The aims of the "action fédératrice" on high-performance numerical applications are to (1) promote and share the knowledge on HPC within the Observatoire de Paris; (2) provide aid to the researchers and engineers involved in HPC; (3) help the emergence of collaborations and large projects.
The first meeting was held on February 13th 2014. The slides of the presentation (in French) can be found here.
For information, to suggest a speaker or a training session, you should contact:
Andrea Ciardi (LERMA): firstname.lastname@example.org
Philippe Grandclement (LUTh): email@example.com
Videos of past seminars are available under SCIENTIFIC ACTIVITIES > videos of presentations
We are pleased to announce that Petros Tzeferacos, Associate Director and Code Group Lead of the Flash Center for Computational Science, will give a seminar and a tutorial talk on the FLASH code (http://flash.uchicago.edu/site/flashcode/) on Thursday 28th of May.
There is only a limited number of places and registration is obligatory. If you wish to participate please fill in this form: http://goo.gl/forms/JfiwXYWzB0
Location: Université Pierre et Marie Curie, LERMA, 4 Place Jussieu, 75005, Paris.
Date Thursday 28th of May from 9:30
Seminar, Thursday 28th of May 9:30 - 10:30
FLASH: a powerful tool for designing and interpreting HEDP experiments
With the advent of exceedingly capable compute resources, numerical modeling is a tool that the academic High Energy Density Physics community simply cannot pass up. In this seminar we will present and showcase the FLASH code, a free, publicly available, multi-physics framework that empowers academic groups, allowing them to simulate laboratory experiments with high fidelity. More specifically, we will recount the physics capabilities of the code and their applicability to laser-driven experiments; we will then present three experiments/case studies in Laboratory Astrophysics that aim to model turbulent amplification of magnetic fields in a controlled laboratory environment. This mechanism is ubiquitous in the Universe and can have significant implications to astrophysical processes that are encountered at disparate scales, from the formation of galaxy clusters to expanding supernova remnants. The experiments have been modeled and designed using the FLASH code, and are excellent examples of how FLASH can benefit academic experimental groups in designing and planning an experimental campaign, as well as analyzing the outcome of said experiments.
Tutorial, Thursday 28th of May 11:00 - 13:00
FLASH code, an open source framework for HEDP
In this tutorial we will be presenting the machinery of the FLASH code in the context High Energy Density Physics simulations. We will go through the infrastructure details of the code, the physics units it offers, and give “quick start” tips that will guide a new user in the process of acquiring and setting up the code. We will then discuss basic concepts of gas-dynamics numerical simulations; demonstrate how to build and run test problems offered with the FLASH code; and conclude with a step-by-step guide for creating new simulations of experimental platforms, pertinent to laser-driven experiments. This tutorial assumes a basic understanding of scrip and algorithm development, as well as a modicum of experience in UNIX based architectures.
SEMINAR: From Making Clouds to Making Outflows: Star Formation With the AstroBEAR AMR Multiphysics Code
Wednesday, 15th of October 2014 at 14:00. Salle de l'Atelier, site de Paris.
Adam Frank, Professor of Astrophysics, University of Rochester
Understanding star formation from the creation of molecular clouds to the feedback of stellar outflows on those clouds requires the development of state of the art Adaptive Mesh Refinement MHD multi-physics codes. In this talk I present two new studies using the AstroBEAR 2.0 code. In the first, I show MHD colliding flow simulations relevant to the formation of molecular clouds. Our models articulate the role of shear and field strength in the flows on the development of cloud properties such as mass, trapped magnetic flux and mixing. In the second study we present new simulations of Triggered Star Formation in which a shock sweeping over a magnetized cloud core triggers both collapse and the formation of an MHD outflow. I will also present a short overview of the code focusing on algorithmic challenges and strategies.
Jeremy Foulon, Institut du Calcul et de la Simulation, UPMC
PETSc is an open-source software/library design to allow engineers and scientists to perform large-scale numerical simulations rapidly and efficiently. PETSc has enabled simulations by industry, universities, and research laboratories, from desktop systems to the most powerful supercomputers. During the presentation, we will present an overview of capabilities and functionnalities from Petsc and some software which have choosen to use it.
Ivan Hubeny est invité par le labex Plas@Par
INTRODUCTION TO NON-LTE RADIATIVE TRANSFER
Ivan Hubeny, University of Arizona
In this lecture meant above all for students I will discuss the basic concepts of the physical description of the interaction between radiation and matter, as suitable for astrophysical applications, such as stellar atmospheres, accretion disks, and H II regions. The radiation is described through the radiative transfer equations which is formulated through phenomenological absorption and emission coefficients. In a general non-equilibrium situation these coefficients depend on the radiation field, and the set of atomic level populations. One is thus faced with a complex non-local, non-linear problem of solving simultaneously the radiative transfer equation and the kinetic equilibrium equations for level populations. I will briefly outline modern numerical methods for solving this problem efficiently.
Organisateur: Alain Coulais - firstname.lastname@example.org
Pierre Kestener, CEA
A wide range of major astrophysical problems can be investigated by means of computational fluid dynamics methods, and performing numerical simulations of Magneto-Hydrodynamics (MHD) flows using realistic setup parameters can be very challenging. We will first report on technical expertise gained in developing code Ramses-GPU designed for efficient use of large cluster of GPUs in solving MHD flows. We will illustrate how challenging state-of-the-art highly resolved simulations requiring hundreds of GPUs can provide new insights into real case applications: (1) the study of the Magneto-Rotational Instability and (2) high Mach number MHD turbulent flows.
This conference is co-financed by the labex Plas@Par and UPMC.
Date : Thursday March 2014, 27th from 9:00 to 18:00.
Location : Institut Poincaré, Amphithéâtre Darboux, 11, rue Pierre et Marie Curie, 75005 Paris (how to go there : http://www.ihp.fr/en/guide)
Program and registration are available on the Plas@Par website : http://plasapar.upmc.fr/?page_id=1124
Direct on line registration at: https://docs.google.com/forms/d/1a7VOD7Pgijfhtbq5ZhZf3w3NGD1JYDdJZkCo9KEgPR0/viewform.
In order to organize coffee breaks and free lunch on site, you should register before March 21st.
Compte rendu de la Journée sur la "Visualisation collaborative"
Le compte rendu de la Journee sur la Visualisation collaborative du 7 Nov 2013 (Neurospin-CEA) est accessible en ligne: