Presentation is loading. Please wait.

Presentation is loading. Please wait.

Monitoring for Data Quality Status of monitoring Histogramm analysis tools Data quality qualification.

Similar presentations


Presentation on theme: "Monitoring for Data Quality Status of monitoring Histogramm analysis tools Data quality qualification."— Presentation transcript:

1 Monitoring for Data Quality Status of monitoring Histogramm analysis tools Data quality qualification

2 On line Monitoring Normal tool when taking data – Set of histogramm from OlivierD – Profile histogramm for different energy bin Mean energy per channel For TAE events Kernel and asymetry as defined by Yasmine – Around 1MB histogram /run Analysed per run/fill – Using histogramm saver and adder (should be available - compatibility problem 32/64 bits) – Monitoring and calibration farm concerned

3 Histogramm Analysis - Work with the MonObject Dim service – commissioned and tested at the pit - Uses OMalib v1r3 - Input interface : - Saver will write savesets for each Monitoring Taskthere should be a DIM service with a name like /UTgid/SAVESET LOCATION/Taskname containing the saveset file name. - Integrating the Savers with the Monitoring Farm is now the highest priority.

4 Tool Histogramm analysis Standard Algorithms for automatic checks available: - CheckXRange check that all Histogram entries are in given range - CheckMeanAndSigma check that mean and standard deviation are in given ranges - GaussFit fitted mean and standard deviation in given ranges - CheckHolesAndSpikes check for holes and spikes with respect to reference or fitted polynomial - CompareToReference goodness of fit test wrt reference (2 or Kolmogorov) New ones can be added on request ( G.Giacomo )

5 Tools for monitoring Publish rate onto PVSS ( on-line) ( JF Menou) – To be done from counters – Convert to rate

6 Data quality scheme Data Quality group & meeeting ( P;Koppenburg ) Twiki page available Determine quality Duck, maybe, good Rating per sub-detector foreseen ( ~90 bits ) Using Savanah Not all files will be processed Rating based on On line results to be automatised for next years. Comparison to reference Input from Calo piquet Results from calibration farm Calibration should be updated Unstable channels high rate Results from Brunel Rating may be modified after Brunel

7 Data quality qualification Input from monitoring farm : Missing detector parts Noisy – dead channel rate Channel flux ( digit – energy ) -> warnings at the level of the data manager in future years For TAE event timing rating Input from Brunel histogramm ( 1MB) Energy flux /cluster ( neutral, charged) Energy spectrum / type ID Sum Et spectrum

8 Data quality qualification Input from monitoring farm : Missing detector parts Noisy – dead channel rate Channel flux ( digit – energy ) -> warnings at the level of the data manager in future years For TAE event timing rating Input from Brunel histogramm ( 1MB) Energy flux /cluster ( neutral, charged) Energy spectrum / type ID Sum Et spectrum Brunel hitrograms may be analysed with similar tools than monitoring farm.

9 Calorimeter specific checks Linked to data quality Particle id efficiency Calibration on collected data : Et spectrum Pi0 mass ( with or without tracking ) E/p for electron ( need alignment ) For HCAL ( charged flux ) SPD ( minimum ionising ) Prepare histogramm reference Reference programs to be run regurlarly.


Download ppt "Monitoring for Data Quality Status of monitoring Histogramm analysis tools Data quality qualification."

Similar presentations


Ads by Google