Abstract observation base class. More...

#include <GObservation.hpp>

Inheritance diagram for GObservation:

Classes
class	model_func

class	npred_func

class	npred_kern

class	npred_spec_kern

Public Member Functions
	GObservation (void)
	Void constructor.

	GObservation (const GObservation &obs)
	Copy constructor.

virtual	~GObservation (void)
	Destructor.

virtual GObservation &	operator= (const GObservation &obs)
	Assignment operator.

virtual void	clear (void)=0
	Clear object.

virtual GObservation *	clone (void) const =0
	Clones object.

virtual std::string	classname (void) const =0
	Return class name.

virtual void	response (const GResponse &rsp)=0

virtual const GResponse *	response (void) const =0

virtual std::string	instrument (void) const =0

virtual double	ontime (void) const =0

virtual double	livetime (void) const =0

virtual double	deadc (const GTime &time=GTime()) const =0

virtual void	read (const GXmlElement &xml)=0

virtual void	write (GXmlElement &xml) const =0

virtual std::string	print (const GChatter &chatter=NORMAL) const =0
	Print content of object.

virtual GEvents *	events (void)
	Return events.

virtual const GEvents *	events (void) const
	Return events (const version)

virtual void	events (const GEvents &events)
	Set event container.

virtual double	likelihood (const GModels &models, GVector gradients, GMatrixSparse curvature, double *npred) const
	Compute likelihood function.

virtual double	model (const GModels &models, const GEvent &event, GVector *gradients=NULL) const
	Return model value and (optionally) gradients.

virtual GVector	model (const GModels &models, GMatrixSparse *gradients=NULL) const
	Return vector of model values and (optionally) gradients.

virtual double	nobserved (void) const
	Return total number of observed events.

virtual double	npred (const GModels &models, GVector *gradients=NULL) const
	Return total number (and optionally gradients) of predicted counts for all models.

virtual double	npred (const GModel &model) const
	Return total number of predicted counts for one model.

virtual double	model_grad (const GModel &model, const GModelPar &par, const GEvent &event) const
	Returns parameter gradient of model for a given event.

virtual GVector	model_grad (const GModel &model, const GModelPar &par) const
	Returns parameter gradients of model for all events.

virtual double	npred_grad (const GModel &model, const GModelPar &par) const
	Returns parameter gradient of Npred.

virtual void	remove_response_cache (const std::string &name)
	Remove response cache for specific model.

virtual double	grad_step_size (const GModelPar &par) const
	Return gradient step size for a given model parameter.

bool	has_events (void) const
	Signal if observation has events.

bool	has_gradient (const GModel &model, const GModelPar &par) const
	Check whether a model parameter has an analytical gradient.

void	name (const std::string &name)
	Set observation name.

void	id (const std::string &id)
	Set observation identifier.

void	statistic (const std::string &statistic)
	Set optimizer statistic.

const std::string &	name (void) const
	Return observation name.

const std::string &	id (void) const
	Return observation identifier.

const std::string &	statistic (void) const
	Return optimizer statistic.

void	computed_gradient (const GModel &model, const GModelPar &par) const
	Signals that an analytical gradient was computed for a model parameter.

Public Member Functions inherited from GBase
virtual	~GBase (void)
	Destructor.

Protected Member Functions
void	init_members (void)
	Initialise class members.

void	copy_members (const GObservation &obs)
	Copy class members.

void	free_members (void)
	Delete class members.

virtual double	likelihood_poisson_unbinned (const GModels &models, GVector gradients, GMatrixSparse curvature, double *npred) const
	Evaluate log-likelihood function for Poisson statistic and unbinned analysis (version with working arrays)

virtual double	likelihood_poisson_binned (const GModels &models, GVector gradients, GMatrixSparse curvature, double *npred) const
	Evaluate log-likelihood function for Poisson statistic and binned analysis (version with working arrays)

virtual double	likelihood_gaussian_binned (const GModels &models, GVector gradients, GMatrixSparse curvature, double *npred) const
	Evaluate log-likelihood function for Gaussian statistic and binned analysis (version with working arrays)

virtual bool	use_event_for_likelihood (const int &index) const
	Check whether bin should be used for likelihood analysis.

virtual double	npred_spec (const GModel &model, const GTime &obsTime) const
	Integrates spatially integrated Npred kernel spectrally.

Protected Attributes
std::string	m_name
	Observation name.

std::string	m_id
	Observation identifier.

std::string	m_statistic
	Optimizer statistic.

GEvents *	m_events
	Pointer to event container.

std::vector< std::string >	m_pars_with_gradients

Detailed Description

Abstract observation base class.

This class provides an abstract interface for an observation. The observation collects information about the instrument, holds the measured events, and provides information about the analysis definition.

The response() method returns a pointer to the response function. The derived classes have to make sure that this method never returns NULL.

