Validating astrophysical simulation codes

A. Calder, J. Dursi, B. Fryxell, T. Plewa, G. Weirs, T. Dupont, H. Robey, J. Kane, P. Drake, B. Remington, G. Dimonte, J. Hayes (+5 others)
2004 Computing in science & engineering (Print)  
M uch of the contemporary research in astrophysics necessarily involves the development of models and simulation technology to help us understand the complex phenomena underlying astrophysical events. As with any modeling approach, numerical models and simulation codes developed for astrophysical applications must be thoroughly verified and validated to demonstrate their accuracy and assess their credibility. Although verification and validation (V&V) is maturing as a discipline due to its
more » ... ine due to its importance in fields such as computational fluid dynamics (CFD), until recently it has received scant attention in the astrophysical literature. In astrophysics, validation is a challenging problem because many of the conditions of interest can't be reproduced in terrestrial laboratories. In addition, astrophysical settings such as the interiors of stars are not directly observable. Even if this were not a problem, most astrophysical events happen at extreme distances, making observation difficult if not impossible. The complexity of astrophysical events also makes validating an astrophysical simulation code difficult. Astrophysical events typically involve many complex interacting physical processes, all of which must be included in a realistic model. Direct numerical simulations (for example, resolving all of a problem's scales of interest) are often beyond the capabilities of current computers, requiring the development of approximate methods and descriptions of the physical processes on scales smaller than the simulation can track (subgrid models). Validation requires identifying the physical Copublished by the IEEE CS and the AIP Astrophysical simulations model phenomena that can't be fully reproduced terrestrially. Validation then requires carefully devising feasible experiments with the relevant physics. The authors describe validating simulations against experiments that probe fluid instabilities, nuclear burning, and radiation transport, and then discuss insights from-and the limitations of-these tests.
doi:10.1109/mcse.2004.44 fatcat:s6na7flkongfxlu7f5vb4vvdfm