Detailed Description

Header file for likelihood functions used by LALInference codes.

LALInferenceLikelihood contains all the necessary routines to compute the likelihood from a template (computed with LALInferenceTemplate) and the data (initialised with LALInferenceReadData).

Likelihood functions follow the basic naming convention: LALInference<type_of>LogLikelihood()

Takes as input:

a pointer to a LALInferenceVariable structure containing the parameters to compute the likelihood for,
a pointer to a LALInferenceIFOData structure containing the linked list of interferometer data,
a pointer to the LALInferenceModel model structure, containing the template function and model.

Outputs as a REAL8 the natural logarithm value of the likelihood, as defined by:

\[ Likelihood(\vec{x}|\vec{\lambda},M)=\exp(-\tfrac{1}{2}<\vec{x}-\vec{h_M}(\vec{\lambda})|\vec{x}-\vec{h_M}(\vec{\lambda})>) \]

where: \(<x|y>=4Re\left ( \int \frac{\tilde{x}\,\tilde{y}^*}{S_f}\, df \right )\)

Note that the likelihood is reported unnormalised.

Prototypes
REAL8	LALInferenceUndecomposedFreqDomainLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)
	(log-) likelihood function. More...

REAL8	LALInferenceZeroLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)
	For testing purposes (for instance sampling the prior), likelihood that returns 0.0 = log(1) every time. More...

REAL8	LALInferenceComputeFrequencyDomainOverlap (LALInferenceIFOData data, COMPLEX16Vector freqData1, COMPLEX16Vector *freqData2)
	Computes the <x\|y> overlap in the Fourier domain. More...

COMPLEX16	LALInferenceComputeFrequencyDomainComplexOverlap (LALInferenceIFOData dataPtr, COMPLEX16Vector freqData1, COMPLEX16Vector *freqData2)
	Computes the complex <x\|y> overlap. More...

REAL8	LALInferenceNullLogLikelihood (LALInferenceIFOData *data)
	Identical to LALInferenceFreqDomainNullLogLikelihood, but returns the likelihood of a null template. More...

REAL8	LALInferenceFreqDomainStudentTLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)
	Student-t (log-) likelihood function as described in Roever/Meyer/Christensen (2011): "Modelling coloured residual noise in gravitational-wave signal processing. More...

REAL8	LALInferenceCorrelatedAnalyticLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)
	An analytic likeilhood that is a correlated Gaussian in 15 dimensions. More...

REAL8	LALInferenceBimodalCorrelatedAnalyticLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)
	An analytic likeilhood that is two correlated Gaussians in 15 dimensions. More...

REAL8	LALInferenceRosenbrockLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)
	15-D Rosenbrock log(L) function (see Eq (3) of http://en.wikipedia.org/wiki/Rosenbrock_function . More...

REAL8	LALInferenceMarginalisedPhaseLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)

REAL8	LALInferenceMarginalisedTimePhaseLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)

double	LALInferenceMarginalDistanceLogLikelihood (double dist_min, double dist_max, double OptimalSNR, double d_inner_h, int cosmology, int margphi)
	Compute delta-log-likelihood for given distance min, max and OptimalSNR and d_inner_h when evaluated at 1Mpc cosmology: 0 = Euclidean distance prior , 1 = uniform in comoving volume margphi: 0 = use gaussian likelihood, 1 = phase-marginalised bessel likelihood. More...

REAL8	LALInferenceMarginalisedTimeLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)
	Returns the log-likelihood marginalised over the time dimension from the prior min to the prior max. More...

void	LALInferenceInitLikelihood (LALInferenceRunState *runState)
	Initialisation function which reads runState->commaneLine and sets up the likelihood function accordingly. More...

LALInferenceVariables	LALInferenceGetInstrinsicParams (LALInferenceVariables *currentParams)
	Get the intrinsic parameters from currentParams. More...

REAL8	LALInferenceFastSineGaussianLogLikelihood (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)
	fast SineGaussian likelihood for LIB More...

void	LALInferenceNetworkSNR (LALInferenceVariables currentParams, LALInferenceIFOData data, LALInferenceModel *model)
	Calculate the SNR across the network. More...

Function Documentation

◆ LALInferenceUndecomposedFreqDomainLogLikelihood()

REAL8 LALInferenceUndecomposedFreqDomainLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

(log-) likelihood function.

Returns the non-normalised logarithmic likelihood.

Required (‘currentParams’) parameters are:

"rightascension" (REAL8, radian, 0 <= RA <= 2pi)
"declination" (REAL8, radian, -pi/2 <= dec <=pi/2)
"polarisation" (REAL8, radian, 0 <= psi <= ?)
"distance" (REAL8, Mpc, >0)
"time" (REAL8, GPS sec.)

Definition at line 313 of file LALInferenceLikelihood.c.

◆ LALInferenceZeroLogLikelihood()

REAL8 LALInferenceZeroLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

For testing purposes (for instance sampling the prior), likelihood that returns 0.0 = log(1) every time.

Activated with the –zeroLogLike command flag.

◆ LALInferenceComputeFrequencyDomainOverlap()

REAL8 LALInferenceComputeFrequencyDomainOverlap	(	LALInferenceIFOData *	data,
		COMPLEX16Vector *	freqData1,
		COMPLEX16Vector *	freqData2
	)

Computes the <x|y> overlap in the Fourier domain.

Definition at line 1490 of file LALInferenceLikelihood.c.

