The aim of this proposed work is the development of new scalable algorithms and software to address a number of outstanding problems arising in scientific computing applications. In particular,computationally complex simulations that demand the solution of multi-scale or multi-physics models will benefit. A significant component of this research will be the collaboration with applications scientists considering problems of this nature.
The project PIs have had significant experience and success in the development of scalable algorithms for the fundamental problems inherent in unstructured mesh calculations. The realization of this toolkit will be a significant resource enabling educators and scientists to make better use of the large-scale computing infrastructure. This proposed work is driven through signi.cant collaborations by the PIs with other NSF-funded initiatives at Universities,on-going work at the National Laboratories,and with Industry. These collaborations will promote widespread use of the toolkit as well as feedback to ensure that the toolkit developed is relevant to user needs. The toolkit will be disseminated into the public domain, as well as through direct collaboration, to ensure widespread use.
