21. In Situ Data Steering on Sedimentation Simulation with Provenance Data
Authors: Vítor Silva (Federal University of Rio de Janeiro)José Camata (Federal University of Rio de Janeiro)Daniel de Oliveira (Fluminense Federal University)Alvaro L.G.A. Coutinho (Federal University of Rio de Janeiro)Patrick Valduriez (Inria)Marta Mattoso (Federal University of Rio de Janeiro)
Abstract: Parallel adaptive mesh refinement and coarsening (AMR) are optimal strategies for tackling large-scale simulations. libMesh is an open-source finite-element library that supports parallel AMR and is used in multiphysics applications. In complex simulation runs, users have to track quantities of interest (residuals, errors estimates, etc.) to control as much as possible the execution. However, this tracking is typically done only after the simulation ends. This paper presents DfAnalyzer, a solution based on provenance data to extract and relate strategic simulation data for online queries. We integrate DfAnalyzer to libMesh and ParaView Catalyst, so that queries on quantities of interest are enhanced by in situ visualization.
Two-page extended abstract: pdf