.");
}
@@ -320,6 +328,8 @@ void IsGISAXSTools::drawOutputDataDifference1D(const OutputData<double> &left, c
/* ************************************************************************* */
void IsGISAXSTools::drawOutputDataRelativeDifference2D(const OutputData<double> &left, const OutputData<double> &right, const std::string &draw_options, const std::string &histogram_title)
{
+ assert(&left);
+ assert(&right);
if(!gPad) {
throw NullPointerException("IsGISAXSTools::drawOutputDataDifference2D -> Error! No canvas exists.");
}
@@ -342,6 +352,8 @@ void IsGISAXSTools::drawOutputDataRelativeDifference2D(const OutputData<double>
/* ************************************************************************* */
void IsGISAXSTools::drawOutputDataChi2Difference2D(const OutputData<double> &left, const OutputData<double> &right, const std::string &draw_options, const std::string &histogram_title)
{
+ assert(&left);
+ assert(&right);
if(!gPad) {
throw NullPointerException("IsGISAXSTools::drawOutputDataDifference2D -> Error! No canvas exists.");
}
@@ -465,6 +477,7 @@ void IsGISAXSTools::exportOutputDataInVectors2D(const OutputData<double> &output
, std::vector<std::vector<double > > &v_axis0
, std::vector<std::vector<double > > &v_axis1)
{
+ assert(&output_data);
if (output_data.getRank() != 2) return;
const IAxis *p_axis0 = output_data.getAxis(0);
@@ -511,6 +524,8 @@ void IsGISAXSTools::exportOutputDataInVectors2D(const OutputData<double> &output
/* ************************************************************************* */
TLine *IsGISAXSTools::getOutputDataScanLine(const OutputData<double> &data)
{
+ assert(&data);
+
if(data.getNdimensions() != 2) throw LogicErrorException("IsGISAXSTools::getOutputDataScanLine() -> Error! Number of dimensions should be 2");
double x1(0), x2(0), y1(0), y2(0);
if( data.getAxis(NDetector2d::ALPHA_AXIS_NAME) && data.getAxis(NDetector2d::ALPHA_AXIS_NAME)->getSize() == 1) {
@@ -540,6 +555,8 @@ TLine *IsGISAXSTools::getOutputDataScanLine(const OutputData<double> &data)
/* ************************************************************************* */
TH1D *IsGISAXSTools::getOutputDataScanHist(const OutputData<double> &data, const std::string &hname)
{
+ assert(&data);
+
if(data.getNdimensions() != 2) throw LogicErrorException("IsGISAXSTools::getOutputDataScanHist() -> Error! Number of dimensions should be 2");
// one of axis should have dimension 1
if( (data.getAxis(NDetector2d::ALPHA_AXIS_NAME) && data.getAxis(NDetector2d::ALPHA_AXIS_NAME)->getSize() != 1) && (data.getAxis(NDetector2d::PHI_AXIS_NAME) && data.getAxis(NDetector2d::PHI_AXIS_NAME)->getSize() != 1))
@@ -578,6 +595,7 @@ TH1D *IsGISAXSTools::getOutputDataScanHist(const OutputData<double> &data, const
/* ************************************************************************* */
OutputData<double > *IsGISAXSTools::createNoisyData(const OutputData<double> &exact_data, double noise_factor)
{
+ assert(&exact_data);
OutputData<double > *real_data = exact_data.clone();
OutputData<double>::iterator it = real_data->begin();
while (it != real_data->end()) {
@@ -594,6 +612,7 @@ OutputData<double > *IsGISAXSTools::createNoisyData(const OutputData<double> &ex
OutputData<double > *IsGISAXSTools::createDataWithGaussianNoise(const OutputData<double> &exact_data, double sigma)
{
+ assert(&exact_data);
OutputData<double > *real_data = exact_data.clone();
OutputData<double>::iterator it = real_data->begin();
while (it != real_data->end()) {
@@ -608,6 +627,8 @@ OutputData<double > *IsGISAXSTools::createDataWithGaussianNoise(const OutputData
void IsGISAXSTools::drawOutputDataComparisonResults(const OutputData<double> &data, const OutputData<double> &reference, const std::string &name, const std::string &title)
{
+ assert(&data);
+ assert(&reference);
TCanvas *c1 = DrawHelper::instance().createAndRegisterCanvas(name, title);
c1->Divide(2,2);
diff --git a/App/src/ROOTMinimizer.cpp b/App/src/ROOTMinimizer.cpp
index b24af7286c200c88814368f042587f61e141cea6..45f7efc9bfe5a46a4390c80eaf5e00e2846c17b2 100644
--- a/App/src/ROOTMinimizer.cpp
+++ b/App/src/ROOTMinimizer.cpp
@@ -10,21 +10,20 @@
/* ************************************************************************* */
+// ROOTMinimizer c-tor
//
+// some usefull staff
+// see http://root.cern.ch/phpBB3/viewtopic.php?f=15&t=14230&p=61216&hilit=minimizer+precision#p61216
+// see http://root.cern.ch/phpBB3/viewtopic.php?f=3&t=9181&hilit=precision+tolerance
/* ************************************************************************* */
ROOTMinimizer::ROOTMinimizer(const std::string &minimizer_name, const std::string &algo_type)
: m_minimizer_name(minimizer_name)
, m_algo_type(algo_type)
- , m_minfunc(0)
- , m_minfunc_element(0)
+ , m_chi2_func(0)
+ , m_gradient_func(0)
{
-
if( !isValidNames(m_minimizer_name, m_algo_type) ) throw LogicErrorException("ROOTMinimizer::ROOTMinimizer() -> Error! Wrong minimizer initialization parameters.");
- // see http://root.cern.ch/phpBB3/viewtopic.php?f=15&t=14230&p=61216&hilit=minimizer+precision#p61216
- // see http://root.cern.ch/phpBB3/viewtopic.php?f=3&t=9181&hilit=precision+tolerance
- //ROOT::Math::MinimizerOptions::SetDefaultTolerance(0.1);
-
if( m_minimizer_name == "GSLMultiFit") {
// hacked version of ROOT's GSL Levenberg-Marquardt minimizer
m_root_minimizer = new ROOT::Patch::GSLNLSMinimizer(2);
@@ -32,20 +31,14 @@ ROOTMinimizer::ROOTMinimizer(const std::string &minimizer_name, const std::strin
m_root_minimizer = ROOT::Math::Factory::CreateMinimizer(minimizer_name, algo_type );
}
if( !m_root_minimizer ) throw NullPointerException("ROOTMinimizer::ROOTMinimizer() -> Error! Can't create minimizer.");
-
-// m_root_minimizer->SetMaxFunctionCalls(20000);
-// m_root_minimizer->SetMaxIterations(20000);
-// m_root_minimizer->SetPrintLevel(4);
-// m_root_minimizer->SetTolerance(0.01);
-// m_root_minimizer->SetPrecision(1e-6);
}
ROOTMinimizer::~ROOTMinimizer()
{
delete m_root_minimizer;
- delete m_minfunc;
- delete m_minfunc_element;
+ delete m_chi2_func;
+ delete m_gradient_func;
}
@@ -149,21 +142,18 @@ void ROOTMinimizer::minimize()
/* ************************************************************************* */
// set fcn function for minimizer
/* ************************************************************************* */
-// FIXME ROOTMinimizer::setFunction Implement Multiple inheretiance in ROOTMinimizerElementFunction ;)
-void ROOTMinimizer::setFunction(function_chi2_t fun_chi2, size_t nparameters, function_gradient_t fun_gradient, size_t ndatasize)
+void ROOTMinimizer::setChiSquaredFunction(function_chi2_t fun_chi2, size_t nparameters)
{
- if( isGradientBasedAgorithm() ) {
- std::cout << " ROOTMinimizer::setFunction() -> XXX 1.1 making ROOTMinimizerElementFunction " << std::endl;
- delete m_minfunc_element;
- m_minfunc_element = new ROOTMinimizerElementFunction(fun_gradient, nparameters, ndatasize);
- m_root_minimizer->SetFunction(*m_minfunc_element);
+ delete m_chi2_func;
+ m_chi2_func = new ROOTMinimizerChiSquaredFunction(fun_chi2, nparameters);
+ if( !isGradientBasedAgorithm() ) m_root_minimizer->SetFunction(*m_chi2_func);
+}
- } else {
- std::cout << " ROOTMinimizer::setFunction() -> XXX 1.2 making ROOTMinimizerFunction" << std::endl;
- delete m_minfunc;
- m_minfunc = new ROOTMinimizerFunction(fun_chi2, nparameters);
- m_root_minimizer->SetFunction(*m_minfunc);
- }
+void ROOTMinimizer::setGradientFunction(function_gradient_t fun_gradient, size_t nparameters, size_t ndatasize)
+{
+ delete m_gradient_func;
+ m_gradient_func = new ROOTMinimizerGradientFunction(fun_gradient, nparameters, ndatasize);
+ if( isGradientBasedAgorithm() ) m_root_minimizer->SetFunction(*m_gradient_func);
}
