Symbolic Shape Analysis Diploma Thesis

Shape analysis deals with the synthesis of invariants for programs manipulating heap-allocated data structures. Explicit shape analysis algorithms do not scale very well. This work proposes a framework for symbolic shape analysis that addresses this problem. Our contribution is a framework that allows to abstract programs with heap-allocated data symbolically by Boolean programs. For this purpose, we combine abstraction techniques from shape analysis with ideas from predicate abstraction. Our framework is parameterized by a set of abstraction predicates. We propose a class of predicates that can be used to analyze reachability properties for linked data structures. This class may potentially be used for automated abstraction refinement.