Presentation is loading. Please wait.

Presentation is loading. Please wait.

ID Week 13 th of October 2014 Per Johansson Sheffield University.

Similar presentations


Presentation on theme: "ID Week 13 th of October 2014 Per Johansson Sheffield University."— Presentation transcript:

1 ID Week 13 th of October 2014 Per Johansson Sheffield University

2  Will touch briefly upon past experiences but will concentrate on the future and Run 2  Cover both offline and online  Bring up some issues we might need to look at in no specific order 2

3  Main losses: ◦ Pixel: Standby, HVscan ◦ SCT: ROD_OUT_2, Global_standby, Crate_out ◦ TRT: Nodata_06, Desync ◦ ID Global: 97.5%: tracking coverage (SCT ROD plus...), beamspot  Data Quality paper: ◦ https://cds.cern.ch/record/1633621 https://cds.cern.ch/record/1633621  2013 workshop ◦ https://indico.cern.ch/event/221713/ https://indico.cern.ch/event/221713/  Summary presentation of workshop ◦ https://indico.cern.ch/event/256916/session/3/contribution/9/material/ slides/0.pdf

4  M4 7– 11 th of July ◦ Managed to run the monitoring, and some of the DQM was also included and running stable in rel. 19 ◦ However Pixel/SCT not powered so only integration work  M5 8-12 th of September ◦ For SCT/Pixel priority of integration with powered detectors ◦ Monitoring not really exercised ◦ Some basic standalone monitoring tools working  M6 started today ◦ Work on ID Global and detector monitoring, DQM and standalone monitoring tools  Should aim to be in a good state for M7 starting 24 th of November, ending on 7 th of December 4

5  Histogram Review to be done when packages are working as intended  In Run-1 probably too many unneeded plots ◦ consumes CPU and memory  Too many meaningless checks (wrong algorithms, bad thresholds, etc.) and perhaps not enough good checks ◦ Can the histogram be designed better? ◦ Much lumi information is available - can the histogram be made robust to changes in trigger or lumi/conditions? ◦ Luminosity-dependent rebooking, etc...  Can the offline code be made easier to maintain? ◦ Can its performance be improved?  Are similar plots, e.g. coverage/timing plots made across related sub- systems in a unified way? ◦ Which *need* monitoring online? ◦ Are the checks reliable in all conditions, to avoid spurious red flags?  What should ID ACR shifter watch and what should central DQ desk?  Is the documentation clear and up-to-date? 5

6  In run 1 there were 3 detector shifters and one for the ID Global part ◦ Covering tracking, vertexing, beamspot and alignment  Some issues that were raised: ◦ Communication Global – subdetector shifters ◦ The ID Global DQ checks should occur after subdetector checks ◦ Lack of understanding from the detectors about ID decisions made ◦ Duplication of effort between Global and subdetectors ◦ Lack of ID Global experts  Merging these shifts into one was under way during Run 1and there were several successful trial periods with one shifter doing all parts (in addition to the normal ones as crosscheck)  New twiki DQ pages were used to present the shifter with all the information needed ◦ https://twiki.cern.ch/twiki/bin/viewauth/AtlasDQ/ ◦ Not working at the moment?! ◦ Needs some attention – content, histograms, instructions ◦ Webdisplay used previously, contains all histograms, shifter, expert, etc, and is still there of course 6

7  Defects need to be reviewed  How to decide if a tolerable or intolerable defect should be set? ◦ Obviously should depend if data is good for physics ◦ Not obviously how to do this, would need feedback from physics groups  An example: for SCT when >100 modules/2 RODs out = intolerable defect in Run 1  However this don’t necessarily means data is bad, one ROD out on each endcap would translate into one hit loss on track trough that quadrant, 2 RODS on barrel would be worse due to eta/phi module layout  Should aim to find and decide in the ID global part if there is a loss of tracking in the affected region and set appropriate defect  Of course still the question of how big a loss leads to an intolerable defect 7

8  The prompt calibration loop ◦ Derived calibrations in 36 hour loop which starts after run finished ◦ Status?!  PIX/SCT/TRT calibrations to be moved into the Tier-0 system if stand-alone processing if robust enough ◦ Question raised during recent DPC meeting  DCS Calculator ◦ PVSS -> COOL issues in CONDBR2 ◦ Question if using names/channel numbers to write it and if using names or channel numbers to access data offline ◦ Writing by names -> channel# get random assignment 8

9  Lots of things to do!  First make sure monitoring online and offline is running ok, and DQM and various other tools we use  Then start reviewing histograms, defects, shifter tools online and offline, etc,.. and we should not forget Documentation! 9


Download ppt "ID Week 13 th of October 2014 Per Johansson Sheffield University."

Similar presentations


Ads by Google