Class containing direct analysis (in optimization) based on neural network optimization. More...

Inheritance diagram for IG::Neural::AnalysisFileServerNeural:

Collaboration diagram for IG::Neural::AnalysisFileServerNeural:

Public Member Functions
	AnalysisFileServerNeural (string workingDirectoryPath)
	Constructor, sets object's working directory to the specified path.
virtual void	Analyse (IAnalysisResults analysisData)
	Performs analysis - calculates requested results and writes them to the provided data structure.
override void	ClearMessages ()
	Clears all messages in the neural network approximator's file client/server directory.
virtual void	ServerAnalyse ()
	Performs server-side direct analysis. Reads analysis input from standard location, calculates output and writes it to the standard location.
virtual void	ClientCalculateAnalysisResults (AnalysisResults anRes)
	Calculates analysis results by using the analysis server.
virtual void	ClientTestCalculateAnalysisResults (string inputFilePath, bool reqObjective, bool reqConstraints, bool reqObjectiveGradient, bool reqConstraintGradients, string outputFilePath)
	Performs client-side test calculation of analysis response.
Protected Member Functions
virtual void	AnalysisToApproximationInput (IVector anInput, ref IVector approximationInput)
	Converts analysis input parameters to approximation input parameters.
virtual void	ApproximationToAnalysisOutput (IVector approxOuptut, IAnalysisResults anResults)
	Converts approximation output to direct analysis results. Sets objective function to sum of squares of approximated values. WARNING: This method should be overridden in derived classes.
Properties
OptFileManager	OptimizationFileManager `[get, set]`
int	NumParameters `[get, set]`
	Number of parameters.
int	NumObjectives `[get, set]`
	Number of objective functions (normally 1 for this type, but can be 0).
int	NumConstraints `[get, set]`
	Number of constraints.
int	NumEqualityConstraints `[get, set]`
	Number of equality constraints.
Private Attributes
OptFileManager	_optFileManager

Detailed Description

Class containing direct analysis (in optimization) based on neural network optimization.

Constructor & Destructor Documentation

IG::Neural::AnalysisFileServerNeural::AnalysisFileServerNeural ( string workingDirectoryPath ) [inline]

Constructor, sets object's working directory to the specified path.

Parameters:

workingDirectoryPath

Member Function Documentation

virtual void IG::Neural::AnalysisFileServerNeural::Analyse ( IAnalysisResults analysisData ) [inline, virtual]

Performs analysis - calculates requested results and writes them to the provided data structure.

Parameters:

analysisData Data structure where analysis request parameters are obtained and where analysis results are written.

Implements IG::Num::IAnalysis.

virtual void IG::Neural::AnalysisFileServerNeural::AnalysisToApproximationInput	(	IVector	anInput,
		ref IVector	approximationInput
	)		`[inline, protected, virtual]`

Converts analysis input parameters to approximation input parameters.

Parameters:

anInput	Input parameters for direct analysis.
approximationInput	Input parameters for response approximation.

virtual void IG::Neural::AnalysisFileServerNeural::ApproximationToAnalysisOutput	(	IVector	approxOuptut,
		IAnalysisResults	anResults
	)		`[inline, protected, virtual]`

Converts approximation output to direct analysis results.
Sets objective function to sum of squares of approximated values.
WARNING:
This method should be overridden in derived classes.

Parameters:

approxOuptut	Vector of approximated output values.
anResults	Direct analysis results.

override void IG::Neural::AnalysisFileServerNeural::ClearMessages ( ) [inline, virtual]

Clears all messages in the neural network approximator's file client/server directory.

Reimplemented from IG::Neural::ApproximationFileServerNeural.

virtual void IG::Neural::AnalysisFileServerNeural::ServerAnalyse ( ) [inline, virtual]

Performs server-side direct analysis. Reads analysis input from standard location, calculates output and writes it to the standard location.

virtual void IG::Neural::AnalysisFileServerNeural::ClientCalculateAnalysisResults ( AnalysisResults anRes ) [inline, virtual]

Calculates analysis results by using the analysis server.

Parameters:

inputParameters	Intput parameters for which approximation is calculated.
outputValues	Vector where approximation output values are stored.

virtual void IG::Neural::AnalysisFileServerNeural::ClientTestCalculateAnalysisResults	(	string	inputFilePath,
		bool	reqObjective,
		bool	reqConstraints,
		bool	reqObjectiveGradient,
		bool	reqConstraintGradients,
		string	outputFilePath
	)		`[inline, virtual]`

Performs client-side test calculation of analysis response.

Parameters:

inputFilePath	Path to the JSON file where input parameters are read from. The file pointed at must exist.
reqObjective	Flag indicating whether objective function must be calculated.
reqConstraints	Flag indicating whether constraint functions must be calculated.
reqGradObjective	Flag indicating whether objective function gradientmust be calculated.
reqGradOConstraints	Fleg indicating whether constraint function gradients must be calculated.
outputFilePath	Path of a file where the calculated analysis response in JSON is written to. It can be null or empty string, in this case response is not written to a file (but it is output on console).