An important and resource-intensive class of computation codes consists of simulators for physical systems. Today, most simulation codes are written in general-purpose imperative languages such as C or FORTRAN. Unfortunately, such languages encourage the programmer to focus her attention on details of how the computation is performed, rather than on the system being modeled.

This paper reports the design and implementation of a novel notion of an object for a physical modeling language called Acumen. A key idea underlying the language's design is encouraging a programming style that enables a "globally parallel, locally imperative" view of the world. The language is also being designed to preserve deterministic execution even when the underlying computation is performed on a highly parallel platform. Our main observation with the initial study is that extensive and continual experimental evaluation is crucial for keeping the language design process informed about bottlenecks for parallel execution.