#include <EcalUncalibRecHitMultiFitAlgo.h>

Public Member Functions
void	disableErrorCalculation ()

	EcalUncalibRecHitMultiFitAlgo ()

EcalUncalibratedRecHit	makeRecHit (const EcalDataFrame &dataFrame, const EcalPedestals::Item aped, const EcalMGPAGainRatio aGain, const SampleMatrixGainArray &noisecors, const FullSampleVector &fullpulse, const FullSampleMatrix &fullpulsecov, const BXVector &activeBX)
	compute rechits More...

void	setAddPedestalUncertainty (double x)

void	setDoPrefit (bool b)

void	setDynamicPedestals (bool b)

void	setGainSwitchUseMaxSample (bool b)

void	setMitigateBadSamples (bool b)

void	setPrefitMaxChiSq (double x)

void	setSelectiveBadSampleCriteria (bool b)

void	setSimplifiedNoiseModelForGainSwitch (bool b)

	~EcalUncalibRecHitMultiFitAlgo ()

Private Attributes
double	_addPedestalUncertainty

bool	_computeErrors

bool	_doPrefit

bool	_dynamicPedestals

bool	_gainSwitchUseMaxSample

bool	_mitigateBadSamples

double	_prefitMaxChiSq

PulseChiSqSNNLS	_pulsefunc

PulseChiSqSNNLS	_pulsefuncSingle

bool	_selectiveBadSampleCriteria

bool	_simplifiedNoiseModelForGainSwitch

BXVector	_singlebx

Detailed Description

Amplitude reconstucted by the multi-template fit

Author: J.Bendavid, E.Di Marco

Definition at line 21 of file EcalUncalibRecHitMultiFitAlgo.h.

Constructor & Destructor Documentation

EcalUncalibRecHitMultiFitAlgo::EcalUncalibRecHitMultiFitAlgo ( )

Definition at line 8 of file EcalUncalibRecHitMultiFitAlgo.cc.

References _pulsefuncSingle, _singlebx, PulseChiSqSNNLS::disableErrorCalculation(), BXVector< T >::resize(), PulseChiSqSNNLS::setMaxIters(), and PulseChiSqSNNLS::setMaxIterWarnings().

                                                              : 
   _computeErrors(true),
   _doPrefit(false),
   _prefitMaxChiSq(1.0),
   _dynamicPedestals(false),
   _mitigateBadSamples(false),
   _selectiveBadSampleCriteria(false),
   _addPedestalUncertainty(0.),
   _simplifiedNoiseModelForGainSwitch(true),
   _gainSwitchUseMaxSample(false){
     
   _singlebx.resize(1);
   _singlebx << 0;
   
   _pulsefuncSingle.disableErrorCalculation();
   _pulsefuncSingle.setMaxIters(1);
   _pulsefuncSingle.setMaxIterWarnings(false);
     
 }

EcalUncalibRecHitMultiFitAlgo::~EcalUncalibRecHitMultiFitAlgo ( )

inline

Definition at line 27 of file EcalUncalibRecHitMultiFitAlgo.h.

References makeRecHit().

27 { };

Member Function Documentation

void EcalUncalibRecHitMultiFitAlgo::disableErrorCalculation ( )

inline

Definition at line 29 of file EcalUncalibRecHitMultiFitAlgo.h.

References _computeErrors.

Referenced by EcalUncalibRecHitWorkerMultiFit::set().

29 { _computeErrors = false; }

EcalUncalibRecHitMultiFitAlgo::_computeErrors

bool _computeErrors

Definition: EcalUncalibRecHitMultiFitAlgo.h:42

EcalUncalibratedRecHit EcalUncalibRecHitMultiFitAlgo::makeRecHit	(	const EcalDataFrame &	dataFrame,
		const EcalPedestals::Item *	aped,
		const EcalMGPAGainRatio *	aGain,
		const SampleMatrixGainArray &	noisecors,
		const FullSampleVector &	fullpulse,
		const FullSampleMatrix &	fullpulsecov,
		const BXVector &	activeBX
	)

compute rechits

Definition at line 29 of file EcalUncalibRecHitMultiFitAlgo.cc.

References _addPedestalUncertainty, _computeErrors, _doPrefit, _dynamicPedestals, _gainSwitchUseMaxSample, _mitigateBadSamples, _prefitMaxChiSq, _pulsefunc, _pulsefuncSingle, _selectiveBadSampleCriteria, _simplifiedNoiseModelForGainSwitch, _singlebx, funct::abs(), EcalMGPASample::adc(), CustomPhysics_cfi::amplitude, PulseChiSqSNNLS::BXs(), PulseChiSqSNNLS::ChiSq(), PulseChiSqSNNLS::disableErrorCalculation(), PulseChiSqSNNLS::DoFit(), PulseChiSqSNNLS::Errors(), flags, EcalMGPAGainRatio::gain12Over6(), EcalMGPAGainRatio::gain6Over1(), ecalMGPA::gainId(), EcalMGPASample::gainId(), EcalDataFrame::hasSwitchToGain1(), EcalDataFrame::hasSwitchToGain6(), EcalDataFrame::id(), EcalDataFrame::isSaturated(), RecoTauDiscriminantConfiguration::mask, hpstanc_transforms::max, EcalDataFrame::MAXSAMPLES, EcalPedestal::mean_x1, EcalPedestal::mean_x12, EcalPedestal::mean_x6, muonCSCDigis_cfi::pedestal, EcalPedestal::rms_x1, EcalPedestal::rms_x12, EcalPedestal::rms_x6, simplePhotonAnalyzer_cfi::sample, EcalDataFrame::sample(), EcalUncalibratedRecHit::setAmplitudeError(), mps_update::status, and PulseChiSqSNNLS::X().

