Inital Short Course information can be found below. Addtional details will be posted soon.
Courses are taught by internationally known experts. Instructors typically include an overview of the state of the art of their topic, and highlight their own research, but also include the current work of others. It is intended to be a “course” in the traditional sense of enabling attendees to go forth and produce new results of their own, rather than simply use existing knowledge. This year we are having two short course tracks, each with two classes. One track is traditional “core” meshing topics, and the other is topics that we believe would “enrich” the perspective of meshing researchers beyond what they are most familiar with. The goal of the core topics is to bring attention to the state of the art, so that attendees would be positioned to contribute directly to that topic. The goal of the enrichment topics is to make attendees aware of exciting knowledge from nearby fields that could bring a new set of tools, math, and perspectives to meshing research. Both tracks are suitable for both new and experienced meshing researchers.
The IMR short courses will be held Monday, September 18, 2017. Courses are taught by internationally known experts in the field of Mesh Generation. Instructors will address practical issues in the design and implementation of both structured and unstructured mesh generation codes.
To register for the short courses, mark the appropriate boxes on the registration form.
Title: An introduction to curved high-order mesh generation
Biography: Josep Sarrate is currently an associate professor at the Laboratori de Càlcul Numèric (LaCàN) of the Universitat Politècnica de Catalunya (UPC). He got his bachelor degree in Physics from Universitat de Barcelona (UB) in 1985, and his PhD in Physical Sciences in 1996 from Universitat Politècnica de Catalunya (UPC). His research focuses on numerical modelling in applied sciences and engineering. Looking at the discretizations needed by these models, he has contributed to the development of several algorithms for semi-structured and fully unstructured quadrilateral and hexahedral mesh generation. In addition, he has also collaborated in the development of algorithms to generate curved high-order meshes for CAD models.
Abstract: During the last two decades, special effort has been focus on the development of high-order methods to solve partial differential equations. Nowadays, these methods exhibit several features that make them competitive in a wide range of applications. For instance, they converge exponentially with the order of the approximating polynomial when the exact solution of the PDE is smooth and without singularities, they can deliver higher accuracy with a lower computational cost than low order methods in several applications or they provide lower numerical dissipation and dispersion than linear solvers.
However, to achieve the theoretical advantages of unstructured high-order methods the geometry has to be approximated with high-order accuracy as well. Therefore, special attention has been focused on the development of algorithms to generate curved high-order meshes that match the geometry boundary.
In this course, we will classify and review the main strategies to generate these curved high-order meshes. We will identify the main issues that have to be addressed in order to generate them and how they are treated by different methods. In addition, we will also analyse procedures to assess whether a curved element is geometrically valid. Finally, we will discuss the challenges that curved high-order methods must address in the near future.
Title: An Introduction to Polyhedral Meshing
Biography: Stefano Paoletti is currently Director of Meshing Research at Siemens SISW. Previously worked for the IBM ECSEC (European Center for Scientific and Engineering Computing) at Rome’s IBM Scientific Center where he started his focus on CFD and mesh generation. Was software developer and manager at CD-adapco in the mesh generation department for the last 20 years. His main interest is the polyhedral mesh generation in its various declinations. He was the main developer and the leader of the polyhedral meshing group in CD-adapco for the Star-CCM+ suite.
Abstract: This short course will introduce polyhedral meshing. We will start with the definition and a brief history of the main methodologies used today for polyhedral mesh generation. The relative merits of a few generation algorithms will be reviewed and attention will be paid to explain the benefits and the difficulties that can be encountered by those new to this particular form of mesh generation. Finally, an insight of where the polyhedral mesh generation is going will be provided.
Title: Challenges in Meshing and Form-Finding for Architectural Geometry
Biography: Amir Vaxman is an assistant professor in the Geometric Computing Group in The Department of Information and Computing Sciences at Utrecht University, The Netherlands. He was a postdoctoral fellow at TU Vienna, working with Helmut Pottmann in the Geometric Modeling and Industrial Geometry group. He received his PhD from the Technion-IIT under the supervision of Gill Barequet. His research focuses on vector and directional-field design, and architectural geometry, with an emphasis on polyhedral meshes.
Abstract: There are two major concepts in geometric shape design for architecture: the combinatorics, which is the connected set of elements that tessellate and represent the object, and the geometry, which is the embedding of the object in space. Both concepts are dependent and complementary, and often equally important both visually and structurally.
Designing shapes and meshes for architectural realization must meet both functional and aesthetic demands: on the functional side, the meshes must adhere to constraints such as stability, face planarity, prescribed space measures, uniform elements for cost effectiveness, and ease of fabrication and assembly. On the aesthetic side, the design process must be flexible enough to meet the vision of the architect, and provide full freedom of expression and intuitive user control. In addition, designers and architects constantly seek to work with novel and unconventional meshes, such as polyhedral patterns, circular arcs, and more.
I will give an account of the mathematics and algorithms behind these design methods. They are derived from notions in discrete differential geometry and numerical optimization. The course will focus on two aspects: geometry-from combinatorics, which is about finding the form of a shape to address user input on a fixed mesh, and combinatorics-from-geometry, which is about creating meshes that fit a given geometric shape design, and adhere prescribed constraints
Title: GPU Programming for Meshing Algorithms
Nancy Hitschfeld Biography: Nancy Hitschfeld received her BSc and MSc degrees in Computer Science from the University of Chile in 1984 and 1987, respectively. She received a PhD in Applied Sciences (Technischen Wissenschaften) from the Swiss Federal Institute of Technology (ETH-Zurich) in 1993. Currently, she works as an associate professor at the Computer Science Department (DCC) of the Faculty of Physical and Mathematical Sciences at the University of Chile. Her main research interests include geometric modeling, polygonal and polyhedral meshes, and parallel algorithms (GPU computing), for problem solving in computational science and engineering applications. She received two best papers awards, one of them with Cristóbal Navarro for improving the efficiency of thread mapping onto triangular domains for GPU architectures. Cristóbal and Nancy were awarded an NVidia GPU Research Center in 2016.
Cristóbal A. Navarro Biography: Cristóbal A. Navarro is an assistant professor at the Institute of Informatics of Universidad Austral de Chile. He received his PhD in Computer Science from the University of Chile in 2015. His research interests include High Performance Computing, Computational Physics and Real-time Computer Graphics. He is currently the main researcher of a Fondecyt Postdoc project aimed at studying efficient thread mapping techniques for GPU computing.
Abstract: This course focuses on how to represent and manipulate triangular meshes with GPU Computing in order to accelerate geometrical computations. Its target audience is those motivated by, or with a background in, computational geometry but who are new to GPU computing. The course has been designed to be highly practical and illustrative, so that the attendees can get a clear idea of which computing patterns, designs and strategies are the ones that will produce efficient GPU performance. Fundamental concepts of GPU computing will be introduced along with relevant techniques to make optimal use of GPU hardware, such as thread branching, coalesced memory, effective use of shared memory, thread mappings onto a triangular mesh, dynamic memory allocation, thread-safe exclusion mechanisms for neighboring triangles, and indeterminate decisions and traversals over the discrete graph structure of unstructured meshes. Part of the course includes a hands-on tutorial where attendees can apply the techniques to (1) transform any triangulation into a quasi-Delaunay triangulation, and (2) simplify meshes using a parallel edge-collapse technique.