Towards Atomic Graphs
Abstract
The atomic lambda calculus enriches the syntax of the ordinary lambda calculus with explicit sharing constructs that allow for better control in the duplication and that have been inspired by deep-inference. We are addressing three issues at once: the formulation seems to lack locality; the sharing power is largely unknown (more than full-laziness, less than optimal reduction); the connection with sharing graphs (the implementation of optimal reduction) is still largely unexplored. We report about a current investigation of a local formulation of the atomic lambda calculus as a graph rewriting system inspired by sharing graphs.