How beamlines are produced

Intention of this document

Step 1: Produce histograms

Beamlines using vertices are produced from fitting profile histograms of the vertex position in x or y versus z. Beamlines made using the d-phi algorithm use histograms were the impact parameter of a track with respect to the origin (d) is plotted versus the track angle phi.

These histograms are produced by FitBeamModule. We are running FitBeamModule as part of the standard CDF reconstructionn (Production) or we run a separate job (makeBeamHists) just producing the histograms on already reconstructed data.

• 1.1. Histograms out of production

  Production runs FitBeamModule. The output files of the different nodes get concatenated for each data stream separately. No specific action is needed. The files show up in:
  
  fcdflnx3:/cdf/scratch/cdfprod0/'Production version'/'Stream'/hist
  
  

• 1.2 Making histograms with makeBeamHists

  In the standard mode of operation the beamline histograms filling by Production can be used to derive beamlines. However, in case the production version is buggy the histograms might not be usable. An example for this case is Production 4.8.4a which is being used in CDF since October 2002. In the particular version the final track refit (after pattern recognition) is buggy causing the beamline to be shifted by 20 to 25 micrometer. Therefore, we have to refit the tracks with a corrected offline version and fill new beamline histograms. This can be conveniently done using the makeBeamHists executable. Here we explain how that can be done.

  • 1.2.1 Create the makeBeamHists executable

    You need to go through these steps only once at the beginning or when you need to change the offline version you are using or if you are required to pick up the latest version of one of the libraries used. In this case, the tagged versions of the packages to add to the release will be different. Make sure you really need a new binary before proceeding.)
    You should have access to the cdfopr account on fcdflnx2/3. You should login as cdfopr and work in the directory /cdf/opr/cdfopr/beamlines so that everyone shares the same code.
    1. Setup an offline release

       source ~cdfsoft/cdf2.cshrc
setup cdfsoft2 4.9.1hpt1
newrel -t 4.9.1hpt1 4.9.1hpt1_beam

       

    2. Check out TrackingUserMods, VertexMods and VertexAlg tagged with v4_9_1_beam (these were tagged from the head on Jan 25, 2003).

       addpkg TrackingUserMods v4_9_1_beam
addpkg VertexMods v4_9_1_beam
addpkg VertexAlg v4_9_1_beam

       

    3. Make the VertexAlg and VertexMods libraries.

       gmake VertexAlg.nobin; gmake VertexMods.nobin
       

    4. Modify the makefile for the test binaries in TrackingUserMods such that only makeBeamHists is left as an test binaries, i.e. remove all other binaries from the TBINS line. The line should now look like:

       TBINS = makeBeamHists

    5. Build the makeBeamHists executable

       gmake TrackingUserMods.tbin

  • 1.2.2 Run the job

    1. Obtain a ticket for cdfopr to be used in the CAF

       ~cdfopr/ticket


    2. Run the job you only have to run the following script:

       ./VertexMods/scripts/goBeam.csh

       This script will do the following steps for you:

       1. Creates a working directory to submit your CAF jobs inside your current release directory.

          cd /cdf/opr/cdfopr/beamlines/4.9.1hpt1_beam/
mkdir caf/input (if does not exist yet)
mkdir caf/output (if does not exist yet)


       2. Copies the executable and the libraries to the CAF input directory
       3. Copies the tcl script for makeBeamHists
       4. Copies the CAF shell script
       5. Creates the run list to be processed (only if running on fcdflnx2/3)

          cp ../../VertexMods/scripts/createRunList.sh .
./createRunList.sh
mv procruns.txt run_list.tcl


       6. Do the special steps to access static cafdatasets. Edit script to modify this.

          setenv CAF_CURRENT caf2
cafdataset gjet08 > files.tcl


       7. Start the CAF GUI

          source ~cdfsoft/cdf2.cshrc
setup cdfsoft2 development
CafGui


       8. Fill out the CafGui form.
          

          script: ./pvtxrun.sh $ 200000
          Fill in the number of jobs in the upper right corner.
          Specify the output directory and file name "beamHists_$.tgz"
          Specify the queue to be "cdfpr" by default.
          

    3. Look into run_list.tcl and find out how many runs you need to process. Use that number for the max section number.

    4. Submit the jobs to the CAF.

  • 1.2.5 Copy results

    The CAF software will put the results for the refit jobs to the result directory you specified in the GUI.
    1. Untar and unzip the result files

       cp 4.9.1hpt1_beam/VertexMods/scripts/untarbeam.sh
./untarbeam.sh 'runnum begin' 'runnum end'


    2. Make sure you kerberos principal was put into the .k5login of the cdfprod0 account on fcdflnx2/3.

    3. Copy the root result files to a specific area on fcdflnx3. The area must fit the offline version you are using, e.g. scp \*.root cdfprod0:/cdf/scratch/cdfprod0/trackRefit4.9.1_b/

Step 2: Fit the beamlines from the histograms

Step 3: Fill gaps

Step 4: Create used sets