@@ -204,8 +194,8 @@ void ROOTMinimizer::printResults() const
std::cout << std::setw(25) << std::left << " Status " << ": " << m_root_minimizer->Status() << " '" << minimizerStatus[m_root_minimizer->Status()] << "'" << std::endl;
std::cout << std::setw(25) << std::left << " IsValidError " << ": " << m_root_minimizer->IsValidError() << " '" << validErrorStatus[m_root_minimizer->Status()] << "'" <<std::endl;
std::cout << std::setw(25) << std::left << " NCalls" << ": " << m_root_minimizer->NCalls() << std::endl;
- if(m_minfunc_element) {
- std::cout << std::setw(25) << std::left << " NCallsElement " << ": " << m_minfunc_element->NCalls() << std::endl;
+ if(m_gradient_func) {
+ std::cout << std::setw(25) << std::left << " NCallsElement " << ": " << m_gradient_func->NCalls() << std::endl;
}
std::cout << std::setw(25) << std::left << " MinValue" << ": " << std::scientific << std::setprecision(8) << getMinValue() << std::endl;
std::cout << std::setw(25) << std::left << " Edm" << ": " << std::scientific << std::setprecision(8) << m_root_minimizer->Edm() << std::endl;
diff --git a/App/src/ROOTMinimizerFunction.cpp b/App/src/ROOTMinimizerFunction.cpp
deleted file mode 100644
index cc33dc7511b476522f427ec9b2d56adf6b6dc3be..0000000000000000000000000000000000000000
--- a/App/src/ROOTMinimizerFunction.cpp
+++ /dev/null
@@ -1,129 +0,0 @@
-#include "ROOTMinimizerFunction.h"
-
-//ROOTMinimizerFunction::ROOTMinimizerFunction()
-//{
-//}
-
-
-///* *************************************************************************** */
-//// evaluation of single data element residual
-///* *************************************************************************** */
-//double ROOTMinimizerElementFunction::DataElement(const double *pars, unsigned int i_selected_element, double *gradient) const
-//{
-// bool parameters_changed = parametersHaveChanged(pars);
-// for(size_t i=0; i<m_ndims; ++i) {
-// std::cout << pars[i] << " ";
-// }
-// std::cout << " parameters_changed:" << parameters_changed << " selected_element:" << i_selected_element << " index_prev:" << m_prev_element << std::endl;
-
-// int diff = i_selected_element-m_prev_element;
-// if( diff >1 && ( diff != -1.*(int)m_nelements) ) std::cout << "XXXXX 1.1 Warning! (i-prev>1) Non sequential access to elements!" << std::endl;
-// if( parameters_changed && i_selected_element!= 0 )std::cout << "XXXXX 1.2 Warning! Parameters changed and i!= 0" << std::endl;
-// if( parameters_changed && i_selected_element == m_prev_element )std::cout << "XXXXX 1.2 Warning! Parameters changed and i=prev_element" << std::endl;
-// m_prev_element = i_selected_element;
-
-// if(parameters_changed) {
-// m_fcn(pars);
-// for(size_t i_element=0; i_element<m_nelements; ++i_element) {
-// m_residuals[i_element] = m_element_fcn(pars, i_element, gradient);
-// }
-// }
-
-// if( gradient ) {
-// std::cout << " gradient " << std::endl;
-// if(i_selected_element == 0 || parameters_changed) {
-// std::cout << "g!=0 parameters_changed " << parameters_changed << std::endl;
-// }
-// for(size_t i_par=0; i_par<m_ndims; ++i_par) {
-// gradient[i_par] = m_gradients[i_par][i_selected_element];
-// std::cout << " gradient_result i_par:" << i_par << " " << gradient[i_par] << std::endl;
-// }
-// }
-
-// // ./root/hist/hist/src/TF1.cxx - how root estimate derivative
-
-// m_ncalls_element++;
-// return m_residuals[i_selected_element];
-//}
-
-
-
-//class ROOTMinimizerElementFunction : public ROOT::Math::FitMethodFunction
-//{
-//public:
-// typedef ROOT::Math::BasicFitMethodFunction<ROOT::Math::IMultiGenFunction>::Type_t Type_t;
-
-// ROOTMinimizerElementFunction(IMinimizer::function_chi2_t fcn, size_t ndims, IMinimizer::function_gradient_t element_fcn, size_t nelements)
-// : ROOT::Math::FitMethodFunction(ndims, nelements)
-// , m_fcn(fcn)
-// , m_element_fcn(element_fcn)
-// , m_ndims(ndims)
-// , m_nelements(nelements)
-// , m_ncalls(0)
-// , m_ncalls_element(0)
-// , m_prev_element(0)
-// {
-// m_parameter_values.resize(ndims, 0.0);
-// m_residuals.resize(nelements, 0.0);
-// m_gradients.resize(ndims);
-// for(size_t i_dim=0; i_dim<ndims; ++i_dim) {
-// m_gradients[i_dim].resize(nelements, 0.0);
-// }
-// }
-
-// virtual ~ROOTMinimizerElementFunction(){}
-
-// Type_t Type() const { return ROOT::Math::FitMethodFunction::kLeastSquare; }
-// ROOT::Math::IMultiGenFunction * Clone() const { return new ROOTMinimizerElementFunction(m_fcn, m_ndims, m_element_fcn, m_nelements); }
-
-// //! evaluation of chi2
-// double DoEval(const double * pars) const
-// {
-// double chi2 = 0.0;
-// for(size_t i_data=0; i_data<m_nelements; ++i_data) {
-// chi2 += DataElement(pars, i_data);
-// }
-// ++m_ncalls;
-// std::cout << " ROOTMinimizerElementFunction::DoEval() -> " << chi2 << std::endl;
-// return chi2;
-// }
-
-// //! evaluation of single data element residual
-// double DataElement(const double *pars, unsigned int i_selected_element, double *gradient = 0) const ;
-
-// virtual unsigned int NCalls() const { return m_ncalls; }
-// virtual unsigned int NCallsElement() const { return m_ncalls_element; }
-
-// bool isEqual(double a, double b) const
-// {
-// return std::abs(a-b) < 2.*Numeric::double_epsilon;
-// }
-
-// bool parametersHaveChanged(const double *pars) const
-// {
-// for(size_t i=0; i<m_ndims; ++i) {
-// if( !isEqual(pars[i], m_parameter_values[i])) {
-// std::copy(pars, pars+m_ndims, m_parameter_values.begin());
-// return true;
-// }
-// }
-// return false;
-// }
-
-
-//private:
-
-// IMinimizer::function_chi2_t m_fcn;
-// IMinimizer::function_gradient_t m_element_fcn;
-// size_t m_ndims;
-// size_t m_nelements;
-// mutable std::vector<double > m_parameter_values;
-// mutable size_t m_ncalls;
-// mutable size_t m_ncalls_element;
-// mutable size_t m_prev_element;
-// mutable std::vector<double > m_residuals; // [m_nelements]
-// mutable std::vector<std::vector<double> > m_gradients; // [m_ndims][m_nelements]
-//};
-
-
-
diff --git a/App/src/TestIsGISAXS12.cpp b/App/src/TestIsGISAXS12.cpp
index da0bd1f8f250c31ea8e5f3ae586b7dc6cb96efd9..30c6edbfa931f4bf0ea75915ceaa9453ff9c6800 100644
--- a/App/src/TestIsGISAXS12.cpp
+++ b/App/src/TestIsGISAXS12.cpp
@@ -23,6 +23,7 @@
#include "ResolutionFunction2DSimple.h"
#include "MathFunctions.h"
#include "ROOTMinimizer.h"
+#include "MinimizerTest.h"
#include "OutputDataFunctions.h"
#include "ExperimentConstants.h"
#include "OutputDataIOFactory.h"
@@ -366,7 +367,7 @@ void TestIsGISAXS12::run_test_minimizer()
// our simulation produces numerically same results
m_fitSuite = new FitSuite();
- m_fitSuite->setMinimizer( new TestMinimizer() );
+ m_fitSuite->setMinimizer( new MinimizerTest() );
m_fitSuite->addFitParameter("*Normalizer/scale", 1.31159E+05, 100, AttLimits::limited(1e4, 2e5));
m_fitSuite->addFitParameter("*Normalizer/shift", -8.10009E-02, 1, AttLimits::limited(-10., 20.));
diff --git a/App/src/TestIsGISAXS13.cpp b/App/src/TestIsGISAXS13.cpp
index 115ccd002b285e80dfe968d34756cfaf7b3e79af..7919f3c86bb81814559517ba5e76e99b83b99d39 100644
--- a/App/src/TestIsGISAXS13.cpp
+++ b/App/src/TestIsGISAXS13.cpp
@@ -28,6 +28,7 @@
#include "ExperimentConstants.h"
#include "OutputDataIOFactory.h"
#include "IsGISAXSData.h"
+#include "MinimizerScan.h"
#include <iostream>
#include <fstream>
@@ -80,20 +81,24 @@ void TestIsGISAXS13::run_isgisaxs_fit()
//m_fitSuite->setMinimizer( new ROOTMinimizer("GSLMultiFit", "") );
//m_fitSuite->setMinimizer( new ROOTMinimizer("Fumili", "") );
- m_fitSuite->setMinimizer( new ROOTMinimizer("GSLSimAn") );
-
- m_fitSuite->attachObserver( new FitSuiteObserverPrint(10) );
- m_fitSuite->attachObserver( new FitSuiteObserverDraw(10) );
- ROOT::Math::Minimizer *minim = (dynamic_cast<ROOTMinimizer *>(m_fitSuite->getMinimizer()))->getROOTMinimizer();
