Level: Intermediate/Advanced
Organizer
Ye Zhao, Kent State University
Arie Kaufman, Stony Brook University
Klaus Mueller, Stony Brook University
Nils Thuery, ETH Zurich
Ulrich Rüde, University Erlangen-Nürnberg
Download - Video | Audio
Running Time: 3 hr 4 min
Agenda
Introduction - Presentation Slides [PDF]
Lattice Boltzmann Method and GPU Acceleration, Ye Zhao - Presentation Slides [PDF]
Efficient Implementation and Parallelization, Klaus Iglberger - Presentation Slides [PDF]
Lattice-based flow simulation Applications, Klaus Mueller, Ye Zhao - Presentation Slides [PDF]
Free Surface and Adaptive Simulation, Nils Thuerey - Presentation Slides [PDF]
Abstract
The advances in applying physical models and numerical PDE (Partial Differential Equation) solvers for flowsimulation, along with the rapid increase in computer power, open a new era in a variety of fields and applicationdomains in computer graphics and visualization. As a unique explicit, simple and inherently-parallel scheme, the latticeBoltzmann method (LBM) has developed into a promising numerical method for simulating fluid flows and modelingphysics in fluids. It has achieved great success in the world of computational physics both from the analytical andpractical points of view. The LBM scheme excels due to its very efficient and simple computing process for modelingfluid dynamics even in the presence of very complex boundary conditions, such as arbitrarily-shaped obstacles, movingobjects, free surfaces and the like. Due to its discrete nature, the LBM lends itself well to efficient interface tracking asto adaptive and multi-resolution approaches, which are both critical for flow simulations in realistic graphicsapplications. Moreover, its computational pattern, which is similar to cellular automata, is easily parallelizable. Thismakes the LBM very amenable to acceleration on parallel computers, such as single GPUs (Graphics Processing Unit)and for GPU clusters, enabling it to achieve interactive or real-time flow simulation performance in a scalable fashion.Due to these many merits, the LBM continues to gain momentum and adoption within the computer graphics andvisualization community for the modeling of various fluid phenomena. But despite these positive trends, the power ofthis parallel lattice-based scheme has still not been fully utilized, and we feel that this is mostly due to the lack ofproper education activities in the method’s underlying principles. Fortunately, unlike other simulation methods incomputational physics, the LBM’s ease of use and acceleration, and especially its well-known gentle learning curve,make it relatively easy for users to become proficient. The authors of this tutorial have a long history in applying theLBM for the modeling of a wide variety of flow phenomena. Since the year 2001, these works have lead to numerousresearch papers, presentations and collaborative outreaching projects. We found this period a very enjoyable one andwe are extremely optimistic for these successes to persist in the near and far future. In this course, we hope to share thisenthusiasm with the audience who are interested in interactive and real-time flow simulation and visualization. We areaiming to educate attendees in the proficient use of the LBM in various application domains: for the development ofcomputer graphics and games, for prediction simulation capabilities, and for general computational science andengineering.