I was informed of this paper on a trip to the University of Sussex just before Christmas. Unfortunately, it’s taken me a while to get around to reading it, and therefore posting it to the blog. The paper’s of interest not only to those who are interested in name binding, where interaction nets appear to provide a graphical representation of lambda-terms (and much more), but also those interested in implementing rewrite systems, wherein interaction nets facilitate a very efficient implementation.
By Ian Mackie, from RTA 2004, available from SpringerLink:
This paper presents an efficient implementation of the lambda-calculus using the graph rewriting formalism of interaction nets. Building upon a series of previous works, we obtain one of the most efficient implementations of this kind to date: out performing existing interaction net implementations, as well as other approaches. We conclude the paper with extensive testing to demonstrate the capabilities of this evaluator.