- minim->SetPrintLevel(4);
- minim->SetMaxIterations(400);
- minim->SetMaxFunctionCalls(1000);
- minim->SetTolerance(0.1);
+// m_fitSuite->setMinimizer( new ROOTMinimizer("GSLSimAn") );
+// ROOT::Math::Minimizer *minim = (dynamic_cast<ROOTMinimizer *>(m_fitSuite->getMinimizer()))->getROOTMinimizer();
+// minim->SetPrintLevel(4);
+// minim->SetMaxIterations(400);
+// minim->SetMaxFunctionCalls(1000);
+// minim->SetTolerance(0.1);
//minim->SetPrecision(0.004);
// minim->SetStrategy(1);
// minim->SetPrecision(1.);
+ m_fitSuite->setMinimizer( new MinimizerScan(20) );
+
+
+// m_fitSuite->attachObserver( new FitSuiteObserverPrint(10) );
+// m_fitSuite->attachObserver( new FitSuiteObserverDraw(10) );
+
+
// Migrad
// 0 *Normalizer/scale 1.333150e+05 7.316920e+04 6.781978e-01
// 1 *Normalizer/shift 1.212894e+00 1.469154e+01 2.140932e-01
@@ -125,9 +130,10 @@ void TestIsGISAXS13::run_isgisaxs_fit()
// m_fitSuite->addFitParameter("*Normalizer/shift", 10, 10, AttLimits::limited(0., 20.));
m_fitSuite->addFitParameter("*Normalizer/scale", 1.333150e+05, 1e5, AttLimits::fixed());
m_fitSuite->addFitParameter("*Normalizer/shift", 1.212894e+00, 10, AttLimits::fixed());
- m_fitSuite->addFitParameter("*SampleBuilder/particle_radius", 4.0e+00*Units::nanometer, 0.01*Units::nanometer, AttLimits::limited(1., 10.) );
+ m_fitSuite->addFitParameter("*SampleBuilder/particle_radius", 4.0e+00*Units::nanometer, 0.01*Units::nanometer, AttLimits::limited(2.0, 8.0) );
m_fitSuite->addFitParameter("*SampleBuilder/dispersion_radius", 1.991729e-01, 0.2, AttLimits::fixed() );
- m_fitSuite->addFitParameter("*SampleBuilder/height_aspect_ratio", 9.937262e-01, 1.0, AttLimits::fixed() );
+ //m_fitSuite->addFitParameter("*SampleBuilder/height_aspect_ratio", 9.937262e-01, 1.0, AttLimits::fixed() );
+ m_fitSuite->addFitParameter("*SampleBuilder/height_aspect_ratio", 9.937262e-01, 1.0, AttLimits::limited(0.5, 1.5) );
m_fitSuite->addFitParameter("*SampleBuilder/interf_distance", 1.497518e+01*Units::nanometer, 15.*Units::nanometer, AttLimits::fixed() );
m_fitSuite->addFitParameter("*SampleBuilder/interf_width", 3.023607e+00*Units::nanometer, 3.0*Units::nanometer, AttLimits::fixed() );
@@ -145,51 +151,73 @@ void TestIsGISAXS13::run_isgisaxs_fit()
m_fitSuite->runFit();
- // drawing results
- TCanvas *c2 = new TCanvas("c2","GISASFW fit results",800,600);
- c2->Divide(2,2);
- TLegend *leg1 = new TLegend(0.5,0.6,0.85,0.85);
- leg1->SetBorderSize(1);
- leg1->SetFillStyle(0);
- for(size_t i_set=0; i_set<m_fitSuite->getFitObjects()->size(); ++i_set) {
- c2->cd((int)i_set+1);
- const FitObject *obj = m_fitSuite->getFitObjects()->getObject(i_set);
- TH1D *hreal = IsGISAXSTools::getOutputDataScanHist(*obj->getChiSquaredModule()->getRealData(),"gisasfw_real");
- TH1D *hsimul = IsGISAXSTools::getOutputDataScanHist(*obj->getChiSquaredModule()->getSimulationData(),"gisasfw_simul");
- hreal->SetLineColor(kBlue);
- gPad->SetLogy();
- hreal->DrawCopy();
- hsimul->DrawCopy("same");
- if(i_set==0) leg1->AddEntry(hreal,"GISASFW data","lp");
- if(i_set==0) leg1->AddEntry(hsimul,"GISASFW simul","lp");
+ TCanvas *c1 = new TCanvas("c1","c1",1024,768);
+ c1->cd();
+
+ MinimizerScan *minim = dynamic_cast<MinimizerScan *>(m_fitSuite->getMinimizer());
+ std::cout << "Results " << minim->getMinValue() << std::endl;
+ for(size_t i=0; i<minim->getNumberOfVariables(); ++i) {
+ std::cout << i << " " << minim->getValueOfVariableAtMinimum(i) << std::endl;
}
- c2->cd(1); leg1->Draw();
- c2->cd(2); leg1->Draw();
-
- // drawing ratio
- TLegend *leg2 = new TLegend(0.5,0.6,0.85,0.85);
- leg2->SetBorderSize(1);
- leg2->SetFillStyle(0);
- for(size_t i_set=0; i_set<m_fitSuite->getFitObjects()->size(); ++i_set) {
- c2->cd(3+1);
- const FitObject *obj = m_fitSuite->getFitObjects()->getObject(i_set);
- OutputData<double > *real = obj->getChiSquaredModule()->getRealData()->clone();
- OutputData<double > *simul = obj->getChiSquaredModule()->getSimulationData()->clone();
-
- c2->cd((int)(i_set+3));
- *simul /= *real;
- TH1D *hratio = IsGISAXSTools::getOutputDataScanHist(*simul,"gisasfw_real_simul_ratio");
- hratio->DrawCopy();
- if(i_set==0) {
- leg2->AddEntry(hratio,"GISASFW simul/real","lp");
- }
- delete real;
- delete simul;
+
+ const OutputData<double> *data = minim->getOutputData();
+ if(data->getNdimensions()==1) {
+ TH1 * h1 = IsGISAXSTools::getOutputDataTH123D(*data,"hist");
+ h1->Draw();
+ gPad->SetLogy();
+ } else {
+ TH2D *hist = IsGISAXSTools::getOutputDataTH2D( *data, "hist");
+ hist->Draw("colz");
+ gPad->SetLogz();
}
- c2->cd(3); leg2->Draw();
- c2->cd(4); leg2->Draw();
- c2->Update();
+
+
+// // drawing results
+// TCanvas *c2 = new TCanvas("c2","GISASFW fit results",800,600);
+// c2->Divide(2,2);
+// TLegend *leg1 = new TLegend(0.5,0.6,0.85,0.85);
+// leg1->SetBorderSize(1);
+// leg1->SetFillStyle(0);
+// for(size_t i_set=0; i_set<m_fitSuite->getFitObjects()->size(); ++i_set) {
+// c2->cd((int)i_set+1);
+// const FitObject *obj = m_fitSuite->getFitObjects()->getObject(i_set);
+// TH1D *hreal = IsGISAXSTools::getOutputDataScanHist(*obj->getChiSquaredModule()->getRealData(),"gisasfw_real");
+// TH1D *hsimul = IsGISAXSTools::getOutputDataScanHist(*obj->getChiSquaredModule()->getSimulationData(),"gisasfw_simul");
+// hreal->SetLineColor(kBlue);
+// gPad->SetLogy();
+// hreal->DrawCopy();
+// hsimul->DrawCopy("same");
+// if(i_set==0) leg1->AddEntry(hreal,"GISASFW data","lp");
+// if(i_set==0) leg1->AddEntry(hsimul,"GISASFW simul","lp");
+// }
+// c2->cd(1); leg1->Draw();
+// c2->cd(2); leg1->Draw();
+
+// // drawing ratio
+// TLegend *leg2 = new TLegend(0.5,0.6,0.85,0.85);
+// leg2->SetBorderSize(1);
+// leg2->SetFillStyle(0);
+// for(size_t i_set=0; i_set<m_fitSuite->getFitObjects()->size(); ++i_set) {
+// c2->cd(3+1);
+// const FitObject *obj = m_fitSuite->getFitObjects()->getObject(i_set);
+// OutputData<double > *real = obj->getChiSquaredModule()->getRealData()->clone();
+// OutputData<double > *simul = obj->getChiSquaredModule()->getSimulationData()->clone();
+
+// c2->cd((int)(i_set+3));
+// *simul /= *real;
+// TH1D *hratio = IsGISAXSTools::getOutputDataScanHist(*simul,"gisasfw_real_simul_ratio");
+// hratio->DrawCopy();
+// if(i_set==0) {
+// leg2->AddEntry(hratio,"GISASFW simul/real","lp");
+// }
+// delete real;
+// delete simul;
+// }
+// c2->cd(3); leg2->Draw();
+// c2->cd(4); leg2->Draw();
+
+// c2->Update();
}
diff --git a/Core/Core.pro b/Core/Core.pro
index aa706e4fc4543516ce439c4b4d22ff3eaded24e5..bc1368efd575c5213f35bae9cdc82607c2e25295 100644
--- a/Core/Core.pro
+++ b/Core/Core.pro
@@ -51,6 +51,8 @@ SOURCES += \
Fitting/src/FitSuite.cpp \
Fitting/src/FitSuiteFunctions.cpp \
Fitting/src/FitSuiteParameters.cpp \
+ Fitting/src/MinimizerTest.cpp \
+ Fitting/src/MinimizerScan.cpp \
\
FormFactors/src/FormFactorBox.cpp \
FormFactors/src/FormFactorCrystal.cpp \
@@ -175,6 +177,8 @@ HEADERS += \
Fitting/inc/FitSuiteParameters.h \
Fitting/inc/FitSuiteStrategy.h \
Fitting/inc/IMinimizer.h \
+ Fitting/inc/MinimizerTest.h \
+ Fitting/inc/MinimizerScan.h \
\
FormFactors/inc/FormFactorBox.h \
FormFactors/inc/FormFactorCrystal.h \
diff --git a/Core/Fitting/inc/FitParameterLinked.h b/Core/Fitting/inc/FitParameterLinked.h
index 2482ed781f031bc3c3257a379ae50c919c7f2849..cf3758db22965723ced5d9cf75a72651a2f171c1 100644
--- a/Core/Fitting/inc/FitParameterLinked.h
+++ b/Core/Fitting/inc/FitParameterLinked.h
@@ -30,18 +30,17 @@
class FitParameterLinked : public FitParameter
{
public:
- typedef std::vector<ParameterPool::parameter_t > PoolParameterColl_t;
+ typedef std::vector<ParameterPool::parameter_t > pool_parameters_t;
FitParameterLinked();
