One of the most difficult challenges for component integration is the problem of integrating data structure components (e.g., sparse matrices) with functional components (e.g., linear algebra). This problem is hard because the frequency of invocation of data access methods places a premium on high performance of the component interfaces. The long term-research section of the proposal has taken this as a major focus for the next several years.

To drive this research in directions that are most useful to LANL, we will collaborate with developers on the Marmot code teams to understand how component integration strategies can make their efforts more effective overall. In particular, we will work to define a challenge problem by specifying the interfaces and functionality of components within Marmot that implement abstract meshes on which computations are carried out. These specifications will be developed through a joint study between code developers and computer and computational scientists within LACSI. A goal of this effort is to leverage the telescoping languages strategy for efficient component integration that is the subject of LACSI research.  The ultimate goal is to make it possible for the designer to specify data structures and functionality at a high level of abstraction without sacrificing the efficiency required by production weapons codes.