Computer-assisted hybrid reasoning in simulation and analysis of physical systems

Abstract

The aim of the paper is to advocate the use of hybrid reasoning systems for computer-assisted analysis of physical systems. The paper starts from a critical assessment of classic numerical techniques, with the problem of sensitivity analysis of fuel rod support spring in a nuclear reactor used as an example. Then, the significance and some basic issues concerning qualitative physics methods of analysis of physical systems are discussed. Using the example of the so-called "snap-through" mechanism, the basic principles, advantages and limitations of qualitative simulation technique are shown. Certain future development possibilities are indicated, especially the necessity to formalise the order-oj-magnitude reasoning. The recently developing techniques of diagrammatic reasoning are also introduced, with another mechanical example illustrating sources of their advantages for certain kinds of problems. The significant role of logical (expert-system-like) reasoning techniques and constraint-satisfaction systems is shown as well. Finally, the hybrid reasoning system concept is sketched. Such hybrid systems should integrate quantitative (numerical) analysis, various methods of qualitative analysis as well .as diagrammatic and logical reasoning techniques.