Referenced by EcalUncalibRecHitWorkerMultiFit::run(), and ~EcalUncalibRecHitMultiFitAlgo().

                                                                                                                                                                                                                                                                                                            {
 
   uint32_t flags = 0;
   
   const unsigned int nsample = EcalDataFrame::MAXSAMPLES;
   
   double maxamplitude = -std::numeric_limits<double>::max();
   double maxgainratio = -std::numeric_limits<double>::max();
   unsigned int iSampleMax = 0;
   
   double pedval = 0.;
     
   SampleVector amplitudes;
   SampleGainVector gainsNoise;
   SampleGainVector gainsPedestal;
   SampleGainVector badSamples = SampleGainVector::Zero();
   bool hasSaturation = dataFrame.isSaturated();
   bool hasGainSwitch = hasSaturation || dataFrame.hasSwitchToGain6() || dataFrame.hasSwitchToGain1();
   
   //no dynamic pedestal in case of gain switch, since then the fit becomes too underconstrained
   bool dynamicPedestal = _dynamicPedestals && !hasGainSwitch;
   
   for(unsigned int iSample = 0; iSample < nsample; iSample++) {
         
     const EcalMGPASample &sample = dataFrame.sample(iSample);
     
     double amplitude = 0.;
     int gainId = sample.gainId();
     
     double pedestal = 0.;
     double gainratio = 1.;
     
     if (gainId==0 || gainId==3) {
       pedestal = aped->mean_x1;
       gainratio = aGain->gain6Over1()*aGain->gain12Over6();
       gainsNoise[iSample] = 2;
       gainsPedestal[iSample] = dynamicPedestal ? 2 : -1;  //-1 for static pedestal
     }
     else if (gainId==1) {
       pedestal = aped->mean_x12;
       gainratio = 1.;
       gainsNoise[iSample] = 0;
       gainsPedestal[iSample] = dynamicPedestal ? 0 : -1; //-1 for static pedestal
     }
     else if (gainId==2) {
       pedestal = aped->mean_x6;
       gainratio = aGain->gain12Over6();
       gainsNoise[iSample] = 1;
       gainsPedestal[iSample] = dynamicPedestal ? 1 : -1; //-1 for static pedestals
     }
 
     if (dynamicPedestal) {
       amplitude = (double)(sample.adc())*gainratio;
     }
     else {
       amplitude = ((double)(sample.adc()) - pedestal) * gainratio;
     }
     
     if (gainId == 0) {
        edm::LogError("EcalUncalibRecHitMultiFitAlgo")<< "Saturation encountered.  Multifit is not intended to be used for saturated channels.";
       //saturation
       if (dynamicPedestal) {
         amplitude = 4095.*gainratio;
       }
       else {
         amplitude = (4095. - pedestal) * gainratio;
       }
     }
         
     amplitudes[iSample] = amplitude;
     
     if (gainratio > maxgainratio || gainId==0 || amplitude>maxamplitude) {
     //if (iSample==5) {
       maxamplitude = amplitude;
       maxgainratio = gainratio;
       iSampleMax = iSample;
       pedval = pedestal;
     }
         
   }
 
   double amplitude, amperr, chisq;
   bool status = false;
     
   //special handling for gain switch, where sample before maximum is potentially affected by slew rate limitation
   //optionally apply a stricter criteria, assuming slew rate limit is only reached in case where maximum sample has gain switched but previous sample has not
   //option 1: use simple max-sample algorithm
   if (hasGainSwitch && _gainSwitchUseMaxSample) {
     EcalUncalibratedRecHit rh( dataFrame.id(), maxamplitude, pedval, 0., 0., flags );
     rh.setAmplitudeError(0.);
     for (unsigned int ipulse=0; ipulse<_pulsefunc.BXs().rows(); ++ipulse) {
       int bx = _pulsefunc.BXs().coeff(ipulse);
       if (bx!=0) {
         rh.setOutOfTimeAmplitude(bx+5, 0.0);
       }
     }
     return rh;
   }
 
   //option2: A floating negative single-sample offset is added to the fit
   //such that the affected sample is treated only as a lower limit for the true amplitude
   bool mitigateBadSample = _mitigateBadSamples && hasGainSwitch && iSampleMax>0;
   mitigateBadSample &= (!_selectiveBadSampleCriteria || (gainsNoise.coeff(iSampleMax-1)!=gainsNoise.coeff(iSampleMax)) );
   if (mitigateBadSample) {
     badSamples[iSampleMax-1] = 1;
   }
   
