← All problems

Furstenberg–Kesten conjecture on bigeodesics in two-dimensional first-passage percolation

11 years old · traced to

Let (te)(t_e) be i.i.d. edge weights on the nearest-neighbor edges of Z2\mathbb{Z}^2, with a continuous distribution. A bigeodesic is a doubly infinite path whose finite segments are geodesics for the induced passage-time metric. Furstenberg–Kesten's bigeodesic conjecture. Almost surely, there are no bigeodesics in dimension two. This is a central question in first-passage percolation; the paper proves the absence of bigeodesics with one end in any deterministic direction under differentiability of the limit-shape boundary, but the unrestricted conjecture remains open.

References
  1. Michael Damron and Jack Hanson, “Bigeodesics in first-passage percolation”, arXiv:1512.00804 (2016).

Progress summary

Refreshed
Open · progress reported

The conjecture remains open, although an unverified conditional manuscript claims to extend the known finite-moment result to all continuous edge-weight laws.

Furstenberg’s conjecture, recorded via Kesten, asserts that almost surely no bigeodesic exists for continuous i.i.d. edge weights on the two-dimensional square lattice. The unrestricted statement remains unresolved.

Known results

  • Licea–Newman (1996) ruled out bigeodesics with prescribed asymptotic directions outside a null set.
  • Wehr–Woo (1998) obtained related fixed-direction nonexistence results.
  • Damron–Hanson and Ahlberg–Hoffman proved nonexistence for each deterministic direction where the limit-shape boundary is differentiable, under finite-mean assumptions.
  • Later work constructed direction-specific geodesic graphs without doubly infinite paths, but explicitly did not settle simultaneous absence in all directions.

October 2026 conditional extension

A submitted manuscript argues that the stipulated finite-moment no-bigeodesics theorem extends to every nonatomic edge-weight law, removing the moment assumption and claiming absence of all bigeodesics. This is conditional on the upstream theorem and its proof inputs, and remains unverified; it is not independent validation of the unrestricted conjecture.

Current status (as of October 2026): the unrestricted two-dimensional conjecture remains open; fixed-direction results under limit-shape differentiability are established, while the new all-directions extension is an unverified conditional claim.

Sources

Solutions 1

Conditional planar no-bigeodesics extension to every nonatomic edge-weight law on [0,∞), removing the moment assumption. This relies on the stipulated OpenAI source theorem and its operative proof inputs; it is not an independent resolution.See full solutionHide full solution

Conditional solution: removing the moment assumption from a stipulated planar no-bigeodesics theorem.

The paper Removing the moment assumption from a stipulated planar no-bigeodesics theorem deduces that, for every nonatomic Borel probability law G on [0,∞), iid edge weights on the nearest-neighbor square lattice almost surely admit no bigeodesic, simultaneously over all bi-infinite simple paths, without a moment assumption or prescribed direction.

This conclusion is conditional on source S and its operative proof inputs under their stated hypotheses, as specified in the manuscript. The stipulated source already covers E[min(X₁,X₂,X₃,X₄)²] < ∞. The extension addresses its infinite-moment complement through the lifted limiting-graph coupling and transferred proof estimates. It relies on the source proof machinery, not merely a black-box finite-moment conclusion.

This matches the two-dimensional problem on this page. It makes no claim in higher dimensions and is not an independent validation of the upstream no-bigeodesics argument. In particular, the upstream family212 selected Lean statements concern limit-shape differentiability, not the absence of bigeodesics; the present package also does not formalize the full probability theorem.

The full manuscript and supporting package are available on Zenodo, DOI 10.5281/zenodo.23647122.

Source and review status. The upstream inputs are stipulated at OpenAI/math revision fd4aeeb2ee4fc729c18d98444fed42fd0529eeeb. This is an AI-assisted manuscript with internal project reviews, not external peer review. The supporting Lean checks cover selected lemmas, not the full theorem.