rpdf C++ API Reference¶

class FastConvolutedPandel : public rpdf::PhotoElectronProbability ¶

#include <pandel.h>

Contains functions for calculating the convoluted Pandel function and it’s survival function using approximations of these functions which were optimized to be as fast as possible.

Intended to be used with I3RecoLLH parameter PEProb=”GaussConvoluted”

Public Functions

inline FastConvolutedPandel(const double jitter, const rpdf::IceModel &ice_model)¶

Creates a FastConvolutedPandel object for calculating probabilities from the convoluted pandel function

Parameters:

jitter – the width of the Gaussian which is convoluted with the pandel function
ice_model – a struct containing the description of optical properties of the ice

virtual double pdf(const double t_res, const double eff_distance) const¶

Calculate the convoluted pandel function using the approximation by van Eindhoven et al, by breaking it into 5 regions and evaluating a different approximation in each region.

Parameters:

t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track

Returns:

the probability density of a photon arriving at t=t_res

virtual double sf(const double t_res, const double eff_distance) const¶

Calculate complementary cumulative distribution function of the convoluted pandel function using an approximate integral by D. Chirkin.

The approximation replaces the Gaussian with a box of the same first and second moment. An additional term is added to account for the behavior at the singularity at t=0. This approximation is very accurate. See https://icecube.wisc.edu/~dima/work/WISC/cpdf/a.pdf for more information

Parameters:

t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track

Returns:

the probability of a photon arriving at t>=t_res

Private Members

const double jitter_¶

const rpdf::IceModel ice_model_¶

struct I3HitCache¶

Structure for caching the information needed to calculate the likelihood. This data is the same regardless of the hypothesis so it is computed once per event and cached so it doesn’t have to be calculated for each iteration

Public Members

double total_npe¶: Total number of Photoelectrons observed by the DOM in the event.

double first_pulse_time¶: The time of the first pulsse.

I3Position pos¶: position of the DOM

class I3RecoLLH : public I3EventLogLikelihoodBase ¶

#include <I3RecoLLH.h>

A gulliver likelihood service for reconstructions which use the Pandel function.

It calculates the likelihood of a muon track hypothesis using an analytic description of light propagation through the ice. This analytic description is referred to as the Pandel function. This implementation assumes all ice in the detector has uniform optical properties.

Public Functions

I3RecoLLH(const std::string &input_readout, const std::string &likelihood, const std::string &peprob, const double jitter, const double noise, const rpdf::IceModel &ice)¶

Create a new I3RecoLLH object.

Parameters:

input_readout – The name of the I3RecoPulseSeriesMap to read from the frame
likelihood – The name of the DOM likelihood algorithm to use. Options are “GaussConvoluted” or “UnconvolutedPandel”
peprob – The name of the photoelectron probability calculation to use Options are “SPE1st” or “MPE”
jitter – the width of the Gaussian which is convoluted with the pandel function
noise – The frequency of noise hits to use in the reconstruction
ice_model – A struct containing the description of optical properties of the ice

virtual ~I3RecoLLH()¶

void SetGeometry(const I3Geometry &geo)¶

This is called when the reader gets a new geometry. It saves the geometry for SetEvent() to save the location of each DOM in the hit cache

Parameters:: geo – the Geometry to use to calculate likelihoods

void SetEvent(const I3Frame &frame)¶

Called when a new event occurs, this reads the new event and the pulse map in the frame. and calls SetHitCache()

Parameters:: frame – the frame to extract the I3RecoPulseSeriesMap from

void SetPulseMap(const I3RecoPulseSeriesMap &pulse_map)¶

Calculates the hit cache from the geometry and given pulse map

Parameters:: pulse_map – map containing the pulses for calculating the likelihood

double GetLogLikelihood(const I3EventHypothesis &event_hypothesis)¶

Calculate the log likelihood of an event hypothesis for the current event

Parameters:: event_hypothesis – the gulliver event hypothesis (which contains an I3Particle) with which to calculate the likelihood
Returns:: the likelihood of the hypothesis

unsigned int GetMultiplicity()¶

Returns:: the multiplicity of the event in question: the number of hit DOMs

void SetName(std::string name)¶: changes the name of this particular instance of I3RecoLLH

const std::string GetName() const¶

Returns:: a string which describes this particular instance of I3RecoLLH

SET_LOGGER ("I3RecoLLH")

Private Members

std::string name_¶: A descriptive string representing the instance of this class.

const std::string inputReadout_¶: The name of the I3FrameObject to read from the frame.

const std::string likelihood_¶: The name of the likelihood function to use: SPE1st or MPE.

const std::string peprob_¶: The name of the Photoelectron probability to use: FastConvoluted or Unconvoluted.

const double jitter_¶: The time scale of the DOM jitter to consider when reconstructing.

const double noise_¶: The frequency of noise hits to use in the reconstruction.

const rpdf::IceModel ice_model_¶: The ice model object to store the optical properties in.

rpdf::DOMLikelihoodFunction dom_likelihood_func_¶: Instance of the DOM likelihood function object.

std::shared_ptr<rpdf::PhotoElectronProbability> pe_prob_¶: Instance of the Photoelectron probability function object.

