Extracting logical theories from DEVS models

An investigation is made of the use of relational state representations in modeling and simulation and, in particular, of the use of such representations for reasoning about model behavior. Two possibilities are identified: one is based on automated theorem proving and the other is based on discovering possible events and using this event knowledge with a planning algorithm. It is argued that it would be useful to associate alternative solutions, when they exist, with performance indicators to give an indication of preferable solutions. Model simplification may be achieved by combining a sequence of events as one abstract event which would eliminate intermediate events and the computational cost associated with them. To achieve this, consider logical theories of two states separated from each other by a number of events and characterizing an abstract event between these two theories as if they are immediately related via a state transition relation. It is noted that this proposal suppresses any probabilistic nature of the original models.<<ETX>>