Dr. Rahul Narain of the University of California, Berkeley will speak on "Adaptive Methods for Physical Simulation."
Numerical simulation of complex physical systems is a powerful tool used
in many practical applications including design, engineering, and visual
effects. Many natural phenomena have traditionally been challenging to
simulate efficiently, due to salient features emerging at fine scales.
For example, features such as the turbulent vortices in smoke, wrinkles
in cloth, and creases in metal sheets all have significant impact on
large-scale behaviour, but capturing these features requires finely
resolved simulations. In my talk, I will describe techniques to enable
efficient solutions of such simulation problems through novel
mathematical models and adaptive discretization techniques. In
particular, I will focus on adaptive remeshing for simulating thin
materials such as cloth, paper, and metal sheets. Through remeshing, we
can dynamically adapt the resolution of the simulation mesh to
efficiently represent the state of the material. Emerging features such
as wrinkles and creases are accurately resolved by refining the mesh
while smooth, quiescent regions are represented parsimoniously, using
coarser elements. Compared to fixed discretizations, adaptive meshes
yield substantial reductions in computational cost and make it tractable
to simulate complex, finely detailed behaviours to high fidelity. These
advantages point towards the potential of adaptive techniques to enable
versatile, robust, and accurate solutions to a wide range of challenging
Rahul Narain is a postdoctoral scholar at the University of California,
Berkeley, where he is a member of the Visual Computing Lab. His research
interests lie in physically based modeling and numerical methods for
computer graphics and computational science. He received a B.Tech. from
the Indian Institute of Technology, Delhi in 2006, and an M.S. and a
Ph.D. in Computer Science from the University of North Carolina at
Chapel Hill in 2009 and 2011 respectively. He has received several
awards including an Intel Ph.D. fellowship, and he is the lead architect of
the ARCSim simulation engine that is being used in industry and in
several research labs.
Events are free and open to the public unless otherwise noted.