Abstract
We discuss a PSTuple package, which is intended for creation, accumulation and storage of histograms and tuples in psana project.
Objectives
In psana project we need in package which allows to accumulate data in form of histograms and tuples and save them in file for further analysis. Though it might be based on well-known underlying packages like HBOOK, ROOT, or HippoTuple , we prefer to use an uniform interface with abstract base class, substituting the direct interaction with underlying methods. This intermediate abstract layer provides a flexibility in implementation of new things, for example multithreding in analysis, which algorithm is not yet defined.
Interface
An example of the program interface for this histogram-tuple-management package can be presented as follows. First, the underlying (derived) package needs to be choisen and instantiated. For example for ROOT
HManager *hMan = new RootHist("my-data-file.root"); // for ROOT
Then later in the program we may create ntuple(s) without knowing what is the underlying histogram-tuple-storage management system
Tuple *nt = hMan->ntuple("EXP Data");
and histograms
H1 *pHis1i = hMan->his1i("His1 int title",100,0.,1.); H1 *pHis1f = hMan->his1f("His1 float title",100,0.,1.); H1 *pHis1d = hMan->his1d("His1 double title",100,0.,1.); H2 *pHis2d = hMan->his2d("His2 double title",100,0.,1.,100,0.,1.);
When the ntuple is created, its parameters need to be defined once
TuplePar *pBeamEnergy = nt->parameter("beamEnergy"); // minValue, maxValue, dtype ? TuplePar *pBeamCurrent = nt->parameter("beamCurrent");
Then, in data processing stage, for example for each event, we may accumulate data in created histogram and ntuple objects using pointers
pBeamEnergy ->fill(E); pBeamCurrent->fill(I); nt ->addRow();
or parameter names
nt ->fill("beamEnergy", E); nt ->fill("beamCurrent", I); nt ->addRow();
Histograms can be accumulated using their pointers
pHis1i ->fill(x,[weight]); ... pHis2 ->fill(x,y,[weight]);
where we assume that all parameters like x
, y
, E
, and I
and optional weight
were earlier defined.
When you are all done, write the data into a file:
hMan->write(); delete hMan;
Structure and content of the package
Package PSTuple contains the base abstract class and methods, which have to be implemented in derived classes for HBOOK, ROOT, HippoTuple, etc., i.e. in packages HBookTuple, RootTuple, HippoTuple, etc.
Histogramming in BABAR
#include "HepTuple/Histogram.h" #include "HepTuple/TupleManager.h" HepTupleManager* manager = gblEnv->getGen()->ntupleManager(); HepTuple *_ntuple = manager->ntuple("file-name.root"); _ntuple->column("run", eventID->run(), -99, "Event" ); _ntuple->column("event", eventCounter, -99, "Event" ); _ntuple->dumpData();
Histogramming in CLEO
Histograms
Declare Histogram in your .h file
#include "HistogramInterface/HistogramPackage.h’’ ... HIHist1D *m_trackMom;
Define Histogram in your Processor
m_trackMom = iHistoManager.histogram("track momentum (GeV)",50, 0, 1.5);
Fill Histogram in event() method of your Processor
m_trackMom->fill(momentum);
Load Conversion Module in .tcl script
module sel RootHistogramModule root file root_suez_style_example_data31.root root init
NTuple
Declare Ntuple in your Processor’s .h file
#include "HistogramInterface/HistogramPackage.h’’ #include ‘‘HistogramInterface/HINtupleVarNames.h’’ ... HINtuple *m_trackTuple;
Make an enum of variable names
enum{kpx, kpy, kpz, kd0, kz0, knum_vars};
Define Ntuple in hist_book()
// make an object that holds the names of the variables in your ntuple
HINtupleVarNames knames(knum_vars); knames.addVar(kpx,"px"); knames.addVar(kpy,"py"); knames.addVar(kpz,"pz"); knames.addVar(kd0,"d0"); knames.addVar(kz0,"z0"); m_trackTuple = iHistoManager.ntuple(1,"tracktuple",knum_vars,50000,knames.names());
Fill Ntuple in event() method
trackTuple[kpx] = px; trackTuple[kpy] = py; trackTuple[kpz] = pz; trackTuple[kd0] = d0; trackTuple[kz0] = z0; m_trackTuple->fill(trackTuple);