saved
A Faster Shortest Path Algorithm
Hraness cites a source capture. The source author remains the source.
gist
Geby Jaff describes a ten-agent Claude Opus 5.5 effort that produced C-HD, a formally checked exact shortest-path algorithm for directed graphs with non-negative real weights. In a restricted sparse-density regime, its proven bound improves on Dijkstra and two recent deterministic results by reducing repeated local-search and data-structure work. The result is asymptotic, not empirical: only small correctness simulations were run, constants are enormous, and Bellman–Ford handles other inputs.
ideas
- The gain is regime-specific. C-HD improves the leading bound along a sparse graph profile, with a certified range that does not cover every edge density.
- Local search controls repeated work. Bounded searches, unexplored-leaf accounting, pivots, and edge deletion keep recursive processing from multiplying unnecessarily.
- Formal verification makes the claim inspectable. Lean and the Comparator establish the runtime target, exactness, permitted axioms, and the stated asymptotic comparisons.
- The theorem is not a benchmark. The author ran only small correctness simulations; huge constants mean the proof does not establish a practical speedup.
- A message board made parallel research composable. Ten agents could reorganize roles, share discoveries, challenge claims, and preserve failed approaches while pursuing a reproducible proof.
quotes
“the team had completed a proposed new algorithm for finding exact shortest-path distances in the directed graph setting.”
“it reduces repeated search and data-structure work.”
“The constants in the formal construction are enormous, so this does not establish a practical speedup.”