Towards a Substrate Framework of Computation

Abstract

Abstract. A grand challenge in computing is to establish a substrate computational model that encompasses diverse forms of non-sequential computation. This paper demonstrates how a hypergraph rewriting framework nicely integrates various forms and ingredients of concurrent computation and how simple static analyses help the understanding and optimization of programs. Hypergraph rewriting treats processes and messages in a unified manner, and treats message sending and parameter passing as symmetric reaction between two entities. Specifically, we show how fine-grained strong reduction of the λ-calculus can be concisely encoded into hypergraph rewriting with a small set of primitive operations.

Engineering Grammar-Based Type Checking for Graph Rewriting Languages

“…For our base language LMNtal, the method of [22] deals with 'microscopic' properties by giving capability types, which represent both polarities and sharing of (hyper)links, to individual connection between nodes, whereas LMNtal ShapeType handles 'macroscopic' graph structures, i.e., shapes.…”