FitParameterLinked(const std::string &name, double value, double step, const AttLimits &attlim=AttLimits::limitless(), double error=0.0);
virtual ~FitParameterLinked(){}
//! set given value for all binded parameters
- virtual void setValue(double value)
- {
+ virtual void setValue(double value) {
FitParameter::setValue(value);
- for(PoolParameterColl_t::iterator it=m_parametercoll.begin(); it!=m_parametercoll.end(); ++it) {
- (*it).setValue(value); // setting value for all registered parameters
+ for(pool_parameters_t::iterator it=m_pool_parameters.begin(); it!=m_pool_parameters.end(); ++it) {
+ (*it).setValue(value);
}
}
@@ -55,14 +54,14 @@ public:
friend std::ostream &operator<<(std::ostream &ostr, const FitParameterLinked &m) { m.print(ostr); return ostr; }
protected:
- // disabled copy constructor and assignment operator
- FitParameterLinked(const FitParameterLinked &other);
- FitParameterLinked &operator=(const FitParameterLinked &other);
-
//! print class
void print(std::ostream &ostr) const;
- PoolParameterColl_t m_parametercoll; //! collection of parameters from parameter pools
+ pool_parameters_t m_pool_parameters; //! collection of parameters from parameter pools
+
+private:
+ FitParameterLinked(const FitParameterLinked &);
+ FitParameterLinked &operator=(const FitParameterLinked &);
};
#endif // FITMULTIPARAMETER_H
diff --git a/Core/Fitting/inc/FitSuite.h b/Core/Fitting/inc/FitSuite.h
index 09376b26397672134c198978e654e1804e67ca88..fc08eaa812238b8dd5ceaea083869e5c1a9df56e 100644
--- a/Core/Fitting/inc/FitSuite.h
+++ b/Core/Fitting/inc/FitSuite.h
@@ -70,12 +70,6 @@ public:
//! run fitting which may consist of several minimization rounds
virtual void runFit();
- //! function to minimize
- //double fittingChiSquaredFunction(const double *pars_current_values);
-
- //! provides minimizer with gradients wrt parameters for single data element
- //double fittingGradientFunction(const double *pars_current_values, unsigned int index, double *deriv);
-
//! return reference to the kit with data
FitSuiteObjects *getFitObjects() { return &m_fit_objects; }
diff --git a/Core/Fitting/inc/FitSuiteParameters.h b/Core/Fitting/inc/FitSuiteParameters.h
index d1bb7407912361534ea1b3a6167462012ceb0bf8..7099e46de481888e31fb3efd2d099c44978e5f13 100644
--- a/Core/Fitting/inc/FitSuiteParameters.h
+++ b/Core/Fitting/inc/FitSuiteParameters.h
@@ -15,28 +15,30 @@
//! @date 15.11.2012
#include "Exceptions.h"
+#include "SafePointerVector.h"
#include "FitParameterLinked.h"
#include <vector>
//class Experiment;
class ParameterPool;
+
//- -------------------------------------------------------------------
//! @class FitSuiteParameters
//! @brief Class holds vector of parameters for FitSuite
//- -------------------------------------------------------------------
-class FitSuiteParameters
+class FitSuiteParameters : public SafePointerVector<FitParameter >
{
public:
- typedef std::vector<FitParameter *> parameters_t;
- typedef parameters_t::iterator iterator;
- typedef parameters_t::const_iterator const_iterator;
+// typedef std::vector<FitParameter *> parameters_t;
+// typedef parameters_t::iterator iterator;
+// typedef parameters_t::const_iterator const_iterator;
- FitSuiteParameters();
- virtual ~FitSuiteParameters();
+ FitSuiteParameters() {}
+ virtual ~FitSuiteParameters(){}
- //! clear all defined parameters
- void clear();
+// //! clear all defined parameters
+// void clear();
//! add fit parameter
void addParameter(const std::string &name, double value, double step, const AttLimits &attlim, double error=0.0);
@@ -52,22 +54,22 @@ public:
//! get values of all defined parameters
std::vector<double > getValues() const;
- //! return number of parameters
- size_t size() const { return m_parameters.size(); }
+// //! return number of parameters
+// size_t size() const { return m_parameters.size(); }
- //! return begin of container
- iterator begin() { return m_parameters.begin(); }
- const_iterator begin() const { return m_parameters.begin(); }
+// //! return begin of container
+// iterator begin() { return m_parameters.begin(); }
+// const_iterator begin() const { return m_parameters.begin(); }
- //! return end of container
- iterator end() { return m_parameters.end(); }
- const_iterator end() const { return m_parameters.end(); }
+// //! return end of container
+// iterator end() { return m_parameters.end(); }
+// const_iterator end() const { return m_parameters.end(); }
- //! access to parameters
- const FitParameter *operator[](size_t index) const { return m_parameters[check_index(index)]; }
- FitParameter *operator[](size_t index) { return m_parameters[check_index(index)]; }
- const FitParameter *operator[](std::string name) const { return getParameter(name); }
- FitParameter *operator[](std::string name) { return getParameter(name); }
+// //! access to parameters
+// const FitParameter *operator[](size_t index) const { return m_parameters[check_index(index)]; }
+// FitParameter *operator[](size_t index) { return m_parameters[check_index(index)]; }
+// const FitParameter *operator[](std::string name) const { return getParameter(name); }
+// FitParameter *operator[](std::string name) { return getParameter(name); }
//! linking fit parameters with pool parameters
void link_to_pool(const ParameterPool *pool);
@@ -80,13 +82,13 @@ public:
private:
//! disabled copy constructor and assignment operator
- FitSuiteParameters &operator=(const FitSuiteParameters &other);
- FitSuiteParameters(const FitSuiteParameters &other);
+// FitSuiteParameters &operator=(const FitSuiteParameters &);
+// FitSuiteParameters(const FitSuiteParameters &);
- inline size_t check_index(size_t index) const { return (index < m_parameters.size() ? index : throw OutOfBoundsException("FitSuiteParameters::check_index() -> Index out of bounds") ); }
- parameters_t m_parameters; //! collection of fit parameters
+// inline size_t check_index(size_t index) const { return (index < m_parameters.size() ? index : throw OutOfBoundsException("FitSuiteParameters::check_index() -> Index out of bounds") ); }
+// parameters_t m_parameters; //! collection of fit parameters
- static double m_default_parameter_error;
+// static double m_default_parameter_error;
};
#endif // FITSUITEPARAMETERS_H
diff --git a/Core/Fitting/inc/IMinimizer.h b/Core/Fitting/inc/IMinimizer.h
index 885a8ff8d810c14fd16cb2a25d41f0b1032e2f5a..4b151e72ff312cc938131bdde3f526e87d042b8a 100644
--- a/Core/Fitting/inc/IMinimizer.h
+++ b/Core/Fitting/inc/IMinimizer.h
@@ -14,13 +14,12 @@
//! @author Scientific Computing Group at FRM II
//! @date 05.10.2012
-
-#include "FitParameter.h"
-#include "FitSuiteParameters.h"
#include <boost/function.hpp>
-#include <map>
#include "Exceptions.h"
+class FitParameter;
+class FitSuiteParameters;
+
//- -------------------------------------------------------------------
//! @class IMinimizer
@@ -29,107 +28,108 @@
class IMinimizer
{
public:
- //! signature of function to minimize
+ //! signature of chi squared function to minimize
typedef boost::function<double(const double *)> function_chi2_t;
- //! signature of function to minimize with acess to single element residual and gradient
- typedef boost::function<double(const double *, unsigned int, double *)> function_gradient_t;
- IMinimizer(){}
- virtual ~IMinimizer(){}
-
- //! set parameter
- virtual void setParameter(size_t index, const FitParameter *par) = 0;
- virtual void setParameters(const FitSuiteParameters ¶meters) = 0;
+ //! signature of gradient to minimize with acess to single data element residuals
+ typedef boost::function<double(const double *, unsigned int, double *)> function_gradient_t;
- //! set function to minimize
