Qualitative-numeric simulation with Q3

13 task by allowing intervals to be given as values of the input parameters, and generating the plausible behaviors. A similar application in engineering relates to necessary tolerances of various model parameters. Simply simulating many times with dierent combinations of plausible model parameter values, as in Monte Carlo simulation, is likely to be tedious, perhaps impractical, and could also miss possible and potentially serious (mis)behaviors. This sort of tolerance analysis could also be done automatically. Given a model with precisely specied parameter values, those values could be "blurred" into intervals by a preprocessor, then Q3 could simulate to nd the plausible behaviors. Thus we expect an exciting potential for practical application of qualitative-numeric simulation. Acknowledgements. We would like to acknowledge useful comments on the work reported here by