marlintpc::RowBasedHitFinderProcessor Class Reference

A hit finder for row based (pad) geometries It searches for connected pulses in a row and combines them into a hit. More...

#include <RowBasedHitFinderProcessor.h>

Inheritance diagram for marlintpc::RowBasedHitFinderProcessor:

Public Member Functions
virtual Processor *	newProcessor ()
virtual void	init ()
	Called at the begin of the job before anything is read. More...
virtual void	processRunHeader (lcio::LCRunHeader *run)
	Called for every run.
virtual void	processEvent (lcio::LCEvent *evt)
	Called for every event - the working horse. More...
virtual void	check (lcio::LCEvent *evt)
virtual void	end ()
	Called after data processing for clean up.

Detailed Description

A hit finder for row based (pad) geometries It searches for connected pulses in a row and combines them into a hit.

The position inside the row is calculated by a center-of-gravity w.r.t. to the charge of the individual pulses. The position on the row is fixed by the r/y coordinate of the row. The z-position is inherited from the pulse with the highest charge within the hit.

Author

Christoph Rosemann, DESY

Input - Prerequisites

• TrackerPulse collection
• tpcconddata::TPCConditions for the drift velicity (optionally provided as processor parameter)

  Todo:

  Hit quality flag (including inheritance of pulse flags)

  Handling of dead and noisy channels.

Output

• TrackerHit collection

Parameters

InputTrackerPulses	Name of input TrackerPulses collection (default: TPCPulses)
OutputTrackerHits	Name of output TrackerHits collection (default: TPCHits)
InputTPCConditions	OPTIONAL: Name of the input collection containing the TPC conditions data (default: tpcconddata::TPCConditions::getDefaultColName() )
WriteOutputToStorage:bool	If true the output collection is written to file (default: true)
ChargeConversionIsCalibrated:bool	Set true, if the charge conversion factor from ADC values to primary electrons is known (default:true).
MaxEmptyPadsOverride	Maximum number of consecutive empty pads in a row (default: 1)
MinHitSizeOverride	Minimum size of hit in number of pads (default: 1)
MinMaxPulseSliceHeightOverrid	Minimum size of the maximum pulse slice of the highest pulse in the hit (default: 12.)
MaxTimeSpreadOverride	Maximum time between pulses in a hit [in ns] (default: 200)
VDriftOverride	OPTIONAL: Set drift velocity in case there is no conditions data [in mm/us] (default: 0)
DeadPadsOverride	OPTIONAL: a vector of geometry indices of dead channels, default should be conditions data
ChargeConversionFactorOverride	OPTIONAL The factor to convert adc counts into number of primary electrons (-> should be conditions data)
UseGlobalCoordinates	OPTIONAL If set to true, the output coordinate system will be forced to global/cartesian; otherwise it is the local coordinate system of the pad plan (either polar or cartesian) (default: false)

List of open issues/questions:

• extend debug output
• incorporate damaged pad info
• incorporate inheritance of pulse quality
• separation of hits? +++ needs testing!

Member Function Documentation

void marlintpc::RowBasedHitFinderProcessor::init ( )

virtual

Called at the begin of the job before anything is read.

Use to initialize the processor, e.g. book histograms.

void marlintpc::RowBasedHitFinderProcessor::processEvent ( lcio::LCEvent *  evt )

virtual

Called for every event - the working horse.

the processor is supposed to find "hits" in "rows" input are pulses: (charge,time,moduleID,cellID + qualityFlag) output are hits: ( [x,y,z] + covariance matrix, charge + qualityFlag )

steering input: - minimal hit size

• max number of dead/empty pads within a hit
• max distance in z/time between pulses external input: - dead pads
• drift velocity
• charge conversion factor

structurally this happens in a decomposition of:
(A) sort the pulses by module & row
(B) LOOP: look for pulses that belong together in a row (for all rows in a module, for all modules)
(C) create the "Hit" by calculating all numbers

---

the basic data structure for the hit search is a single row; the row is a simple vector of the Pulse pointers: vector<TrackerPulse*>

this is repeated for every (unique) row on a module -> use a map of these repeat this for every possible module, since the module id is unique use a map

so the final data structure for the hit search is: map<int moduleID, map<int rowID, vector<TrackerPulse*>* > (no typedef is introduced, although arguably this might improve the readability)

a helper class is instantiated to hold the hit candidates: "hitCandidate" the temporary result of the hit candidate finding is stored in a vector of these the key is again a map – since it is stored per /unique/ module

---

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

• RowBasedHitFinderProcessor.h
• RowBasedHitFinderProcessor.cc