- virtual void setFunction(function_chi2_t fun_chi2, size_t nparameters, function_gradient_t fun_gradient = function_gradient_t(), size_t ndatasize = 0) = 0;
+ IMinimizer() { }
+ virtual ~IMinimizer() { }
//! run minimization
virtual void minimize() = 0;
+ //! set internal minimizer parameter
+ virtual void setParameter(size_t index, const FitParameter *par);
+
+ //! set internal minimizer parameters using external parameter list
+ virtual void setParameters(const FitSuiteParameters ¶meters);
+
+ //! set chi squared function to minimize
+ virtual void setChiSquaredFunction(function_chi2_t fun_chi2, size_t nparameters);
+
+ //! set gradient function to minimize
+ virtual void setGradientFunction(function_gradient_t fun_gradient, size_t nparameters, size_t ndatasize);
+
//! get number of variables to fit
- virtual size_t getNumberOfVariables() const = 0;
+ virtual size_t getNumberOfVariables() const;
//! return minimum function value
- virtual double getMinValue() const = 0;
+ virtual double getMinValue() const;
//! return pointer to the parameters values at the minimum
- virtual double getValueOfVariableAtMinimum(size_t i) const = 0;
+ virtual double getValueOfVariableAtMinimum(size_t i) const;
//! return pointer to the parameters values at the minimum
- virtual double getErrorOfVariable(size_t i) const = 0;
+ virtual double getErrorOfVariable(size_t i) const;
//! clear resources (parameters) for consecutives minimizations
- virtual void clear() = 0;
+ virtual void clear();
//! print fit results
- virtual void printResults() const = 0;
-
+ virtual void printResults() const;
};
-
-//- -------------------------------------------------------------------
-//! @class TestMinimizer
-//! @brief Minimizer which calls minimization function once to test whole chain
-//- -------------------------------------------------------------------
-class TestMinimizer : public IMinimizer
+inline void IMinimizer::setParameter(size_t index, const FitParameter *par)
{
-public:
- TestMinimizer(){}
- virtual ~TestMinimizer(){}
-
- //! set variable
- virtual void setParameter(size_t index, const FitParameter *par) { m_values[index] = par->getValue(); }
- virtual void setParameters(const FitSuiteParameters &/*parameters */) { throw NotImplementedException("TestMinimizer::setParameters() -> Error! Not implemented."); }
-
- //! set function to minimize
- virtual void setFunction(function_chi2_t fun_chi2, size_t /* nparameters */, function_gradient_t /* fun_gradient */, size_t /* ndatasize */ ) { m_fcn = fun_chi2; }
-
- //! run minimization
- virtual void minimize()
- {
- std::vector<double > buffer;
- buffer.resize(m_values.size(), 0.0);
- for(std::map<int, double >::iterator it=m_values.begin(); it!= m_values.end(); ++it ) {
- buffer[it->first] = it->second;
- std::cout << " minimize(): " << it->first << " " << it->second << std::endl;
- }
- std::cout << "TestMinimizer::minimize() -> Info. Calling fcn" << std::endl;
- m_fcn(&buffer[0]);
- }
+ (void)index;
+ (void)par;
+ throw NotImplementedException("IMinimizer::setParameter() -> Not implemented.");
+}
- //! get number of variables to fit
- virtual size_t getNumberOfVariables() const { return m_values.size(); }
+inline void IMinimizer::setParameters(const FitSuiteParameters ¶meters)
+{
+ (void)parameters;
+ throw NotImplementedException("IMinimizer::setParameters() -> Not implemented.");
+}
- //! return minimum function value
- virtual double getMinValue() const { throw NotImplementedException("TestMinimizer::getMinValue() -> Not implemented. "); return 0.0; }
+inline void IMinimizer::setChiSquaredFunction(function_chi2_t fun_chi2, size_t nparameters)
+{
+ (void)fun_chi2;
+ (void)nparameters;
+ throw NotImplementedException("IMinimizer::setChiSquaredFunction() -> Not implemented.");
+}
- //! return pointer to the parameters values at the minimum
- virtual double getValueOfVariableAtMinimum(size_t i) const
- {
- std::map<int, double >::const_iterator pos = m_values.find((int)i);
- if(pos != m_values.end()){
- return pos->second;
- } else {
- throw LogicErrorException("TestMinimizer::getValueOfVariableAtMinimum() -> Not found!");
- }
- }
+inline void IMinimizer::setGradientFunction(function_gradient_t fun_gradient, size_t nparameters, size_t ndatasize)
+{
+ (void)fun_gradient;
+ (void)nparameters;
+ (void)ndatasize;
+ throw NotImplementedException("IMinimizer::setGradientFunction() -> Not implemented.");
+}
- //! return pointer to the parameters values at the minimum
- virtual double getErrorOfVariable(size_t /* i*/) const { throw NotImplementedException("TestMinimizer::getMinValue() -> Not implemented. "); return 0.0; }
+inline size_t IMinimizer::getNumberOfVariables() const
+{
+ throw NotImplementedException("IMinimizer::getNumberOfVariables() -> Not implemented.");
+}
- //! clear resources (parameters) for consecutives minimizations
- virtual void clear() { throw NotImplementedException("TestMinimizer::getMinValue() -> Not implemented. "); }
+inline double IMinimizer::getMinValue() const
+{
+ throw NotImplementedException("IMinimizer::getMinValue() -> Not implemented.");
+}
- //! print fit results
- virtual void printResults() const { throw NotImplementedException("TestMinimizer::getMinValue() -> Not implemented. "); }
+inline double IMinimizer::getValueOfVariableAtMinimum(size_t index) const
+{
+ (void)index;
+ throw NotImplementedException("IMinimizer::getValueOfVariableAtMinimum() -> Not implemented.");
+}
-private:
- std::map<int, double > m_values;
- function_chi2_t m_fcn;
-};
+inline double IMinimizer::getErrorOfVariable(size_t index) const
+{
+ (void)index;
+ throw NotImplementedException("IMinimizer::getErrorOfVariable() -> Not implemented.");
+}
+inline void IMinimizer::clear()
+{
+ throw NotImplementedException("IMinimizer::clear() -> Not implemented.");
+}
+inline void IMinimizer::printResults() const
+{
+ throw NotImplementedException("IMinimizer::printResults() -> Not implemented.");
+}
#endif // IMINIMIZER_H
diff --git a/Core/Fitting/inc/MinimizerScan.h b/Core/Fitting/inc/MinimizerScan.h
new file mode 100644
index 0000000000000000000000000000000000000000..dcc455c462d49d7394bd5b71b519c90ccc28560c
--- /dev/null
+++ b/Core/Fitting/inc/MinimizerScan.h
@@ -0,0 +1,68 @@
+#ifndef MINIMIZERSCAN_H
+#define MINIMIZERSCAN_H
+// ********************************************************************
+// * The BornAgain project *
+// * Simulation of neutron and x-ray scattering at grazing incidence *
+// * *
+// * LICENSE AND DISCLAIMER *
+// * Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris *
+// * eget quam orci. Quisque porta varius dui, quis posuere nibh *
+// * mollis quis. Mauris commodo rhoncus porttitor. *
+// ********************************************************************
+//! @file MinimizerScan.h
+//! @brief Definition of MinimizerScan class
+//! @author Scientific Computing Group at FRM II
+//! @date 21.01.2013
+
+
+#include "IMinimizer.h"
+#include "FitSuiteParameters.h"
+#include "SafePointerVector.h"
+#include "OutputData.h"
+
+//- -------------------------------------------------------------------
+//! @class MinimizerScan
+//! @brief Simple scan minimizer
+//- -------------------------------------------------------------------
+class MinimizerScan : public IMinimizer
+{
+public:
+ MinimizerScan(int nbins = 10) : m_parameter_map(0), m_nbins(nbins) { }
+ virtual ~MinimizerScan() { delete m_parameter_map; }
+
+ virtual void minimize();
+
+ virtual void setParameters(const FitSuiteParameters ¶meters);
+
+ virtual void setChiSquaredFunction(function_chi2_t fun_chi2, size_t nparameters);
+
+ virtual void setGradientFunction(function_gradient_t fun_gradient, size_t nparameters, size_t ndatasize);
+
+ virtual size_t getNumberOfVariables() const { return m_fit_parameters.size(); }
