This is a scratchpad area relating to the Knowledge Management aspects of the University of Wisconsin RACE project.
The dataset generation system may be viewed here and the stream mixer may be downloaded here.
In order to focus our initial efforts, we will go through several simple scenarios where knowledge management techniques can be applied to validation of the CMS Level-1 trigger.
The first scenario is a misconfiguration of HoverEcut, a parameter that controls the cut-off level in the trigger for classification of electrons. In the online system, this parameter could be loaded incorrectly by portions of the trigger hardware, but for our initial test, we will only deal with the case of a global misconfiguration, where the whole system has the wrong value. The goal is to look at the trigger output for a stream of events and determine whether it is likely that HoverEcut has been misconfigured during the processing of that stream. We will assume a constant value of HoverEcut for each test.
The following steps may be followed to begin tackling this problem:
This case is the same sort of global parameter misconfiguration as the global HoverEcut case. The misconfiguration may be detected using the same observables in both cases. This allows us to test generalization of the previous solution as well as extending the diagnosis to include the possibility of two underlying sources of error.