Pure, Declarative, and Constructive Arithmetic Relations
Pure, Declarative, and Constructive Arithmetic Relations. Oleg Kiselyov, William E. Byrd, Daniel P. Friedman, and Chung-chieh Shan. FLOPS 2008. (source code)
We present decidable logic programs for addition, multiplication, division with remainder, exponentiation, and logarithm with remainder over the unbounded domain of natural numbers. Our predicates represent relations without mode restrictions or annotations. They are fully decidable under the common, DFS-like, SLD resolution strategy
of Prolog or under an interleaving refinement of DFS…[The] attempts to define decidable multiplication even for the seemingly trivial unary case show the difficulties that become more pronounced as we move to binary arithmetic. We rely on a finite representation of infinite domains, precise instantiatedness analysis, and reasoning about SLD using search trees.
So you’ve read The Reasoned Schemer and were excited about the fact that unlike the built-in operations in Prolog, arithmetic relations (over binary numbers) were fully implemented. For example, addition could also be used for subtraction and multiplication for factoring numbers and for generating all triples of numbers related by multiplication. Now comes this paper to explain the motivation behind some of the more arcane definitions needed to implement arithmetic in a fully relational style, and to prove their properties formally. The paper develops unary and binary arithmetic relations in pure Prolog (with no cuts, negation or introspection).
LtU readers will also be interested in yet another embedding of pure Prolog into Haskell , that the authors offer. It is not meant to be the most optimal or convenient Prolog implementation (it wasn’t even intended to be an implementation of a logic
Also see: DevWeek 2008 Cross Platform Silverlight Demos
Also see: Silverlight 2 Beta 1 Cross Domain Bug
Also see: Stand alone FXCop download
Also see: Natural Sorting in C#
Also see: Determining Whether a File Is an Assembly
Also see: What Are You Destined to Be ?
Also see: Reliability
Also see: Stand alone FXCop download
Also see: Memory Model
Also see: Stand alone FXCop download
Also see: On the Perils of Wikipedia
Also see: Bragging on HDNet
Also see: Important changes to the BASE element for IE 7
Also see: DevWeek 2008 Cross Platform Silverlight Demos
Also see: Silverlight 2 Developer Poster
Also see: When Are Two Algorithms the Same?
Also see: Memory Model
Also see: Why I hate Radio
Also see: Brad Abrams’ pixel8 Interview Podcast posted
Also see: A quick update on me.
Also see: What Are You Destined to Be ?
system). It was explicitly designed to be easier to reason about and so help prove certain properties of SLD or similar evaluation strategies. The main difference of DefinitionTree from other embeddings of Prolog in Haskell has to do with the generation of fresh names for logic variables. In DefinitionTree, name generation is not an effect, and the naming is fully decoupled from the evaluation. The evaluation no longer needs to carry a state for the generation of fresh names, hence the evaluator is easier to reason
about equationally.
http://lambda-the-ultimate.org/node/2697
