High-fidelity numerical simulation of physical systems modeled by time-dependent partial differential equations (PDEs) has been at the center of many technological advances in the last century. However, for engineering applications such as design, control, optimization, data assimilation, and uncertainty quantification, which require repeated model evaluation over a potentially large number of parameters, or initial conditions, these simulations remain prohibitively expensive, even with state-of-art PDE solvers. The necessity of reducing the overall cost for such downstream applications has led to the development of surrogate models, which captures the core behavior of the target system but at a fraction of the cost. In this context, new advances in machine learning provide a new path for developing surrogates models, particularly when the PDEs are not known and the system is advection-dominated. In a nutshell, we seek to find a data-driven latent representation of the state of the system, and then learn the latent-space dynamics. This allows us to compress the information, and evolve in compressed form, therefore, accelerating the models. In this series of lectures, I will present recent advances in two fronts: deterministic and probabilistic modeling latent representations. In particular, I will introduce the notions of hyper-networks, a neural network that outputs another neural network, and diffusion models, a framework that allows us to represent probability distributions of trajectories directly. I will provide the foundation for such methodologies, how they can be adapted to scientific computing, and which physical properties they need to satisfy. Finally, I will provide several examples of applications to scientific computing.[-]

In this talk, I will give an overview of recent successes (and some failures) of combining modern, high order discretization schemes with machine learning submodels and their applications for large scale computations. The primary focus will be on supervised learning strategies, where a multivariate, non-linear function approximation of given data sets is found through a high-dimensional, non-convex optimization problem that is efficiently solved on modern GPUs. This approach can thus for example be employed in cases where current submodels in the discretization schemes currently rely on heuristic data. A prime of example of this is shock detection and shock capturing for high order methods, where essentially all known approaches require some expert user knowledge as guiding input. As an illustrative example, I will show how modern, multiscale neural network architectures originally designed for image segmentation can ameliorate this problem and provide parameter free and grid independent shock front detection on a subelement level. With this information, we can then inform a high order artificial viscosity operator for inner-element shock capturing. In the second part of my talk, I will present data-driven approaches to LES modeling for implicitly filtered high order discretizations. Whereas supervised learning of the Reynolds force tensor based on non-local data can provide highly accurate results that provide higher a priori correlation than any existing closures, a posteriori stability remains an issue. I will give reasons for this and introduce reinforcement learning as an alternative optimization approach. Our experiments with this method suggest that is it much better suited to account for the uncertainties introduced by the numerical scheme and its induced filter form on the modeling task. For this coupled RL-DG framework, I will present discretization-aware model approaches for the LES equations and discuss the future potential of these solver-in-the-loop optimizations.[-]

In this talk, I will give an overview of recent successes (and some failures) of combining modern, high order discretization schemes with machine learning submodels and their applications for large scale computations. The primary focus will be on supervised learning strategies, where a multivariate, non-linear function approximation of given data sets is found through a high-dimensional, non-convex optimization problem that is efficiently solved on modern GPUs. This approach can thus for example be employed in cases where current submodels in the discretization schemes currently rely on heuristic data. A prime of example of this is shock detection and shock capturing for high order methods, where essentially all known approaches require some expert user knowledge as guiding input. As an illustrative example, I will show how modern, multiscale neural network architectures originally designed for image segmentation can ameliorate this problem and provide parameter free and grid independent shock front detection on a subelement level. With this information, we can then inform a high order artificial viscosity operator for inner-element shock capturing. In the second part of my talk, I will present data-driven approaches to LES modeling for implicitly filtered high order discretizations. Whereas supervised learning of the Reynolds force tensor based on non-local data can provide highly accurate results that provide higher a priori correlation than any existing closures, a posteriori stability remains an issue. I will give reasons for this and introduce reinforcement learning as an alternative optimization approach. Our experiments with this method suggest that is it much better suited to account for the uncertainties introduced by the numerical scheme and its induced filter form on the modeling task. For this coupled RL-DG framework, I will present discretization-aware model approaches for the LES equations and discuss the future potential of these solver-in-the-loop optimizations.[-]

In this talk, I will give an overview of recent successes (and some failures) of combining modern, high order discretization schemes with machine learning submodels and their applications for large scale computations. The primary focus will be on supervised learning strategies, where a multivariate, non-linear function approximation of given data sets is found through a high-dimensional, non-convex optimization problem that is efficiently solved on modern GPUs. This approach can thus for example be employed in cases where current submodels in the discretization schemes currently rely on heuristic data. A prime of example of this is shock detection and shock capturing for high order methods, where essentially all known approaches require some expert user knowledge as guiding input. As an illustrative example, I will show how modern, multiscale neural network architectures originally designed for image segmentation can ameliorate this problem and provide parameter free and grid independent shock front detection on a subelement level. With this information, we can then inform a high order artificial viscosity operator for inner-element shock capturing. In the second part of my talk, I will present data-driven approaches to LES modeling for implicitly filtered high order discretizations. Whereas supervised learning of the Reynolds force tensor based on non-local data can provide highly accurate results that provide higher a priori correlation than any existing closures, a posteriori stability remains an issue. I will give reasons for this and introduce reinforcement learning as an alternative optimization approach. Our experiments with this method suggest that is it much better suited to account for the uncertainties introduced by the numerical scheme and its induced filter form on the modeling task. For this coupled RL-DG framework, I will present discretization-aware model approaches for the LES equations and discuss the future potential of these solver-in-the-loop optimizations.[-]

