/*=========================================================================
 *
 *  Copyright NumFOCUS
 *
 *  Licensed under the Apache License, Version 2.0 (the "License");
 *  you may not use this file except in compliance with the License.
 *  You may obtain a copy of the License at
 *
 *         https://www.apache.org/licenses/LICENSE-2.0.txt
 *
 *  Unless required by applicable law or agreed to in writing, software
 *  distributed under the License is distributed on an "AS IS" BASIS,
 *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 *  See the License for the specific language governing permissions and
 *  limitations under the License.
 *
 *=========================================================================*/
#include "itkChiSquareDistribution.h"
#include "itkGaussianDistribution.h"
#include "itkMath.h"

extern "C" double
dgami_(double * a, double * x);

extern "C" double
dgamma_(double * x);

namespace itk
{
namespace Statistics
{
ChiSquareDistribution::ChiSquareDistribution()
{
  m_Parameters = ParametersType(1);
  m_Parameters[0] = 1.0;
}

void
ChiSquareDistribution::SetDegreesOfFreedom(SizeValueType dof)
{
  bool modified = false;

  if (m_Parameters.GetSize() > 0)
  {
    if (Math::NotExactlyEquals(m_Parameters[0], static_cast<double>(dof)))
    {
      modified = true;
    }
  }

  if (m_Parameters.GetSize() != 1)
  {
    m_Parameters = ParametersType(1);
  }

  m_Parameters[0] = static_cast<double>(dof);

  if (modified)
  {
    this->Modified();
  }
}

SizeValueType
ChiSquareDistribution::GetDegreesOfFreedom() const
{
  if (m_Parameters.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << m_Parameters.size() << " parameters.");
  }
  return static_cast<SizeValueType>(m_Parameters[0]);
}

double
ChiSquareDistribution::PDF(double x, SizeValueType degreesOfFreedom)
{
  auto   dof = static_cast<double>(degreesOfFreedom);
  double dofon2 = 0.5 * dof;
  double pdf = 0.0;

  if (x >= 0.0)
  {
    pdf = std::exp(-0.5 * x) * std::pow(x, dofon2 - 1.0) / (std::pow(2.0, dofon2) * dgamma_(&dofon2));
  }

  return pdf;
}

double
ChiSquareDistribution::PDF(double x, const ParametersType & p)
{
  if (p.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << p.size() << " parameters.");
  }
  return ChiSquareDistribution::PDF(x, static_cast<SizeValueType>(p[0]));
}

double
ChiSquareDistribution::CDF(double x, SizeValueType degreesOfFreedom)
{
  // Based on Abramowitz and Stegun 26.4.19 which relates the
  // cumulative of the chi-square to incomplete (and complete) gamma
  // function.
  if (x < 0)
  {
    return 0.0;
  }

  double dofon2 = 0.5 * degreesOfFreedom;
  double xon2 = 0.5 * x;

  return dgami_(&dofon2, &xon2) / dgamma_(&dofon2);
}

double
ChiSquareDistribution::CDF(double x, const ParametersType & p)
{
  if (p.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << p.size() << " parameters.");
  }
  return ChiSquareDistribution::CDF(x, (SizeValueType)p[0]);
}

double
ChiSquareDistribution::InverseCDF(double p, SizeValueType degreesOfFreedom)
{
  if (p <= 0.0)
  {
    return 0.0;
  }
  else if (p >= 1.0)
  {
    return itk::NumericTraits<double>::max();
  }

  double x;
  double dof;
  double nx;

  // Based on Abramowitz and Stegun 26.4.17
  dof = static_cast<double>(degreesOfFreedom);
  nx = GaussianDistribution::InverseCDF(p);

  double f = 2.0 / (9.0 * dof);
  x = dof * std::pow(1.0 - f + nx * std::sqrt(f), 3.0);

  // The approximation above is only accurate for large degrees of
  // freedom. We'll improve the approximation by a few Newton iterations.
  //
  //   0 iterations, error = 10^-1  at 1 degree of freedom
  //   3 iterations, error = 10^-11 at 1 degree of freedom
  //  10 iterations, error = 10^-13 at 1 degree of freedom
  //  20 iterations, error = 10^-13 at 1 degree of freedom
  //
  //   0 iterations, error = 10^-1  at 11 degrees of freedom
  //   3 iterations, error = 10^-8  at 11 degrees of freedom
  //  10 iterations, error = 10^-13 at 11 degrees of freedom
  //  20 iterations, error = 10^-13 at 11 degrees of freedom
  //
  //   0 iterations, error = 10^-1  at 100 degrees of freedom
  //   3 iterations, error = 10^-9  at 100 degrees of freedom
  //  10 iterations, error = 10^-10 at 100 degrees of freedom
  //  20 iterations, error = 10^-9  at 100 degrees of freedom

