Computational proxies: modeling scientific applications in object databases

This paper addresses problems of data management and interoperability for computational science applications. We recount efforts to design and implement a framework for computational experiment management that encapsulates computationally intensive programs in an object-oriented database. The mechanism we designed and implemented, dubbed “computational proxy”, is defined, and a conceptual design for it rendered. We also describe the functional components of the proxy, i.e. the ability to start up and control application invocations and to capture experimental results into the database. We specifically address applications in computational chemistry, but believe our work is applicable to other computational sciences