Belle2::SVD Namespace Reference

Namespace to encapsulate code needed for simulation and reconstrucion of the SVD. More...

Classes
class	DefaultWave
	A functor to provide a simple model of APV25 strip response. More...
class	EmpiricalDistributionFunction
	Empirical distribution function object is basic for mainpulation of probabilities. More...
class	GeoSVDCreator
	The creator for the SVD geometry of the Belle II detector. More...
class	NNWaveFitter
	The class uses a neural network to find a probability distribution of arrival times for a sextet of APX25 signal samples. More...
class	NNWaveFitTool
	The class holds arrays of bins and bin centers, and a wave generator object containing information on the waveform function. More...
class	NoiseMap
	Base Class to represent strip-dependent noise map. More...
class	SensorInfo
	Specific implementation of SensorInfo for SVD Sensors which provides additional sensor specific information. More...
class	SimpleClusterCandidate
	Class representing a cluster candidate during simple clustering of the SVD. More...
struct	stripInCluster
	structure containing the relevant informations of eachstrip of the cluster More...
class	SVDBackgroundModule
	SVD Background module. More...
class	SVDBeamBackHitFilterModule
	The SVDBeamBackHitFilter module. More...
class	SVDClusterizerDirectModule
	SVD Direct Clusterizer. More...
class	SVDDatabaseTestModule
	Return the calibration results for the noises and the constants measured during the svd local runs (charge, ADC pulse, peaking time, pulse width). More...
class	SVDDigitizerModule
	The SVD Digitizer module. More...
class	SVDDQMInjectionModule
	The SVD Occupancy after Injection DQM module. More...
class	SVDMissingAPVsClusterCreatorModule
	SVDMissingAPVsClusterCreatorModule: The SVD MissingAPVsClusterCreator. More...
class	SVDNNClusterizerModule
	SVD NN Clusterizer. More...
class	SVDNNShapeReconstructorModule
	The SVD NNShapeReconstructor. More...
class	SVDPackerModule
	SVDPackerModule: The SVD Raw Hits Creator. More...
class	SVDShaperDigitSorterModule
	The SVDShaperDigitSorter module. More...
class	SVDSignal
	The SVD signal class. More...
class	SVDSimpleClusterizerModule
	SVDSimpleClusterizerModule: The SVD SimpleClusterizer. More...
class	SVDUnpackerModule
	SVDUnpackerModule: The SVD Raw Hits Decoder. More...
class	TauEncoder
	Encoder/decoder for neural network tau values. More...
class	WaveFitter
	Waveform fitter class. More...
class	WaveGenerator
	Waveform generator This is a functor to calculate signal values. More...

Typedefs
typedef std::map< short int, SVDSignal >	StripSignals
	Map of all signals in one sensor.
typedef std::pair< StripSignals, StripSignals >	Sensor
	Signals of u- and v- strips in one sensor.
typedef std::map< VxdID, Sensor >	Sensors
	Map of all signals in all sensors.
typedef std::vector< double >	nnFitterBinData
	Vector of values defined for bins, such as bin times or bin probabilities.
typedef std::vector< double >	nnFitterBins
	Vector of bin edges, nnFitterBinData.size() + 1.
typedef VXD::SensitiveDetector< SVDSimHit, SVDTrueHit >	SensitiveDetector
	The SVD Sensitive Detector class.
typedef double	apvSampleBaseType
	Vector of input samples in float form.
typedef std::array< apvSampleBaseType, nAPVSamples >	apvSamples
	vector od apvSample BaseType objects
typedef std::function< double(double)>	waveFunction
	Wavefrom function type. More...

