Title:Linear-Time Graph Programs without Preconditions

Abstract:We report on a recent breakthrough in rule-based graph programming, which allows us to reach the time complexity of imperative linear-time algorithms. In general, achieving the complexity of graph algorithms in conventional languages using graph transformation rules is challenging due to the cost of graph matching. Previous work demonstrated that with rooted rules, certain algorithms can be executed in linear time using the graph programming language GP 2. However, for non-destructive algorithms that retain the structure of input graphs, achieving linear runtime required input graphs to be connected and of bounded node degree. In this paper, we overcome these preconditions by enhancing the graph data structure generated by the GP 2 compiler and exploiting the new structure in programs. We present three case studies, a cycle detection program, a program for numbering the connected components of a graph, and a breadth-first search program. Each of these programs runs in linear time on both connected and disconnected input graphs with arbitrary node degrees. We give empirical evidence for the linear time complexity by using timings for various classes of input graphs.