Our proposal and the Phase I work completed under it addressed these NASA-identified needs by providing software infrastructure that provides physical scientists a "plug-and-play" architecture in which they can insert their "physics kernels" and exploit very large existing code bases for the computer science aspects of the problem. In particular, our STTR product provides cutting-edge adaptive mesh refinement (AMR) capabilities, and our Phase I results demonstrate the ability of our software architecture to run existing physics code with the newly incorporated AMR driver. Our innovative solution to the problem delivered significant value to NASA at a relatively small cost by combining existing open source tools. In Phase I, we built an interface, which we call Parca, between the Paramesh computational libraries, developed at NASA GSFC to support AMR computations in the area computational hydrodynamics, and the Cactus computational toolkit, which is an infrastructure package developed by Louisiana State University that provides a "plug-and-play" framework for cross-institution and cross-disciplinary scientific codes. Both of these software packages have large user bases in the areas of computational fluid dynamics and numerical general relativity, and both had existing users at NASA GSFC. Prior to our Phase I work, there was no way these user communities to collaborate directly, leading each user group to redevelop software already available in the other user community.
More »