The University of Arizona
Please note that this event has ended!

Scalable and Non-Intrusive Parallel-in-Time with Multigrid

Modeling, Computation, Nonlinearity, Randomness and Waves Seminar

Scalable and Non-Intrusive Parallel-in-Time with Multigrid
Series: Modeling, Computation, Nonlinearity, Randomness and Waves Seminar
Location: Hybrid: Math 402/Online
Presenter: Jacob Schroder, Department of Mathematics, University of New Mexico

The need for parallel-in-time algorithms is currently being driven by the rapidly changing nature of computer architectures.  Future speedups will come through ever increasing numbers of processors (cores), but not faster clock speeds, which are stagnant.  Previously, increasing clock-speeds compensated for traditional sequential time stepping algorithms when the problem size increased. However, this is no longer the case, leading to the sequential time integration bottleneck and the need to parallelize in time.  In this talk, we examine an optimal-scaling parallel time integration method, multigrid reduction in time (MGRIT).  MGRIT applies multigrid to the time dimension by solving the (non)linear systems that arise when solving for multiple time steps simultaneously.  The result is a versatile approach that is nonintrusive and wraps existing time evolution codes.  MGRIT allows for various time discretizations (e.g., Runge-Kutta and multistep) and for adaptive refinement/coarsening in time and space.  Nonlinear problems are handled through full approximation storage (FAS) multigrid.  Some recent theoretical results, as well as practical results for a variety of problems will be presented, e.g., nonlinear diffusion, powergrid systems, advection, and compressible Navier-Stokes.  Additionally, a novel application of time-parallelism to deep learning is presented, where the layer dimension of deep neural networks is explicitly parallelized by an MGRIT method.

Place: Hybrid: Math 402 and
Password: “arizona” (all lower case)