Functions
double	w_expo (double t)
	Gamma waveform, x.exp(-x) This is only historically useful. More...
double	w_poly3 (double t)
	Polynomial waveform, x. More...
double	w_betaprime (double t)
	Beta-prime wave function, x^alpha/(1+x)^beta. More...
template<typename T >
void	zeroSuppress (T &a, double thr)
	pass zero suppression
template<typename T >
bool	pass3Samples (const T &a, double thr)
	pass 3-samples
double	tau_raw2real (double raw_tau)
	Convert Hao's raw tau (integral, in latency units) to correct betaprime scale. More...
	TEST (NNTimeFitter, DISABLED_CompareNetworkCoefficient)
	Compare NN fitter parameter dump from Python with internal data representation in C++ to verify that the network coefficients were transferred without a significant loss of accuracy. More...
	TEST (NNTimeFitter, DISABLED_CompareFits)
	Read a sample of fits from the Python NN suite and check that the C++ NN fitter produces the same results. More...
	TEST (SVDModeByte, Constructor)
	Check that both constructors yield the same result.
	TEST (SVDModeByte, getters)
	Check getters and setters.
	TEST (SVDRecoDigit, ConstructEmpty)
	Check empty object creation and data getters.
	TEST (SVDRecoDigit, ConstructFromContainer)
	Check object creation and data getters using a stl container of probabilities.
	TEST (SVDRecoDigit, ToString)
	Check object creation and data getters using a stl container of probabilities.
	TEST (SVDRecoTimeBase, ConstructEmpty)
	Check empty object creation and data getters.
	TEST (SVDRecoTimeBase, ConstructFromContainer)
	Check object creation and data getters.
	TEST (SVDRecoTimeBase, toString)
	Check object printout.
	TEST (SVDShaperDigit, ConstructEmpty)
	Check empty object creation and data getters.
	TEST (SVDShaperDigit, ConstructFromContainer)
	Check object creation and data getters using a stl container of samples.
	TEST (SVDShaperDigit, ConstructFromCArray)
	Check object creation and data getters using a c-array of samples.
	TEST (SVDShaperDigit, ConstructDefaultTimeMode)
	Check standard object creation without FADC time and mode information.
	TEST (SVDShaperDigit, SampleTrimming)
	test sample trimming
	TEST (SVDShaperDigit, TrimmingFunction)
	test sample trimming
	TEST (SVDSignal, Getters)
	Check object creation and simple object getters.
	TEST (SVDSignal, Constructors)
	Check the different constructors.
	TEST (SVDSignal, Assignment)
	Check assignment operators.
	TEST (SVDSignal, Waveform)
	Check the waveform values and relations.
	TEST (SVDSignal, toString)
	Check the waveform string representation.
	TEST (TNiel, output)
	Check correct functioning of the class.

Variables
geometry::CreatorFactory< GeoSVDCreator >	GeoSVDFactory ("SVDCreator")
	Register the creator.
const std::size_t	nAPVSamples = 6
	Number of APV samples.
const double	dt_APV = 31.44 * Unit::ns
	APV sampling time.
const apvSamples	apvTimeBase
	APV time base - times for the 6 signals. More...

Detailed Description

Namespace to encapsulate code needed for simulation and reconstrucion of the SVD.

Typedef Documentation

waveFunction

typedef std::function<double(double)> waveFunction

Wavefrom function type.

This is the type for a naked wave function, giving a single value for a properly scaled arguemnt. The functions are scaled to 1.0 at mode, location (shift) 0.0, and scale 1.0.

Definition at line 56 of file SVDSimulationTools.h.

Function Documentation

tau_raw2real()

double Belle2::SVD::tau_raw2real ( double  raw_tau )

inline

Convert Hao's raw tau (integral, in latency units) to correct betaprime scale.

Includes scaling and fit adjustmenta

Definition at line 155 of file SVDSimulationTools.h.

{ return 7.32313 * raw_tau; }

TEST() [1/2]

Belle2::SVD::TEST	(	NNTimeFitter	,
		DISABLED_CompareFits
	)

Read a sample of fits from the Python NN suite and check that the C++ NN fitter produces the same results.

This is the result of select_t0_fits.columns: Index([ 'test', 'amplitude', 't0', 'tau', 'sigma', 's1', 's2', 's3', 's4', 's5', 's6', 'normed_tau', 't0_bin', 'abin', 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 'a_fit', 'a_sigma', 'chi2_ndf', 't_fit', 't_sigma', 't_true', 'tau', 'a_true', 't_bin', 'a_bin', 't_diff','a_diff'], dtype='object') Columns labeled by numerals contain probabilities.

Definition at line 55 of file NNFitterTest.cc.

    {
      const size_t max_lines = 100; // maximum number of lines to be read
 
      // Create an instance of the NN fitter and the fitter tool.
      NNWaveFitter fitter("SVDTimeNet_6samples");
      auto fitTool = fitter.getFitTool();
      size_t nProbs = fitTool.getBinCenters().size();
 
      ifstream infile("svd/data/test_sample.csv");
 
      // Read the rows one by one and compare results
      string line;
      getline(infile, line);
 
      for (size_t i_line = 0; i_line < max_lines; i_line++) {
 
        getline(infile, line);
        if (line.size() < 10) break;
        istringstream sline(line);
 
        // Parse header. We want the dimennsion of the probability array.
        // not needed
        string cell;
        getline(sline, cell, ','); // index
        getline(sline, cell, ','); // test
 
