Since, in contrast to other methods, interval computations give results with mathematical certainty, even in the presence of uncertainty in the data, roundoff error, and nonlinearities, they are of unique potential importance throughout science, engineering, and operations research. The work in this project will advance realization of this potential in the following ways. o A modern, portable, standardized library of elementary functions will be developed, compatible with the operator overloading feature of Fortran 90, to be used in conjunction with interval arithmetic computations. o A parsing technique for nonlinear systems of algebraic equations will be implemented to better take advantage of the structure of such systems, and thus avoid much of the overestimation and resulting inefficiency in interval nonlinear system solvers. o The methods will be applied to various practical problems, including robust geometric computation, robot kinematics, chemical kinetics equilibrium, and chemical refinery models.
