The Implementation of Graph Programming Systems

Implementing a graph programming system is not a straightforward task. The challenge not only lies in converting a graphical graph-based program into low-level machine code, but the code must correctly perform the complex and non-deterministic operation of graph transformation through graph rewriting rules. GP is a high-level graph programming language. Users of the GP system can write graph programs using a graphical editor by constructing rule schemata and combining them with control constructs. This report addresses the possible ways to refine the prototype implementation of GP. The current implementation uses a stack-based abstract machine tailored for graph-based computation, which is ideal for handling nondeterminism and easily adaptable to changes in the language. We focus on two main refinements. The first concern is improving GP’s subgraph matching algorithm. We examine the literature of several graph programming languages and their strategies for inspiration. The second involves modifying the current implementation to be able to execute rooted graph programs, a type of graph program that can be executed very efficiently. We present current work concerning the theory behind rooted graph transformation, including a complexity analysis of rooted graph matching and some example rooted programs in GP.