Difference between revisions of "Calibration Train"

From GlueXWiki
Jump to: navigation, search
(→‎Calibration Challenge)
(→‎Procedures)
 
(49 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
__TOC__
 
__TOC__
  
 +
= Processing Overview =
  
−
= Calibration Challenge =
+
* [[Old Calibration Train]]
  
−
We propose to run a "Calibration Challenge" starting the first week of December 2015.
+
== Job Structure ==
  
−
The goal will be to take a run with zeroed out calibrations and to see how many calibrations can be automatically extracted.
+
Currently, two passes are planned:  One automated step, and one to produce outputs for calibration procedures that are still manual.
  
−
Test runs will be performed in the weeks leading up to this challenge.
+
The calibrations/plugins that are run on each pass are:
 +
* '''Pass 1'''
 +
** Run as many calibrations as possible on one file
 +
*# Pass 1: RF_online
 +
*# Pass 2: HLDetectorTiming,TOF_TDC_shift
 +
*# Pass 3: st_tw_corr_auto [don't commit]
 +
*# Pass 4: HLDetectorTiming, CDC_amp, BCAL_TDC_Timing [time offsets, need to update]
 +
* '''Pass 2'''
 +
** Process/skim full run
 +
** Calibrations:  BCAL_attenlength_gainratio, BCAL_LEDonline, CDC_amp,CDC_TimeToDistance, FCALpedestals, FCALpulsepeak, FCAL_TimingOffsets, HLDetectorTiming, imaging, PSC_TW,  PS_timing, pedestals,ST_Propagation_Time
 +
** EVIO skims: FCAL pi0, BCAL pi0, BCAL-LED, FCAL-LED, random, sync
 +
** ROOT skims: TOF_calib
 +
** Other [Monitoring]: BCAL_LED, BCAL_inv_mass, imaging, p2pi_hists, p3pi_hists
 +
* '''Incoming'''
 +
** Tagger/PS workflow
 +
** BCAL LED monitoring
  
−
More details to follow.
+
== Calibration run plan ==
  
−
= Organization =
+
The two priority items are to recalibrate the TOF with the its new running conditions and to verify calibrations with the new fADC250 firmware.  All calibrations should be checked, in any case.
  
−
* The jobs will be submitted every Tuesday at noon, JLab time.
+
# Standalone calibrations (no forward tracking)
−
* The jobs will be run from the gxproj3 account  [parallel use with EventStore jobs]
+
#* ✓ RF time (Paul M.)
 +
#* Pedestals (System owners)
 +
#* BCAL Attenuation length/gain ratio (Mark D.)
 +
#* CDC time to distance (Mike S.)
 +
#* CDC gain (Naomi)
 +
#* ✓ SC timewalk (Mahmoud)
 +
#* TOF timing (Offsets/timewalks) (Beni)
 +
#* ✓ TAGM timing (Alex B.)
 +
#* ✓ TAGH timing (Nathan)
 +
#* ✓ PS timing (Nathan)
 +
#* ✓ Overall timing (rough) (Mike S./Sean)
 +
# Full tracking calibrations
 +
#* ✓ BCAL Effective velocities (George)
 +
#* ✓ Overall timing (Mike S./Sean)
 +
#* ✓ SC Propagation time (Mahmoud)
 +
# ✓ BCAL/FCAL pi0 calibrations (Adesh/Will M.)
  
−
* The output of the jobs will be stored in ...
+
All calibrations except the pi0 calibration should take no more than 1-3 2h runs of data.
  
−
= Run Ranges =  
+
=== Frequency ===
  
−
The following runs will be processed:
+
* Timing will be checked for each run.  The known variations are:
 +
** TOF (run-to-run)
 +
** Tagger (~<day, corrected in run-to-run in the spring)
 +
* CDC gains vary with temperature and pressure, can be averaged over a 1-2 hour run.
  