  // We are trying to find the zero of
  //
  // f(x) = p - chisquarecdf(x) = 0;
  //
  // So,
  //
  // x(n+1) = x(n) - f(x(n)) / f'(x(n))
  //        = x(n) + (p - chisquarecdf(x)) / chisquarepdf(x)
  //
  // Note that f'(x) = - chisquarepdf(x)
  //
  double delta;
  for (unsigned int newt = 0; newt < 10; ++newt)
  {
    delta = (p - ChiSquareDistribution::CDF(x, degreesOfFreedom)) / ChiSquareDistribution::PDF(x, degreesOfFreedom);
    x += delta;
  }

  return x;
}

double
ChiSquareDistribution::InverseCDF(double p, const ParametersType & params)
{
  if (params.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << params.size() << " parameters.");
  }
  return ChiSquareDistribution::InverseCDF(p, static_cast<SizeValueType>(params[0]));
}

double
ChiSquareDistribution::EvaluatePDF(double x) const
{
  if (m_Parameters.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << m_Parameters.size() << " parameters.");
  }
  return ChiSquareDistribution::PDF(x, static_cast<SizeValueType>(m_Parameters[0]));
}

double
ChiSquareDistribution::EvaluatePDF(double x, const ParametersType & p) const
{
  if (p.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << p.size() << " parameters.");
  }
  return ChiSquareDistribution::PDF(x, static_cast<SizeValueType>(p[0]));
}

double
ChiSquareDistribution::EvaluatePDF(double x, SizeValueType degreesOfFreedom) const
{
  return ChiSquareDistribution::PDF(x, degreesOfFreedom);
}

double
ChiSquareDistribution::EvaluateCDF(double x) const
{
  if (m_Parameters.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << m_Parameters.size() << " parameters.");
  }
  return ChiSquareDistribution::CDF(x, static_cast<SizeValueType>(m_Parameters[0]));
}

double
ChiSquareDistribution::EvaluateCDF(double x, const ParametersType & p) const
{
  if (p.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << p.size() << " parameters.");
  }
  return ChiSquareDistribution::CDF(x, static_cast<SizeValueType>(p[0]));
}

double
ChiSquareDistribution::EvaluateCDF(double x, SizeValueType degreesOfFreedom) const
{
  return ChiSquareDistribution::CDF(x, degreesOfFreedom);
}

double
ChiSquareDistribution::EvaluateInverseCDF(double p) const
{
  if (m_Parameters.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << m_Parameters.size() << " parameters.");
  }
  return ChiSquareDistribution::InverseCDF(p, static_cast<SizeValueType>(m_Parameters[0]));
}

double
ChiSquareDistribution::EvaluateInverseCDF(double p, const ParametersType & params) const
{
  if (params.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << params.size() << " parameters.");
  }
  return ChiSquareDistribution::InverseCDF(p, static_cast<SizeValueType>(params[0]));
}

double
ChiSquareDistribution::EvaluateInverseCDF(double p, SizeValueType degreesOfFreedom) const
{
  return ChiSquareDistribution::InverseCDF(p, degreesOfFreedom);
}

double
ChiSquareDistribution::GetMean() const
{
  if (m_Parameters.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << m_Parameters.size() << " parameters.");
  }
  return m_Parameters[0];
}

double
ChiSquareDistribution::GetVariance() const
{
  if (m_Parameters.GetSize() != 1)
  {
    itkGenericExceptionMacro("Invalid number of parameters to describe distribution. Expected 1 parameter, but got "
                             << m_Parameters.size() << " parameters.");
  }
  return 2.0 * m_Parameters[0];
}

void
ChiSquareDistribution::PrintSelf(std::ostream & os, Indent indent) const
{
  Superclass::PrintSelf(os, indent);

  if (m_Parameters.GetSize() > 0)
  {
    os << indent << "Degrees of freedom: " << static_cast<SizeValueType>(m_Parameters[0]) << std::endl;
  }
  else
  {
    os << indent << "Degrees of freedom: (unknown)" << std::endl;
  }
}
} // end of namespace Statistics
} // end namespace itk