Friday, March 14

MS51
Parallel Implicit Approaches in Magnetic Fusion Applications - Part II of II

3:30 PM - 5:30 PM
Room: Roswell

For Part I, see MS45

Magnetic-fusion energy (MFE) devices are large, complex systems to develop a major new consumer energy source. Many MFE plasma models utilize parallel solution of large-scale nonlinear partial differential equations describing the confinement of hot plasma in electromagnetic fields. An enormous range of time-scales must be spanned by these models. Implicit algorithms, e.g., preconditioned Newton-Krylov methods, have demonstrated efficient and scalable solution for such problems. The speakers in this session will discuss issues in algorithms and implementations for the practical and efficient application of implicit methods for a variety of MFE physics components, including edge plasma, MHD, and RF heating.

Organizer: Lois Curfman McInnes
Argonne National Laboratory, U.S.
Hong Zhang
Argonne National Laboratory, U.S.
Tom Rognlien
Lawrence Livermore National Laboratory, U.S.

3:30-3:55 Linear Iterative Solves and the Fusion RF Problem abstract

Lee A. Berry, ; Eduardo F. D'Azevedo and E. F. Jaeger, Oak Ridge National Laboratory, U.S.

4:00-4:25 Preconditioning and Scalability of Implicit Extended MHD Plasma Simulation by FETI-DP Domain Substructuring abstract

Alan H. Glasser, University of Washington, U.S.

4:30-4:55 Fully Nonlinearly Implicit Resistive MHD Solvers with Textbook Multigrid Efficiency abstract

Mark Adams, Columbia University, U.S.; Ravi Samtaney, King Abdullah University of Science & Technology (KAUST), Saudi Arabia; Achi E. Brandt,

Cancelled 5:00-5:25 A Symplectic Iterative Semi-Implicit Algorithm and Its Relation to Preconditioned Fully Implicit Algorithms

Greg Hammett, ; Nuno Loureiro, UKAEA, United Kingdom