DistOpt: A Ptolemy-based Tool to Model and Evaluate the Solution of Optimization Problems in Distributed Environments.
Arturo Losi, Universita' di Cassino - Italy

To implement the solution of optimization problems in a parallel/distributed environment, many architectures, many hardware and software products, as well as different solving methods can be chosen, since the advances in technology and computational methods have vastly increased the number of options available. The knowledge necessary to cope with all the options is so wide, the state-of-the-art is changing so rapidly that a flexible and cooperative software environment is needed to evaluate the impact of any single choice on the performance of the overall system, in order to compare the various alternatives.

To model the solution of optimization problems in a parallel/distributed environment and to evaluate its performance, we have developed a Ptolemy-based tool, founded on the Auxiliary Problem Principle (by G. Cohen), named DistOpt. Thanks to DistOpt, users can easily build a representation of the problem to be solved (also by intefacing to existing application-oriented environments), choose the way the whole problem is partitioned into subproblems and their mathematical models are built, and choose the algorithm for solving each subproblem (also using existing codes).