The method model() returns the probability for an event to be measured with a given instrument direction, a given energy and at a given time, given a source model and an instrument pointing direction. The method npred() returns the total number of expected events within the analysis region for a given source model and a given instrument pointing direction. The methods are defined as virtual and can be overloaded by derived classes that implement instrument specific observations in order to optimize the execution speed for data analysis.

Definition at line 70 of file GObservation.hpp.

Constructor & Destructor Documentation

◆ GObservation() [1/2]

GObservation::GObservation ( void )

Void constructor.

Definition at line 84 of file GObservation.cpp.

References init_members().

◆ GObservation() [2/2]

GObservation::GObservation ( const GObservation & obs )

Copy constructor.

Parameters

[in] obs Observation.

Instantiate the class by copying from an existing observation.

Definition at line 101 of file GObservation.cpp.

References copy_members(), and init_members().

◆ ~GObservation()

GObservation::~GObservation ( void )

virtual

Destructor.

Definition at line 117 of file GObservation.cpp.

References free_members().

Member Function Documentation

◆ classname()

virtual std::string GObservation::classname ( void ) const

pure virtual

Return class name.

Returns: String containing the class name.

Returns the class name for non-abstract classes in a human readable way.

Implements GBase.

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

◆ clear()

virtual void GObservation::clear ( void )

pure virtual

Clear object.

Sets the object to a clean initial state. After calling the method the object will be in the same state as it were if an empty instance of the object would have been created.

Implements GBase.

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

◆ clone()

virtual GObservation * GObservation::clone ( void ) const

pure virtual

Clones object.

Returns: Pointer to deep copy of object.

Creates a deep copy of the object and returns a pointer to the object.

Implements GBase.

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

Referenced by GObservationRegistry::alloc(), GObservations::append(), GObservations::insert(), and GObservations::set().

◆ computed_gradient()

void GObservation::computed_gradient	(	const GModel &	model,
		const GModelPar &	par ) const

Signals that an analytical gradient was computed for a model parameter.

Parameters

[in]	model	Model.
[in]	par	Model parameter.

Signals that an analytical gradient was computed for a model parameter.

Definition at line 1272 of file GObservation.cpp.

References m_pars_with_gradients, model(), and GOptimizerPar::name().

Referenced by GModelData::eval(), GSPIModelDataSpace::eval(), GCTAResponseCube::irf_radial(), and GModelSky::signal_analytical_gradients().

◆ copy_members()

void GObservation::copy_members ( const GObservation & obs )

protected

Copy class members.

Parameters

[in] obs Observation.

Copy members from an observation.

Definition at line 1336 of file GObservation.cpp.

References GEvents::clone(), m_events, m_id, m_name, m_pars_with_gradients, and m_statistic.

Referenced by GObservation(), and operator=().

◆ deadc()

virtual double GObservation::deadc ( const GTime & time = GTime() ) const

pure virtual

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

Referenced by GCOMResponse::irf(), GCTAResponseIrf::irf(), GCTAResponseIrf::irf_diffuse(), GCTAResponseIrf::irf_elliptical(), GCTAResponseIrf::irf_radial(), GCTAResponseIrf::mc(), and GCTAResponseIrf::nirf().

◆ events() [1/3]

void GObservation::events ( const GEvents & events )

virtual

Set event container.

Parameters

[in] events Event container.

Set the event container for this observation by cloning the container specified in the argument.

Definition at line 218 of file GObservation.cpp.

References GEvents::clone(), events(), and m_events.

◆ events() [2/3]

GEvents * GObservation::events ( void )

virtual

Return events.

Exceptions

GException::no_events No events allocated for observation.

Returns pointer to events.

Definition at line 176 of file GObservation.cpp.

References G_EVENTS, and m_events.

Referenced by GCOMObservation::compute_drb_bgdlixa(), GCOMObservation::compute_drb_bgdlixe(), GCOMObservation::compute_drb_bgdlixf(), GCOMObservation::compute_drb_phinor(), GCOMDri::compute_dre(), GCOMDris::compute_drws(), GResponse::convolve(), gammalib::cta_event_cube(), gammalib::cta_event_list(), GModelData::eval(), GModels::eval(), GSPIModelDataSpace::eval(), GResponse::eval_probs(), events(), GCTAModelSpatialLookup::fill_buffer(), GSPIModelDataSpace::GSPIModelDataSpace(), GCOMResponse::irf(), GSPIResponse::irf(), GResponse::irf_composite(), GCOMResponse::irf_diffuse(), GResponse::irf_diffuse(), GSPIResponse::irf_diffuse(), GSPIResponse::irf_diffuse_map(), GCOMResponse::irf_elliptical(), GResponse::irf_elliptical(), GSPIResponse::irf_elliptical(), GSPIResponse::irf_init(), GCOMResponse::irf_ptsrc(), GResponse::irf_ptsrc(), GSPIResponse::irf_ptsrc(), GCOMResponse::irf_radial(), GCTAResponseCube::irf_radial(), GResponse::irf_radial(), GSPIResponse::irf_radial(), GCTAResponseCube::irf_spatial(), GResponse::irf_spatial(), likelihood(), likelihood_gaussian_binned(), likelihood_poisson_binned(), likelihood_poisson_unbinned(), GCTAObservation::load(), GCTAObservation::load(), GLATObservation::load_binned(), GLATObservation::load_unbinned(), GCTAModelRadialAcceptance::mc(), model(), model_grad(), GCOMObservation::npred(), npred(), npred_spec(), GCTAObservation::read(), GCTAObservation::read(), GSPIObservation::read(), GCTAOnOffObservation::set(), GLATMeanPsf::set(), GSPIResponse::set(), GLATMeanPsf::set_map_corrections(), GSPIModelDataSpace::setup(), GCTAObservation::write(), GCTAObservation::write(), and GCTAObservation::write_attributes().

