/***************************************************************************
* SimetricCosWeightChange.h *
* ------------------- *
* copyright : (C) 2014 by Francisco Naveros and Niceto Luque *
* email : fnaveros@ugr.es nluque@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 SIMETRICCOSWEIGHTCHANGE_H_
#define SIMETRICCOSWEIGHTCHANGE_H_
/*!
* \file CosWeightChange.h
*
* \author Francisco Naveros
* \author Niceto Luque
* \date May 2014
*
* This file declares a class which abstracts the behaviour of a exponential-cosinusoidal additive learning rule. When a spike arrive
* for a non-trigger synapse, a LTP method with value a1pre is applied. When a spike arrive for a trigger synapse, a LTD method with kernel
* a2prepre*exp(exponent*t/tau)*cos^2((pi/2)*t/tau), for all non-trigger synapses. This kernel is applied to the previous and future activity
* to the trigger synapse.
*/
#include "./WithoutPostSynaptic.h"
/*!
* \class SimetricCosWeightChange
*
* \brief Cosinusoidal learning rule.
*
* This class abstract the behaviour of a exponential-cosinusoidal additive learning rule. When a spike arrive for a non-trigger synapse,
* a LTP method with value a1pre is applied. When a spike arrive for a trigger synapse, a LTD method with kernel
* a2prepre*exp(exponent*t/tau)*cos^2((pi/2)*t/tau), for all non-trigger synapses. This kernel is applied only to the previous and future
* activity to the trigger synapse.
*
* \author Francisco Naveros
* \author Niceto Luque
* \date May 2014
*/
class SimetricCosWeightChange: public WithoutPostSynaptic{
private:
/*!
* \brief Kernel amplitude in second.
*/
float tau;
/*!
* \brief Exponent
*/
float exponent;
/*!
* \brief Maximum weight change for LTP
*/
float a1pre;
/*!
* \brief Maximum weight change LTD
*/
float a2prepre;
public:
/*!
* \brief Default constructor with parameters.
*
* It generates a new learning rule with its index.
*
* \param NewLearningRuleIndex learning rule index.
*/
SimetricCosWeightChange(int NewLearningRuleIndex);
/*!
* \brief Object destructor.
*
* It remove the object.
*/
virtual ~SimetricCosWeightChange();
/*!
* \brief It initialize the state associated to the learning rule for all the synapses.
*
* It initialize the state associated to the learning rule for all the synapses.
*
* \param NumberOfSynapses the number of synapses that implement this learning rule.
*/
void InitializeConnectionState(unsigned int NumberOfSynapses);
/*!
* \brief It loads the learning rule properties.
*
* It loads the learning rule properties.
*
* \param fh A file handler placed where the Learning rule properties are defined.
* \param Currentline The file line where the handler is placed.
*
* \throw EDLUTFileException If something wrong happens in reading the learning rule properties.
*/
virtual void LoadLearningRule(FILE * fh, long & Currentline) throw (EDLUTFileException);
/*!
* \brief It applies the weight change function when a presynaptic spike arrives.
*
* It applies the weight change function when a presynaptic spike arrives.
*
* \param Connection The connection where the spike happened.
* \param SpikeTime The spike time.
*/
virtual void ApplyPreSynapticSpike(Interconnection * Connection,double SpikeTime);
/*!
* \brief It prints the learning rule info.
*
* It prints the current learning rule characteristics.
*
* \param out The stream where it prints the information.
*
* \return The stream after the printer.
*/
virtual ostream & PrintInfo(ostream & out);
};
#endif /*SIMETRICCOSWEIGHTCHANGE_H_*/