#include <TMetropHasting.h>

Inheritance diagram for URANIE::Calibration::TMetropHasting:

[legend]

Collaboration diagram for URANIE::Calibration::TMetropHasting:

[legend]

Public Member Functions
Constructor and Destructor
	TMetropHasting (URANIE::DataServer::TDataServer tds, URANIE::Relauncher::TRun run, int nS=100, Option_t *option="lhs")

	TMetropHasting (URANIE::DataServer::TDataServer tds, void(fcn)(Double_t , Double_t ), const char varexpinput, const char varexpoutput, int ns=100, Option_t *option="")
	Default Calibration constructor with the function argument: it contains the assessor to be used.

	TMetropHasting (URANIE::DataServer::TDataServer tds, const char fcn, const char varexpinput, const char varexpoutput, int ns=100, Option_t *option="")
	Default Calibration constructor with the function argument: it contains the assessor to be used.

	TMetropHasting (URANIE::DataServer::TDataServer tds, URANIE::Launcher::TCode fcode, int ns=100, Option_t *option="")
	Default Calibration constructor with the code argument: it contains the assessor to be used.

virtual	~TMetropHasting ()
	Default destructor.

evaluate the events
void	computeParameters (Option_t *option="")
	Generate the sample.

void	checktdsParContent ()

Set reference information and initialisation methods
void	setBurnin (int burn)

void	setLag (int lag)

void	setNbDump (int nbDump)

void	setAcceptationRatioRange (double lower, double higher)

string	getDefaultCut ()

void	clearDefaultCut ()

void	setInitialisation (int n, double values, double standDev)
	Initialise the parameters.

void	setInitialisation (vector< double > values, vector< double > standDev)
	Initialise the parameters.

void	setDistanceAndReference (const char funcName, URANIE::DataServer::TDataServer tdsRef, const char input, const char output, const char *weight="")
	Set the distance function and some needed informations.

void	setDistanceAndReference (URANIE::Calibration::TDistanceFunction distFunc, URANIE::DataServer::TDataServer tdsRef, const char input, const char output, const char *weight="")
	Set the distance function and some needed informations.

Visualisation and diagnostic
void	drawTrace (TString sTitre, const char variable="", const char select="1>0", Option_t option="")
	Draws the evolution of parameters as a function of the iterator.

void	drawAcceptationRatio (TString sTitre, const char variable="", const char select="1>0", Option_t option="")
	Draws the evolution of acceptation ratio as a function of the iterator.

void	drawParameters (TString sTitre, const char variable="", const char select="1>0", Option_t option="")
	Draws the parameters as distributions The estimateParamters method has computed the parameters, so this one plots the parameters as distributions.

void	getAutoCorrelation (vector< int > l, vector< double > *out, int cut=0)
	Compute the autocorrelation.

Printing Log
virtual void	printLog (Option_t *option="")
	Prints the log.

Public Member Functions inherited from URANIE::Calibration::TCalibration
	TCalibration (URANIE::DataServer::TDataServer tds, URANIE::Relauncher::TRun run, int ns=1, Option_t *option="")
	Default constructor with the runner argument: it contains the assessor to be used.

	TCalibration (URANIE::DataServer::TDataServer tds, void(fcn)(Double_t , Double_t ), const char varexpinput, const char varexpoutput, int ns=1, Option_t *option="")
	Default Calibration constructor with the function argument: it contains the assessor to be used.

	TCalibration (URANIE::DataServer::TDataServer tds, const char fcn, const char varexpinput, const char varexpoutput, int ns=1, Option_t *option="")
	Default Calibration constructor with the function argument: it contains the assessor to be used.

	TCalibration (URANIE::DataServer::TDataServer tds, URANIE::Launcher::TCode fcode, int ns=1, const char *option="")
	Default Calibration constructor with the code argument: it contains the assessor to be used.

virtual	~TCalibration ()
	Default destructor.

void	estimateParameters (Option_t *option="")

void	estimateCustomResidues (string resName, int n_theta, double *theta)

void	checkReference (URANIE::DataServer::TDataServer tdsRef, const char input, const char output, const char weight)
	Check the consistency of the formation provided.

void	drawResidues (TString sTitre, const char variable="", const char select="1>0", Option_t option="")

void	setLog ()

void	unsetLog ()

void	changeLog ()

Bool_t	getLog ()

Int_t	getID ()

void	setSeed (UInt_t nval)
	Set the seed of the random generator if one is used.