◆ events() [3/3]

const GEvents * GObservation::events ( void ) const

virtual

Return events (const version)

Exceptions

GException::no_events No events allocated for observation.

Returns const pointer to events.

Definition at line 197 of file GObservation.cpp.

References G_EVENTS, and m_events.

◆ free_members()

void GObservation::free_members ( void )

protected

Delete class members.

Definition at line 1357 of file GObservation.cpp.

References m_events.

Referenced by GCOMObservation::clear(), GCTAObservation::clear(), GLATObservation::clear(), GMWLObservation::clear(), GSPIObservation::clear(), operator=(), and ~GObservation().

◆ grad_step_size()

double GObservation::grad_step_size ( const GModelPar & par ) const

inlinevirtual

Return gradient step size for a given model parameter.

Parameters

[in] par Model parameter

Returns: Gradient step size

Returns the step size that is used for numerical gradient computation for a given model parameter par.

The step size is fixed to a value of 0.0002, which for spatial parameters corresponds to an angular resolution of about 1 arcsec. Parameters that need numerical gradient computations are typical angles, hence fixing this value to an accuracy of 1 arcsec should be more than sufficient for all considered cases.

Note, however, that numerical precision issues may occur ata this angular resolution, and depending on how the response is computed for a given instrument, the instrument specific interface may overload this parameter by different values.

Reimplemented in GCOMObservation, and GSPIObservation.

Definition at line 347 of file GObservation.hpp.

Referenced by model_grad(), model_grad(), and npred_grad().

◆ has_events()

bool GObservation::has_events ( void ) const

inline

Signal if observation has events.

Returns: True if observation contains events.

Definition at line 239 of file GObservation.hpp.

References m_events.

◆ has_gradient()

bool GObservation::has_gradient	(	const GModel &	model,
		const GModelPar &	par ) const

Check whether a model parameter has an analytical gradient.

Parameters

[in]	model	Model.
[in]	par	Model parameter.

Returns: True if model parameter is free and has an analytical gradient.

Checks whether a model parameter is free and has an analytical gradient.

Definition at line 1236 of file GObservation.cpp.

References has_gradient(), GOptimizerPar::is_free(), m_pars_with_gradients, model(), and GOptimizerPar::name().

Referenced by has_gradient(), model(), and model().

◆ id() [1/2]

void GObservation::id ( const std::string & id )

inline

Set observation identifier.

Parameters

[in] id Observation identifier.

Set identifier of the observation.

Definition at line 268 of file GObservation.hpp.

References id(), and m_id.

Referenced by GObservations::append(), gammalib::cta_obs(), gammalib::cta_rsp_irf(), GModels::eval(), GModels::eval(), GCTACubeBackground::fill(), GCTACubeEdisp::fill(), GCTACubeExposure::fill(), GCTACubePsf::fill(), GCTACubeEdisp::fill_cube(), GCTACubeExposure::fill_cube(), GCTACubePsf::fill_cube(), GObservations::insert(), GCOMResponse::irf_ptsrc(), GCTAResponseCube::irf_spatial(), GResponse::irf_spatial(), GResponse::irf_spatial(), GCTAModelAeffBackground::npred(), GCTAModelCubeBackground::npred(), GCTAModelIrfBackground::npred(), GCTAModelSkyCube::npred(), GCTAResponseIrf::nroi(), GObservations::read(), GCTAOnOffObservation::set(), and GObservations::set().

◆ id() [2/2]

const std::string & GObservation::id ( void ) const

inline

Return observation identifier.

Returns: Observation identifier.

Definition at line 308 of file GObservation.hpp.

References m_id.

Referenced by GCOMObservation::compute_drb_bgdlixa(), GCOMObservation::compute_drb_bgdlixe(), GCOMObservation::compute_drb_bgdlixf(), GCOMObservation::compute_drb_phinor(), id(), GMWLObservation::load(), GCOMObservation::obs_id(), GCTAObservation::read_attributes(), and remove_response_cache().

◆ init_members()

void GObservation::init_members ( void )

protected

Initialise class members.

