Marsha J. Berger

Creative contributor to computer science and applied mathematics. Devised a new methodology for the three-dimensional computation of complicated flows around aircraft and in airborne dispersal. Based on adaptive refinement of Cartesian grids (Adaptive Mesh Refinement or AMR), created hierarchical, multilevel methods with astonishing resolution at acceptable cost, and automation that uses computational resources efficiently. Applied methods to astrophysics, civil engineering, atmospheric sciences, groundwater flow, and numerical relativity using the software AMRCLAW, which has also been developed for computational cardiology and biological fluid dynamics. Contributions include generation of locally refined Cartesian grids, schemes that maintain conservation at grid interfaces, and the high-performance parallel implementation of AMR. Helped use self-consistent rescaling to determine singularity structure in nonlinear parabolic problems. Applications include inviscid flow and multiphase flow. Uses computational geometry and adaptive precision arithmetic to generate grids for full aircraft shapes; error estimators automatically determine where such grids should be refined.