Pulsar::ProfileShiftFit Class Reference

Calculates profile shifts by fitting to a template/standard. More...

#include <ProfileShiftFit.h>

Inheritance diagram for Pulsar::ProfileShiftFit:

Public Types
enum	Error_Method { Traditional_Chi2 , MCMC_Variance , Numerical }
	Method to use for calculating shift error.

Public Member Functions
	ProfileShiftFit ()
	Default constructor.
	~ProfileShiftFit ()
	Destructor.
void	reset ()
	Reset everything to zero, free internal memory.
void	set_nharm (unsigned nh)
	Set the number of harmonics to use for fit.
unsigned	get_nharm ()
	Get the number of harmonics currently in use.
void	set_error_method (Error_Method e)
	Set the error method.
void	set_mcmc_iterations (int nit)
	Set number of iterations for MCMC errors.
int	get_mcmc_iterations () const
	Get number of iterations.
void	set_standard (const Profile *p)
	Set the standard or template profile to use.
void	set_Profile (const Profile *p)
	Set the data profile to use.
void	compute ()
	Run the fit.
Tempo::toa	toa (const Integration *)
	Return a TOA object for the current fit.
Estimate< double >	get_shift ()
	Get the resulting shift.
Estimate< double >	get_scale ()
	Get the resulting scale factor.
double	get_mse ()
	Get the resulting Mean Squared Error (per fit DOF)
double	get_sigma2 ()
	Get the resulting noise per harmonic.
double	get_snr ()
	Get the resulting SNR.
double	get_reduced_chisq () const
	Get the reduced chi-squared.
double	get_effective_duty_cycle () const
	Get the effective duty cycle of the standard.
void	apply_scale_and_shift (Profile *p)
	Determine, then apply a shift a scale to data profile.
Public Member Functions inherited from Reference::Able
	Able (const Able &)
Able &	operator= (const Able &)
unsigned	get_reference_count () const
Public Member Functions inherited from Reference::HeapTracked
	HeapTracked (const HeapTracked &)
HeapTracked &	operator= (const HeapTracked &)
bool	__is_on_heap () const

Public Attributes
int	mcmc_trials
	Current MCMC accept stats.
int	mcmc_accept
bool	choose_maximum_harmonic
	Set true when set_standard should choose the maximum harmonic.

Protected Member Functions
void	init ()
	Initialize vars.
void	choose_nharm ()
	Choose the maximum number of harmonics using LastHarmonic.
double	ccf (double phi)
	Evaluate ccf at phase shift phi.
double	dccf (double phi)
	Evaluate d(ccf)/d(phi) at phase shift phi.
double	d2ccf (double phi)
	Evaluate d^2(ccf)/d(phi)^2 at phase shift phi.
double	log_shift_pdf (double phi)
	Returns log of a value proportional to the posterior shift PDF.
double	log_shift_pdf_pos (double phi)
	Same as above, but with scale>0 prior.
void	error_traditional ()
	Calculate "traditional" parameter uncertainties.
void	error_numerical ()
	Calculate "traditional" parameter uncertainties.
void	error_mcmc_pdf_var ()
	Calculate shift uncertainty as posterior PDF variance using MCMC.
Protected Member Functions inherited from Reference::Able
Handle *	__reference (bool active) const
void	__dereference (bool auto_delete=true) const

Protected Attributes
unsigned	nharm
	Number of harmonics to use.
unsigned	effective_nharm
	Effective number of harmonics = min (nharm, nbins_prof/2-1)
Reference::To< const Profile >	std
	Current template profile.
unsigned	nbins_std
	Current template nbin.
float *	fstd
	R2C FFT of current template.
double	std_pow
	Template normalization factor.
Reference::To< const Profile >	prof
	Current data profile.
unsigned	nbins_prof
	Number of bins in profile.
float *	fprof
	R2C FFT of current profile.
float *	fccf
	Current freq domain cross-correlation function.
unsigned	nbins_ccf
	Number of bins in ccf.
Error_Method	err_meth
	Error method to use.
int	mcmc_it
	Number of iterations to use for MCMC.
bool	computed
	Have valid results been computed.
double	shift
	Current shift result.
double	eshift
	Current shift error.
double	correction
	Correction for profile nbins != template nbins.
double	scale
	Current scale result.
double	escale
	Current scale error.
double	sigma2
	Current sigma^2 estimate.
double	mse
	Current fit MSE.
double	chi2
	Current fit Chi^2 (not reduced)
unsigned	dof
	Current degrees of freedom.
double	snr
	Current profile SNR.

Additional Inherited Members
Static Public Member Functions inherited from Reference::Able
static size_t	get_instance_count ()
Static Public Member Functions inherited from Reference::HeapTracked
static void *	operator new (size_t size, void *ptr=0)
static void	operator delete (void *location, void *ptr)
static void	operator delete (void *location)
static size_t	get_heap_queue_size ()

Detailed Description

Calculates profile shifts by fitting to a template/standard.

This class essentially reimplements the standard "Taylor fftfit" timing algorithm (used in PSRCHIVE's PhaseGradientShift method, etc). This has several benefits over the existing implementation in model_profile.C:

1. It caches the FFT'd template, resulting in about half as many FFT calls (when timing many profiles) versus the old routines.
2. It uses only the standard PSRCHIVE FFT interface, and avoids the old Fortran "fccf" routines.
3. The Chi2 minimization procedure is much more straightforward - at least in the author's opinion ;)
4. Additional information (fitted amplitude, profile sigma, etc) is returned in a cleaner, more extensible object oriented way.
5. Supports multiple/alternate phase shift error calculation algorithms.

Member Function Documentation

apply_scale_and_shift()

void Pulsar::ProfileShiftFit::apply_scale_and_shift

(

Profile *

p

)

Determine, then apply a shift a scale to data profile.

Convenience function to determine, then apply, scale and shift.

References compute(), Pulsar::Profile::rotate_phase(), Pulsar::Profile::scale(), scale, set_Profile(), and shift.

get_effective_duty_cycle()

double Pulsar::ProfileShiftFit::get_effective_duty_cycle

(

)

const

Get the effective duty cycle of the standard.

Refer to Equation 13 of van Straten (2006) or Equation B1 of Downs & Reichley (1983) [these equations are Fourier transform pairs]. The effective duty cycle is obtained when gain, g, and noise, sigma, both equal unity.

References fstd, Pulsar::ProfileStats::get_onpulse_nbin(), Pulsar::ProfileStats::get_total(), Estimate< typename T, typename U >::get_value(), nharm, and Pulsar::ProfileStats::set_profile().

---

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

• ProfileShiftFit.h
• ProfileShiftFit.C