The step size for gradient computation is fixed by default to 0.0002 degrees, which corresponds to about 1 arcsec for parameters that are given in degrees. The reasoning behind this value is that parameters that use numerical gradients are typically angles, such as for example the position, and we want to achieve arcsec precision with this method.

Definition at line 1313 of file GObservation.cpp.

References m_events, m_id, m_name, m_pars_with_gradients, and m_statistic.

Referenced by GCOMObservation::clear(), GCTAObservation::clear(), GLATObservation::clear(), GMWLObservation::clear(), GSPIObservation::clear(), GObservation(), GObservation(), and operator=().

◆ instrument()

virtual std::string GObservation::instrument ( void ) const

pure virtual

◆ likelihood()

double GObservation::likelihood	(	const GModels &	models,
		GVector *	gradients,
		GMatrixSparse *	curvature,
		double *	npred ) const

virtual

Compute likelihood function.

Parameters

[in]	models	Models.
[in,out]	gradients	Pointer to gradients.
[in,out]	curvature	Pointer to curvature matrix.
[in,out]	npred	Pointer to Npred value.

Returns: Likelihood.

Computes the likelihood for a specified set of models. The method also returns the gradients, the curvature matrix, and the number of events that are predicted by all models.

Reimplemented in GCTAOnOffObservation.

Definition at line 247 of file GObservation.cpp.

References events(), G_LIKELIHOOD, likelihood_gaussian_binned(), likelihood_poisson_binned(), likelihood_poisson_unbinned(), npred(), statistic(), and gammalib::toupper().

◆ likelihood_gaussian_binned()

double GObservation::likelihood_gaussian_binned	(	const GModels &	models,
		GVector *	gradients,
		GMatrixSparse *	curvature,
		double *	npred ) const

protectedvirtual

Evaluate log-likelihood function for Gaussian statistic and binned analysis (version with working arrays)

Parameters

[in]	models	Models.
[in,out]	gradients	Gradient.
[in,out]	curvature	Curvature matrix.
[in,out]	npred	Number of predicted events.

Returns: Likelihood value.

This method evaluates the -(log-likelihood) function for parameter optimisation using binned analysis and Poisson statistic. The -(log-likelihood) function is given by

\[ L = 1/2 \sum_i (n_i - e_i)^2 \sigma_i^{-2} \]

where the sum is taken over all data space bins, \(n_i\) is the observed number of counts, \(e_i\) is the model and \(\sigma_i\) is the statistical uncertainty. This method also computes the parameter gradients \(\delta L/dp\) and the curvature matrix \(\delta^2 L/dp_1 dp_2\) and also updates the total number of predicted events m_npred.

Definition at line 1795 of file GObservation.cpp.

References GMatrixSparse::add_to_column(), GMatrixSparse::column_to_vector(), GEventBin::counts(), GVectorSparse::data(), GVectorSparse::elements(), GEventBin::error(), events(), GVectorSparse::inx(), gammalib::is_infinite(), minerr, minmod, model(), npred(), GEventBin::size(), GEvents::size(), GVector::size(), GMatrixSparse::transpose(), and use_event_for_likelihood().

Referenced by likelihood().

◆ likelihood_poisson_binned()

double GObservation::likelihood_poisson_binned	(	const GModels &	models,
		GVector *	gradients,
		GMatrixSparse *	curvature,
		double *	npred ) const

protectedvirtual

Evaluate log-likelihood function for Poisson statistic and binned analysis (version with working arrays)

Parameters

[in]	models	Models.
[in,out]	gradients	Gradient.
[in,out]	curvature	Curvature matrix.
[in,out]	npred	Number of predicted events.

Returns: Likelihood value.

This method evaluates the -(log-likelihood) function for parameter optimisation using binned analysis and Poisson statistic. The -(log-likelihood) function is given by

\[ L=-\sum_i n_i \log e_i - e_i \]

where the sum is taken over all data space bins, \(n_i\) is the observed number of counts and \(e_i\) is the model. This method also computes the parameter gradients \(\delta L/dp\) and the curvature matrix \(\delta^2 L/dp_1 dp_2\) and also updates the total number of predicted events m_npred.

Definition at line 1535 of file GObservation.cpp.

References GMatrixSparse::add_to_column(), GMatrixSparse::column_to_vector(), GEventBin::counts(), GVectorSparse::data(), GVectorSparse::elements(), events(), GVectorSparse::inx(), gammalib::is_infinite(), minmod, model(), npred(), GEventBin::size(), GEvents::size(), GVector::size(), GMatrixSparse::transpose(), and use_event_for_likelihood().

Referenced by likelihood().

◆ likelihood_poisson_unbinned()

double GObservation::likelihood_poisson_unbinned	(	const GModels &	models,
		GVector *	gradients,
		GMatrixSparse *	curvature,
		double *	npred ) const

protectedvirtual

Evaluate log-likelihood function for Poisson statistic and unbinned analysis (version with working arrays)

Parameters

[in]	models	Models.
[in,out]	gradients	Gradient.
[in,out]	curvature	Curvature matrix.
[in,out]	npred	Number of predicted events.

