School of Computing Science

PhD Theses

On the Mechanisation of the Logic of Partial Functions (2013)

The full text of this thesis is available from the Newcastle University Library website: https://theses.ncl.ac.uk/dspace/handle/10443/2229

Lovert, M.J., School of Computing Science, University of Newcastle upon Tyne

It is well known that partial functions arise frequently in formal reasoning about programs. A partial function may not yield a value for every member of its domain. Terms that apply partial functions thus may not denote, and coping with such terms is problematic in two-valued classical logic. A question is raised: how can reasoning about logical formulae that can contain references to terms that may fail to denote (partial terms) be conducted formally? Over the years a number of approaches to coping with partial terms have been documented. Some of these approaches attempt to stay within the realm of two-valued classical logic, while others are based on non-classical logics. However, as yet there is no consensus on which approach is the best one to use. A comparison of numerous approaches to coping with partial terms is presented based upon formal semantic definitions. One approach to coping with partial terms that has received attention over the years is the Logic of Partial Functions (LPF), which is the logic underlying the Vienna Development Method. LPF is a non-classical three-valued logic designed to cope with partial terms, where both terms and propositions may fail to denote. As opposed to using concrete un defined values, undefinedness is treated as a "gap", that is, the absence of a defined value. LPF is based upon Strong Kleene logic, where the interpretations of the logical operators are extended to cope with truth value "gaps". Over the years a large body of research and engineering has gone into the development of proof based tool support for two-valued classical logic. This has created a major obstacle that affects the adoption of LPF, since such proof support cannot be carried over directly to LPF. Presently, there is a lack of direct proof support for LPF. An aim of this work is to investigate the applicability of mechanised (automated) proof support for reasoning about logical formulae that can contain references to partial terms in LPF. The focus of the investigation is on the basic but fundamental two-valued classical logic proof procedure: resolution and the associated technique proof by contradiction. Advanced proof techniques are built on the foundation that is provided by these basic fundamental proof techniques. Looking at the impact of these basic fundamental proof techniques in LPF is thus the essential and obvious starting point for investigating proof support for LPF. The work highlights the issues that arise when applying these basic techniques in LPF, and investigates the extent of the modifications needed to carry them over to LPF. This work provides the essential foundation on which to facilitate research into the modification of advanced proof techniques for LPF.