This year’s International Meshing Roundtable was held just down the road from Pointwise, in Austin, Texas on 11-14 October 2015. Now in its 24th incarnation, the IMR is the event for mesh generation as it brings together meshing researchers and practitioners from government, academic, and industrial organizations worldwide.
Sandia National Labs began hosting the IMR back in 1992 and remains the event’s primary host, benefactor, and organizer. Each year’s event is run by a volunteer steering committee consisting. The AIAA’s Meshing, Visualization, and Computational Environments technical committee is an active co-sponsor and always has representation on the steering committee. Plus, the IMR solicits sponsorships each year to help provide financial support for students and postdocs. [In the interest of full disclosure, Pointwise sponsors the IMR, I have served on its steering committee, and I’m a member of AIAA’s MVCE. So forgive me if I’m biased.]
The four short courses at this year’s IMR were:
- Jonathan Shewchuk from Berkeley on Tetrahedral Mesh Improvement and Dynamic Meshing [always an excellent presentation]
- Jessica Zhang from Carnegie Mellon on Image-Based Mesh Generation and Volumetric T-spline Modeling for Isogeometric Analysis
- John Verdicchio from Cambridge Flow Solutions on Commercial Mesh Generation – Why It’s Different to Being at University. [I’m told this was a lively course and it led to a follow-on discussion in the Open Spaces session.]
- Matthew Staten from Sandia on Introduction to Quad and Hex Mesh Generation and Modification
Invited speaker Thierry Coupez opened the IMR with a rousing presentation on his work in adaptive meshing (the goal being to make a mesh that’s unique to the solution).
INRIA’s Adrien Loseille presented their work on fast adaptive meshing. Their target benchmark performance is 1 billion cells in 15 minutes on 100 processors because that hardware is representative of what an engineering organization might typically have. [In other words, let’s not target a massively parallel HPC system that only 1% of users have access to.]
The IMR’s professional development session is something I enjoy listening to. It’s targeted for students and postdocs so they can gain insight into meshing as a career. [RUN AWAY! No seriously, it’s about where their field is going and how they can get a job.] This year’s session was centered around the results of the pre-conference survey the IMR conducted of meshing professionals. [I plan to write more about the survey later.] Sadly, there were few questions from the audience and it makes me wonder how we can better serve the needs of students. The session’s general conclusions were:
- Hex meshing remains the primary need/want.
- Funding for academic research in meshing is spotty.
- The physics of the solution need to be linked to the mesh (i.e. adaption)
- Meshing needs to be cost effective, robust, and cheap. [Better, faster, and cheaper? I thought we could only get 2 of the 3.]
- We need to strive toward providing a mesh that balances the quality of the simulation and the resources required to compute it.
One of the panelists (I’ve forgotten whom) said that experience with parallel programming will help you get a job.
Omar Ghattas from the University of Texas gave an extremely interesting presentation on melting of the Antarctic ice sheet. His work may have been the most complicated simulation I have ever seen and another attendee remarked that Dr. Ghattas was easily the smartest person in the room because each attendee could probably cite at least one thing he said that went over their head. But consider what he’s trying to compute. He wants to understand what’s happening at the ice-ground interface because that’s key to understanding whether melting will accelerate, decelerate or remain at the same rate. But that’s a boundary condition. So the inverse problem becomes one of simulating the ice flowing and melting to match satellite observations to determine what the boundary condition actually is. [At least, that’s how I understood it.] Antarctica’s ice is melting at the rate of 200 billion gallons per year. This is expected to lead to a half meter rise in sea levels by the year 2070. This rise will impact 136 coastal cities, 150 million people, and $35 trillion of assets.
Sam Landier’s presentation of Boolean Operations on Arbitrary Polyhedral Meshes delved deeply into work that’s available in ONERA’s Cassiopee package. Their work always fascinates me.
Open Spaces consisted of self-organized 1-hour discussions on topics decided upon by attendees on-the-spot. Think of them like birds-of-a-feather sessions but with no advance notice of what they’ll be. I don’t have a good feel for how effective they were.
For the third consecutive year, the IMR has hosted a meshing contest for which they supply CAD models and mesh criteria and evaluate the suitability of the submitted meshes. (The CAD models are still online should you want to try them yourself.)
This year they supplied two geometries: a CAD model of an electric guitar and a 3D scan of the human abdomen. We at Pointwise meshed both and then struggled to figure out how to use both of them on a poster. Fortunately, we were inspired by the fact that the IMR overlapped by one day with the Austin City Limits music festival.
Even more fortunately, Pointwise’s mesh was named winner of the meshing contest. [We’ll probably write more about the meshes at a future date.]
- Best Poster: High-Order Unstructured Curved Mesh Generation Using the Winslow Equations, Meire Fortunato and Per-Olof Persson, University of California at Berkeley
- Best Student Paper: On Indecomposable Polyhedra and the Number of Steiner Points, Nadja Goerigk and Hang Si, Weierstrass Institute
- Best Technical Paper: Parallel Generation of Large-Size Adapted Meshes, Adrien Loseille, Victorien Menier, Frederic Alauzet, INRIA
This year’s IMR Fellow is Dr. Tim Tautges from CD-adapco.
The IMR is growing. Next year’s event expands to five days, Mon-Fri, 26-30 September 2016 in Washington, DC. Short courses will be held on Monday. I’m told that the expanded event will provide more opportunities for meshing implementations and applied meshing.
Complete conference proceedings are available online.
Caveat 1: I admit to being a terrible photographer.
Caveat 2: Dear Presenter, please put down the laser pointer. It’s not helping. In fact, it’s making things worse. Much, much worse.