9361232 Taylor Existing commercial modeling and simulation software packages do not adequately predict the behavior of complex computer- controlled systems being developed and fielded today. These "hybrid" systems involve discontinuous phenomena such as mechanical parts engaging and disengaging (e.g., spacecraft docking), control units switching, and software components reconfiguring. Unfortunately, these effects are handled inaccurately and sometimes even missed completely. Recent research in nonlinear systems has revealed that there are many circumstances where this deficiency presents a major barrier to showing that a system will perform as required, or conversely that a newly designed system may have safety-critical defects. For example, a hybrid system may be prone to oscillation or to chaotic behavior (seemingly random motion) and the simulator will not be able to predict these problems in time to avoid costly accidents or redesign. We propose in Phase I to develop both a rigorous modeling language for representing such phenomena, and integration algorithms that will simulate discontinuous nonlinear hybrid systems with complete accuracy. The language will be developed according to modern standards, to facilitate correct modeling practices. The integration algorithms will rigorously catch and hand all discontinuous events. A commercial package will be produced in Phases II and III. ***
