Using failure to guide inductive proof
Andrew Ireland, Alan Bundy · Edinburgh Research Explorer (University of Edinburgh) · 1992
Lemma discovery and generalization are two of the major hurdles in automating inductive proof. This paper addresses aspects of these related problems. We build upon rippling, a heuristic which plays a pivotal role in guiding inductive proof. Rippling provides a high-level explanation of how to control the search for a proof. We demonstrate how this high-level explanation can be exploited productively when a proof attempt fails. In particular we show how failure can be used to focus the search for lemmas and generalizations.