1. CMS Pixel Data Quality Monitoring
Petra Merkel
Purdue University, West Lafayette, IN, USA
For the CMS Pixel Group

2. CMS Pixel Data Quality Monitoring - Petra Merkel, Purdue University
Barrel
Endcaps
Data quality monitoring
The CMS Pixel detector is highly granulated (1440 modules containing 66M pixels)  need for automated data quality monitoring (DQM)
The DQM system in the CMS experiment has been developed within the CMS Software framework
online: identify major problems in real time for prompt action (use subset of data, ~5Hz)
offline: detect reconstruction and calibration problems (full statistics, but limited granularity)
ROOT histograms are filled for a range of quantities. They are subsequently summarized and automatically evaluated. Problems result in warnings and alarms, which will be investigated further by Pixel experts.
In particular we monitor readout errors, raw charge deposition information, as well as reconstructed hits, both on and off tracks.
Experience during global cosmic ray data taking shows that we are thus able to detect, both, with fast turn-around (online), as well as high precision (offline), data corruption, mis-configuration and mis-calibration of the detector, as well as newly broken modules and dead or noisy pixels.
Interactive geometrical Maps
[mean raw charge]
Data
corruption
Calibrated cluster charge
Mean cluster charge [ke-]
Automated
quality test
Endcap modules
5/28/2009
CMS Pixel Data Quality Monitoring - Petra Merkel, Purdue University

3. CMS Pixel Data Quality Monitoring - Petra Merkel, Purdue University
Data certification
Central CMS DQM GUI
All histograms can be displayed both in the central CMS DQM graphical user interface (GUI), as well as in the Pixel specific expert GUI and in interactive svg maps.
The main data characteristics are filled into trend plots, which monitor the behaviour of the detector over time.
In order to certify the data for Physics analyses a complex workflow has been put in place.
comparison to reference histograms to detect unexpected behaviour
application of cuts to automatically spot outliers
visual inspection by shifters
The results of this evaluation are then combined to a final Data Quality Flag, which is stored for each run into a data base. The underlying intermediate results, which were used to obtain this final flag are also stored.
Pixel DQM Expert GUI
Trend Monitoring
Run number
5/28/2009
CMS Pixel Data Quality Monitoring - Petra Merkel, Purdue University