Project

General

Profile

Bug #1692

Tracker Reconstruction not proceeding beyond Digits for real data

Added by Dobbs, Adam over 6 years ago. Updated about 6 years ago.

Status:
Closed
Priority:
Normal
Assignee:
Category:
Tracker
Target version:
Start date:
17 June 2015
Due date:
% Done:

100%

Estimated time:
Workflow:
New Issue

Description

Issue to track progress on fixing the tracker reconstruction for real data. Now we have data and the unpacker is working the reconstruction is producing digits. No subsequent cluster are produced however.


Files

#1

Updated by Dobbs, Adam over 6 years ago

I am currently looking at run 6989, file 6989.000. The data can be obtained from:

http://www.hep.shef.ac.uk/research/mice/uploads/MDR3_data_temp/

It can be analysed using the MAUS trunk (not release) branch with a command similar to the following:

python bin/analyze_data_offline.py --daq_data_path /vols/mice3/adobbs/maus/data/6989/ --daq_data_file 6989.000  --Number_of_DAQ_Events 5 --Enable_VLSB_Unpacking=True

At present I am adding cerr statements to the code to see what is failing...

#2

Updated by Dobbs, Adam over 6 years ago

Found a first failure mode in cluster recon. A large number of initial digits are found, 185426, of which 4026 are found to be suitable for forming into clusters (pass the npe cut). This leads to failure as a maximum number of seeds is specified via ConfigurationDefaults, with a default value of SciFiClustExcept = 100. Will try raising this value to 10000 and see what happens.

Question (Ed?): How many digits are expecting per DAQ event and spill? Is this number reasonable?

#3

Updated by Dobbs, Adam over 6 years ago

Raising the allowed number of seeds allowed the recon to progress, but second error quickly appeared:

DEBUG: SciFiClusterRec: Found good cluster
Traceback (most recent call last):
  File "/vols/mice3/adobbs/maus/maus-shared/cluster_fix/src/common_py/ErrorHandler.py", line 159, in HandleCppException
    raise CppError(error_message)
ErrorHandler.CppError: Failed to find SciFi plane in _geometry_map. at SciFiClusterRec::process_cluster

No clusters produced in the final ROOT file. Investigating geometry handling next...

#4

Updated by Dobbs, Adam over 6 years ago

Aside: The code is running spectacularly slowly. 16mins for 5 DAQ events. Will need to investigating that soon too.

#5

Updated by Dobbs, Adam over 6 years ago

Received advice from C. Heidt:

  • Speed - turn on VLSB zero suppression: --Do_VLSB_Zero_Suppression=True
  • Change VLSB threshold from 60 to 40: --VLSB_Zero_Suppression_Threshold=40
  • Add an npe cut in RealDataDigitization of 4
  • Use a specified simulation geometry not the default: --simulation_geometry_file=Stage4.dat

This lead to a massive speed up and the generation of spacepoints. Someone at RAL buy Heidt a beer.

Will try to get this code ready for control room deployment for this evening run, and then move on to looking for tracks.

#6

Updated by Rajaram, Durga over 6 years ago

Does one need special map & calibration for this? Vanilla trunk from the repository gives empty scifi events

#7

Updated by Dobbs, Adam over 6 years ago

Apologies Durga, it does indeed, I will create a dev branch to play with asap, and then the trunk update after that...

#8

Updated by Dobbs, Adam over 6 years ago

OK, lp:phuccj/maus/merge-candidate has a version which should now produce spacepoints, though the Stage4.dat geometry file must be specified.

#9

Updated by Dobbs, Adam over 6 years ago

ROOT file with tracker only recon data, 100 spills, with spacepoints, for run 6989 added.

#10

Updated by Rajaram, Durga over 6 years ago

Following up on some concerns about whether the unpacking unpacks the rest of the detectors along with the tracker data.
The above run [ 6989 ] contains only tracker data.
I looked at 7002 which has other detectors in it as well.
Attaching root output from MAUSing 200 spills, see all detectors including tracker coming through.
They were processed with the dev branch and cards listed in comments 5,8

#11

Updated by Dobbs, Adam over 6 years ago

Confirmed - able to see tracker and other detector data. In addition, using this run to mock the onrec the .maus_dev branch now works on both onrec02 and onrec03 (though still with crashes). Not seeing good straights tracks yet, next thing to attend to...

#12

Updated by Dobbs, Adam about 6 years ago

  • Status changed from Open to Closed
  • % Done changed from 0 to 100

We now have straight tracks in MAUS 0.9.6. Closing the issue.

#13

Updated by Dobbs, Adam about 6 years ago

  • Target version changed from Future MAUS release to MAUS-v0.9.6

Also available in: Atom PDF