Skyline-First Traversal Enables Multi-Criteria Graph Search Without Heuristics

A new study on arXiv (2604.19807) shows that you can handle scheduling and termination in multi-criteria graph searches using just Pareto geometry—no need for heuristics, scalarization, or population methods. The researchers employed constrained cost models and finite cost grids, along with Markovian transitions and a nonzero progress measure. By concentrating on the first Pareto layer, or skyline, they achieved a steady reduction in discrete completion potential. This ensures consistent progress. Moreover, a vector lower-bound certificate sets a stopping rule that guarantees coverage of all remaining paths. This work redefines Pareto dominance from a simple pruning tool to an active regulatory system.