I3GeometryConstPtr geoptr_¶: A pointer to the geometry to store.

std::vector<I3HitCache> hit_cache_¶: a list of the pertinent information for each hit in an event stored in a vector for fast access

class I3RecoLLHFactory : public I3ServiceFactory ¶

#include <I3RecoLLHFactory.h>

A I3Service factory to install an I3RecoLLH into the context

Public Functions

I3RecoLLHFactory(const I3Context &context)¶

virtual ~I3RecoLLHFactory()¶

virtual bool InstallService(I3Context &services)¶: Installs this service in the context

virtual void Configure()¶: Gets the parameters defined by the steering file from the context and initializes the instance of I3RecoLLH

Private Functions

SET_LOGGER ("I3RecoLLHFactory")

Private Members

std::string inputreadout_¶: The name of the I3FrameObject to read from the frame.

std::string likelihood_¶: The name of the likelihood function to use: SPE1st or MPE.

double jitter_¶: The time scale of the DOM jitter to consider when reconstructing.

double noiseProb_¶: The frequency of noise hits to use in the reconstruction.

std::string name_¶: the name to install the I3RecoLLH instance in the context

std::string peprob_¶: The name of the Photoelectron probability to use: FastConvoluted or Unconvoluted.

I3EventLogLikelihoodBasePtr llh_¶: pointer to the instance of the object installed in the context

rpdf::IceModel ice_model_¶: instance of the ice model parameters to use

struct IceModel¶

#include <geometry.h>

This structure holds all the relevant parameters to describe the optical properties of ice used for reconstruction

Public Functions

inline IceModel(const double a, const double t, const double s, const double p1, const double p0c0, const double p0c1, const double p0c2)¶

IceModel is mostly a dumb struct, however the constructor computes rho from the other parameters

Parameters:

absorption_length – The distance a photon can go before being absorbed on average
tau_scale – The pandel tau parameter, represents the non-linear behavior of multiple photon scatters
scattering_length – The distance a photon can go on average before being absorbed
P1 – The coefficient in front of the distance when calculating effective distance
P0_CS0 – The constant coefficient when calculating effective distance
P0_CS1 – The coefficient in front of cos when calculating effective distance
P0_CS2 – The coefficient in front of cos^2 when calculating effective distance

Public Members

double absorption_length¶: The distance a photon can go before being absorbed on average.

double tau_scale¶: The pandel tau parameter, represents the non-linear behavior of multiple photon scatters.

double scattering_length¶: The distance a photon can go on average before being absorbed.

double P1¶: The coefficient in front of the distance when calculating effective distance.

double P0_CS0¶: The constant coefficient when calculating effective distance.

double P0_CS1¶: The coefficient in front of cos when calculating effective distance.

double P0_CS2¶: The coefficient in front of cos^2 when calculating effective distance.

double rho¶: rho is a combination of the other optical properties, it is the scale parameter in the Pandel function

struct MPEfunc¶

#include <pandel.h>

Public Functions

double operator()(const PhotoElectronProbability &p, const double t_res, const double eff_distance, const double Npe) const¶

Boost.Function for calculating the MPE likelihood for an individual DOM.

MPE is the likelihood of observing the first photon at t=t_res given that the DOM saw Npe photoelectrons total. This is intended for use with I3RecoLLH parameter DOMLikelihood=MPE.

Parameters:

p – the class representing the photoelectron probability calculation
t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track
Npe – the number of photoelectrons observed by this DOM in this event

Returns:

the MPE likelihood for this DOM

class PhotoElectronProbability¶

#include <pandel.h>

Base Class for Photoelectron Probabilities. I3RecoLLH uses a plugable system to pick which PE probability to use. Those object inherent from this class.

Subclassed by rpdf::FastConvolutedPandel, rpdf::UnconvolutedPandel

Public Functions

inline virtual ~PhotoElectronProbability()¶

virtual double pdf(const double t, const double d) const = 0¶

virtual function for the probability density of a photon arriving at t=t_res at a distance d from the track

Parameters:

t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track

Returns:

the probability density of a photon arriving at t=t_res

virtual double sf(const double t, const double d) const = 0¶

virtual function for complementary cumulative distribution function: the probability of observing a hit at t_res or later a distance eff_distance away from the track

Parameters:

t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track

Returns:

the probability of a photon arriving at t>=t_res

struct SPEfunc¶

#include <pandel.h>

Public Functions

double operator()(const PhotoElectronProbability &p, const double t_res, const double eff_distance, const double Npe) const¶

Boost.Function for calculating the SPE1st likelihood for an individual DOM.

