부산대학교 도서관

Current practice in stellar evolution is to employ one-dimensional calculations that quantitatively apply only to a minority of the observed stars (single non-rotating stars, or well detached binaries). Even in these systems, astrophysicists are dependent on approximations to handle complex three-dimensional processes like convection. Understanding the structure of binary stars, like those that lead to the Type Ia supernovae used to measure the expansion of the universe, are grossly non-spherical and await a 3D treatment. To approach very large problems like multi-dimensional modeling of stars, the Lawrence Livermore National Laboratory has invested in massively parallel computers and invested even more in developing the algorithms to utilize them on complex physics problems. We have leveraged skills from across the lab to develop a 3D stellar evolution code, Djehuty (after the Egyptian god for writing and calculation) that operates efficiently on platforms with thousands of nodes, with the best available physical data (opacities, EOS, etc.). Djehuty has incorporated all basic physics for modeling stars including an accurate equation of state, radiation transport by diffusion, thermonuclear reaction rates, and hydrodynamics, and we have begun testing it in a number of applications.
Comment: 14 pages, 8 figures. Proceedings of the 3D Stellar Evolution workshop held in Livermore, July 22-26 2002. ASP Conf. Series vol. 293