+
+ virtual double getMinValue() const;
+
+ virtual double getValueOfVariableAtMinimum(size_t i) const;
+
+ virtual void printResults() const;
+
+
+ void setNbins(int nbins) { m_nbins = nbins; }
+ size_t getNbins() const { return m_nbins; }
+
+ const OutputData<double > *getOutputData() { return m_parameter_map; }
+
+private:
+
+ void construct_parameter_map();
+ void set_parvalues_to_minimum();
+
+ OutputData<double > *m_parameter_map;
+ size_t m_nbins; //! number of bins per one parameter
+ //SafePointerVector<FitParameter > m_fit_parameters;
+ FitSuiteParameters m_fit_parameters;
+ std::vector<double> m_parvalues_at_minimum;
+ function_chi2_t m_fcn;
+};
+
+#endif // MINIMIZERSCAN_H
diff --git a/Core/Fitting/inc/MinimizerTest.h b/Core/Fitting/inc/MinimizerTest.h
new file mode 100644
index 0000000000000000000000000000000000000000..8bfbe325d9bc6f0c1bb90e95c3af58817a94f438
--- /dev/null
+++ b/Core/Fitting/inc/MinimizerTest.h
@@ -0,0 +1,53 @@
+#ifndef MINIMIZERTEST_H
+#define MINIMIZERTEST_H
+// ********************************************************************
+// * The BornAgain project *
+// * Simulation of neutron and x-ray scattering at grazing incidence *
+// * *
+// * LICENSE AND DISCLAIMER *
+// * Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris *
+// * eget quam orci. Quisque porta varius dui, quis posuere nibh *
+// * mollis quis. Mauris commodo rhoncus porttitor. *
+// ********************************************************************
+//! @file MinimizerTest.h
+//! @brief Definition of MinimizerTest class
+//! @author Scientific Computing Group at FRM II
+//! @date 21.01.2013
+
+#include "IMinimizer.h"
+#include "FitSuiteParameters.h"
+#include <map>
+
+
+//- -------------------------------------------------------------------
+//! @class MinimizerTest
+//! @brief Minimizer which calls minimization function once to test whole chain
+//- -------------------------------------------------------------------
+class MinimizerTest : public IMinimizer
+{
+public:
+ MinimizerTest(){}
+ virtual ~MinimizerTest(){}
+
+ //! run minimization
+ virtual void minimize();
+
+ //! set variable
+ virtual void setParameter(size_t index, const FitParameter *par) { m_values[index] = par->getValue(); }
+
+ //! set function to minimize
+ virtual void setChiSquaredFunction(function_chi2_t fun_chi2, size_t nparameters) { (void)nparameters, m_fcn = fun_chi2; }
+
+ //! get number of variables to fit
+ virtual size_t getNumberOfVariables() const { return m_values.size(); }
+
+ //! return pointer to the parameters values at the minimum
+ virtual double getValueOfVariableAtMinimum(size_t index) const;
+
+private:
+ std::map<int, double > m_values;
+ function_chi2_t m_fcn;
+};
+
+#endif // MINIMIZERTEST_H
+
diff --git a/Core/Fitting/src/FitParameterLinked.cpp b/Core/Fitting/src/FitParameterLinked.cpp
index 27d4382e781949cf97b8d812f482aacae1cba88e..ad91792ea0b539be94db1d6a473b798a7cbe86f6 100644
--- a/Core/Fitting/src/FitParameterLinked.cpp
+++ b/Core/Fitting/src/FitParameterLinked.cpp
@@ -20,7 +20,7 @@ FitParameterLinked::FitParameterLinked(const std::string &name, double value, do
void FitParameterLinked::addParameter(ParameterPool::parameter_t par)
{
if( !par.isNull() ) {
- m_parametercoll.push_back(par);
+ m_pool_parameters.push_back(par);
} else {
throw LogicErrorException("FitMultiParameter::addParameter() -> Attempt to add null parameter");
}
@@ -35,8 +35,8 @@ void FitParameterLinked::addMatchedParametersFromPool(const ParameterPool *pool,
std::string wildcard_to_use = getName();
if( !wildcard.empty()) wildcard_to_use = wildcard;
- PoolParameterColl_t matched_pars = pool->getMatchedParameters(wildcard_to_use);
- m_parametercoll.insert(m_parametercoll.end(), matched_pars.begin(), matched_pars.end());
+ pool_parameters_t matched_pars = pool->getMatchedParameters(wildcard_to_use);
+ m_pool_parameters.insert(m_pool_parameters.end(), matched_pars.begin(), matched_pars.end());
if( matched_pars.empty() ) {
throw LogicErrorException("FitMultiParameter::addMatchedParametersFromPool() -> Error! Failed to add anything from pool using wildcard '"+wildcard_to_use+"'");
@@ -48,7 +48,7 @@ void FitParameterLinked::addMatchedParametersFromPool(const ParameterPool *pool,
void FitParameterLinked::print(std::ostream &ostr) const
{
FitParameter::print(ostr);
- ostr << "FitParameterLinked '" << getName() << "'" << " value:" << m_value << " collsize:" << m_parametercoll.size();
+ ostr << "FitParameterLinked '" << getName() << "'" << " value:" << m_value << " collsize:" << m_pool_parameters.size();
// if(m_parametercoll.size() ) {
// ostr << " addresses: ";
// for(parametercoll_t::const_iterator it=m_parametercoll.begin(); it!=m_parametercoll.end(); it++) {
diff --git a/Core/Fitting/src/FitSuite.cpp b/Core/Fitting/src/FitSuite.cpp
index f5913009fd14b9f85028b3dbcd5d4881d399dddc..2caf51c5c107e36bf31f93b2c61898696dc93bb0 100644
--- a/Core/Fitting/src/FitSuite.cpp
+++ b/Core/Fitting/src/FitSuite.cpp
@@ -86,13 +86,11 @@ void FitSuite::link_fit_parameters()
void FitSuite::minimize()
{
// initializing minimizer with fitting functions
- //IMinimizer::function_chi2_t fun_chi2 = boost::bind(&FitSuite::fittingChiSquaredFunction, this, _1);
- //IMinimizer::function_gradient_t fun_gradient = boost::bind(&FitSuite::fittingGradientFunction, this, _1, _2, _3);
-
IMinimizer::function_chi2_t fun_chi2 = boost::bind(&FitSuiteChiSquaredFunction::evaluate, &m_function_chi2, _1);
- IMinimizer::function_gradient_t fun_gradient = boost::bind(&FitSuiteGradientFunction::evaluate, &m_function_gradient, _1, _2, _3);
+ m_minimizer->setChiSquaredFunction( fun_chi2, m_fit_parameters.size());
- m_minimizer->setFunction( fun_chi2, m_fit_parameters.size(), fun_gradient, m_fit_objects.getSizeOfDataSet() );
+ IMinimizer::function_gradient_t fun_gradient = boost::bind(&FitSuiteGradientFunction::evaluate, &m_function_gradient, _1, _2, _3);
+ m_minimizer->setGradientFunction( fun_gradient, m_fit_parameters.size(), m_fit_objects.getSizeOfDataSet() );
// initializing minimizer's parameters with the list of local fit parameters
m_minimizer->setParameters(m_fit_parameters);
@@ -108,7 +106,7 @@ void FitSuite::minimize()
bool FitSuite::check_prerequisites() const
{
if( !m_minimizer ) throw LogicErrorException("FitSuite::check_prerequisites() -> Error! No minimizer found.");
- if( !m_fit_objects.size() ) throw LogicErrorException("FitSuite::check_prerequisites() -> Error! No experiment defined");
+ if( !m_fit_objects.size() ) throw LogicErrorException("FitSuite::check_prerequisites() -> Error! No experiment/data description defined");
if( !m_fit_parameters.size() ) throw LogicErrorException("FitSuite::check_prerequisites() -> Error! No fit parameters defined");
return true;
}
@@ -149,38 +147,3 @@ void FitSuite::runFit()
}
-/* ************************************************************************* */
-// function to minimize
-/* ************************************************************************* */
-//double FitSuite::fittingChiSquaredFunction(const double *pars_current_values)
-//{
-// std::cout << "FitSuite::functionToMinimize() -> Info" << std::endl;
-// // set fitting parameters to values suggested by the minimizer
-// m_fit_parameters.setValues(pars_current_values);
-
-// // run simulations
-// m_fit_objects.runSimulation();
-
-// // caclulate chi2 value
-// double chi_squared = m_fit_objects.getChiSquaredValue((int)m_fit_parameters.getNfreeParameters());
-
-// notifyObservers();
-// m_n_call++;
-// return chi_squared;
-//}
-
-
-/* ************************************************************************* */
-// provides minimizer with gradients wrt parameters for single data element
