Publications (3)0 Total impact
[show abstract] [hide abstract]
ABSTRACT: A new problem in scientific computing is the merging of existing simulation models to create new, higher fidelity combined models. This has been a driving force in climate modeling for nearly a decade now, and fusion energy, space weather modeling are starting to integrate different sub-physics into a single model. Through component-based software engineering, an interface supporting this coupling process provides a way to invoke the sub-model through the common interface which the top model uses, then a coupled model turns into a higher level model. In addition to allowing applications to switch among linear solvers, a linear solver interface is also needed for the model coupling. A linear solver interface helps in creating solvers for the integrated multi-physics simulation that combines separate codes, and can use each code’s native and specialized solver for the sub-problem corresponding to each physics sub-model. This paper presents a new approach on coupling multi-physics codes in terms of coupled solver, and shows the successful proof for coupled simulation through the implicit solve. KeywordsParallel model coupling-Component architecture and interface-Sparse matrix computation04/2010: pages 503-518;
Conference Proceeding: A HPC sparse solver interface for scalable multilevel methods.Proceedings of the 2009 Spring Simulation Multiconference, SpringSim 2009, San Diego, California, USA, March 22-27, 2009; 01/2009
Concurrency and Computation: Practice and Experience. 01/2008; 20:1439-1454.