2009 Spin QA: 500GeV Part II

The purpose of this follow-up analysis to the Spin QA for the 2009 500GeV run is to investigate fills in which production or zdc_polarimetry runs were discarded and determine why they had to be discarded and if it is possible to assign them a spin pattern. To help with this task, I created two new tickle scripts, spinBitCrawler.tcl and crawlerComp.tcl. The spinBitCrawler script starts at the beginning of a fill and quiries the cdev database every N seconds and prints out the polarization pattern as well as generating a .v124 file which contains information about the intended fill pattern and polarization pattern. The crawlerComp script looks at the .v124 files and determines which sequential pairs do not match. Using these two scripts, I can pinpoint any wierd behavior in cdev which may cause a fill to be bad.

This investigation will focus on two fills: F10449 and F10536. See the original QA page for details.

F10449:

F10449 contains 35 runs between 10087050 and 10087100 and spans times 9:49:11 to 19:35:52 on 3/28/09. In the previous analysis, I found that I had to remove 7 runs from the end of the fill to obtain a consistent spin pattern from beginning to end. Table 1 lists the removed runs.

Table 1: List of removed runs along with runtype, start and end times (bnl time) and status.

The output of the crawlerComp script shows a discrepency between the files coresponding to times 18:04:58 and 18:15:23. The output of spinBitCrawler around these times shows that timestamp 18:04:58 has the polarization pattern B + - + - - + - +  Y + + - - + + - -  which is consistent with the beginning of the run, while timestamp 18:15:23 has polarization pattern B - + - + + - + -  Y + + - - + + - -. The cdev database is filled every 5 min so there is a gap between the two timestamps above. For some reason, my script would not read this intermediate timestamp, but I looked at it interactively and found that timestamp 18:10:04 has polarization pattern B - + - + + - + -  Y + + - - + + - -. When doing the original analysis, the script which compiles the cdev information defines a fill as lasting from 20 min before the first run to 10 min after the last run so I needed to remove the runs above to get a consistent spin pattern for this fill. Performing the original analysis on runs 10087096 - 10087100 shows that these runs have a consistent polarization pattern B - + - + + - + -  Y + + - - + + - -  as I would expect given the timing of the polarization pattern shift.

I also checked the RICH scalars to make sure the runs I had to remove during the original analysis belonged to fill F10449. The relevant plot is in Fig 1.

Fig 1: The RICH scalars for F10449.

With the evidence presented above, I believe that it is reasonable to conclude that the runs listed above can be grouped with F10449 having polarization pattern B + - + - - + - +  Y + + - - + + - - .

F10536:

F10536 contains 17 runs between 10103026 and 10103048 and spans times 7:23:18 to 13:30:57 on 4/13/09. In the previous analysis, I found that I had to remove 8 runs from the end of the fill to obtain a consistent spin pattern from beginning to end. Table 2 lists the removed runs.

Table 2: List of removed runs along with runtype, start and end times (bnl time) and status.

The output of the crawlerComp script shows a discrepency between the files corresponding to times 10:50:26 and 10:55:29. The output of the spinBitCrawler script around these times shows that timestamp 10:50:26 has the polarization pattern B - + - + + - + -  Y + + - - + + - -  which is consistent with the beginning of the run, while timestamp 10:55:29 has polarization pattern B - + - + + - + -  Y eeeeeeee. The 'e' indicates that the polarization bit for a given bunch was zero. Indeed, if we look at the raw output from the cdev database for timestamp 10:55:29, we see that the polarizationFillPatternS field for the yellow beam contains all zeros. This is opposed to the output for timestamp 10:50:26 in which the polarizationFillPatternS field for the yellow beam contains 1's and -1's in most (i%3)'th spots. Performing the original analysis on runs 10103042 - 10103048 shows that these runs have a consistent polarization pattern B - + - + + - + -  Y eeeeeeee  as I would expect given the timing of the polarization pattern shift.

I also checked the RICH scalars to make sure the runs I had to remove during the original analysis belonged to fill F10536. The relevant plot is in Fig 2.

Fig 2: The RICH scalars for F10536

With the evidence presented above, I believe that it is reasonable to conclude that the runs listed above can be grouped with F10536 having polarization pattern B - + - + + - + -  Y + + - - + + - - .

Modified Good List:

If its decided that we do want to include the runs that I have listed above with thier respecitve fills for the purpose of uploading spin information, a modified list of good runs will be required. This list can be found here.

This list is the same as the last good list found in my previous analysis exept for the changes listed. For F10449: The last run number has changed from 10087087 to 10087100. The number of runs has changed from 28 to 35 and the end unix time has changed from 1238276352 to 1238283352. For F10536: The last run number has changed from 10103038 to 10103048. The number of runs has changed from 9 to 17 and the end unix time has changed from 1239633725 to 1239643857.