UInt_t	getSeed ()
	Get the seed of the random generator if one is used.

const char *	getMethodName ()
	Get the method name.

void	setObservationCovarianceMatrix (TMatrixD &mat)
	Set the observatiton covariance matrix.

URANIE::Calibration::TDistanceFunction *	getDistanceFunction ()
	Return the distance function.

Int_t	getNPar ()
	Get the number of parameters to be calibrated.

URANIE::DataServer::TDataServer *	getEvaluationTDS ()
	Get the tds in which evaluation will be performed.

Public Attributes
vector< double >	_values
	Vector of values to be tested.

vector< double >	_vstd
	Vector of standard deviation.

int	_burnin
	The warm-up or burn-in.

int	_lag
	The lag value.

int	_nbDump
	Frequency to which the algo dump a line.

double	_lowAccRange
	loweracceptation ratio bound to increase _vstd

double	_higAccRange
	higheracceptation ratio bound to decrease _vstd

bool	_mbGDVersion
	Use the Guillaume Damblin.

bool	_bcleaningAtt
	Do not store the underlying att.

Public Attributes inherited from URANIE::Calibration::TCalibration
URANIE::DataServer::TDataServer *	_tdsPar
	TDS containing parameters properties (parameters that should be calibrated)

URANIE::DataServer::TDataServer *	_tdsObs
	TDS containing observations used for calibration.

URANIE::DataServer::TDataServer *	_tdsEval
	TDS containing a priori / a posteriori evaluations.

ELauncher	_nLauncher
	The type of launcher.

TString	_sFunctionName
	The Name of the evaluatuor.

TString	_sEI
	The Name of input.

TString	_sEO
	The Name of output.

URANIE::Launcher::TCode *	_code
	The tcode.

URANIE::Relauncher::TRun *	_run
	Pointer to the runner to be used.

void(*	_pFunction )(double , double )
	Function pointer.

vector< URANIE::DataServer::TStochasticAttribute * >	_vatt
	internal vector of stochastic attribute for some methods

Protected Member Functions
private methods
void	logPriorPdf (double &ret)
	Logarithm of the prior.

void	parseOption (Option_t *option="")
	Read the possible options.

Protected Member Functions inherited from URANIE::Calibration::TCalibration
void	checkCanvasCreation (bool newcan)
	Create a canvas if needed.

void	initInputs ()
	Initialise some common inputs.

void	initResults (vector< string > *ParsedLines)
	Initialise some common inputs.

void	computeAPosterioriForDistribution ()
	Compute the a posteriori residual for many-solutions method.

void	setMethodName (const char *str)
	Set the Method name.

Additional Inherited Members
Public Types inherited from URANIE::Calibration::TCalibration
enum	ELauncher { kCode , kFunction , kRun , kUnknown }

Protected Attributes inherited from URANIE::Calibration::TCalibration
URANIE::Calibration::TDistanceFunction *	_dFunc
	Pointer to chosen distance function.

URANIE::DataServer::TDSNtupleD *	_evTuple
	Pointer to the eval ntuple.

TList *	_listOfParameters
	List of the parameters to be calibrated.

TCanvas *	_canvas
	Canvas object to deal with.

TObjArray *	_drawingGarbageCollector
	Garbage collector for prints.

int	_nSam
	The number of sample in a posteriori distributions.

int	_nObs
	The number of observations in the reference database.

int	_nIterMax
	The maximum number of iteration allowed (meaning total number of code estimation is _nIterMax * _nObs)

int	_nPar
	Dimension of the parameters.

int	_nVar
	Dimension of the output and references to be compared with.

int	_nSeed
	The seed of the random generator.

TString	_sMethodName
	The method name.

TString	_referenceName
	The reference name.

TString	_outputName
	The output name.

vector< string >	_vrefName
	The reference names.

vector< string >	_voutName
	The output names.

TString	_weightName
	The weight name.

TMatrixD	_mObsCovMat
	Observation Covariance matrix.

bool	_buseMatrix
	Use matrix instead of vectors in the Distance Function.

bool	_bsaveAll
	Whether all evaluations should be saved, not only a priori and a posteriori.

bool	_bdontKeepAgreement
	Remove the agreement attribute from the tdsPar object.

bool	_buseMode
	Use Mode instead of Mean.

