/***************************************************************************
 *                           IntegrationMethod.h                           *
 *                           -------------------                           *
 * copyright            : (C) 2013 by Francisco Naveros                    *
 * email                : fnaveros@atc.ugr.es                              *
 ***************************************************************************/

/***************************************************************************
 *                                                                         *
 *   This program is free software; you can redistribute it and/or modify  *
 *   it under the terms of the GNU General Public License as published by  *
 *   the Free Software Foundation; either version 3 of the License, or     *
 *   (at your option) any later version.                                   *
 *                                                                         *
 ***************************************************************************/

#ifndef INTEGRATIONMETHOD_H_
#define INTEGRATIONMETHOD_H_

/*!
 * \file IntegrationMethod.h
 *
 * \author Francisco Naveros
 * \date May 2013
 *
 * This file declares a class which abstracts all integration methods in CPU. This methods can
 * be fixed step or variable step (adaptative).
 */

#include <string>

#include "../../include/simulation/Utils.h"
#include "../../include/spike/EDLUTFileException.h"



using namespace std;
class TimeDrivenNeuronModel;


#define MAX_VARIABLES 15 

/*!
 * \class IntegrationMethod
 *
 * \brief Integration methods in CPU
 *
 * This class abstracts the behavior of all integration methods for neurons in a 
 * time-driven spiking neural network.
 * It includes internal model functions which define the behavior of integration methods
 * (initialization, calculate next value, ...).
 * This is only a virtual function (an interface) which defines the functions of the
 * inherited classes.
 *
 * \author Francisco Naveros
 * \date May 2013
 */
class IntegrationMethod {
	protected:

		/*!
		 * \brief Time driven neuron model associated to this integration method.
		*/
		TimeDrivenNeuronModel * model;

	public:

		/*!
		 * \brief Number of state variables for each cell.
		*/
		int N_NeuronStateVariables;

		/*!
		 * \brief Number of state variables witch are calculate with a differential equation for each cell.
		*/
		int N_DifferentialNeuronState;

		/*!
		 * \brief Number of state variables witch are calculate with a time dependent equation for each cell.
		*/
		int N_TimeDependentNeuronState;


		/*!
		 * \brief PredictedElapsedTime. This vector contains only one element for fixed step itegration method or one
		 * element for each neuron for variable step integration method.
		*/
		double ElapsedTime;


		/*!
		 * \brief Integration method type.
		*/
		string IntegrationMethodType;


		/*!
		 * \brief Constructor with parameters.
		 *
		 * It generates a new IntegrationMethod object.
		 *
		 * \param NewModel time driven neuron model associated to this integration method.
		 * \param integrationMethodType integration method type.
		 * \param N_neuronStateVariables total number of state variable for each neuron
		 * \param N_differentialNeuronState number of state variables that are diffined by a differential ecuation.
		 * \param N_timeDependentNeuronState number of state variables that are not diffined by a differential ecuation.
		 * \param jacobian set to true if the integration method calculates the jacobian of the neuron model differential 
		 *  equation and must be allocat memory for this one (only used in the BDF methods).
		 * \param inverse set to true if the integration method calculates the inverse of a matrix and must be allocated 
		 *  memory for this method(only used in the BDF methods).
		 */
		IntegrationMethod(TimeDrivenNeuronModel* NewModel, string integrationMethodType, int N_neuronStateVariables, int N_differentialNeuronState, int N_timeDependentNeuronState, bool jacobian, bool inverse);


		/*!
		 * \brief Class destructor.
		 *
		 * It destroys an object of this class.
		 */
		~IntegrationMethod();

		
		/*!
		 * \brief It calculate the new neural state variables for a defined elapsed_time.
		 *
		 * It calculate the new neural state variables for a defined elapsed_time.
		 *
		 * \param index for method with memory (e.g. BDF1ad, BDF2, BDF3, etc.).
		 * \param NeuronState neuron state variables of one neuron.
		 * \param elapsed_time integration time step.
		 */
		virtual void NextDifferentialEcuationValue(int index, float * NeuronState, float elapsed_time) = 0;


		/*!
		 * \brief It prints the integration method info.
		 *
		 * It prints the current integration method characteristics.
		 *
		 * \param out The stream where it prints the information.
		 *
		 * \return The stream after the printer.
		 */
		virtual ostream & PrintInfo(ostream & out) = 0;


		/*!
		 * \brief It gets the integration method type.
		 *
		 * It gets the integration method type.
		 *
		 * \return The integration method type.
		 */
		string GetType();


		/*!
		 * \brief It initialize the state of the integration method for method with memory (e.g. BDF1ad, BDF2, BDF3, etc.).
		 *
		 * It initialize the state of the integration method for method with memory (e.g. BDF1ad, BDF2, BDF3, etc.).
		 *
		 * \param N_neuron number of neurons in the neuron model.
		 * \param inicialization vector with initial values.
		 */
		virtual void InitializeStates(int N_neurons, float * inicialization) = 0;


		/*!
		 * \brief It calculate numerically the Jacobian .
		 *
		 * It calculate numerically the Jacobian.
		 *
		 * \param NeuronState neuron state variables of one neuron.
		 * \param jancum vector where is stored the Jacobian.
		 * \param elapsed_time integration method step.
		 */
		void Jacobian(float * NeuronState, float * jacnum, float elapsed_time);

		/*!
		 * \brief It calculate the inverse of a square matrix using Gauss-Jordan Method.
		 *
		 * It calculate the inverse of a square matrix using Gauss-Jordan Method.
		 *
		 * \param a pointer to the square matrix.
		 * \param ainv pointer to the inverse of the square matrix.
		 */
		void invermat(float *a, float *ainv);


		/*!
		 * \brief It reset the state of the integration method for method with memory (e.g. BDF1ad, BDF2, BDF3, etc.).
		 *
		 * It reset the state of the integration method for method with memory (e.g. BDF1ad, BDF2, BDF3, etc.).
		 *
		 * \param index indicate witch neuron must be reseted.
		 */
		virtual void resetState(int index) = 0;


		/*!
		 * \brief It loads the integration method parameters.
		 *
		 * It loads the integration method parameters from the file that define the parameter of the neuron model.
		 *
		 * \param Pointer to a neuron description file (*.cfg). At the end of this file must be included 
		 *  the integration method type and its parameters.
		 * \param Currentline line inside the neuron description file where start the description of the integration method parameter. 
		 *
		 * \throw EDLUTFileException If something wrong has happened in the file load.
		 */
		virtual void loadParameter(FILE *fh, long * Currentline) throw (EDLUTFileException) = 0 ;

};

#endif /* INTEGRATIONMETHOD_H_ */