← All problems

Fence conjecture of Dumas and Wolf

Let P⊂RP2P\subset\mathbb{RP}^2 be a properly convex projective polygon with vertices v1,…,vnv_1,\ldots,v_n, where n≥5n\ge 5, indexed cyclically. For each ii, let Δi=conv⁡(vi−1,vi,vi+1)\Delta_i=\operatorname{conv}(v_{i-1},v_i,v_{i+1}) be the triangle at the consecutive triple of vertices, and let F(P)=P∖⋃i=1nint⁡(Δi)‾F(P)=\overline{P\setminus\bigcup_{i=1}^n\operatorname{int}(\Delta_i)} be the region obtained by cutting off these triangles. Then every zero of the Pick differential qPq_P of PP lies in int⁡F(P)\operatorname{int}F(P). Equivalently, the n−3n-3 distinguished zeros of PP, counted with multiplicity, all lie in the interior of F(P)F(P).

References

Progress summary

Refreshed
Open · solution reported

An October 2026 preprint claims to prove the conjecture and extend it beyond polygons, but the result has not been independently checked.

Dumas and Wolf proposed the Fence conjecture for convex nn-gons in the projective plane in 2014. It concerns the locations of the n−3n-3 distinguished zeros associated with the polygon.

October 2026 claimed proof

Erxiao Wang’s preprint claims a complete proof for convex projective polygons and a stronger extension to arbitrary properly convex domains, using sharp disk estimates. The claim is unrefereed and remains unverified.

Current status (as of October 2026): The Fence conjecture is claimed solved by Wang’s preprint, but that proof has not been independently verified; the strengthened general-domain statement is likewise unverified.

Sources

Solutions 0

No solutions have been posted yet.