Returns: Likelihood value.

This method evaluates the -(log-likelihood) function for parameter optimisation using unbinned analysis and Poisson statistic. The -(log-likelihood) function is given by

\[ L = N_{\rm pred} - \sum_i \log e_i \]

where \(N_{\rm pred}\) is the number of events predicted by the model, and the sum is taken over all events. This method also computes the parameter gradients \(\delta L/dp\) and the curvature matrix \(\delta^2 L/dp_1 dp_2\).

Definition at line 1402 of file GObservation.cpp.

References GMatrixSparse::add_to_column(), GMatrixSparse::column_to_vector(), events(), gammalib::is_infinite(), minmod, model(), npred(), GEvents::size(), GVector::size(), GMatrixSparse::transpose(), and use_event_for_likelihood().

Referenced by likelihood().

◆ livetime()

virtual double GObservation::livetime ( void ) const

pure virtual

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

◆ model() [1/2]

double GObservation::model	(	const GModels &	models,
		const GEvent &	event,
		GVector *	gradients = NULL ) const

virtual

Return model value and (optionally) gradients.

Parameters

[in]	models	Model container.
[in]	event	Observed event.
[out]	gradients	Pointer to gradient vector (optional).

Returns: Model value.

Exceptions

GException::invalid_value Dimension of gradient vector mismatches number of parameters.

Implements generic model and gradient evaluation for an observation. The model gives the probability for an event to occur with a given instrument direction, at a given energy and at a given time. The gradient is the parameter derivative of this probability. If NULL is passed for the gradient vector, then gradients will not be computed.

The method will only operate on models for which the list of instruments and observation identifiers matches those of the observation. Models that do not match will be skipped.

Definition at line 337 of file GObservation.cpp.

References GModel::eval(), GModel::eval_indices(), GOptimizerPar::factor_gradient(), G_MODEL1, GModel::has_eval_indices(), has_gradient(), instrument(), GOptimizerPar::is_free(), GModel::is_valid(), m_pars_with_gradients, model(), model_grad(), GModels::npars(), GModel::size(), GModels::size(), GVector::size(), and gammalib::str().

Referenced by computed_gradient(), GCOMObservation::drm(), has_gradient(), likelihood_gaussian_binned(), likelihood_poisson_binned(), likelihood_poisson_unbinned(), model(), model_grad(), model_grad(), GCOMObservation::npred(), npred(), npred_grad(), npred_spec(), and GCTAOnOffObservation::set().

◆ model() [2/2]

GVector GObservation::model	(	const GModels &	models,
		GMatrixSparse *	gradients = NULL ) const

virtual

Return vector of model values and (optionally) gradients.

Parameters

[in]	models	Model container.
[out]	gradients	Pointer to sparse gradient matrix.

Returns: Vector of model values.

Exceptions

GException::invalid_argument Gradient matrix mismatches number of events or model parameters.

Returns the model values for each event in the observation.

If gradients is not NULL, the matrix contains on output the model factor gradients for all events. Each row of the gradients matrix corresponds to one event, the columns correspond to the parameters of the models container.

The method is optimised for speed, using sparse vectors and matrices.

Definition at line 491 of file GObservation.cpp.

References GMatrixSparse::colfill_flush(), GMatrixSparse::colfill_init(), GMatrixSparse::colfill_set_column(), GMatrixSparse::column(), GMatrixSparse::column_to_vector(), GMatrixBase::columns(), GModel::eval(), GModel::eval_indices(), events(), G_MODEL2, GModel::has_eval_indices(), has_gradient(), instrument(), GOptimizerPar::is_free(), GModel::is_valid(), m_pars_with_gradients, model_grad(), GOptimizerPar::name(), GModels::npars(), GMatrixBase::rows(), GEvents::size(), GModel::size(), GModels::size(), GMatrixSparse::stack_destroy(), GMatrixSparse::stack_init(), and gammalib::str().

◆ model_grad() [1/2]

GVector GObservation::model_grad	(	const GModel &	model,
		const GModelPar &	par ) const

virtual

Returns parameter gradients of model for all events.

Parameters

[in]	model	Model.
[in]	par	Model parameter.

This method uses a robust but simple difference method to estimate parameter gradients that have not been provided by the model. We use here a simple method as this method is likely used for spatial model parameters, and the spatial model may eventually be noisy due to numerical integration limits.

Definition at line 998 of file GObservation.cpp.

References events(), GOptimizerPar::factor_max(), GOptimizerPar::factor_min(), GOptimizerPar::factor_value(), grad_step_size(), GOptimizerPar::has_max(), GOptimizerPar::has_min(), GOptimizerPar::is_free(), and model().

◆ model_grad() [2/2]

double GObservation::model_grad	(	const GModel &	model,
		const GModelPar &	par,
		const GEvent &	event ) const

virtual

Returns parameter gradient of model for a given event.

