Front-to-Attractors: Modifying the Front-to-Front Heuristic in Bidirectional Search

In three linesNew bidirectional search heuristic: front-to-attractors (F2A) reduces pairwise evaluations by 11.2x vs front-to-front (F2F) while maintaining optimality. F2A uses a dynamically maintained set of attractors instead of evaluating all opposite frontier states, achieving 4.8x fewer node expansions than front-to-end.

Summary generated by Claude — human-verified