Third edition of the Mmg day

The 3rd Mmg day will be held on 13th December (within the FreeFem++ days).

It will take place at:

Corridor 13-14, Room 101,
Jussieu campus,
4 place Jussieu,
75005 Paris (France).

The registration is free but mandatory: https://www.mmgtools.org/mmg-simplicial-remesher/mmg-day-2018-subscription-form

The purpose of the day is to:

  • present news regarding the platform and its community;
  • discover how users use the Mmg software;
  • foster exchanges between Mmg users and developpers.

Program

14h00 – 14h15: Welcome speech. Algiane Froehly (Consortium Mmg)
14h15 – 14h45: Mesh adaptation for Large Eddy Simulation for compressible combustion. Gabriel STAFFELBACH (CERFACS)

The objective of this presentation is to highlight state of the art usage of mesh adaptation using the MMG library suit for Large Eddy Simulation at CERFACS. The software is used for improved numerical results but also to improve user experience in our code GUI. Recent examples using MMG on reactive and non-reactive flows will be presented.

14h45 – 15h15: Shape optimization based on a level-set mesh evolution method. Florian FEPPON (Ecole polytechnique)

We present a implementation coupling FreeFem++, python and mmg2d that allows to treat a variety of multiphysics shape optimization problems with a level set based mesh evolution method. In the context of a model weakly coupling thermal fluid and elasticity state equations, our implementation allows the user to specify arbitrary objective functions and constraints. For the optimization, we rely on our recent nullspace gradient flow algorithm which allows to solve conveniently the mathematical program with multiple equality and inequality constraints. Illustrations on various test cases will be shown.

15h15 – 15h45: Software development for the computation of maximal probability domains (MPDs). Jeremy DALPHIN (ISCD, Sorbonne Université)

In this talk, we will be interested in a quantum chemistry model, which yields a three-dimensional vizualization of interactions between molecules. These Maximal Probability Domains (MPDs) are the local maximizers of a shape optimization problem. We will present the numerical strategy adopted to obtain MPDs and how it heavily relies on the mmg3d software for both an accurate discretization of the level-set isosurface and a mesh adaptation to the chemistry encoded in the wave function. Finally, some results will be shown to conclude the presentation. This is a joint work with Pascal Frey, Yannick Privat, Benoit Braida, Andreas Savin, and Charles Dapogny.

15h45 – 16h15: Break

16h15 – 16h45: Mmg for 3D graphics. Loïc NORGEOT (Institut Carnot-SMILES,Sorbonne Université)

MMGtools is great for two-dimensionnal mesh optimization, implicit domain meshing, isovalue discretization… Serious mathematical concepts, with plenty of serious academic and industrial applications. But this toolbox can also provide a huge value for more “amusing” applications: video game creation, web publishing, 3D printing, VR & AR applications… This presentation will double up as a short tutorial, showcasing ways to intuitively use mmgs to reprocess 3D models within Blender (an open-source 3D graphics software). If you wish to follow the hands-on session, make sure to install Blender (and mmgs) on your computer beforehand. And although using a mouse with Blender is not mandatory, it should definitely make your life easier if you have never used this software before!

16h45 – 17h15: Surface reconstruction for tumor segmentation in medical imaging. Olivier GALLINATO (SOPHiA Genetics)

SOPHiA Genetics develops SOPHiA, the universal technology for Data-Driven Medicine. SOPHiA AI continuously learns from thousands of patients’ genomic profiles and experts’ knowledge to improve patients’ diagnostics and treatment options. Recently, SOPHiA extends towards radiomics by introducing SOPHiA Radiomics, a groundbreaking application that analyzes medical images and combines them with biological and clinical data to provide to the radiologist more information than what is visible to the trained eye. This powerful radiomic solution is designed to go beyond the usual RECIST and PERCIST criteria, to deliver finer biomarkers and support clinicians in choosing the right therapeutic strategy for a given patient. SOPHiA Radiomics analysis begins with the segmentation of medical images that produces a precise contouring of the tumor thanks to our proprietary algorithms. We use MMG3D and MMGS as external libraries to refine our 3D meshes representing the tumor segmentation in two applications. The first application is the manual correction of the tumor contouring. The user can modify the delineation on each 2D slice of the medical image and MMG tools are used to build the corrected 3D mesh thanks to level-set discretization and surface refinement. The second application is the segmentation of complex tumors. Sometimes, several segmentation processes are required to catch all parts of complex tumors. MMG makes it possible to merge the different segmentations into a unique 3D mesh thanks again to level-set discretization and surface refinement. These powerful tools allows us to provide accurate 3D tumor modeling and then accurate radiomics evaluation to clinicians.

17h15 – 17h45: Practical cases on the integration Kratos Multiphysics-MMG. Vicente MATAIX FERRANDIZ (CIMNE)

The previous year our team (CIMNE, Prof. R. Rossi) presented the way we coupled our framework, Kratos Multiphysics, with MMG. This year we want to present the advances and applications obtained with the MMG library in the last year. Many new developers have started to use the library and incorporate into their simulations different re-meshing techniques. Between this applications we can find the use in embedded CFD simulations, DEM-FEM iteraction, MMC (Multilevel Monte Carlo), among others.

Important dates

November 23 – Presentation submission deadline
December 3 – Acceptance notification
December 7 – Mmg day subscription deadline (has been extended)

The Mmg councils will be held with the consortium members during the morning.

Comments are closed.