New techniques for improving the accuracy of gradients for particle and grid simulations of magnetohydrodynamic turbulence.

American Museum Natural History, New York NY

Investigators

Abstract

AST-0612724 Maron In astrophysics, computing the flow of magnetized gases is central to understanding systems ranging from stellar formation and evolution to black hole accretion disks. However, how well a fluid modeling algorithm performs depends greatly on how well it can compute derivatives or gradients. This project will implement two new types of algorithms, one for particles using polynomial regression techniques, and one for grids using high-order, volume-centered, constrained transport, combined with tuned finite difference coefficients. When used with a de-aliasing correction, this yields near spectral quality in a finite difference code, allowing large parallel computations at relatively low cost. These algorithms will be available as independent modules, but also as part of the publicly distributed and broadly used codes GADGET2 and Pencil, in order to promote their wide use in the astrophysical community. The work will integrate applied mathematical techniques and astrophysical simulations. Research results will be incorporated into Space Shows of the Hayden Planetarium at the American Museum of Natural History, viewed locally by a million visitors a year, including over a hundred thousand New York City school children, and into training provided to hundreds of teachers annually.