Bool_t	_blog
	Boolean for edit the log.

Constructor & Destructor Documentation

◆ TMetropHasting() [1/4]

URANIE::Calibration::TMetropHasting::TMetropHasting	(	URANIE::DataServer::TDataServer *	tds,
		URANIE::Relauncher::TRun *	run,
		int	nS = `100`,
		Option_t *	option = `"lhs"`
	)

Constructor usual where nS is set to be the number of sample produced by the algorithm.

Parameters

tds	the pointer of the TDataServer. It must contains attribuutes that inherit from TStochasticAttribute
run	the runner that would be used to produce evaluations
nS	size of the sample to be generated [100 by default]

Referenced by ClassImp().

◆ TMetropHasting() [2/4]

URANIE::Calibration::TMetropHasting::TMetropHasting	(	URANIE::DataServer::TDataServer *	tds,
		void()(Double_t , Double_t *)	fcn,
		const char *	varexpinput,
		const char *	varexpoutput,
		int	ns = `100`,
		Option_t *	option = `""`
	)

Default Calibration constructor with the function argument: it contains the assessor to be used.

Parameters

tds	: the dataserver that contains no data but only one attribute per parameter to be calibrated
fcn	: the pointer to a function
ns	: number of sample to be generated (depending on the method)
varexpinput	: the input variable for the function in the correct order (both input and parameters)
varexpoutput	: the output of the function in the correct order

◆ TMetropHasting() [3/4]

URANIE::Calibration::TMetropHasting::TMetropHasting	(	URANIE::DataServer::TDataServer *	tds,
		const char *	fcn,
		const char *	varexpinput,
		const char *	varexpoutput,
		int	ns = `100`,
		Option_t *	option = `""`
	)

Default Calibration constructor with the function argument: it contains the assessor to be used.

Parameters

tds	: the dataserver that contains no data but only one attribute per parameter to be calibrated
fcn	: the name of the function
ns	: number of sample to be generated (depending on the method)
varexpinput	: the input variable for the function in the correct order (both input and parameters)
varexpoutput	: the output of the function in the correct order

◆ TMetropHasting() [4/4]

URANIE::Calibration::TMetropHasting::TMetropHasting	(	URANIE::DataServer::TDataServer *	tds,
		URANIE::Launcher::TCode *	fcode,
		int	ns = `100`,
		Option_t *	option = `""`
	)

Default Calibration constructor with the code argument: it contains the assessor to be used.

Parameters

tds	: the dataserver that contains no data but only one attribute per parameter to be calibrated
code	: the code object that will be runned
ns	: number of sample to be generated (depending on the method)

◆ ~TMetropHasting()

virtual URANIE::Calibration::TMetropHasting::~TMetropHasting ( )

virtual

Default destructor.

Referenced by ClassImp().

Member Function Documentation

◆ checktdsParContent()

void URANIE::Calibration::TMetropHasting::checktdsParContent ( )

virtual

Check the content of the input dataserver

Implements URANIE::Calibration::TCalibration.

Referenced by ClassImp().

◆ clearDefaultCut()

void URANIE::Calibration::TMetropHasting::clearDefaultCut ( )

inline

References setBurnin(), and setLag().

◆ computeParameters()

void URANIE::Calibration::TMetropHasting::computeParameters ( Option_t * option = "" )

virtual

Generate the sample.

Parameters

option

possble options are

"clean": remove internal branche from the tdsPar object at the end
"gd": another implementation of this algorithm

Implements URANIE::Calibration::TCalibration.

Referenced by ClassImp().

◆ drawAcceptationRatio()

void URANIE::Calibration::TMetropHasting::drawAcceptationRatio	(	TString	sTitre,
		const char *	variable = `"*"`,
		const char *	select = `"1>0"`,
		Option_t *	option = `""`
	)

Draws the evolution of acceptation ratio as a function of the iterator.

Parameters

sTitre	title of the plot if specified
variable	list of variables that should be plotted
select	select that can be applied on the parameters or the iterator (burnin, selection for autocorrelation)
option	possible option for this plot "nonewcanvas": don't create a new canvas, but use the one active

Referenced by ClassImp().