   //compute noise covariance matrix, which depends on the sample gains
   SampleMatrix noisecov;
   if (hasGainSwitch) {
     std::array<double,3> pedrmss = {{aped->rms_x12, aped->rms_x6, aped->rms_x1}};
     std::array<double,3> gainratios = {{ 1., aGain->gain12Over6(), aGain->gain6Over1()*aGain->gain12Over6()}};
     if (_simplifiedNoiseModelForGainSwitch) {
       int gainidxmax = gainsNoise[iSampleMax];
       noisecov = gainratios[gainidxmax]*gainratios[gainidxmax]*pedrmss[gainidxmax]*pedrmss[gainidxmax]*noisecors[gainidxmax];
       if (!dynamicPedestal && _addPedestalUncertainty>0.) {
         //add fully correlated component to noise covariance to inflate pedestal uncertainty
         noisecov += _addPedestalUncertainty*_addPedestalUncertainty*SampleMatrix::Ones();
       }
     }
     else {
       noisecov = SampleMatrix::Zero();
       for (unsigned int gainidx=0; gainidx<noisecors.size(); ++gainidx) {
         SampleGainVector mask = gainidx*SampleGainVector::Ones();
         SampleVector pedestal = (gainsNoise.array()==mask.array()).cast<SampleVector::value_type>();
         if (pedestal.maxCoeff()>0.) {
           //select out relevant components of each correlation matrix, and assume no correlation between samples with
           //different gain
           noisecov += gainratios[gainidx]*gainratios[gainidx]*pedrmss[gainidx]*pedrmss[gainidx]*pedestal.asDiagonal()*noisecors[gainidx]*pedestal.asDiagonal();
           if (!dynamicPedestal && _addPedestalUncertainty>0.) {
             //add fully correlated component to noise covariance to inflate pedestal uncertainty
             noisecov += gainratios[gainidx]*gainratios[gainidx]*_addPedestalUncertainty*_addPedestalUncertainty*pedestal.asDiagonal()*SampleMatrix::Ones()*pedestal.asDiagonal();
           }
         }
       }
     }
   }
   else {
     noisecov = aped->rms_x12*aped->rms_x12*noisecors[0];
     if (!dynamicPedestal && _addPedestalUncertainty>0.) {
       //add fully correlated component to noise covariance to inflate pedestal uncertainty
       noisecov += _addPedestalUncertainty*_addPedestalUncertainty*SampleMatrix::Ones();
     }
   }
   
   //optimized one-pulse fit for hlt
   bool usePrefit = false;
   if (_doPrefit) {
     status = _pulsefuncSingle.DoFit(amplitudes,noisecov,_singlebx,fullpulse,fullpulsecov,gainsPedestal,badSamples);
     amplitude = status ? _pulsefuncSingle.X()[0] : 0.;
     amperr = status ? _pulsefuncSingle.Errors()[0] : 0.;
     chisq = _pulsefuncSingle.ChiSq();
     
     if (chisq < _prefitMaxChiSq) {
       usePrefit = true;
     }
   }
   
   if (!usePrefit) {
   
     if(!_computeErrors) _pulsefunc.disableErrorCalculation();
     status = _pulsefunc.DoFit(amplitudes,noisecov,activeBX,fullpulse,fullpulsecov,gainsPedestal,badSamples);
     chisq = _pulsefunc.ChiSq();
     
     if (!status) {
       edm::LogWarning("EcalUncalibRecHitMultiFitAlgo::makeRecHit") << "Failed Fit" << std::endl;
     }
 
     unsigned int ipulseintime = 0;
     for (unsigned int ipulse=0; ipulse<_pulsefunc.BXs().rows(); ++ipulse) {
       if (_pulsefunc.BXs().coeff(ipulse)==0) {
         ipulseintime = ipulse;
         break;
       }
     }
     
     amplitude = status ? _pulsefunc.X()[ipulseintime] : 0.;
     amperr = status ? _pulsefunc.Errors()[ipulseintime] : 0.;
   
   }
   
   double jitter = 0.;
   
   EcalUncalibratedRecHit rh( dataFrame.id(), amplitude , pedval, jitter, chisq, flags );
   rh.setAmplitudeError(amperr);
   
   if (!usePrefit) {
     for (unsigned int ipulse=0; ipulse<_pulsefunc.BXs().rows(); ++ipulse) {
       int bx = _pulsefunc.BXs().coeff(ipulse);
       if (bx!=0 && std::abs(bx)<100) {
         rh.setOutOfTimeAmplitude(bx+5, status ? _pulsefunc.X().coeff(ipulse) : 0.);
       }
       else if (bx==(100+gainsPedestal[iSampleMax])) {
         rh.setPedestal(status ? _pulsefunc.X().coeff(ipulse) : 0.);
       }
     }
   }
   
   return rh;
 }