Reasoning with Simplifying Assumptions: A Methodology and Example

Abstract

Simplifying assumptions are a powerful technique for dealing with complexity, which is used in all branches of science and engineering. This work develops a formal account of this technique in the context of heuristic search and automated reasoning. We also present a methodology for choosing appropriate simplifying assumptions in specific domains, and demonstrate the use of this methodology with an example of reasoning about typed partial functions in an automated programming assistant. soning, and to present a methodology for choosing appropriate simplifying assumptions in specific domains. The techniques discussed here are also closely related to techniques for reasoning with default assumptions, and non-monotonic reasoning generally. However, whereas most of the current work in this field (see [2,1]) focusses on the logical prop-erties of these types of reasoning, this work emphasizes methodological and pragmatic issues. In particular, other current work does not address the questions of how to choose default assump-tions and what specific control mechanisms are necessary to rea-son effectively with such assumptions.