-/* ************************************************************************* */
-//double FitSuite::fittingGradientFunction(const double *pars_current_values, unsigned int index, double *deriv)
-//{
-// throw 1;
-// (void)pars_current_values;
-// (void) deriv;
-// double residual = m_fit_objects.getResidualValue(index);
-// return residual;
-//}
-
-
-
diff --git a/Core/Fitting/src/FitSuiteParameters.cpp b/Core/Fitting/src/FitSuiteParameters.cpp
index ee5a764be166d8136a76604650c38c71c3be5418..ef5e90d49d743bf76c903d5878481d17e1daca92 100644
--- a/Core/Fitting/src/FitSuiteParameters.cpp
+++ b/Core/Fitting/src/FitSuiteParameters.cpp
@@ -2,83 +2,136 @@
#include "Experiment.h"
-double FitSuiteParameters::m_default_parameter_error=0.001;
+//double FitSuiteParameters::m_default_parameter_error=0.001;
-FitSuiteParameters::FitSuiteParameters()
-{
-}
+//FitSuiteParameters::FitSuiteParameters()
+//{
+//}
-FitSuiteParameters::~FitSuiteParameters()
-{
- clear();
-}
+//FitSuiteParameters::~FitSuiteParameters()
+//{
+// clear();
+//}
// clear all defined parameters
-void FitSuiteParameters::clear()
-{
- for(parameters_t::iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it) {
- delete (*it);
- }
- m_parameters.clear();
-}
+//void FitSuiteParameters::clear()
+//{
+//// for(parameters_t::iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it) {
+// for(parameters_t::iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it) {
+// delete (*it);
+// }
+// m_parameters.clear();
+//}
// add fit parameter
+//void FitSuiteParameters::addParameter(const std::string &name, double value, double step, const AttLimits &attlim, double error)
+//{
+// for(parameters_t::const_iterator it = m_parameters.begin(); it!=m_parameters.end(); ++it) {
+// if( (*it)->getName() == name ) throw LogicErrorException("FitSuiteParameters:addtFitParameter() -> Error. Existing parameter '"+name+"'");
+// }
+// // defining default parameter error
+//// if(error == 0.0) error = value*m_default_parameter_error;
+// m_parameters.push_back(new FitParameterLinked(name, value, step, attlim, error) );
+//}
+
void FitSuiteParameters::addParameter(const std::string &name, double value, double step, const AttLimits &attlim, double error)
{
- for(parameters_t::const_iterator it = m_parameters.begin(); it!=m_parameters.end(); ++it) {
+ for(const_iterator it = begin(); it!=end(); ++it) {
if( (*it)->getName() == name ) throw LogicErrorException("FitSuiteParameters:addtFitParameter() -> Error. Existing parameter '"+name+"'");
}
// defining default parameter error
- if(error == 0.0) error = value*m_default_parameter_error;
- m_parameters.push_back(new FitParameterLinked(name, value, step, attlim, error) );
+// if(error == 0.0) error = value*m_default_parameter_error;
+ push_back(new FitParameterLinked(name, value, step, attlim, error) );
}
+
// return const fit parameter with given name
+//const FitParameter *FitSuiteParameters::getParameter(const std::string &name) const
+//{
+// for(parameters_t::const_iterator it = m_parameters.begin(); it!=m_parameters.end(); ++it) {
+// if( (*it)->getName() == name ) return (*it);
+// }
+// throw LogicErrorException("FitSuiteParameters::getFitParameter() -> Error. No parameter with name '"+name+"'");
+//}
const FitParameter *FitSuiteParameters::getParameter(const std::string &name) const
{
- for(parameters_t::const_iterator it = m_parameters.begin(); it!=m_parameters.end(); ++it) {
+ for(const_iterator it = begin(); it!=end(); ++it) {
if( (*it)->getName() == name ) return (*it);
}
throw LogicErrorException("FitSuiteParameters::getFitParameter() -> Error. No parameter with name '"+name+"'");
}
+
+//// return fit parameter with given name
+//FitParameter *FitSuiteParameters::getParameter(const std::string &name)
+//{
+// for(parameters_t::iterator it = m_parameters.begin(); it!=m_parameters.end(); ++it) {
+// if( (*it)->getName() == name ) return (*it);
+// }
+// throw LogicErrorException("FitSuiteParameters::getFitParameter() -> Error. No parameter with name '"+name+"'");
+//}
// return fit parameter with given name
FitParameter *FitSuiteParameters::getParameter(const std::string &name)
{
- for(parameters_t::iterator it = m_parameters.begin(); it!=m_parameters.end(); ++it) {
+ for(iterator it = begin(); it!=end(); ++it) {
if( (*it)->getName() == name ) return (*it);
}
throw LogicErrorException("FitSuiteParameters::getFitParameter() -> Error. No parameter with name '"+name+"'");
}
+
// set values for all defined parameters
// FIXME FitSuiteParameters::setValues remove check for attempt to set parameter values
+//void FitSuiteParameters::setValues(const double *pars_values)
+//{
+// if( !valuesAreDifferrent(pars_values) ) {
+// std::cout << "FitSuiteParameters::setValues() -> Warning! Small or absent change in parameter values." << std::endl;
+// for(size_t i=0; i<m_parameters.size(); i++) std::cout << (m_parameters[i]->getValue() -pars_values[i]) << " " << Numeric::areAlmostEqual(m_parameters[i]->getValue(), pars_values[i]) << std::endl;
+// }
+// size_t index(0);
+// for(parameters_t::iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it) (*it)->setValue(pars_values[index++]);
+//}
void FitSuiteParameters::setValues(const double *pars_values)
{
if( !valuesAreDifferrent(pars_values) ) {
std::cout << "FitSuiteParameters::setValues() -> Warning! Small or absent change in parameter values." << std::endl;
- for(size_t i=0; i<m_parameters.size(); i++) std::cout << (m_parameters[i]->getValue() -pars_values[i]) << " " << Numeric::areAlmostEqual(m_parameters[i]->getValue(), pars_values[i]) << std::endl;
+ for(size_t i=0; i<size(); i++) std::cout << ((*this)[i]->getValue() -pars_values[i]) << " " << Numeric::areAlmostEqual( (*this)[i]->getValue(), pars_values[i]) << std::endl;
}
size_t index(0);
- for(parameters_t::iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it) (*it)->setValue(pars_values[index++]);
+ for(iterator it=begin(); it!=end(); ++it) (*it)->setValue(pars_values[index++]);
}
-
+//void FitSuiteParameters::setValues(const std::vector<double> &pars_values)
+//{
+// if(pars_values.size() != m_parameters.size() ) throw OutOfBoundsException("FitSuiteParameters::setValues() -> Wrong size of array with parameter values");
+// setValues(&pars_values[0]);
+//}
void FitSuiteParameters::setValues(const std::vector<double> &pars_values)
{
- if(pars_values.size() != m_parameters.size() ) throw OutOfBoundsException("FitSuiteParameters::setValues() -> Wrong size of array with parameter values");
+ if(pars_values.size() != size() ) throw OutOfBoundsException("FitSuiteParameters::setValues() -> Wrong size of array with parameter values");
setValues(&pars_values[0]);
}
+
+//std::vector<double > FitSuiteParameters::getValues() const
+//{
+// std::vector<double > result;
+// for(parameters_t::const_iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it)
+// {
+// result.push_back((*it)->getValue());
+// }
+// return result;
+//}
+
std::vector<double > FitSuiteParameters::getValues() const
{
std::vector<double > result;
- for(parameters_t::const_iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it)
+ for(const_iterator it=begin(); it!=end(); ++it)
{
result.push_back((*it)->getValue());
}
@@ -86,10 +139,19 @@ std::vector<double > FitSuiteParameters::getValues() const
}
+//size_t FitSuiteParameters::getNfreeParameters() const
+//{
+// size_t result(0);
+// for(parameters_t::const_iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it)
+// {
+// if( !(*it)->isFixed() ) result++;
+// }
+// return result;
+//}
size_t FitSuiteParameters::getNfreeParameters() const
{
size_t result(0);
- for(parameters_t::const_iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it)
