Time Clustering

MCH Time Clustering

Regroups (aka merges) ROFs which have compatible times, making ROFs that span larger durations.

o2-mch-digits-to-timeclusters-workflow

Take as input the list of all digits (Digit) in the current time frame, with the data description "DIGITS", and the list of ROF records (ROFRecord) pointing to the digits associated to each interaction, with the data description "DIGITROFS". Send a new list of ROF records that combine all the digits correlated in time within a user-defined time window, with the data description "TIMECLUSTERROFS".

The time clustering can be configured using --configKeyValues :

o2-mch-digits-to-timeclusters-workflow --configKeyValues="MCHTimeClustering.maxClusterWidth=xx;MCHTimeClustering.peakSearchNbins=yy"

maxClusterWidth allows to set the width of the time correlation window. The time clustering is based on a brute-force peak search algorithm, which arranges the input digits into coarse time bins. The number of bins in one time cluster window can be set via the peakSearchNbins parameter.