        // true values. Read from the file, though not used.
        getline(sline, cell, ',');
        [[maybe_unused]] double true_amp = stod(cell);
        getline(sline, cell, ',');
        [[maybe_unused]] double true_t0 = stod(cell);
        getline(sline, cell, ',');
        [[maybe_unused]] double width = stod(cell);
        getline(sline, cell, ',');
        [[maybe_unused]] double noise = stod(cell);
 
        // normalized samples
        apvSamples normedSamples;
        for (size_t iSample = 0; iSample < nAPVSamples; ++iSample) {
          getline(sline, cell, ',');
          normedSamples[iSample] = stod(cell);
        }
 
        // not needed
        getline(sline, cell, ',');
        getline(sline, cell, ',');
        getline(sline, cell, ',');
 
        // probabilities
        nnFitterBinData ProbsPy(nProbs);
        for (size_t iSample = 0; iSample < nProbs; ++iSample) {
          getline(sline, cell, ',');
          ProbsPy[iSample] = stod(cell);
        }
 
        // fit results
        getline(sline, cell, ',');
        double fitPy_amp = stod(cell);
        getline(sline, cell, ',');
        double fitPy_ampSigma = stod(cell);
        getline(sline, cell, ',');
        [[maybe_unused]] double fitPy_chi2 = stod(cell);
        getline(sline, cell, ',');
        double fitPy_t0 = stod(cell);
        getline(sline, cell, ',');
        double fitPy_t0Sigma = stod(cell);
 
        // now do the Cpp fit
        const shared_ptr<nnFitterBinData> ProbsCpp = fitter.getFit(normedSamples, width);
        for (size_t iBin = 0; iBin < nProbs; ++iBin)
          EXPECT_NEAR((*ProbsCpp)[iBin], ProbsPy[iBin], 5.0e-3);
 
        double t0_cpp, t0_err_cpp;
        tie(t0_cpp, t0_err_cpp) = fitTool.getTimeShift(*ProbsCpp);
        EXPECT_NEAR(t0_cpp, fitPy_t0, 5);
        EXPECT_NEAR(t0_err_cpp, fitPy_t0Sigma, 2);
 
        double amp_cpp, amp_err_cpp, chi2_cpp;
        tie(amp_cpp, amp_err_cpp, chi2_cpp) = fitTool.getAmplitudeChi2(normedSamples, t0_cpp, width);
        EXPECT_NEAR(amp_cpp, fitPy_amp, 1.0);
        EXPECT_NEAR(amp_err_cpp, fitPy_ampSigma, 0.1);
        // FIXME: This is calculated slightly differently in Python, and it shows.
        // EXPECT_NEAR(chi2_cpp, fitPy_chi2, 1);
      }
    }

TEST() [2/2]

Belle2::SVD::TEST	(	NNTimeFitter	,
		DISABLED_CompareNetworkCoefficient
	)

Compare NN fitter parameter dump from Python with internal data representation in C++ to verify that the network coefficients were transferred without a significant loss of accuracy.

FIXME (Oct 12,2017: This test fails on buildbot, but ran successfully on all machines I tried.

Definition at line 37 of file NNFitterTest.cc.

w_betaprime()

double Belle2::SVD::w_betaprime ( double  t )

inline

Beta-prime wave function, x^alpha/(1+x)^beta.

This is the function for general use.

Parameters

t	Properly scaled time, (t - t0)/tau.
tau	Waveform width

Returns

6x2 Eigen matrix, times in 1st column, signals in the 2nd.

Definition at line 89 of file SVDSimulationTools.h.

w_expo()

double Belle2::SVD::w_expo ( double  t )

inline

Gamma waveform, x.exp(-x) This is only historically useful.

Use beta-prime instead.

Parameters

t	properly scaled time, (t-t0)/tau

Returns

Waveform value t * exp(1 - t) for t > 0, else 0

Definition at line 63 of file SVDSimulationTools.h.

w_poly3()

double Belle2::SVD::w_poly3 ( double  t )

inline

Polynomial waveform, x.

(1-x)^2 This is used in signal calibration in SVD hardware tests. Do not use for signal simulation.

Parameters

t	properly scaled time, (t-t0)/tau

Returns

Waveform value 27/4 * t * (1 - t)^2 for t in <0,1>, else 0

Definition at line 75 of file SVDSimulationTools.h.

Variable Documentation

apvTimeBase

const apvSamples apvTimeBase

Initial value:

=
    {{ -dt_APV, 0.0, dt_APV, 2 * dt_APV, 3 * dt_APV, 4 * dt_APV}}

APV time base - times for the 6 signals.

Definition at line 44 of file SVDSimulationTools.h.