4:00 pm Tuesday, August 26, 2014
ICES seminar: Sparse Optimization for Elliptic Obstacle Problems by Giang Tran (UCLA) in POB 6.304
We construct an efficient numerical scheme for solving obstacle problems in divergence form. The numerical method is based on a reformulation of the obstacle in terms of an L1-like penalty on the variational problem. The problem can be solved by decoupling it into a non-differentiable term and a strictly convex smooth term. For the first term, the minimizers are computed directly using shrink-like operators. For the smooth term, we use either a conjugate gradient method or an accelerated gradient descent method to quickly solve the subproblem. Our formulation is applied to classical elliptic obstacle problems as well as some related free boundary problems, including the two-phase membrane problem, the divisible sandpile and the Hele-Shaw model. One advantage of the proposed method is that the reformulation is an exact regularizer in the sense that for large (but finite) penalty parameter, we recover the exact solution. Moreover, the free boundary inherent in the obstacle problem arises naturally in our energy minimization without any need for problem specific or complicated discretization. This is the joint work with S. Osher, H. Schaeffer, R. Caflisch, and W. Feldman. Submitted by
|
|