|
Graham WhiteI am a lecturer in the Department of Computer Science, Queen Mary, University of London. Here is my curriculum vitae.PublicationsHere are some of my publications, arranged under common themes; for a complete list of publications, see my curriculum vitae.Invariance"Lewis, Possible Worlds, and Non-Integrability", in Dialectica 54 (2000), pp. 133-137. looks at David Lewis's account of counterfactual conditionals, and shows that it suffers unexpected difficulties when confronted with a real-world example (namely, where possible worlds are the trajectories of a classical mechanical system). The basic problem is that, in the real world, you would want the constructs of your theory to be invariant under the symmetries of the system you are studying: for the majority of dynamical systems, Lewis' closeness relation cannot be thus invariant.A closely related problem is whether the predictions made by a theory are independent of the vocabulary used to describe that theory: in Intensionality and Circumscription, given at NMR2002, I argue that the Situation Calculus (a standard framework for reasoning about action) is unexpectedly vocabulary-dependent. In the slides, I continues the argument by contrasting the situation calculus with the McCain-Turner theory, which is thus invariant Though I don't emphasise this in the paper, the implementation in "A Linear Meta-Interpreter for Reasoning about Action and Change" is functorial enough to be invariant under change of vocabulary. Modal Formulations of Non-Monotonic LogicOne way of guaranteeing invariance is to formulate one's theory using operators which (in category-theoretic terms) are functorial, or which (in logical terms) are modal operators. I have a modal formulation of the McCain-Turner theory, which I talked about at JELIA 02; the paper is in LNAI 2424, and here are the slides.One can also use these techniques to handle more mainstream treatments of the frame problem, for example Reiter's: I gave a talk at the Workshop on Formal Pragmatics at Verona (September 2003), which describes this: I have a forthcoming article in Fundamenta Informatica on this. Mathematical PracticeMuch of the above argument is derived from things that practising mathematicians (and particularly applied mathematicians) do, and from the constraints (particularly symmetry constraints) that they impose on their formulations; it doesn't come from a foundational view.My chapter on "The Philosophy of Programming Languages" for the forthcoming Blackwell Guide to the Philosopy of Computing and Information, ed. L. Floridi continues this by describing the semantics of programming languages in these terms, and nudging the reader a bit in the direction of category theory. CausalityI have written the article on Medieval Theories of Causality for the (on-line) Stanford Encyclopaedia of Philosophy Though this work may seem rather remote from contemporary interests, the middle ages deserves attention: the work I describe is technically sophisticated, and attempts to handle common sense reasoning about causality.Linear LogicI have also done some work on reasoning about action using linear logic. My paper "Simulation, Theory, and Cut Elimination" (The Monist 82 (1999) 165-184) shows that a principled distinction can be drawn between simulating an action and reasoning about that action. (This answers a question that goes back to the philosophical debate about simulation.)I have also a series of papers which attempt to give a linear logical treatment of what the AI community calls ramification within a linear-logical account of action. Ramification is the problem of dealing, not just with the direct effects of an action, but also with the cascade of changes brought about by events triggered by the direct effects. Ramification can be accomodated by starting with the usual treatment of action in linear logic, and then applying some fairly standard category-theoretical constructions to deal with the ramification. "Simulation, Ramification, and Linear Logic", February 1999, in the Linkoping Electronic Articles in Computer and Information Science 3 (1998) no 11; ISSN 1401-9841. Available from http://www.ep.liu.se/ea/cis/1998/011/ "Actions, Ramification and Linear Modalities", August 1998, in the Linkoping Electronic Articles in Computer and Information Science 3 (1998) no 12; ISSN 1401-9841. Available from http://www.ep.liu.se/ea/cis/1998/012/. Previously, it appeared as QMW Department of Computer Science Technical Report 749, August 1998; available from ftp://ftp/dcs.qmul.ac.uk/pub/applied_logic/graham/ramMod.pdf "Ramification and Linear Logic", in the proceedings of the IJCAI 99 Workshop on Practical Reasoning and Rationality, available from ftp://ftp.dcs.qmul.ac.uk/pub/applied_logic/graham/prrWorkshop.pdf Finally, this paper shows that linear logic can provide a basis for extremely efficient implementations: "A Linear Meta-Interpreter for Reasoning about Action and Change", to appear in the Logic Journal of the IGPL. TeachingThe courses I currently teach are:
|