Office: 6-225E Keller Hall
Phone: (612) 625-2013
Email: narain@umn.edu

Rahul Narain rhymes with "car full o' lion"

I'm an assistant professor in the Department of Computer Science and Engineering at the University of Minnesota, Twin Cities. My research focuses on numerical methods and mathematical models for physics-based animation, though I'm also interested in applications to design, robotics, and scientific computing.

Previously, I received a B.Tech. from the Indian Institute of Technology Delhi and an M.S. and a Ph.D. from the University of North Carolina at Chapel Hill advised by Ming C. Lin, and did a postdoc at the University of California, Berkeley working with James F. O'Brien.

Research

For the benefit of prospective students, I've written little blurbs about some of the topics I'm interested in, with selected publications and questions I want to work on in the future. If any of these sound exciting, or you have your own ideas along similar lines, get in touch! I'm looking for enthusiastic students who have a stomach for maths and think physics is cool.

Adaptive remeshing: Lots of physical phenomena show fine localized detail, from folds and wrinkles in cloth to crack patterns in shattering objects. Naturally, we'd like to refine the mesh resolution where there is detail and use coarser elements in smooth regions. The tricky part is anticipating where detail is going to emerge and refining sufficiently in advance, so you don't lose all the interesting dynamics.

Ideas for future work: • How can we express anticipatory refinement in a general, principled way that applies to other physical systems? • How can we determine how much refinement is necessary before the benefits become visually imperceptible? • Are some discretizations of internal forces more compatible with remeshing than others?

Fluids etc.: Computer graphics has a rich history of techniques for simulating smoke, water, and other Newtonian fluids. I like to think about ways to simulate nontraditional fluid-like substances — like sand and oobleck — and to simulate fluids in nontraditional ways — like vorticity-based and boundary element methods.

Ideas for future work: • Can we simulate liquids using a primarily surface-based representation? • How accurately can granular and non-Newtonian fluids be simulated in real time?

Crowd simulation: Crowds of pedestrians can in certain situations be well approximated as a continuous fluid-like system, particularly when the crowd is large and dense. In these cases we can apply the tools of continuum mechanics to model the behaviour of the crowd, enabling efficient simulations and revealing new insights.

Ideas for future work: Actually, I haven't thought seriously about crowd simulation for a while, so I don't have any ideas right now, but if you do we can certainly talk!

Publications

Here's my full list of publications.

Software