Many physical phenomena deal with a fluid interacting with a moving rigid or deformable structure. These kinds of problems have a lot of important applications, for instance, in aeroelasticity, biomechanics, hydroelasticity, sedimentation, etc. From the analytical point of view as well as from the numerical point of view they have been studied extensively over the past years. We will mainly focus on viscous fluid interacting with an elastic structure. The purpose of the present lecture is to present an overview of some of the mathematical and numerical difficulties that may be encountered when dealing with fluid–structure interaction problems such as the geometrical nonlinearities or the added mass effect and how one can deal with these difficulties.[-]

Many physical phenomena deal with a fluid interacting with a moving rigid or deformable structure. These kinds of problems have a lot of important applications, for instance, in aeroelasticity, biomechanics, hydroelasticity, sedimentation, etc. From the analytical point of view as well as from the numerical point of view they have been studied extensively over the past years. We will mainly focus on viscous fluid interacting with an elastic structure. The purpose of the present lecture is to present an overview of some of the mathematical and numerical difficulties that may be encountered when dealing with fluid–structure interaction problems such as the geometrical nonlinearities or the added mass effect and how one can deal with these difficulties.[-]

At the end of the 70', Littlejohn [1, 2, 3] shed new light on what is called the Gyro-Kinetic Approximation. His approach incorporated high-level mathematical concepts from Hamiltonian Mechanics, Differential Geometry and Symplectic Geometry into a physical affordable theory in order to clarify what has been done for years in the domain. This theory has been being widely used to deduce the numerical methods for Tokamak and Stellarator simulation. Yet, it was formal from the mathematical point of view and not directly accessible for mathematicians.
This talk will present a mathematically rigorous version of the theory. The way to set out this Gyro-Kinetic Approximation consists of the building of a change of coordinates that decouples the Hamiltonian dynamical system satisfied by the characteristics of charged particles submitted to a strong magnetic field into a part that concerns the fast oscillation induced by the magnetic field and a other part that describes a slower dynamics.
This building is made of two steps. The goal of the first one, so-called "Darboux Algorithm", is to give to the Poisson Matrix (associated to the Hamiltonian system) a form that would achieve the goal of decoupling if the Hamiltonian function does not depend on one given variable. Then the second change of variables (which is in fact a succession of several ones), so-called "Lie Algorithm", is to remove the given variable from the Hamiltonian function without changing the form of the Poisson Matrix.
(Notice that, beside this Geometrical Gyro-Kinetic Approximation Theory, an alternative approach, based on Asymptotic Analysis and Homogenization Methods was developed in Frenod and Sonnendrücker [5, 6, 7], Frenod, Raviart and Sonnendrücker [4], Golse and Saint-Raymond [9] and Ghendrih, Hauray and Nouri [8].)[-]

Parametric PDEs arise in key applications ranging from parameter optimization, inverse state estimation, to uncertainty quantification. Accurately solving these tasks requires an efficient treatment of the resulting sets of parametric PDE solutions that are generated when parameters vary in a certain range. These solution sets are difficult to handle since their are embedded in infinite dimensional spaces, and present a complex structure. They need to be approximated with numerically efficient reduction techniques, usually called Model Order Reduction methods. The techniques need to be adapted both to the nature of the PDE, and to the given application task. In this course, we will give an overview of linear and nonlinear model order reduction methods when applied to forward and inverse problems. We will particularly emphasize on the role played by nonlinear approximation and geometrical PDE properties to address classical bottlenecks.[-]

Parametric PDEs arise in key applications ranging from parameter optimization, inverse state estimation, to uncertainty quantification. Accurately solving these tasks requires an efficient treatment of the resulting sets of parametric PDE solutions that are generated when parameters vary in a certain range. These solution sets are difficult to handle since their are embedded in infinite dimensional spaces, and present a complex structure. They need to be approximated with numerically efficient reduction techniques, usually called Model Order Reduction methods. The techniques need to be adapted both to the nature of the PDE, and to the given application task. In this course, we will give an overview of linear and nonlinear model order reduction methods when applied to forward and inverse problems. We will particularly emphasize on the role played by nonlinear approximation and geometrical PDE properties to address classical bottlenecks.[-]