Parameters

[in]	model	Model.
[in]	par	Model parameter.
[in]	event	Event.

This method uses a robust but simple difference method to estimate parameter gradients that have not been provided by the model. We use here a simple method as this method is likely used for spatial model parameters, and the spatial model may eventually be noisy due to numerical integration limits.

Definition at line 916 of file GObservation.cpp.

References GDerivative::difference(), GOptimizerPar::factor_max(), GOptimizerPar::factor_min(), GOptimizerPar::factor_value(), grad_step_size(), GOptimizerPar::has_max(), GOptimizerPar::has_min(), GOptimizerPar::is_free(), GDerivative::left_difference(), model(), and GDerivative::right_difference().

Referenced by model(), and model().

◆ name() [1/2]

void GObservation::name ( const std::string & name )

inline

Set observation name.

Parameters

[in] name Observation name.

Set name of the observation.

Definition at line 253 of file GObservation.hpp.

References m_name, and name().

Referenced by gammalib::cta_obs(), gammalib::cta_rsp_irf(), GCTACubeBackground::fill(), GCTACubeEdisp::fill(), GCTACubeExposure::fill(), GCTACubePsf::fill(), GCTACubeEdisp::fill_cube(), GCTACubeExposure::fill_cube(), GCTACubePsf::fill_cube(), GCOMResponse::irf_ptsrc(), GObservations::read(), and GCTAOnOffObservation::set().

◆ name() [2/2]

const std::string & GObservation::name ( void ) const

inline

Return observation name.

Returns: Observation name.

Definition at line 296 of file GObservation.hpp.

References m_name.

Referenced by GCOMObservation::compute_drb_bgdlixa(), GCOMObservation::compute_drb_bgdlixe(), GCOMObservation::compute_drb_bgdlixf(), GCOMObservation::compute_drb_phinor(), GCOMObservation::drm(), GMWLObservation::load(), name(), GCOMObservation::print(), GCTAObservation::print(), GLATObservation::print(), GMWLObservation::print(), GSPIObservation::print(), remove_response_cache(), GCOMObservation::response(), GLATObservation::response(), GMWLObservation::response(), GSPIObservation::response(), and GLATObservation::write().

◆ nobserved()

double GObservation::nobserved ( void ) const

virtual

Return total number of observed events.

Returns: Total number of observed events.

Returns the total number of observed events. If the observation does not contain any events the method returns zero.

Reimplemented in GCTAOnOffObservation.

Definition at line 703 of file GObservation.cpp.

References m_events, nobserved(), and GEvents::number().

Referenced by nobserved().

◆ npred() [1/2]

double GObservation::npred ( const GModel & model ) const

virtual

Return total number of predicted counts for one model.

Parameters

[in] model Gamma-ray source model.

Exceptions

GException::gti_invalid Good Time Interval is invalid.

Computes

\[ N_{\rm pred} = \int_{\rm GTI} \int_{E_{\rm bounds}} \int_{\rm ROI} P(p',E',t') \, dp' \, dE' \, dt' \]

of the event probability

\[ P(p',E',t') = \int \int \int S(p,E,t) \times R(p',E',t'|p,E,t) \, dp \, dE \, dt \]

where \(S(p,E,t)\) is the source model, \(R(p',E',t'|p,E,t)\) is the instrument response function, \(p'\) is the measured photon direction, \(E'\) is the measured photon energy, \(t'\) is the measured photon arrival time, \(p\) is the true photon arrival direction, \(E\) is the true photon energy, and \(t\) is the true photon arrival time.

This method performs the time integration over the Good Time Intervals \({\rm GTI}\). Note that MET is used for the time integration interval. This, however, is no specialisation since npred_grad_kern_spec::eval() converts the argument back in a GTime object by assuming that the argument is in MET, hence the correct time system will be used at the end by the method.

Reimplemented in GCOMObservation.

Definition at line 836 of file GObservation.cpp.

References events(), G_NPRED, GEvents::gti(), instrument(), model(), npred(), npred_spec(), ontime(), GIntegral::romberg(), and gammalib::str().

◆ npred() [2/2]

double GObservation::npred	(	const GModels &	models,
		GVector *	gradients = NULL ) const

virtual

Return total number (and optionally gradients) of predicted counts for all models.

Parameters

[in]	models	Models.
[out]	gradients	Model parameter gradients (optional).

Exceptions

GException::invalid_argument Dimension of gradient vector mismatches number of parameters.

Returns the total number of predicted counts within the analysis region. If NULL is passed for the gradient vector then gradients will not be computed.

The method will only operate on models for which the list of instruments and observation identifiers matches those of the observation. Models that do not match will be skipped.

Definition at line 736 of file GObservation.cpp.

References G_NPRED, instrument(), GModel::is_valid(), GModels::npars(), npred(), npred_grad(), GModel::size(), GModels::size(), GVector::size(), and gammalib::str().