◆ drawParameters()

void URANIE::Calibration::TMetropHasting::drawParameters	(	TString	sTitre,
		const char *	variable = `"*"`,
		const char *	select = `"1>0"`,
		Option_t *	option = `""`
	)

virtual

Draws the parameters as distributions The estimateParamters method has computed the parameters, so this one plots the parameters as distributions.

Parameters

sTitre	title of the plots if specified
variable	list of variables that should be plotted
select	select that can be applied on the parameters or the iterator (burnin, selection for autocorrelation)
option	possible option for this plot "nonewcanvas": don't create a new canvas, but use the one active "vertical" : if there are several parameters to be shown put them on top of another instead of side by side

Reimplemented from URANIE::Calibration::TCalibration.

Referenced by ClassImp().

◆ drawTrace()

void URANIE::Calibration::TMetropHasting::drawTrace	(	TString	sTitre,
		const char *	variable = `"*"`,
		const char *	select = `"1>0"`,
		Option_t *	option = `""`
	)

Draws the evolution of parameters as a function of the iterator.

Parameters

sTitre	title of the plot if specified
variable	list of variables that should be plotted
select	select that can be applied on the parameters or the iterator (burnin, selection for autocorrelation)
option	possible option for this plot "nonewcanvas": don't create a new canvas, but use the one active

Referenced by ClassImp().

◆ getAutoCorrelation()

void URANIE::Calibration::TMetropHasting::getAutoCorrelation	(	vector< int >	l,
		vector< double > *	out,
		int	cut = `0`
	)

Compute the autocorrelation.

Parameters

l	vector of steps (reading 1 events out of l)
out	vector of estimations (one per step)
cut	burn-up value (events ignored from 0 to cut)

Referenced by ClassImp().

◆ getDefaultCut()

string URANIE::Calibration::TMetropHasting::getDefaultCut ( )

inline

References _burnin, _lag, and URANIE::Calibration::TCalibration::_tdsPar.

Referenced by ClassImp().

◆ logPriorPdf()

void URANIE::Calibration::TMetropHasting::logPriorPdf ( double & ret )

protected

Logarithm of the prior.

Referenced by ClassImp().

◆ parseOption()

void URANIE::Calibration::TMetropHasting::parseOption ( Option_t * option = "" )

protectedvirtual

Read the possible options.

On top of the other, one can call

"savealleval": keep all estimation in the _tdsEval (Extremely dangerous as it will keep _nIter branches with _nObs estimations. This option might slow down the process.
"usematrix" : option use to specify that computation is done with matrices instead of vector<vector<double> >. Shoud be used only if one is using an home-made TDistanceFunction

Reimplemented from URANIE::Calibration::TCalibration.

Referenced by ClassImp().

◆ printLog()

virtual void URANIE::Calibration::TMetropHasting::printLog ( Option_t * option = "" )

virtual

Prints the log.

Reimplemented from URANIE::Calibration::TCalibration.

Referenced by ClassImp().

◆ setAcceptationRatioRange()

void URANIE::Calibration::TMetropHasting::setAcceptationRatioRange	(	double	lower,
		double	higher
	)

Referenced by ClassImp().

◆ setBurnin()

void URANIE::Calibration::TMetropHasting::setBurnin ( int burn )

inline

References _burnin.

Referenced by clearDefaultCut().

◆ setDistanceAndReference() [1/2]

void URANIE::Calibration::TMetropHasting::setDistanceAndReference	(	const char *	funcName,
		URANIE::DataServer::TDataServer *	tdsRef,
		const char *	input,
		const char *	output,
		const char *	weight = `""`
	)

virtual

Set the distance function and some needed informations.

Parameters

funcName	name of the distance function chosen among the already implemented ones: "LS" for least square "weighterLS" for weighted least square "relativeLS" for relative least square "L1" for L1 norm "Mahalanobis" for Mahalanobis distance function
tdsRef	: the dataserver that contains all needed information detailled below input: the input attributes used of the assessors in run reference: the reference attributes that will be used to compared the newly done estimations for every iterations weight: a weight to reweight every event, one-by-one (an experimental uncertainty for instance)

Reimplemented from URANIE::Calibration::TCalibration.

Referenced by ClassImp().

