1. Валидация TRT DCS CONDITIONS SERVICE Евгений Солдатов, НИЯУ МИФИ “Physics&Computing in ATLAS” – 22/09/2010

2. Very often during the run some of conditions inside TRT can change, for instance, the HV lines can have incorrect HV and do not work entirely or partially in the proper way. It can be connected with shortcut on the line or short (long) trips. Reconstruction algorithms need to have offline information about the status of the concrete straw and conditions in different parts of detector. If HV line has wrong voltage or some another conditions influence on the straw, algorithms should take into account these conditions for reconstruction. Information about HV, Current and other conditions is stored in PVSS database (within Point 1 firewall). The TRT DCS Conditions Service provides this information to the reconstruction from PVSS database trough COOL Conditions Database (visible to the T0 computer farm). Introduction to the TRT DCS Conditions Service “Physics&Computing in ATLAS” – 22/09/2010

3. What do we have… TRT DCS Conditions Service at this moment can work only with HV (but it can work with other conditions in future as well). It flags each straw, which takes part into the reconstruction (flags are an enum with room for a few additional possibilities): GREEN (1) – HV is in normal range; YELLOW (2) – additional flag for flexibility (for example, if we can use straw hits only for tracking, but not for particle identification); RED (3) – HV is in the wrong range (HV line is offline or there is a trip on it); NOINFO (0) – HV is not available (some problems with software or with information in database). And produces some monitoring histograms for each part of the TRT subdetector inside TRT_DCS_Monitoring.root file. “Physics&Computing in ATLAS” – 22/09/2010

4. What should we do As any software package, TRT DCS Conditions Service, can have some bugs inside and before its using we must validate it: check, that all its results are correct. For this goal we should compare real PVSS trends with results of the TRT DCS Conditions Service (right work of flagging HV lines) during several time periods within different behavior of some HV lines: Static picture of HV map (1 event from run will be enough for this); Dynamic picture (work of service during appearance and disappearance of the trip). We will use run with short HV trip on the line (several minutes long period). From monitoring histograms we can obtain the duration of this period and compare with its duration on PVSS trends. “Physics&Computing in ATLAS” – 22/09/2010

5. What have been done: Software part For validation we do not need to do full reconstruction procedure. We need only to repeat tracking for several LBs of the definite run. So we can use ESDs data for it, because in comparison with RAW data: It is simpler to obtain for definite run; Its size for processing is much less. For this goal InDetRecExample package has been chosen. We improved it to have a possibility to use ESD data, not only RAW data. DbExplorer package has been used to obtain information from PVSS database (time in PVSS is ‘CEST’-2 during summer). “Physics&Computing in ATLAS” – 22/09/2010

6. What have been done : Static picture (I) Static picture of the HV map – run 141811 (900 GeV collisions) From PVSS we see two offline HV lines in EndCap A. No more offline lines in other parts of detector. (Sun Dec 06 2009 18:01 − 23:50 CEST time) From DbExplorer: “Physics&Computing in ATLAS” – 22/09/2010

7. From TRT DCS Conditions Service: What have been done : Static picture (II) (One event from run 141811) We have full match in EC A. Has been found software overflow error (error has been corrected) and 6 red lines in the end of distribution for Barrel, which are HV "crates" and don't actually correspond to straw HV lines, so all is correct. “Physics&Computing in ATLAS” – 22/09/2010

8. What have been done : Dynamic picture (I) Run 152878 (7 TeV collisions) For validation we’ve chosen LBs 473-482 (17:49:42 - 18:09:52 CEST time) of the run 152878. ( Tue Apr 13 2010 00:19 − 20:21 CEST time) Data stream: physics_CosmicCaloEM (0.602 Hz, 0.4%, 67.3 GB/run, 1.6 MB/evt). Total processed events: 692. 248/692*20(min)~7.17 (min) – duration of the trip from histograms of TRT DCS Conditions Service – good coincidence. We have 7-minutes trip on the line HVC/S7S8/WA3/1T. PVSS trend: “Physics&Computing in ATLAS” – 22/09/2010

9. What have been done : Dynamic picture (II) There is a trend of the average voltage during chosen time period. Let’s calculate its value from the PVSS picture on previous slide: 1530*(20-7)/20~994 V (approximately what do we see on the trend below) “Physics&Computing in ATLAS” – 22/09/2010

10. What have been done : Dynamic picture (III) One more trip, but now in the barrel.Run 158801 (7 TeV collisions) ( Fri Jul 09 2010 00:05 − 11:12 CEST time) There is 6-minutes trip on the line HVB/S11/M1/A3. PVSS trend: For validation we’ve chosen LBs 314-321 (10:14:05 - 10:30:13 CEST time) of the run 158801. Data stream: physics_CosmicCalo stream (5.661 Hz, 3.3%, 389.4 GB/run, 1.8 MB/evt). Total processed events: 8896. (65371-8896*1-6*8896)/8896* *16(min)=3099/8896*16(min)=5.6(min) One more HVB line is permanently off since 3 May 2010; 6 – HV “crates” in the end of the distribution (see above). “Physics&Computing in ATLAS” – 22/09/2010

11. What have been done : Dynamic picture (IV) Here is a trend for the average voltage during chosen time period. The situation now is more complicated. On the PVSS picture (previous slide) we have not zero voltage during trip, but approximately 600 V. After calculation we have: (1530*16-6*(1530-600))/16~1181 V (approximately what do we see on the trend below) “Physics&Computing in ATLAS” – 22/09/2010

12. Conclusions & What to do Method of TRT DCS Conditions Service validation has been developed, necessary software has been chosen and improved. Full validation of the setting RED flag to the HV lines in runs with static and dynamic number of the tripped HV lines has been done. The next step is to study cases of set NOINFO flags to HV lines and as much as possible to decrease its number. (In some of shown runs there are several NOINFO lines in different parts of detector.) After that there will be a possibility to provide other information from PVSS database (like Temperature or Current) using this service to the reconstruction algorithms. “Physics&Computing in ATLAS” – 22/09/2010

13. Back-up “Physics&Computing in ATLAS” – 22/09/2010

14. More about Conditions database This is the scheme of data flow through the archiving chain from the PVSS application to the TRT DCS Conditions Service: “Physics&Computing in ATLAS” – 22/09/2010

15. More about run 141811 Here is a trend of the average voltage in the run. Two offline HV line give gaps in the trend: “Physics&Computing in ATLAS” – 22/09/2010

16. 2-d map of HV lines during trip: More about run 152878 We can know precise number of processed events from this picture (because there is one permanently offline HV line: HVA/S19S20/WA4/1T) Tripped HV line “Physics&Computing in ATLAS” – 22/09/2010

17. More about run 158801 We can know precise number of processed events from this picture (because there is one permanently offline HV line: HVA/S19S20/WA4/1T) 3-d map of HV lines during trip: Tripped HV line “Physics&Computing in ATLAS” – 22/09/2010