−
'''RunPeriod-2015-03'''
+
All other calibrations have been seen (so far) to be stable on a ~several week timescale
−
* 2931, 3079, 3179, 3180, 3183, 3185
+
  
−
= Calibrations =
+
=== To-dos ===
  
−
=== Job Requirements ===
+
# <strike>Finish skim improvements</strike>
 +
# Automate constant -> CCDB pipeline
 +
# Implement tracking database
  
−
Each calibration process should include the following:
+
= Procedures =
−
* Plugin stored in standard location
+
−
** https://halldsvn.jlab.org/repos/trunk/sim-recon/src/plugins/Calibration
+
−
* ROOT/other scripts stored in standard location
+
−
** https://halldsvn.jlab.org/repos/trunk/sim-recon/src/scripts/calibrations ?
+
−
** Individual git repos?
+
−
* Output to files
+
−
* QA routines
+
  
−
=== What is Being Run ===  
+
=== How to start a new run period ===
  
−
The following plugins are currently being run:
+
# Edit configuration file, e.g., $CALIBRATION_TRAIN/configs/data.config
−
* RF_online (RF signal)
+
#* Change job name to the current run period and set resource limits
−
* BCAL_TDC_Timing 
+
# Edit $CALIBRATION_TRAIN/template/job_wrapper.sh
−
* HLDetectorTiming 
+
#* Set run period and version number
−
* PSC_TW
+
# Set up the files and directories
 +
#* e.g. python setup_run.py configs/data.config
 +
# Create workflows
 +
#* swif create -workflow GXCalib-2017-01-pass1
 +
#* swif create -workflow GXCalib-2017-01-pass2
 +
# Create SQLite CCDB
 +
#* $CCDB_HOME/scripts/mysql2sqlite/mysql2sqlite.sh -hhallddb.jlab.org -uccdb_user ccdb | sqlite3 ccdb.sqlite
 +
#* mv ccdb.sqlite somewhere
 +
# Launch pass1 jobs
 +
#* python run_jobs_p1.py 2017-01 run_lists/f17.test
  
−
Working on adding:
+
= Calibration Outputs =
−
* BCAL_attenlength_gainratio
+
−
* BCAL gains - /work/halld/home/wmcginle/Gain_Calib
+
−
* TOF calibrations - https://halldsvn.jlab.org/repos/trunk/home/zihlmann/TOF_calib/
+
  
−
= Code =
+
== RunPeriod-2017-01 ==
  
−
The current code can be obtained at
+
{| class="wikitable"
−
<syntaxhighlight>
+
! Run Range
−
svn co https://halldsvn.jlab.org/repos/trunk/home/sdobbs/calibration_train/
+
! Version tag
−
</syntaxhighlight>
+
! Total # Jobs
 +
! Notes
 +
! Total Skim Size (TB)
 +
! BCAL pi0 (GB)
 +
! BCAL LED (GB)
 +
! FCAL pi0 (GB)
 +
! FCAL LED (GB)
 +
! PS Skim (TB)
 +
! TOF Skim (TB)
 +
|-
 +
| 30274 - 30621
 +
| ver01
 +
| 13362
 +
|
 +
| 11.2
 +
| 338
 +
| 74
 +
| 862
 +
| 147
 +
| 6.36
 +
| 1.63
 +
|-
 +
| 30622 - 30959
 +
| ver02
 +
| 13783
 +
| looser BCAL pi0 cuts
 +
| 23.3
 +
| 2379
 +
| 131
 +
| 1703
 +
| 297
 +
| 12.7
 +
| 3.36
 +
|-
 +
| 30960 -
 +
| ver03
 +
| 14869
 +
| looser BCAL pi0 cuts
 +
| 13.4
 +
| 1368
 +
| 71
 +
| 1000
 +
| 142
 +
| 7.47
 +
| 1.99
 +
|}
  
−
The ROOT library directory can be found at:
+
* Skim files can be found in the following directory: /cache/halld/RunPeriod-2017-01/calib/ver01
−
...
+
** BCAL-LED - BCAL LED triggered events
 +
** BCAL_pi0 - BCAL pi0 candidates
 +
** FCAL-LED - FCAL LED triggered events
 +
** FCAL_pi0 - FCAL pi0 candidates
 +
** PS - PS triggered events
 +
** random - random (out-of-time) triggered events
 +
** sync - TS sync events
 +
** TOF - TOF calibration ROOT skim
  