◆ LALInferenceComputeFrequencyDomainComplexOverlap()

COMPLEX16 LALInferenceComputeFrequencyDomainComplexOverlap	(	LALInferenceIFOData *	dataPtr,
		COMPLEX16Vector *	freqData1,
		COMPLEX16Vector *	freqData2
	)

Computes the complex <x|y> overlap.

Definition at line 1517 of file LALInferenceLikelihood.c.

◆ LALInferenceNullLogLikelihood()

REAL8 LALInferenceNullLogLikelihood ( LALInferenceIFOData * data )

Identical to LALInferenceFreqDomainNullLogLikelihood, but returns the likelihood of a null template.

Used for normalising.

Definition at line 1543 of file LALInferenceLikelihood.c.

◆ LALInferenceFreqDomainStudentTLogLikelihood()

REAL8 LALInferenceFreqDomainStudentTLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

Student-t (log-) likelihood function
as described in Roever/Meyer/Christensen (2011):
"Modelling coloured residual noise
in gravitational-wave signal processing.

"
Classical and Quantum Gravity, 28(1):015010.
http://dx.doi.org/10.1088/0264-9381/28/1/015010
http://arxiv.org/abs/0804.3853
Returns the non-normalised logarithmic likelihood.

Required (‘currentParams’) parameters are:

"rightascension" (REAL8, radian, 0 <= RA <= 2pi)
"declination" (REAL8, radian, -pi/2 <= dec <=pi/2)
"polarisation" (REAL8, radian, 0 <= psi <= ?)
"distance" (REAL8, Mpc, > 0)
"time" (REAL8, GPS sec.)

This function is essentially the same as the
"UndecomposedFreqDomainLogLikelihood()" function.
The additional parameter to be supplied is the (REAL8)
degrees-of-freedom parameter (nu) for each Ifo.
The additional "df" argument gives the corresponding
d.f. parameter for each element of the "*data" list.
The names of "df" must match the "->name" slot of
the elements of "data".

(TODO: allow for d.f. parameter to vary with frequency,
i.e., to be a set of vectors corresponding to
frequencies)

Definition at line 1480 of file LALInferenceLikelihood.c.

◆ LALInferenceCorrelatedAnalyticLogLikelihood()

REAL8 LALInferenceCorrelatedAnalyticLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

An analytic likeilhood that is a correlated Gaussian in 15 dimensions.

Definition at line 203 of file LALInferenceAnalyticLikelihood.c.

◆ LALInferenceBimodalCorrelatedAnalyticLogLikelihood()

REAL8 LALInferenceBimodalCorrelatedAnalyticLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

An analytic likeilhood that is two correlated Gaussians in 15 dimensions.

Definition at line 268 of file LALInferenceAnalyticLikelihood.c.

◆ LALInferenceRosenbrockLogLikelihood()

REAL8 LALInferenceRosenbrockLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

15-D Rosenbrock log(L) function (see Eq (3) of http://en.wikipedia.org/wiki/Rosenbrock_function .

Definition at line 349 of file LALInferenceAnalyticLikelihood.c.

◆ LALInferenceMarginalisedPhaseLogLikelihood()

REAL8 LALInferenceMarginalisedPhaseLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

Definition at line 1561 of file LALInferenceLikelihood.c.

◆ LALInferenceMarginalisedTimePhaseLogLikelihood()

REAL8 LALInferenceMarginalisedTimePhaseLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

Definition at line 1643 of file LALInferenceLikelihood.c.

◆ LALInferenceMarginalDistanceLogLikelihood()

double LALInferenceMarginalDistanceLogLikelihood	(	double	dist_min,
		double	dist_max,
		double	OptimalSNR,
		double	d_inner_h,
		int	cosmology,
		int	margphi
	)

Compute delta-log-likelihood for given distance min, max and OptimalSNR and d_inner_h when evaluated at 1Mpc cosmology: 0 = Euclidean distance prior , 1 = uniform in comoving volume margphi: 0 = use gaussian likelihood, 1 = phase-marginalised bessel likelihood.

Definition at line 1408 of file LALInferenceLikelihood.c.

◆ LALInferenceMarginalisedTimeLogLikelihood()

REAL8 LALInferenceMarginalisedTimeLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

Returns the log-likelihood marginalised over the time dimension from the prior min to the prior max.

See https://dcc.ligo.org/LIGO-T1400460 for the details.

Definition at line 1633 of file LALInferenceLikelihood.c.

◆ LALInferenceInitLikelihood()

void LALInferenceInitLikelihood ( LALInferenceRunState * runState )

Initialisation function which reads runState->commaneLine and sets up the likelihood function accordingly.

Can choose between Gaussian, Student-t, marginalised phase likelihoods

Definition at line 102 of file LALInferenceLikelihood.c.

◆ LALInferenceGetInstrinsicParams()

LALInferenceVariables LALInferenceGetInstrinsicParams ( LALInferenceVariables * currentParams )

Get the intrinsic parameters from currentParams.

Definition at line 236 of file LALInferenceLikelihood.c.

◆ LALInferenceFastSineGaussianLogLikelihood()

REAL8 LALInferenceFastSineGaussianLogLikelihood	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

fast SineGaussian likelihood for LIB

Definition at line 1653 of file LALInferenceLikelihood.c.

◆ LALInferenceNetworkSNR()

void LALInferenceNetworkSNR	(	LALInferenceVariables *	currentParams,
		LALInferenceIFOData *	data,
		LALInferenceModel *	model
	)

Calculate the SNR across the network.

Definition at line 1866 of file LALInferenceLikelihood.c.