The Magic of Pushing Extrema into Recursion: Simple, Powerful Datalog Programs

We introduce a novel query optimization method based on pushing extrema and other integrity constraints into recursion. This optimization produces an efficient operational semantics for the goals declaratively specified by the original program that is stratified with respect to negation and aggregates. Complex algorithms can now be specified via simpler programs, and then implemented via optimized rules that use min and max aggregates in the seminaive fixpoint computation to achieve safe and efficient execution. This optimization dovetails with recent advances that entail the use of monotonic aggregates in recursion and leads to efficient implementation, as demonstrated by the scalable Datalog systems we have recently developed.