| Mvd Development Status [message #4230]
||Mon, 14 May 2007 17:05
Registered: May 2007
Location: GSI, Darmstadt
This thread will cover the development communication for the PANDA Mvd in PandaRoot. I'll try to update this first post to have a summary what is done and what is next. Comments on the code itself shall be the central part of the discussion here.
Here is a list of items [09/2008], which have to be done:
- implement root based geometry file [DONE]
- channel mapping [DONE]
- frontend emulation [DONE]
- parameter definitions & database [DONE]
- noise emulation [DONE]
- timestamp ordering for digis [RK]
- multiple hit digi merging [RK]
- Cluster Reconstruction
- alignment [ - ]
- channel back-mapping [DONE]
- cluster finding [DONE]
- output data structure [partly done/additions needed] (genfit compatible)
- Ideal reconstruction [RK (done) is it needed]
- timeordered clusterfinding [RK]
- Tracking/Pattern recognition
- ideal track finding [DONE]
- conformal mapping? [ - ]
- riemann algorithm [DONE]
- local PID [DONE]
- Event mixing
- event time distance sampling [ - ]
- electronics delay & clock [ - ]
More has to come during the discussion.
The following macros are a bit old. They will be revised.
Mvd MC simulation (Geant Transport)
CbmDetector *Mvd = new MvdDetector("MVD", kTRUE);
The output data are MvdPoint
objects (inherited from CbmPoint).
To access them manually you might want to have a look into macro/mvd/anaMvdSim.C.
The Mvd Digitization is almost done. There is only a nice Parameter handling missing and the hard coded values have to be removed. To use the digitization with both Pixel an Strip detectors put the corresponding hitproducers into your macro.
// ====== Hit Producers ======
// ----- MVD Strip hit producer ---------------------------------------
int topFE = 10, botFE = 4, nrFEChannels = 128;
= new MvdStripHitProducer(topPitch, botPitch, orient, skew,
topFE, botFE, nrFEChannels, threshold, noise);
// ----- MVD Pixel hit producer ---------------------------------------
Double_t lx=0.01, ly=0.01, threshold=000, noise=00;
MvdHybridHitProducer* mvdPixProd = new MvdHybridHitProducer(lx,ly,threshold,noise);
// ===== End of HitProducers =====
There will be later a combined task for everything, including the parameter readout from an external ascii file. The output are MvdDigiPixel and MvdDigiStrip data containing mainly Channel and Frontend numbers and Charge.
There is a factory switching the different types of Reconstructions. So far there is only the "Simple" Clusterfinder which is under construction now.
However there is the Monte-Carlo-Clustering there which generates MvdClusters directly from the MvdPoints and which applies a gauss smearing. Here is how you shall use it:
// ----- Ideal Cluster Producers ---------------------------------------
MvdIdealRecoTask* mvdirec = new MvdIdealRecoTask(0.01,0.01,0.005);//(SigmaX,SigmaY,SigmaZ)
The libraries have to have this particular order
until librecotasks doesn't depend on the libtpcreco anymore. Following the tpc example I made a second library, libMvdReco, which depends on libMvd, libgenfit and librecotasks.
The output are MvdCluster and the three parameters are the smearing in the local sensor system
with z being the sensor thickness.
The ideal tracking is started. But we need still some communication with the tracking group and we have to see how to put this together wit the GEANE stuff.
MvdIdealTrackingTask* mvdmctrk = new MvdIdealTrackingTask();
Thoughts on PID
Tobias Baldauf, a student at Dresden, made a task for doing some particle identification on dE/dx for only Mvd Monte-Carlo Points. It is like a MC truth based trackfinding combining the energyloss information.
The corresponding task is MvdIdealPidTask and the Output is MvdPidCand objects.
Double_t trunc=0; //switch for different treatments of calculating the dE/dx mean for a Track
MvdPidIdealTask* mvdpid = new MvdPidIdealTask(trunc);
Have fun, Ralf.
[Updated on: Tue, 16 September 2008 14:28]
Report message to a moderator