•  

High-Performance Python-based Simulations of Pressure and Temperature Waves in a Trace Gas Sensor

A session at Python for High Performance and Scientific Computing (PyHPC 2013)

  • Brian Brennan
  • Robert C. Kirby
  • John Zweck
  • Susan E. Minkoff

Monday 18th November, 2013

9:30am to 10:10am (MST)

We present a coupled model of temperature and pressure waves applicable to photoacoustic trace gas sensors. We discretize this model with finite elements using the Pythonbased FEniCS project. To validate the generated code, we observe optimal convergence rates to a plane wave solution in one and two dimensions. Through the petsc4py package, we use the scalable LU solver MUMPS and preconditioned Krylov methods to perform the numerical linear algebra on a workstation and explore scaling results results on the Baylor University cluster Kodiak. Finally, we use the automated mesh adaptivity of FEniCS to optimize the computation of heat flux along a portion of the boundary, arriving at comparable accuracy to uniform refinement using a factor of forty fewer cells.

About the speakers

This person is speaking at this event.
Brian Brennan

Department of Mathematics, Baylor University

This person is speaking at this event.
Robert C. Kirby

Baylor University

This person is speaking at this event.
John Zweck

University of Texas at Dallas

This person is speaking at this event.
Susan E. Minkoff

University of Texas at Dallas

Next session in 505

10:30am Doubling the Performance of Python/NumPy with less than 100 SLOC by Brian Vinter, Mads Ruben Burgdorff Kristensen, Kenneth Skovhede and Simon A. F. Lund

Coverage of this session

Sign in to add slides, notes or videos to this session

Tell your friends!

When

Time 9:30am10:10am MST

Date Mon 18th November 2013

Session Hash Tag

#SC13

Short URL

lanyrd.com/sctbgh

Official event site

www.dlr.de/sc/pyhpc2013

View the schedule

Share

See something wrong?

Report an issue with this session