2009 Spin QA: 500GeV

On this page I detail the procedure used for doing the spin QA. I also list the steps I have taken to fix fills that had inconsistent spin data in the cdev database.
 

Method:

First, I query the cdev database and compile a list of runs (runlist) spanning a desired time range. This is done with the script loopMainAll.tcl. This script records the fill number, trigger setup name, shift status, which detectors where in, as well as several trigger quantities for each run. The script automatically excludes runs less than a minute as well as runs with certain trigger setup names (ie runs containing ped or pulse).

After the runlist is generated, it is analyzed using the script spbitSeeker.tcl which collects information about each fill, including start/stop times, number of runs in the fill, number of bunches, and spin bunch pattern. For each fill, the script also records the spin bunch pattern for the first and last run and puts them in seperate files.

Finally, a script called begEndComp.tcl is run, which compares the spin patter in the first and last run of each fill. It is assumed that the spin bunch pattern is stable throughout the course of a fill, so if the patterns match in the first and last run, the fill is considered to have a known bunch pattern and is put in the good list. If, however, the patterns do not match, the fill is flaged and put into the bad list. I then take a closer look at each fill on the bad list. Using a combination of runlog entries, RICH scalar information, and cdev queries, I try to determine which runs are causing the spin pattern mismatch. I then remove those runs from the runlist manually and repeat steps two and three until all fills have consistent spin bunch patterns.

500GeV Analysis First Pass:

In the first pass, I only try to fix fills that Jan Balewski has said he would like to analyze for the W program.

The original runlist I generated spanning Fill 10382 to Fill 10536 (3/17/2009 - 4/13/2009) is here.

The list of fills with consistent spin data in the first and last run is here.

The list of fills with inconsistent spin data in the first and last run is here.

Note: In the description of the spin bunch pattern some spaces contain the letter 'e' or 'c'. These occur only when the number of bunches is less than 100. An 'e' was used if the bunch(modulo 8) was consistently empty while a 'c' was used if a bunch(modulo 8) was filled more than once but with different spin states. These fills should not be placed with fills with patterns P1-P4 for systematic purposes. For more detail see Jim's 2006 spin pattern page.

Jan Balewski sent a list of fills that he was going to analyze, so for my first attempt to correct the fills on the bad list, I focused only on those fills which Jan was going to look at. These fills were: F10449, F10536, F0(10404), F10508, F10517, F10527, and F10532. Table 1 lists the runs that I had to remove from each fill in order to obtain consistent spin information in a fill.

Table 1: List of runs removed as well as trigger setup name and status to give consistent spin bunch pattern for a particular fill.

                      *No runs were removed in F0. An extra fill header was inserted into the middle of F10404. I removed the extra header so run 10080057 would appear in F10404 as it should.

The runlist containing the first set of changes I made to fix the fills with inconsistent spin data is here.

The list of fills with consistent spin data in the first and last run after my first set of corrections is here.

The list of fills still haveing inconsistent spin data in the first and last run after my first set of corrections is here.

500GeV Analysis Second Pass:

In the second pass, I look at those fills that remain on the bad list after the first pass corrections. There are seven bad fills left: F10393, F10472, F10524, F10394, F10395, F10416, F10516. None of these fills had production data and many didn't have any consistent spin data and had to be completely removed. Table 2 lists the runs removed from each fill.

Table 2: List of runs removed as well as trigger setup name and status to give consistent spin bunch pattern for a particular fill.

The final runlist containing all corrections can be found here.

The final list of fills with consistent spin information between the first and last run after all corrections can be found here.

The final list of bad fills is empty. All fills have been accounted for.