Paper detail

Virtual Data in CMS Analysis

The use of virtual data for enhancing the collaboration between large groups of scientists is explored in several ways: - by defining ``virtual'' parameter spaces which can be searched and shared in an organized way by a collaboration of scientists in the course of their analysis; - by providing a mechanism to log the provenance of results and the ability to trace them back to the various stages in the analysis of real or simulated data; - by creating ``check points'' in the course of an analysis to permit collaborators to explore their own analysis branches by refining selections, improving the signal to background ratio, varying the estimation of parameters, etc.; - by facilitating the audit of an analysis and the reproduction of its results by a different group, or in a peer review context. We describe a prototype for the analysis of data from the CMS experiment based on the virtual data system Chimera and the object-oriented data analysis framework ROOT. The Chimera system is used to chain together several steps in the analysis process including the Monte Carlo generation of data, the simulation of detector response, the reconstruction of physics objects and their subsequent analysis, histogramming and visualization using the ROOT framework.