Adam Larat

Florence Drui, Alexandru Fikl, Pierre Kestener et al.

Many physical problems involve spatial and temporal inhomogeneities that require a very fine discretization in order to be accurately simulated. Using an adaptive mesh, a high level of resolution is used in the appropriate areas while keeping a coarse mesh elsewhere. This idea allows to save time and computations, but represents a challenge for distributed-memory environments. The MARS project (for Multiphase Adaptative Refinement Solver) intends to assess the parallel library p4est for adaptive mesh, in a case of a finite volume scheme applied to two-phase flows. Besides testing the library's performances, particularly for load balancing, its user-friendliness in use and implementation are also exhibited here. First promising 3D simulations are even presented.

Mohamed Essadki, Jonathan Jung, Adam Larat et al.

This article describes the implementation of an all-in-one numerical procedure within the runtime StarPU. In order to limit the complexity of the method, for the sake of clarity of the presentation of the non-classical task-driven programming environnement, we have limited the numerics to first order in space and time. Results show that the task distribution is efficient if the tasks are numerous and individually large enough so that the task heap can be saturated by tasks which computational time covers the task management overhead. Next, we also see that even though they are mostly faster on graphic cards, not all the tasks are suitable for GPUs, which brings forward the importance of the task scheduler. Finally, we look at a more realistic system of conservation laws with an expensive source term, what allows us to conclude and open on future works involving higher local arithmetic intensity, by increasing the order of the numerical method or by enriching the model (increased number of parameters and therefore equations).

Adam Larat

The roots of Discontinuous Galerkin (DG) methods is usually attributed to Reed and Hills in a paper published in 1973 on the numerical approximation of the neutron transport equation [18]. In fact, the adventure really started with a rather thoroughfull series of five papers by Cockburn and Shu in the late 80's [7, 5, 9, 6, 8]. Then, the fame of the method, which could be seen as a compromise between Finite Elements (the center of the method being a weak formulation) and Finite Volumes (the basis functions are defined cell-wise, the cells being the elements of the primal mesh) increased and slowly investigated successfully all the domains of Partial Differential Equations numerical integration. In particular, one can cite the ground papers for the common treatment of convection-diffusion equations [4, 3] or the treatment of pure elliptic equations [2, 17]. For more information on the history of Discontinuous Galerkin method, please refer to section 1.1 of [15]. Today, DG methods are widely used in all kind of manners and have applications in almost all fields of applied mathematics. (TODO: cite applications and structured/unstructured meshes, steady/unsteady, etc...). The methods is now mature enough to deserve entire text books, among which I cite a reference book on Nodal DG Methods by Henthaven and Warburton [15] with the ground basis of DG integration, numerical analysis of its linear behavior and generalization to multiple dimensions. Lately, since 2010, thanks to a ground work of Zhang and Shu [26, 27, 25, 28, 29], Discontinuous Galerkin methods are eventually able to combine high order accuracy and certain preservation of convex constraints, such as the positivity of a given quantity, for example. These new steps forward are very promising since it brings us very close to the "Ultimate Conservative Scheme", [23, 1].