−
= Older Information =
+
<!--
 +
ver01:
 +
/mss/halld/RunPeriod-2017-01/calib/ver01/hists
 +
total = 1673.9482654 GB
 +
/mss/halld/RunPeriod-2017-01/calib/ver01/BCAL-LED
 +
total = 74.0351243354 GB
 +
/mss/halld/RunPeriod-2017-01/calib/ver01/BCAL_pi0
 +
total = 338.330394119 GB
 +
/mss/halld/RunPeriod-2017-01/calib/ver01/FCAL-LED
 +
total = 146.729467098 GB
 +
/mss/halld/RunPeriod-2017-01/calib/ver01/FCAL_pi0
 +
total = 862.115063164 GB
 +
/mss/halld/RunPeriod-2017-01/calib/ver01/PS
 +
total = 6364.3464742 GB
 +
/mss/halld/RunPeriod-2017-01/calib/ver01/TOF
 +
total = 1636.12805751 GB
 +
/mss/halld/RunPeriod-2017-01/calib/ver01/random
 +
total = 128.437043864 GB
 +
/mss/halld/RunPeriod-2017-01/calib/ver01/sync
 +
total = 3.4401245974 GB
  
−
== Proposal ==
 
  
−
* Use case: Streamline tape library usage, provide common environment for production and development
+
ver02:
−
** Example: Users develop plugins on their favorite files/run, use this for running over larger data
+
−
* Run every week (Wednesday to avoid conflict with monitoring?)
+
−
* Uses subset of runs
+
−
* Users provide:
+
−
** DANA plugin
+
−
** Optional post-processing scripts to be run after every/all runs
+
−
*** Curated in SVN
+
−
* Results stored in standard location(s)
+
−
** Possible results: ROOT files, images, calibration constants, web pages
+
−
* Uses SWIF? (buzzword compliance)
+
−
* Uses GlueX project account (gxproj3? 4? 5?)
+
  
−
== Runs to Use ==
 
  
−
Several possibilities:
+
ver03:
−
 
+
-->
−
# Large/Popular Runs
+
−
#* Fall 2014: 1514, 1769, 1777, 1787, 1803, 1807, 1810, 1825, 1847, 1852, 1854, 1871, 1872, 2138, 2206, 2207, 2209, 2223, 2397
+
−
#* Spring 2015: 2931, 3079, 3179, 3180, 3183, 3185
+
−
# Other big sets of runs:
+
−
#* Fall 2014: 1516-1520, 1769-1777, 1900, 2179-2184, 2205, 2228, 2407-2409, 2416-2420
+
−
#* Look for groups of runs with similar conditions
+
−
# Others
+
−
 
+
−
 
+
−
 
+
−
This also raises the question: what are the good runs?
+
−
 
+
−
One proposal:
+
−
* Tag each run in monitoring database with one of three values:
+
−
*# 0 - non-production quality
+
−
*# 1 - production quality
+
−
*# 2 - production quality, used for calibrations
+
−
* Idea is that > 1 means that the run is good to use for physics
+
−
** Finer grained information can be stored in RCDB
+
−
* To determine production quality, develop quality metrics for each subdetector, use combination of quality metrics and eye test
+

Latest revision as of 23:53, 16 January 2018

Processing Overview

Job Structure

Currently, two passes are planned: One automated step, and one to produce outputs for calibration procedures that are still manual.

The calibrations/plugins that are run on each pass are:

  • Pass 1
    • Run as many calibrations as possible on one file
    1. Pass 1: RF_online
    2. Pass 2: HLDetectorTiming,TOF_TDC_shift
    3. Pass 3: st_tw_corr_auto [don't commit]
    4. Pass 4: HLDetectorTiming, CDC_amp, BCAL_TDC_Timing [time offsets, need to update]
  • Pass 2
    • Process/skim full run
    • Calibrations: BCAL_attenlength_gainratio, BCAL_LEDonline, CDC_amp,CDC_TimeToDistance, FCALpedestals, FCALpulsepeak, FCAL_TimingOffsets, HLDetectorTiming, imaging, PSC_TW, PS_timing, pedestals,ST_Propagation_Time
    • EVIO skims: FCAL pi0, BCAL pi0, BCAL-LED, FCAL-LED, random, sync
    • ROOT skims: TOF_calib
    • Other [Monitoring]: BCAL_LED, BCAL_inv_mass, imaging, p2pi_hists, p3pi_hists
  • Incoming
    • Tagger/PS workflow
    • BCAL LED monitoring

Calibration run plan

The two priority items are to recalibrate the TOF with the its new running conditions and to verify calibrations with the new fADC250 firmware. All calibrations should be checked, in any case.