Referenced by GObservation::npred_func::eval(), GCTAOnOffObservation::likelihood(), likelihood(), GCTAOnOffObservation::likelihood_cstat(), likelihood_gaussian_binned(), likelihood_poisson_binned(), likelihood_poisson_unbinned(), GCTAOnOffObservation::likelihood_wstat(), npred(), and npred().

◆ npred_grad()

double GObservation::npred_grad	(	const GModel &	model,
		const GModelPar &	par ) const

virtual

Returns parameter gradient of Npred.

Parameters

[in]	model	Gamma-ray source model.
[in]	par	Model parameter.

Computes

\[ \frac{{\rm d} N_{\rm pred}}{{\rm d} a_i} \]

where

\[ N_{\rm pred} = \int_{\rm GTI} \int_{E_{\rm bounds}} \int_{\rm ROI} P(p',E',t') \, dp' \, dE' \, dt' \]

is the integral of the event probability

\[ P(p',E',t') = \int \int \int S(p,E,t) \times R(p',E',t'|p,E,t) \, dp \, dE \, dt \]

and \(a_i\) is the model parameter \(i\). Furthermore \(S(p,E,t)\) is the source model, \(R(p',E',t'|p,E,t)\) is the instrument response function, \(p'\) is the measured photon direction, \(E'\) is the measured photon energy, \(t'\) is the measured photon arrival time, \(p\) is the true photon arrival direction, \(E\) is the true photon energy, and \(t\) is the true photon arrival time.

This method uses a robust but simple difference method to estimate parameter gradients that have not been provided by the model. We use here a simple method as this method is likely used for spatial model parameters, and the spatial model may eventually be noisy due to numerical integration limits.

Definition at line 1139 of file GObservation.cpp.

References GDerivative::difference(), GOptimizerPar::factor_max(), GOptimizerPar::factor_min(), GOptimizerPar::factor_value(), grad_step_size(), GOptimizerPar::has_max(), GOptimizerPar::has_min(), GOptimizerPar::is_free(), GDerivative::left_difference(), model(), and GDerivative::right_difference().

Referenced by npred().

◆ npred_spec()

double GObservation::npred_spec	(	const GModel &	model,
		const GTime &	obsTime ) const

protectedvirtual

Integrates spatially integrated Npred kernel spectrally.

Parameters

[in]	model	Gamma-ray source model.
[in]	obsTime	Measured photon arrival time.

Exceptions

GException::erange_invalid Energy range is invalid.

Computes

\[ N_{\rm pred}(t') = \int_{E_{\rm bounds}} \int_{\rm ROI} P(p',E',t') \, dp' \, dE' \]

of the event probability

\[ P(p',E',t') = \int \int \int S(p,E,t) \times R(p',E',t'|p,E,t) \, dp \, dE \, dt \]

where \(S(p,E,t)\) is the source model, \(R(p',E',t'|p,E,t)\) is the instrument response function, \(p'\) is the measured photon direction, \(E'\) is the measured photon energy, \(t'\) is the measured photon arrival time, \(p\) is the true photon arrival direction, \(E\) is the true photon energy, and \(t\) is the true photon arrival time.

This method performs the energy intergration over the energy boundaries \(E_{\rm bounds}\).

Definition at line 2067 of file GObservation.cpp.

References GEvents::ebounds(), GEbounds::emax(), GEbounds::emin(), events(), GIntegral::fixed_iter(), gammalib::is_infinite(), gammalib::is_notanumber(), GEnergy::MeV(), model(), GIntegral::romberg(), and GEbounds::size().

Referenced by npred().

◆ ontime()

virtual double GObservation::ontime ( void ) const

pure virtual

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

Referenced by npred().

◆ operator=()

GObservation & GObservation::operator= ( const GObservation & obs )

virtual

Assignment operator.

Parameters

[in] obs Observation.

Returns: Observation.

Assign observation.

Definition at line 141 of file GObservation.cpp.

References copy_members(), free_members(), and init_members().

Referenced by GCOMObservation::operator=(), GCTAObservation::operator=(), GCTAOnOffObservation::operator=(), GLATObservation::operator=(), GMWLObservation::operator=(), and GSPIObservation::operator=().

◆ print()

virtual std::string GObservation::print ( const GChatter & chatter = NORMAL ) const

pure virtual

Print content of object.

Parameters

[in] chatter Chattiness (defaults to NORMAL).

Returns: String containing the content of the object.

Formats the content in a standard way and puts this content in a C++ string that is returned.

Implements GBase.

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

◆ read()

virtual void GObservation::read ( const GXmlElement & xml )

pure virtual

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

Referenced by GObservations::read().

◆ remove_response_cache()

void GObservation::remove_response_cache ( const std::string & name )

virtual

Remove response cache for specific model.

Parameters

[in] name Model name.

Removes response cache for model name from current observation.

Definition at line 1214 of file GObservation.cpp.

