Implicit-Explicit Time Integration for Multiscale Physics

Speaker Name: 
Dr. Debojyoti Ghosh
Speaker Title: 
Postdoctoral Research Staff Member
Speaker Organization: 
Center for Applied Scientific Computing at Lawrence Livermore National Laboratory
Start Time: 
Wednesday, December 6, 2017 - 11:00am
End Time: 
Wednesday, December 6, 2017 - 12:00pm
Location: 
BE 358
Organizer: 
Dongwook Lee, Nic Brummell, Pascale Garaud

Abstract:

Physical phenomena are often characterized by a large range of temporal scales. Unsteady simulations of these phenomena using explicit time integration may be inefficient if the fastest time scales in the model are not of practical interest. Implicit time integration can be unconditionally stable; however, it may be expensive and inefficient for time scales that are needed to be resolved. Implicit-explicit (IMEX) time integration allows for the explicit integration of time scales that are of interest, while faster scales are integrated implicitly. This talk will focus on two such applications: atmospheric flows and tokamak edge plasma dynamics. The first part of the talk will focus on a characteristic-based decomposition of the Euler equations that separates the acoustic modes from the convective ones. This allows the integration of the acoustic waves implicitly, while the convective terms are integrated explicitly. The second part of the talk will describe our efforts at simulating collisional magnetized plasmas in the edge region of the tokamak, where the collisional timescales are much faster than particle transport. The convergence, accuracy, and computational expense of the high-order IMEX methods will be presented for both these applications.

 

Bio:

Debojyoti Ghosh is a postdoctoral research staff member at the Center for Applied Scientific Computing at Lawrence Livermore National Laboratory.  His current research focuses on developing high-order numerical methods for fusion plasma simulations. Prior to this, he was a postdoctoral researcher at the Mathematics & Computer Science Division at the Argonne National Laboratory, where he worked on implicit-explicit methods for atmospheric flow simulations. He completed his Ph.D. in Applied Mathematics and Scientific Computing from the University of Maryland, and his dissertation focused on high-resolution methods for compressible turbulent flows, specifically rotorcraft wake flows. He obtained his M. Tech. and B. Tech. degrees in Aerospace Engineering from the Indian Institute of Technology Bombay.