Category: seminaire passé

🧑‍🏫 Serge Van Criekingen, IDRIS, France
🌎 March 2020

Abstract

Parallel Schwarz-type domain decomposition methods are based on an iterative process where, at each iteration, a local solve is simultaneously performed on each of the (possibly overlapping) subdomains, using interface values previously computed on neighboring subdomains. The reference method in this framework is the Restricted Additive Schwarz (RAS) method, implemented as a preconditioner in the PETSc library. Using existing PETSc tools, we here implement two improvements to this method: a new coarse correction to obtain a two-level scalable method, as well as optimized transmission conditions, resulting in an Optimized 2-level Restricted Additive Schwarz method.

The first improvement, namely the introduction of a coarse correction to insure scalability, is wellknown and due to that fact that, in the case of elliptic problems, information is only transferred from each subdomain to its direct neighbors at each iteration of a 1-level method such as RAS. This makes the number of iterations grow with the number of subdomains. Scalability is achieved by introducing a coarse grid on which a reduced-size calculation is performed, yielding a coarse correction at each iteration of the solution process. Such a 2-level method permits global propagation of the iterative corrections throughout the entire domain, leading to the scalability of the method. Many choices for the coarse grid point locations are possible, and we here follow a method introduced by M.J. Gander et al. yielding a reduced number of iterations.

The second improvement, namely optimized transmission conditions, stems from the idea that the transmission conditions used in the iterative process at subdomain interfaces can also be chosen such as to reduce the number of iterations. In our case, we consider Robin transmission conditions instead of the classical Dirichlet ones, i.e. a well-chosen combination of Dirichlet and Neumann values at subdomain interfaces. A good choice of the Robin coefficient representing the relative weight of Dirichlet and Neumann values permits minimizing the number of iterations, which led to the name Optimized Schwarz Methods.

We combine these two improvements and apply them to a 2D Laplace test case up to 16,384 CPU cores. We obtain substantially improved computation times, comparable to the ones obtained with the multigrid library HYPRE interfaced by PETSc. This is significant in that Schwarz-type domain decomposition methods were up to now not considered competitive with multigrid methods on this type of problem. Furthermore, we extend the method to non-symmetric problems, adding an advection term to the Laplacian, and investigate various ways of adapting the coarse space.

🧑‍🏫 Victor Alessandrini, Maison de la Simulation, France
🌎 March 2019

Abstract:

There is today increasing consensus on the fact that quantum computing – an emerging data processing technology – may in the future play a significant role (with not yet fully understood boundaries) in high performance scientific computing. The simulation of quantum computers on standard computing platforms is today a necessary step to understand, assess, and develop quantum algorithms for computation, paving the way for the eventual future adoption of this disruptive technology. Emulation software will remain useful for some time to validate results of the earlier quantum computing platforms, and to help tuning application quantum codes. We will present a fully portable C++ emulation library under development for more than one year. The library disposes of three interfaces: a shared memory version running on SMP nodes, a MPI extension enabling access to a larger number of qubits, and a Python wrapper of the C++ library. A significant pedagogical effort is implemented in the documentation. Besides the traditional C++ class documentation, we are producing strategic higher level documentation and pedagogical papers on applications, including Jupyter notebooks. This seminar will propose first an introduction to quantum computing, explaining the fundamental differences with classical computing, underlining the strong features as well as the limiting bottlenecks of the technology. We will present next a short and very high level overview of the emulation library, focusing on a few major strategic choices. Then, the issues involved in quantum algorithms will be illustrated with a quantum chemistry example used in the validation tests of the software. Finally, we will conclude with a rapid overview of different areas in which quantum algorithm research is evolving (quantum chemistry, condensed matter physics, combinatorial optimization, fault tolerant quantum computing,…).

🧑‍🏫 Julien CAPUL, CEA-DAM, France
🌎 February 2019

PaDaWAn is an infrastructure providing in-memory data exchange between applications and a simple configuration model to switch from a file-based workflow to an in-transit workflow. The infrastructure is currently based on CEA-DAM Hercule parallel I/O libray by providing an ABI-compatible library to intercept simulation data in a transparent way and to facilitate integration into existing simulation codes and tools.

🧑‍🏫 Yann Meurdesoif, Researcher at the Laboratoire des Sciences du Climat et l’Environnement (LSCE)
🌎 January 2013

Abstract:

XIOS est un nouvel outil développé à l’IPSL (Institut Pierre Simon Laplace) destiné à gérer efficacement les sorties fichiers des modèles de simulations climatiques. Il vise deux principaux objectifs :

• Souplesse d’utilisation, grâce à une description externe des I/O sous forme d’un fichier XML hiérarchisé et d’une API fortran permettant de compléter ces informations à partir des codes.
• Performance : des cœurs de calculs sont exclusivement dédiés à la gestion des écritures via une technologie client/serveur, les données étant transmises des clients (code de calcul) vers les serveurs (cœurs I/O) par des communications MPI asynchrones non bloquantes. L’écriture des fichiers se fait alors en tâche de fond, en même temps que les calculs. Lors des phases d’écriture, XIOS exploite également le parallélisme du système de fichier via l’utilisation des bibliothèques netcdf4/hdf5 parallèle, permettant à plusieurs processus d’écrire simultanément dans un même fichier, et agrégeant ainsi la bande passante I/O vers les supports matériels.

Slides: