ParameterDistribution.cpp

Go to the documentation of this file.

// ************************************************************************** //
//
//  BornAgain: simulate and fit scattering at grazing incidence
//
//! @file      Param/Distrib/ParameterDistribution.cpp
//! @brief     Implements class ParameterDistribution.
//!
//! @homepage  http://www.bornagainproject.org
//! @license   GNU General Public License v3 or higher (see COPYING)
//! @copyright Forschungszentrum Jülich GmbH 2018
//! @authors   Scientific Computing Group at MLZ (see CITATION, AUTHORS)
//
// ************************************************************************** //
 
#include "Param/Distrib/ParameterDistribution.h"
#include "Base/Types/Exceptions.h"
#include "Param/Distrib/Distributions.h"
 
ParameterDistribution::ParameterDistribution(const std::string& par_name,
                                             const IDistribution1D& distribution,
                                             size_t nbr_samples, double sigma_factor,
                                             const RealLimits& limits)
    : IParameterized("ParameterDistribution"), m_name(par_name), m_nbr_samples(nbr_samples),
      m_sigma_factor(sigma_factor), m_limits(limits), m_xmin(1.0), m_xmax(-1.0)
{
    mP_distribution.reset(distribution.clone());
    if (m_sigma_factor < 0.0)
        throw Exceptions::RuntimeErrorException(
            "ParameterDistribution::ParameterDistribution() -> Error."
            "sigma factor cannot be negative");
    if (nbr_samples == 0)
        throw Exceptions::RuntimeErrorException(
            "ParameterDistribution::ParameterDistribution() -> Error."
            "Number of samples can't be zero.");
}
 
ParameterDistribution::ParameterDistribution(const std::string& par_name,
                                             const IDistribution1D& distribution,
                                             size_t nbr_samples, double xmin, double xmax)
    : IParameterized("ParameterDistribution"), m_name(par_name), m_nbr_samples(nbr_samples),
      m_sigma_factor(0.0), m_xmin(xmin), m_xmax(xmax)
{
    mP_distribution.reset(distribution.clone());
    if (m_sigma_factor < 0.0) {
        throw Exceptions::RuntimeErrorException(
            "ParameterDistribution::ParameterDistribution() -> Error."
            "sigma factor cannot be negative");
    }
    if (nbr_samples == 0) {
        throw Exceptions::RuntimeErrorException(
            "ParameterDistribution::ParameterDistribution() -> Error."
            "Number of samples can't be zero.");
    }
    if (xmin >= xmax) {
        throw Exceptions::RuntimeErrorException(
            "ParameterDistribution::ParameterDistribution() -> Error."
            "xmin>=xmax");
    }
}
 
ParameterDistribution::ParameterDistribution(const ParameterDistribution& other)
    : IParameterized("ParameterDistribution"), m_name(other.m_name),
      m_nbr_samples(other.m_nbr_samples), m_sigma_factor(other.m_sigma_factor),
      m_linked_par_names(other.m_linked_par_names), m_limits(other.m_limits), m_xmin(other.m_xmin),
      m_xmax(other.m_xmax)
{
    mP_distribution.reset(other.mP_distribution->clone());
}
 
ParameterDistribution::~ParameterDistribution() = default;
 
ParameterDistribution& ParameterDistribution::operator=(const ParameterDistribution& other)
{
    if (this != &other) {
        this->m_name = other.m_name;
        m_nbr_samples = other.m_nbr_samples;
        m_sigma_factor = other.m_sigma_factor;
        mP_distribution.reset(other.mP_distribution->clone());
        m_linked_par_names = other.m_linked_par_names;
        m_limits = other.m_limits;
        m_xmin = other.m_xmin;
        m_xmax = other.m_xmax;
    }
    return *this;
}
 
ParameterDistribution& ParameterDistribution::linkParameter(std::string par_name)
{
    m_linked_par_names.push_back(par_name);
    return *this;
}
 
size_t ParameterDistribution::getNbrSamples() const
{
    if (mP_distribution && mP_distribution->isDelta())
        return 1;
    return m_nbr_samples;
}
 
std::vector<ParameterSample> ParameterDistribution::generateSamples() const
{
    if (m_xmin < m_xmax)
        return mP_distribution->equidistantSamplesInRange(m_nbr_samples, m_xmin, m_xmax);
    else
        return mP_distribution->equidistantSamples(m_nbr_samples, m_sigma_factor, m_limits);
}
 
const IDistribution1D* ParameterDistribution::getDistribution() const
{
    return mP_distribution.get();
}
 
IDistribution1D* ParameterDistribution::getDistribution()
{
    return mP_distribution.get();
}