◆ setDistanceAndReference() [2/2]

void URANIE::Calibration::TMetropHasting::setDistanceAndReference	(	URANIE::Calibration::TDistanceFunction *	distFunc,
		URANIE::DataServer::TDataServer *	tdsRef,
		const char *	input,
		const char *	output,
		const char *	weight = `""`
	)

virtual

Set the distance function and some needed informations.

Parameters

distFunc	a pointer to a TDistanceFunction object (not usual, recommended only when dealing with a home-made distance unction compiled on the spot)
tdsRef	: the dataserver that contains all needed information detailled below input: the input attributes used of the assessors in run reference: the reference attributes that will be used to compared the newly done estimations for every iterations weight: a weight to reweight every event, one-by-one (an experimental uncertainty for instance)

Reimplemented from URANIE::Calibration::TCalibration.

◆ setInitialisation() [1/2]

void URANIE::Calibration::TMetropHasting::setInitialisation	(	int	n,
		double *	values,
		double *	standDev
	)

Initialise the parameters.

Parameters

n	number of parameters
values	array of values (size must equal to n)
standDev	array of standard deviation (size must equal to n)

Referenced by ClassImp().

◆ setInitialisation() [2/2]

void URANIE::Calibration::TMetropHasting::setInitialisation	(	vector< double >	values,
		vector< double >	standDev
	)

Initialise the parameters.

Parameters

values	vector of values (size must equal to n)
standDev	vector of standard deviation (size must equal to n)

◆ setLag()

void URANIE::Calibration::TMetropHasting::setLag ( int lag )

inline

References _lag.

Referenced by clearDefaultCut().

◆ setNbDump()

void URANIE::Calibration::TMetropHasting::setNbDump ( int nbDump )

inline

References _nbDump.

Member Data Documentation

◆ _bcleaningAtt

bool URANIE::Calibration::TMetropHasting::_bcleaningAtt

Do not store the underlying att.

Referenced by ClassImp().

◆ _burnin

int URANIE::Calibration::TMetropHasting::_burnin

The warm-up or burn-in.

Referenced by getDefaultCut(), and setBurnin().

◆ _higAccRange

double URANIE::Calibration::TMetropHasting::_higAccRange

higheracceptation ratio bound to decrease _vstd

Referenced by ClassImp().

◆ _lag

int URANIE::Calibration::TMetropHasting::_lag

The lag value.

Referenced by getDefaultCut(), and setLag().

◆ _lowAccRange

double URANIE::Calibration::TMetropHasting::_lowAccRange

loweracceptation ratio bound to increase _vstd

Referenced by ClassImp().

◆ _mbGDVersion

bool URANIE::Calibration::TMetropHasting::_mbGDVersion

Use the Guillaume Damblin.

Referenced by ClassImp().

◆ _nbDump

int URANIE::Calibration::TMetropHasting::_nbDump

Frequency to which the algo dump a line.

Referenced by ClassImp(), and setNbDump().

◆ _values

vector<double> URANIE::Calibration::TMetropHasting::_values

Vector of values to be tested.

Referenced by ClassImp().

◆ _vstd

vector<double> URANIE::Calibration::TMetropHasting::_vstd

Vector of standard deviation.

Referenced by ClassImp().

Documentation / Developer's manual

Available modules

Public Member Functions

Public Attributes

Protected Member Functions

Additional Inherited Members

Constructor & Destructor Documentation

◆ TMetropHasting() [1/4]

◆ TMetropHasting() [2/4]

◆ TMetropHasting() [3/4]

◆ TMetropHasting() [4/4]

◆ ~TMetropHasting()

Member Function Documentation

◆ checktdsParContent()

◆ clearDefaultCut()

◆ computeParameters()

◆ drawAcceptationRatio()

◆ drawParameters()

◆ drawTrace()

◆ getAutoCorrelation()

◆ getDefaultCut()

◆ logPriorPdf()

◆ parseOption()

◆ printLog()

◆ setAcceptationRatioRange()

◆ setBurnin()

◆ setDistanceAndReference() [1/2]

◆ setDistanceAndReference() [2/2]

◆ setInitialisation() [1/2]

◆ setInitialisation() [2/2]

◆ setLag()

◆ setNbDump()

Member Data Documentation

◆ _bcleaningAtt

◆ _burnin

◆ _higAccRange

◆ _lag

◆ _lowAccRange

◆ _mbGDVersion

◆ _nbDump

◆ _values

◆ _vstd