References id(), name(), remove_response_cache(), and response().

Referenced by remove_response_cache().

◆ response() [1/2]

virtual void GObservation::response ( const GResponse & rsp )

pure virtual

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

Referenced by GModelSky::eval(), GModelSky::eval(), and GModelSky::npred().

◆ response() [2/2]

virtual const GResponse * GObservation::response ( void ) const

pure virtual

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

Referenced by remove_response_cache().

◆ statistic() [1/2]

void GObservation::statistic ( const std::string & statistic )

inline

Set optimizer statistic.

Parameters

[in] statistic Optimizer statistic.

Set optimizer statistic for the observation.

Definition at line 283 of file GObservation.hpp.

References m_statistic, and statistic().

◆ statistic() [2/2]

const std::string & GObservation::statistic ( void ) const

inline

Return optimizer statistic.

Returns: Optimizer statistic.

Definition at line 320 of file GObservation.hpp.

References m_statistic.

Referenced by GCTAOnOffObservation::likelihood(), likelihood(), GCTAOnOffObservation::model_background(), GCOMObservation::print(), GCTAObservation::print(), GCTAOnOffObservation::print(), GLATObservation::print(), GMWLObservation::print(), GSPIObservation::print(), GCTAObservation::read(), GCTAOnOffObservation::read(), statistic(), GCTAObservation::write(), and GCTAOnOffObservation::write().

◆ use_event_for_likelihood()

bool GObservation::use_event_for_likelihood ( const int & index ) const

protectedvirtual

Check whether bin should be used for likelihood analysis.

Parameters

[in] index Event index.

Returns: True.

This is a dummy virtual method that allows implementation of a hook for event selection in the likelihood computation. The dummy method always returns true.

Reimplemented in GCOMObservation.

Definition at line 1960 of file GObservation.cpp.

Referenced by likelihood_gaussian_binned(), likelihood_poisson_binned(), and likelihood_poisson_unbinned().

◆ write()

virtual void GObservation::write ( GXmlElement & xml ) const

pure virtual

Implemented in GCOMObservation, GCTAObservation, GCTAOnOffObservation, GLATObservation, GMWLObservation, and GSPIObservation.

Member Data Documentation

◆ m_events

GEvents* GObservation::m_events

protected

Pointer to event container.

Definition at line 226 of file GObservation.hpp.

◆ m_id

std::string GObservation::m_id

protected

Observation identifier.

Definition at line 224 of file GObservation.hpp.

Referenced by copy_members(), id(), id(), init_members(), GCTAOnOffObservation::print(), and GCOMObservation::read().

◆ m_name

std::string GObservation::m_name

protected

Observation name.

Definition at line 223 of file GObservation.hpp.

Referenced by copy_members(), GCOMObservation::GCOMObservation(), GCOMObservation::GCOMObservation(), init_members(), name(), name(), GCTAOnOffObservation::print(), GCOMObservation::read(), and GCOMObservation::read_attributes().

◆ m_pars_with_gradients

std::vector<std::string> GObservation::m_pars_with_gradients

mutableprotected

Definition at line 229 of file GObservation.hpp.

Referenced by computed_gradient(), copy_members(), has_gradient(), init_members(), model(), and model().

◆ m_statistic

std::string GObservation::m_statistic

protected

Optimizer statistic.

Definition at line 225 of file GObservation.hpp.

Referenced by copy_members(), GMWLObservation::init_members(), init_members(), statistic(), and statistic().

The documentation for this class was generated from the following files:

include/GObservation.hpp
src/obs/GObservation.cpp

Classes

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ GObservation() [1/2]

◆ GObservation() [2/2]

◆ ~GObservation()

Member Function Documentation

◆ classname()

◆ clear()

◆ clone()

◆ computed_gradient()

◆ copy_members()

◆ deadc()

◆ events() [1/3]

◆ events() [2/3]

◆ events() [3/3]

◆ free_members()

◆ grad_step_size()

◆ has_events()

◆ has_gradient()

◆ id() [1/2]

◆ id() [2/2]

◆ init_members()

◆ instrument()

◆ likelihood()

◆ likelihood_gaussian_binned()

◆ likelihood_poisson_binned()

◆ likelihood_poisson_unbinned()

◆ livetime()

◆ model() [1/2]

◆ model() [2/2]

◆ model_grad() [1/2]

◆ model_grad() [2/2]

◆ name() [1/2]

◆ name() [2/2]

◆ nobserved()

◆ npred() [1/2]

◆ npred() [2/2]

◆ npred_grad()

◆ npred_spec()

◆ ontime()

◆ operator=()

◆ print()

◆ read()

◆ remove_response_cache()

◆ response() [1/2]

◆ response() [2/2]

◆ statistic() [1/2]

◆ statistic() [2/2]

◆ use_event_for_likelihood()

◆ write()

Member Data Documentation

◆ m_events

◆ m_id

◆ m_name

◆ m_pars_with_gradients

◆ m_statistic