It takes the time of the first from the hit series and simply returns the PDF for that hit. Intended for use with I3RecoLLH parameter DOMLikelihood=SPE1st.

Parameters:

p – the class representing the photoelectron probability calculation
t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track
Npe – the number of photoelectrons observed by this DOM in this event

Returns:

the SPE1st Likelihood for this DOM

class UnconvolutedPandel : public rpdf::PhotoElectronProbability ¶

#include <pandel.h>

contains functions for calculating the pandel function without any convolution. Intended for use with I3RecoLLH with parameter PEProb=”UnconvolutedPandel”

Public Functions

inline UnconvolutedPandel(const rpdf::IceModel &ice_model)¶

Construct an UnconvolutedPandel object

Parameters:: ice_model – ice struct containing the description of the ice

virtual double pdf(const double t, const double d) const¶

Calculate the unconvouted pandel function using boost’s implementation of the gamma distribution.

Parameters:

t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track

Returns:

the probability density of a photon arriving at t=t_res

virtual double sf(const double t, const double d) const¶

calculate complementary cumulative distribution function of the unconvouted pandel function using gsl’s implementation of the gamma distribution

Parameters:

t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track

Returns:

the probability of a photon arriving at t>=t_res

Private Members

const rpdf::IceModel ice_model_¶

namespace [anonymous]

namespace rpdf¶

Typedefs

typedef boost::function<double(const PhotoElectronProbability &p, const double t_res, const double d_eff, const double Npe)> DOMLikelihoodFunction¶: function declaration for the DOM Likelihoods

Functions

double effective_distance(const double distance, const double cos_eta, const IceModel &ice_model)¶

Calculate the effective distance between an emitter and a DOM due to scattering. This is a small correction for DOMs orientated away from the source.

Parameters:

distance – the actual distance traveled by the photon
cos_eta – the cosine of the angle between the photon’s trajectory and the direction the DOM is pointing
ice_model – the object holding the description of the ice

Returns:

the equivalent distance that would have experienced the same amount of scattering if the DOM was orientated toward the photon path

double cherenkov_time(const double d_track, const double d_approach)¶

Calculate the time Cherenkov photon arrives at a position d_track along the track, to a DOM located d_approach from the track.

Parameters:

d_track – The distance along the track from the vertex to the point of closest approach
d_approach – The distance of closest approach between the DOM and track

Returns:

the time at which the Cherenkov photon arrives at the DOM

std::pair<double, double> muon_geometry(const I3Position &om, const I3Particle &track, const IceModel &ice_model)¶

This calculates the two geometrical parameters needed by the pandel function to calculate the PDF.

Because the calculations relating to these function are very tightly coupled they are both calculated in a single function

Parameters:

om – the position of the optical module being hit
track – the muon track hypothesis
ice_model – the model used to describe the optical properties of the ice

Returns:

t_hit the difference in time between the hit and the expected time of a direct photon

Returns:

effective_distance the effective distance between the track hypothesis and the OM including scattering needed to hit an OM facing the other way

double pandel_pdf(const double t_res, const double eff_distance, const IceModel &ice_model)¶

Pure function implementation of the Pandel PDF: the probability of observing a hit at t_res a distance eff_distance away from the track

Parameters:

t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track
ice_model – a struct containing the description of optical properties of the ice

Returns:

the probability density of a photon arriving at t=t_res

double pandel_sample(const double eff_distance, const IceModel &ice_model, I3RandomService &rng)¶

Sample a photon arrival time from the Pandel PDF.

Parameters:

eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track
ice_model – a struct containing the description of optical properties of the ice
rng – an I3RandomService to use to sample random numbers

Returns:

a time residual relative to the direct Cherenkov time at which the photon hits a DOM (see t_res above).

double pandel_sf(const double t_res, const double eff_distance, const IceModel &ice)¶

Pure function implementation of the Pandel complementary cumulative distribution function: the probability of observing a hit at t_res or later a distance eff_distance away from the track

Parameters:

t_res – the time residual of the of a photon which hits a DOM
eff_distance – the distance of the DOM from the hypothesis track accounting for extra scatters caused by the PMT looking away from the track
ice_model – struct containing the description of optical properties of the ice

Returns:

the probability of a photon arriving at t>=t_res

double gslConvoluted1F1Diff(const double xi, const double eta)¶

double gslConvolutedU(const double xi, const double eta)¶

double fastConvolutedHyperg(const double xi, const double eta)¶

Compute the hypergeometric portion of the Gaussian convoluted Pandel PDF.

