Establishing the Viability and Efficacy of In Situ Reduction Via Lagrangian Representations for Time-Dependent Vector Fields

Date and time: 
Mon, May 18 2020 - 2:00pm
Sudhanshu Sane
University of Oregon
  • Hank Childs (Chair)
  • Boyana Norris
  • Brittany Erickson
  • Leif Karlstrom (Earth Sciences)

Exploratory visualization and analysis of time-dependent vector fields or flow fields generated by large scientific simulations is increasingly challenging on modern supercomputers. Traditional time-dependent flow visualization is performed using an Eulerian representation of the vector field and requires both a high spatial and temporal resolution to be accurate. Although the computational capabilities of modern supercomputers enable us to perform simulations at a high spatial resolution, the limited I/O bandwidth and large data set sizes prevent us from storing data as frequently as we require to perform accurate exploration. A potential solution is calculating the Lagrangian representation of a vector field using in situ processing. A Lagrangian-based approach offers improved accuracy- storage propositions, but introduces an in situ encumbrance on the simulation code. This dissertation presents necessary research to establish the viability and efficacy of using in situ Lagrangian analysis to enable exploration of large time-dependent vector fields.