+ for(const_iterator it=begin(); it!=end(); ++it)
{
if( !(*it)->isFixed() ) result++;
}
@@ -101,11 +163,21 @@ size_t FitSuiteParameters::getNfreeParameters() const
/* ************************************************************************* */
// linking fit parameters with pool parameters
/* ************************************************************************* */
+//void FitSuiteParameters::link_to_pool(const ParameterPool *pool)
+//{
+// // linking fit parameter with whose pool parameters which match name of fit parameter
+// // going through all fit parameters defined
+// for(parameters_t::iterator it = m_parameters.begin(); it!= m_parameters.end(); ++it) {
+// FitParameterLinked *par = dynamic_cast<FitParameterLinked *>((*it));
+// if( !par ) throw LogicErrorException("FitSuiteParameters::link_to_experiment() -> Error! Can't cast to FitParameterLinked.");
+// par->addMatchedParametersFromPool(pool);
+// }
+//}
void FitSuiteParameters::link_to_pool(const ParameterPool *pool)
{
// linking fit parameter with whose pool parameters which match name of fit parameter
// going through all fit parameters defined
- for(parameters_t::iterator it = m_parameters.begin(); it!= m_parameters.end(); ++it) {
+ for(iterator it = begin(); it!= end(); ++it) {
FitParameterLinked *par = dynamic_cast<FitParameterLinked *>((*it));
if( !par ) throw LogicErrorException("FitSuiteParameters::link_to_experiment() -> Error! Can't cast to FitParameterLinked.");
par->addMatchedParametersFromPool(pool);
@@ -113,10 +185,18 @@ void FitSuiteParameters::link_to_pool(const ParameterPool *pool)
}
+//bool FitSuiteParameters::valuesAreDifferrent(const double *pars_values, double tolerance_factor) const
+//{
+// size_t index(0);
+// for(parameters_t::const_iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it) {
+// if( !Numeric::areAlmostEqual(pars_values[index++], (*it)->getValue(), tolerance_factor )) return true;
+// }
+// return false;
+//}
bool FitSuiteParameters::valuesAreDifferrent(const double *pars_values, double tolerance_factor) const
{
size_t index(0);
- for(parameters_t::const_iterator it=m_parameters.begin(); it!=m_parameters.end(); ++it) {
+ for(const_iterator it=begin(); it!=end(); ++it) {
if( !Numeric::areAlmostEqual(pars_values[index++], (*it)->getValue(), tolerance_factor )) return true;
}
return false;
diff --git a/Core/Fitting/src/MinimizerScan.cpp b/Core/Fitting/src/MinimizerScan.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..7d79340b160342e1862be81402dbbc8f22213ae7
--- /dev/null
+++ b/Core/Fitting/src/MinimizerScan.cpp
@@ -0,0 +1,101 @@
+#include "MinimizerScan.h"
+#include <algorithm>
+
+
+/* ************************************************************************* */
+//
+/* ************************************************************************* */
+void MinimizerScan::minimize()
+{
+ construct_parameter_map();
+
+ // scanning values of fit parameters
+ for(OutputData<double>::iterator it = m_parameter_map->begin(); it!=m_parameter_map->end(); ++it) {
+ for(size_t i_axis=0; i_axis<m_parameter_map->getNdimensions(); ++i_axis) {
+ size_t xbin = m_parameter_map->toCoordinate(it.getIndex(), i_axis);
+ double value = (*m_parameter_map->getAxis(i_axis))[xbin];
+ std::string parname = m_parameter_map->getAxis(i_axis)->getName();
+ m_fit_parameters.getParameter(parname)->setValue(value);
+ }
+ std::vector<double> current_values=m_fit_parameters.getValues();
+ (*it) = m_fcn(¤t_values[0]);
+ }
+
+ set_parvalues_to_minimum();
+}
+
+
+/* ************************************************************************* */
+// Construct N dimensional space over all fit parameters with lower and upper limits
+// defined.
+/* ************************************************************************* */
+void MinimizerScan::construct_parameter_map()
+{
+ delete m_parameter_map;
+ m_parameter_map = new OutputData<double>;
+ for(size_t i_par=0; i_par < m_fit_parameters.size(); i_par++ ) {
+ const FitParameter *par = m_fit_parameters[i_par];
+ if( par->hasLowerLimit() && par->hasUpperLimit() ) {
+ AxisDouble axis(par->getName(), m_nbins, par->getLowerLimit(), par->getUpperLimit());
+ m_parameter_map->addAxis(axis);
+ }
+ }
+ m_parameter_map->setAllTo(0.0);
+}
+
+
+void MinimizerScan::set_parvalues_to_minimum()
+{
+ assert(m_parameter_map);
+ OutputData<double>::iterator it = std::min_element(m_parameter_map->begin(), m_parameter_map->end());
+ for(size_t i_axis=0; i_axis<m_parameter_map->getNdimensions(); ++i_axis) {
+ size_t xbin = m_parameter_map->toCoordinate(it.getIndex(), i_axis);
+ double value = (*m_parameter_map->getAxis(i_axis))[xbin];
+ std::string parname = m_parameter_map->getAxis(i_axis)->getName();
+ m_fit_parameters.getParameter(parname)->setValue(value);
+ }
+
+}
+
+
+double MinimizerScan::getMinValue() const
+{
+ assert(m_parameter_map);
+ return *std::min_element(m_parameter_map->begin(), m_parameter_map->end());
+}
+
+
+void MinimizerScan::setGradientFunction(function_gradient_t fun_gradient, size_t nparameters, size_t ndatasize)
+{
+ (void) fun_gradient;
+ (void) nparameters;
+ (void) ndatasize;
+}
+
+
+double MinimizerScan::getValueOfVariableAtMinimum(size_t index) const
+{
+ return m_fit_parameters[index]->getValue();
+}
+
+
+void MinimizerScan::printResults() const
+{
+ std::cout << "MinimizerScan::printResults() -> Info" << std::endl;
+}
+
+
+void MinimizerScan::setChiSquaredFunction(function_chi2_t fun_chi2, size_t nparameters)
+{
+ (void)nparameters;
+ m_fcn = fun_chi2;
+}
+
+
+void MinimizerScan::setParameters(const FitSuiteParameters ¶meters)
+{
+ m_fit_parameters.clear();
+ for(size_t i_par = 0; i_par<parameters.size(); ++i_par) {
+ m_fit_parameters.push_back(new FitParameter( *parameters[i_par] ) );
+ }
+}
diff --git a/Core/Fitting/src/MinimizerTest.cpp b/Core/Fitting/src/MinimizerTest.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..8bd15a9b75c8383c682c0516bc0f72481fdf8e24
--- /dev/null
+++ b/Core/Fitting/src/MinimizerTest.cpp
@@ -0,0 +1,27 @@
+#include "MinimizerTest.h"
+
+
+// run minimization
+void MinimizerTest::minimize()
+{
+ std::vector<double > buffer;
+ buffer.resize(m_values.size(), 0.0);
+ for(std::map<int, double >::iterator it=m_values.begin(); it!= m_values.end(); ++it ) {
+ buffer[it->first] = it->second;
+ std::cout << " minimize(): " << it->first << " " << it->second << std::endl;
+ }
+ std::cout << "TestMinimizer::minimize() -> Info. Calling fcn" << std::endl;
+ m_fcn(&buffer[0]);
+}
+
+
+// return pointer to the parameters values at the minimum
+double MinimizerTest::getValueOfVariableAtMinimum(size_t index) const
+{
+ std::map<int, double >::const_iterator pos = m_values.find((int)index);
+ if(pos != m_values.end()){
+ return pos->second;
+ } else {
+ throw LogicErrorException("TestMinimizer::getValueOfVariableAtMinimum() -> Not found!");
+ }
+}
diff --git a/Core/Tools/inc/SafePointerVector.h b/Core/Tools/inc/SafePointerVector.h
index 6ae4143d39bacacc4db0be9ffa7d092393b7b8d3..7110a591892ae2cc8385337c9a87ced652be6f9b 100644
--- a/Core/Tools/inc/SafePointerVector.h
+++ b/Core/Tools/inc/SafePointerVector.h
@@ -30,7 +30,7 @@ public:
typedef typename std::vector<T *>::const_iterator const_iterator;
SafePointerVector();
SafePointerVector(const SafePointerVector &other);
- ~SafePointerVector();
+ virtual ~SafePointerVector();
SafePointerVector &operator=(const SafePointerVector &right);
size_t size() const;
diff --git a/Tests/UnitTests/TestCore/TestCore b/Tests/UnitTests/TestCore/TestCore
index f4d604aaee1ed3bc5d972dd439bf5b150c90cda4..24bf8fa1f1d558f78c018fd07a67d30b5a67689b 100755
Binary files a/Tests/UnitTests/TestCore/TestCore and b/Tests/UnitTests/TestCore/TestCore differ