Both multigrid and domain decomposition methods are so called optimal solvers for Laplace type problems, but how do they compare? I will start by showing in what sense these methods are optimal for the Laplace equation, which will reveal that while both multigrid and domain decomposition are iterative solvers, there are fundamental differences between them. Multigrid for Laplace's equation is a standalone solver, while classical domain decomposition methods like the additive Schwarz method or Neumann-Neumann and FETI methods need Krylov acceleration to work. I will explain in detail for each case why this is so, and then also present modifications so that Krylov acceleration is not necessary any more. For overlapping methods, this leads to the use of partitions of unity, while for non-overlapping methods, the coarse space can be a remedy. Good coarse spaces in domain decomposition methods are very different from coarse spaces in multigrid, due to the very aggressive coarsening in domain decomposition. I will introduce the concept of optimal coarse spaces for domain decomposition in a sense very different from the optimal above, and then present approximations of this coarse space. Together with optimized transmission conditions, this leads to a two level domain decomposition method of Schwarz type which is competitive with multigrid for Laplace's equation in wallclock time.[-]

Iterative methods for linear systems were invented for the same reasons as they are used today,namely to reduce computational cost. Gauss states in a letter to his friend Gerling in 1823: 'you will in the future hardly eliminate directly, at least not when you have more than two unknowns'.
Richardson's paper from 1910 was then very influential, and is a model of a modern numerical analysis paper: modeling, discretization, approximate solution of the discrete problem,and a real application. Richardson's method is much more sophisticated that how it is usually presented today, and his dream became reality in the PhD thesis of Gene Golub.
The work of Stiefel, Hestenes and Lanczos in the early 1950 sparked the success story of Krylov methods, and these methods can also be understood in the context of extrapolation, pioneered by Brezinski and Sidi, based on seminal work by Wynn.
This brings us to the modern iterative methods for solving partial differential equations,which come in two main classes: domain decomposition methods and multigrid methods. Domain decomposition methods go back to the alternating Schwarz method invented by Herman Amandus Schwarz in 1869 to close a gap in the proof of Riemann's famous Mapping Theorem. Multigrid goes back to the seminal work by Fedorenko in 1961, with main contributions by Brandt and Hackbusch in the Seventies.
I will show in my presentation how these methods function on the same model problem ofthe temperature distribution in a simple room. All these methods are today used as preconditioners for Krylov methods, which leads to the most powerful iterative solvers currently knownfor linear systems.[-]

In this talk, we present a three-level variant of the parareal algorithm that uses three propagators at the fine, intermediate and coarsest levels. The fine and intermediate levels can both be run in parallel, only the coarsest level propagation is completely sequential. We interpret our algorithm as a variant of three-level MGRIT, and we present a convergence analysis that uses parareal-type assumptions, i.e., those that involve Lipschitz constants on the propagators. We present numerical experiments to illustrate how sharp the estimates are for various time dependent problems.[-]

I will give an elementary introduction of basic deep learning models and training algorithms from a mathematical viewpoint. In particular, I will relate some basic deep learning models with finite element and multigrid methods. I will also touch on some advanced topics to demonstrate the potential of new mathematical insight and analysis for improving the efficiency of deep learning technologies and, in particular, for their application to numerical solution of partial differential equations.[-]

The parallel-in-time integration of wave-type equations is well known to be a difficult task. When applying classical waveform-relaxation (WR) and parareal type methods, one generally experiences rapid error growth before reaching convergence in a finite number of iterations. This negative behavior prevents, in general, the successful application of these domain decomposition methods. In this talk, the focus is on WR-type methods. Classical WR convergence analyses use classical Laplace/Fourier techniques. However, these approaches provide analyses for unbounded time intervals, and do not allow one to describe precisely the WR converge behavior on finite time intervals. In this talk, we present a novel analysis based on the methods of characteristics, which allows us, on the one hand, to obtain a detailed characterization of the error growth along with the iterations and, on the other hand, to introduce a new parallel-in-time computational strategy. Numerical experiments support our new theoretical and numerical findings. This is a joint work with Martin J. Gander and Ilario Mazzieri.[-]

When thinking about parallel in time schemes, one often tends to view time as a variable to discretize within a numerical scheme (that usually involves a time marching strategy). In this talk, I propose to review alternative strategies where time can be seen as a parameter so that computing the PDE solution at a given time would consist in evaluating closed formulas or in solving tasks of very low computational cost that do not involve any time marching. This type of approach is by nature entirely parallelizable. It can be achieved by either leveraging analytic formulas (whose existence strongly depends on the nature of the PDE), or by learning techniques such as model order reduction. For the later strategy, convection dominated problems are challenging (just like in classical PinT schemes such as parareal) and I will present recent contributions to address this type of problems.[-]

This talk consists of two parts, one elementary and one related to the solution of complicated systems of evolutionary partial differential equations. In the first part we show how preconditioning for all-at-once descriptions of linear time-dependent differential equations can defeat the everpresent danger of high-index nilpotency associated with the principle of causality. In particular we will describe some theory for periodic preconditioning of initial value problems that establishes it as a viable Parallelin-time (PinT) approach. The second part builds on much excellent work on PinT methods for scalar parabolic PDEs such as the diffusion equation to propose PinT methods for more complicated evolutionary PDE systems. We will explain the idea with reference to the time-dependent incompressible Stokes and Navier-Stokes equations and indicate it's more broad applicability. This part of the talk is joint with Federico Danieli and Ben Southworth.[-]