Pach–Tardos conjecture

For every acyclic matrix pattern PP, there exists a constant CPC_P such that the extremal function satisfies Ex⁡(P,n)=O ⁣(n(log⁡n)CP)\operatorname{Ex}(P,n)=O\!\left(n(\log n)^{C_P}\right) as n→∞n\to\infty.

References

Primary source

arXiv

Additional references

Progress summary

Refreshed
Claimed solved

A new unrefereed preprint claims to prove the conjecture, overturning a 2024 claimed refutation, but the result is not independently verified.

Pach and Tardos conjectured in 2005 that every acyclic matrix pattern PP has near-linear extremal function, specifically Ex⁡(P,n)=O(nlog⁡CPn)\operatorname{Ex}(P,n)=O(n\log^{C_P}n).

Known results

  • The conjecture had been established for all patterns of weight at most 55 and all but two patterns of weight 66.
  • Pettie and Tardos, 2024, claimed sharp bounds Ex⁡(Pt,n)=Θ ⁣(n(log⁡nlog⁡log⁡n)t)\operatorname{Ex}(P_t,n)=\Theta\!\left(n\left(\frac{\log n}{\log\log n}\right)^t\right) for alternating patterns.
  • Their 2024 preprint claimed counterexamples S0,S1S_0,S_1 with Ex⁡(Si,n)≥n2Ω(log⁡n)\operatorname{Ex}(S_i,n)\ge n2^{\Omega(\sqrt{\log n})}, refuting the conjecture.

September 17, 2026 claimed proof

On September 17, 2026, Lior Gishboliner and Xiangyu Li submitted Proof of the Pach-Tardos conjecture, claiming for every acyclic PP that Ex⁡(n,P)≤n1+OP(1/log⁡log⁡n)\operatorname{Ex}(n,P)\le n^{1+O_P(1/\log\log n)}. This directly conflicts with the 2024 refutation claim and is supported only by an unrefereed preprint.

Current status (as of September 2026): A new preprint claims the conjecture is proved, while the earlier counterexample claim and the new proof have not been independently verified, so the exact mathematical status remains unresolved.

Sources

Solutions 0

No solutions have been posted yet.