Bidirectional Search for Longest Paths: Case for Front-to-Front Heuristics
Researchers have proposed a new algorithm, BiXDFBnB, for solving longest-path problems and improving bidirectional search efficiency. The algorithm adapts the Single-Frontier Bidirectional Search framework to the Generalized Longest Simple Path setting.
BiXDFBnB is applied to several types of longest-path problems, including Longest Simple Path, Snakes, and Coil-in-the-Box, and has been shown to reduce the number of node expansions and improve overall runtime in some cases[1]. A new heuristic class, front-to-attractors (F2A), has also been introduced to modify the front-to-front heuristic in bidirectional search. F2A estimates the distance from a state to a small, dynamically maintained set of attractors in the opposite search direction, reducing computational cost while maintaining optimality guarantees[2]. According to the research, F2A reduces the number of pairwise evaluations by up to 11.2x compared to front-to-front (F2F) heuristics and achieves 4.8x fewer node expansions than front-to-end (F2E) heuristics on average[2]. The papers detailing these advancements were submitted to arXiv in June 2026, with one submission on 4 Jun 2026[2] and another likely around the same time[1].
research-papertool-release
Background sources we checked (2)
- arxiv.org ↗ Bidirectional heuristic search can potentially reduce search effort for problems amenable to backward search. Therein, it is well-known that front-to-front heuristics can reduce the number of node expansions, but their overhead is so high that overall runtime almost always increa…
- en.wikipedia.org ↗ An algorithm is a fundamental set of rules or defined procedures that are typically designed and used to be a simpler way to solve a specific problem or a broad set of problems. Simply speaking, algorithms define different processes, sets of rules and regulations, or methodologie…