  1. Standalone calibrations (no forward tracking)
    • ✓ RF time (Paul M.)
    • Pedestals (System owners)
    • BCAL Attenuation length/gain ratio (Mark D.)
    • CDC time to distance (Mike S.)
    • CDC gain (Naomi)
    • ✓ SC timewalk (Mahmoud)
    • TOF timing (Offsets/timewalks) (Beni)
    • ✓ TAGM timing (Alex B.)
    • ✓ TAGH timing (Nathan)
    • ✓ PS timing (Nathan)
    • ✓ Overall timing (rough) (Mike S./Sean)
  2. Full tracking calibrations
    • ✓ BCAL Effective velocities (George)
    • ✓ Overall timing (Mike S./Sean)
    • ✓ SC Propagation time (Mahmoud)
  3. ✓ BCAL/FCAL pi0 calibrations (Adesh/Will M.)

All calibrations except the pi0 calibration should take no more than 1-3 2h runs of data.

Frequency

  • Timing will be checked for each run. The known variations are:
    • TOF (run-to-run)
    • Tagger (~<day, corrected in run-to-run in the spring)
  • CDC gains vary with temperature and pressure, can be averaged over a 1-2 hour run.

All other calibrations have been seen (so far) to be stable on a ~several week timescale

To-dos

  1. Finish skim improvements
  2. Automate constant -> CCDB pipeline
  3. Implement tracking database

Procedures

How to start a new run period

  1. Edit configuration file, e.g., $CALIBRATION_TRAIN/configs/data.config
    • Change job name to the current run period and set resource limits
  2. Edit $CALIBRATION_TRAIN/template/job_wrapper.sh
    • Set run period and version number
  3. Set up the files and directories
    • e.g. python setup_run.py configs/data.config
  4. Create workflows
    • swif create -workflow GXCalib-2017-01-pass1
    • swif create -workflow GXCalib-2017-01-pass2
  5. Create SQLite CCDB
    • $CCDB_HOME/scripts/mysql2sqlite/mysql2sqlite.sh -hhallddb.jlab.org -uccdb_user ccdb | sqlite3 ccdb.sqlite
    • mv ccdb.sqlite somewhere
  6. Launch pass1 jobs
    • python run_jobs_p1.py 2017-01 run_lists/f17.test

Calibration Outputs

RunPeriod-2017-01

Run Range Version tag Total # Jobs Notes Total Skim Size (TB) BCAL pi0 (GB) BCAL LED (GB) FCAL pi0 (GB) FCAL LED (GB) PS Skim (TB) TOF Skim (TB)
30274 - 30621 ver01 13362 11.2 338 74 862 147 6.36 1.63
30622 - 30959 ver02 13783 looser BCAL pi0 cuts 23.3 2379 131 1703 297 12.7 3.36
30960 - ver03 14869 looser BCAL pi0 cuts 13.4 1368 71 1000 142 7.47 1.99
  • Skim files can be found in the following directory: /cache/halld/RunPeriod-2017-01/calib/ver01
    • BCAL-LED - BCAL LED triggered events
    • BCAL_pi0 - BCAL pi0 candidates
    • FCAL-LED - FCAL LED triggered events
    • FCAL_pi0 - FCAL pi0 candidates
    • PS - PS triggered events
    • random - random (out-of-time) triggered events
    • sync - TS sync events
    • TOF - TOF calibration ROOT skim


Retrieved from "https://halldweb1.jlab.org/wiki/index.php?title=Calibration_Train&oldid=84847"