CoolProp/dev/fitter/Fitter.cpp

#include <iostream>
#include "Eigen/Dense"
#include "Helmholtz.h"

class EOSFitter;
#include "DataTypes.h"
#include "Fitter.h"

EOSFitter::EOSFitter()
{
	this->Tr = Tr;
	this->rhor = rhor;
	this->R = R;
};
double EOSFitter::dA_dDelta(double log_tau, double log_delta, double delta, int i)
{
	return alphar.dA_dDelta(log_tau, log_delta, delta, i);
};
double EOSFitter::d2A_dTau2(double log_tau, double log_delta, double delta, int i)
{
	return alphar.d2A_dTau2(log_tau, log_delta, delta, i);
};
double EOSFitter::dalphar_dDelta(double log_tau, double log_delta, double delta)
{
	double summer = 0;
	for (unsigned int i = 0; i < alphar.n.size(); i++)
	{
		summer += alphar.n[i]*alphar.dA_dDelta(log_tau, log_delta, delta, i);
	}
	return summer;
};
double EOSFitter::d2alphar_dDelta2(double log_tau, double log_delta, double delta)
{
	double summer = 0;
	for (unsigned int i = 0; i < alphar.n.size(); i++)
	{
		summer += alphar.n[i]*alphar.d2A_dDelta2(log_tau, log_delta, delta, i);
	}
	return summer;
};
double EOSFitter::d2alphar_dTau2(double log_tau, double log_delta, double delta)
{
	double summer = 0;
	for (unsigned int i = 0; i < alphar.n.size(); i++)
	{
		summer += alphar.n[i]*alphar.d2A_dTau2(log_tau, log_delta, delta, i);
	}
	return summer;
};
double EOSFitter::d2alpha0_dTau2(double tau, double delta)
{
	double summer = 0;
	for (std::vector<phi_BC*>::iterator it = alpha0.begin(); it != alpha0.end(); it++){
		summer += (*it)->dTau2(tau,delta);
	}	
	return summer;
};
double EOSFitter::d2alphar_dDelta_dTau(double log_tau, double log_delta, double delta)
{
	double summer = 0;
	for (unsigned int i = 0; i < alphar.n.size(); i++)
	{
		summer += alphar.n[i]*alphar.d2A_dDelta_dTau(log_tau, log_delta, delta, i);
	}
	return summer;
};
/// Set the coefficients in the EOS
void EOSFitter::set_n(const std::vector<double> &n)
{
	alphar.n = n;
};

void EOSFitter::solve_for_n(std::vector<double> &n, bool non_linear_terms_enabled)
{
	Eigen::MatrixXd A = Eigen::MatrixXd::Random(21, 21);
	Eigen::VectorXd Q = Eigen::VectorXd::Random(21);

	// Build the A matrix and the Q vector
	for (int i = 1; i <= A.rows(); i++)
	{
		// The i-th row of the A matrix (Span 2000 Eq. 4.9)
		for (int j = 1; j <= A.cols(); j++)
		{
			// The entry for the j-th column and i-th row
			double summer = 0;
			for (unsigned int m = 0; m < linear_data_points.size(); m++)
			{
				LinearExperimentalDataPoint &pt = *linear_data_points[m];
				summer += (pt.a_i(i)*pt.a_i(j))/pow(pt.variance,(int)2);
			}
			if (non_linear_terms_enabled)
			{
				for (unsigned int m = 0; m < nonlinear_data_points.size(); m++)
				{
					NonlinearExperimentalDataPoint &pt = *nonlinear_data_points[m];
					summer += (pt.a_i(i)*pt.a_i(j))/pow(pt.variance,(int)2);
				}
			}
			A(i-1,j-1) = summer;
		}
		// The i-th entry in the Q column vector
		double summer = 0;
		for (unsigned int m = 0; m < linear_data_points.size(); m++)
		{
			LinearExperimentalDataPoint &pt = *linear_data_points[m];
			summer += (pt.a_i(i)*pt.a_0())/pow(pt.variance,(int)2);
		}
		if (non_linear_terms_enabled)
		{
			for (unsigned int m = 0; m < nonlinear_data_points.size(); m++)
			{
				NonlinearExperimentalDataPoint &pt = *nonlinear_data_points[m];
				summer += (pt.a_i(i)*pt.a_0(n))/pow(pt.variance,(int)2);
			}
		}
		Q(i-1) = summer;
	}

	Eigen::VectorXd N = A.colPivHouseholderQr().solve(Q);
	for (unsigned int i = 0; i < n.size()-1; i++)
	{
		n[i+1] = N(i);
	}

	double relative_error = (A*N - Q).norm() / Q.norm();
};

double EOSFitter::sum_squares(std::vector<double> &n, bool non_linear_terms_enabled)
{
	double summer = 0;
	for (unsigned int m = 0; m < linear_data_points.size(); m++)
	{
		LinearExperimentalDataPoint &pt = *linear_data_points[m];
		summer += pt.sum_squares(n);
	}
	if (non_linear_terms_enabled)
	{
		for (unsigned int m = 0; m < nonlinear_data_points.size(); m++)
		{
			NonlinearExperimentalDataPoint &pt = *nonlinear_data_points[m];
			summer += pt.sum_squares(n);
		}
	}
	return summer;
}