1F1(0.5*xi,0.5,0.5*eta*eta)/gamma(0.5*(xi+1)) - sqrt2*eta*1F1(0.5*(xi+1),1.5,0.5*eta*eta)/gamma(0.5*xi) for 0.5*eta*eta < 100 and xi < 5 using the hypergeometric power series with the denominator precomputed. This is >10x faster than GSL.

Variables

IceModel H0 (98., 596.0, 36.93, 0.9045, 4.249,-6.629, 5.430): AMANDA ice model 0.

IceModel H1 (98., 578.0, 35.04, 0.8682, 3.615,-5.081, 5.015): AMANDA ice model 1.

IceModel H2 (98., 556.7, 33.29, 0.8395, 3.094,-3.946, 4.636): AMANDA ice model 2, this is what is reported in the AMANDA muon reconstruction paper. It is the default ice model in I3RecoLLH.

IceModel H3 (98., 542.2, 31.72, 0.8106, 2.683,-3.090, 4.436): AMANDA ice model 3.

IceModel H4 (98., 526.3, 29.94, 0.7790, 2.139,-0.9614, 4.020): AMANDA ice model 4.

namespace constants¶

Variables

const double C_VACUUM = 2.997924e-1¶: Speed of light in vacuum, less precision than what is found in I3Constants because this is what ipdf did.

const double N_ICE_G = I3Constants::n_ice_group ¶: Group index of refraction of ice.

const double C_ICE_G = C_VACUUM / N_ICE_G ¶: group velocity in ice

const double SIN_CHERENKOV = sin(I3Constants::theta_cherenkov)¶: the sine of the Cherenkov angle

const double TAN_CHERENKOV = tan(I3Constants::theta_cherenkov)¶: the tangent of the Cherenkov angle

const double EFF_TAN_CHERENKOV = N_ICE_G / SIN_CHERENKOV - 1 / TAN_CHERENKOV ¶: a grouping of constants that is always the same in the calculation. It is called the effective tangent of the Cherenkov angle because It is that if the group and phase angles are the same

namespace std: STL namespace.

file geometry.cxx

#include “dataclasses/physics/I3Particle.h”

#include “rpdf/geometry.h”

This file contains functions pertaining to geometrical calculations needed for reconstructing muons.

Author: Kevin Meagher
Date: January 2018

file geometry.h

#include <utility>

#include “dataclasses/I3Constants.h”

functions for calculating geometrical parameters for muon reconstructions

Author: Kevin Meagher
Date: January 2018

file I3RecoLLH.cxx

#include “gulliver/I3EventHypothesis.h”

#include “rpdf/I3RecoLLH.h”

Provides a Gulliver likelihood service for muon based reconstructions.

Author: Kevin Meagher
Date: January 2018

file I3RecoLLH.h

#include “dataclasses/geometry/I3Geometry.h”

#include “gulliver/I3EventLogLikelihoodBase.h”

#include “rpdf/pandel.h”

#include “rpdf/geometry.h”

Provides a Gulliver likelihood service for muon based reconstructions.

Author: Kevin Meagher
Date: January 2018

file I3RecoLLHFactory.cxx: #include “rpdf/I3RecoLLH.h”

#include “rpdf/I3RecoLLHFactory.h”

file I3RecoLLHFactory.h

#include “icetray/I3ServiceFactory.h”

#include “gulliver/I3EventLogLikelihoodBase.h”

#include “rpdf/geometry.h”

Provides a Gulliver likelihood service for muon based reconstructions.

Author: Kevin Meagher
Date: January 2018

file pandel.cxx

#include <gsl/gsl_math.h>

#include <gsl/gsl_rng.h>

#include <gsl/gsl_randist.h>

#include <gsl/gsl_sf_gamma.h>

#include <gsl/gsl_sf_hyperg.h>

#include <gsl/gsl_sf_erf.h>

#include <gsl/gsl_errno.h>

#include <boost/math/distributions/gamma.hpp>

#include “dataclasses/physics/I3Particle.h”

#include “rpdf/geometry.h”

#include “rpdf/pandel.h”

functions for calculating photoelectron probabilities for muon reconstructions

Author: Kevin Meagher
Date: January 2018

file pandel.h

#include <boost/function.hpp>

#include “dataclasses/physics/I3RecoPulse.h”

#include “phys-services/I3RandomService.h”

#include “rpdf/geometry.h”

functions for calculating photoelectron probabilities for muon reconstructions

Author: Kevin Meagher
Date: January 2018

dir icetray

dir private

dir public

dir rpdf

dir rpdf

dir rpdf

rpdf C++ API Reference¶

Table of Contents

Previous topic

Next topic

This Page