← All problems

Mendel–Naor optimal scaling problem for S₁

For each integer n≥1n\ge 1 and each even integer m≥4m\ge 4, let CS1(m,n)C_{S_1}(m,n) be the least constant such that every mapping f:Zmn→S1f:\mathbb{Z}_m^n\to S_1 satisfies

(∑j=1nEx∥f(x+m2ej)−f(x)∥12)1/2≤CS1(m,n)(Ex,ε∥f(x+ε)−f(x)∥12)1/2,\left(\sum_{j=1}^{n}\mathbb{E}_{x}\left\|f\left(x+\frac{m}{2}e_j\right)-f(x)\right\|_1^2\right)^{1/2}\le C_{S_1}(m,n)\left(\mathbb{E}_{x,\varepsilon}\left\|f(x+\varepsilon)-f(x)\right\|_1^2\right)^{1/2},

where xx is uniform on Zmn\mathbb{Z}_m^n and ε\varepsilon is independently uniform on {−1,0,1}n\{-1,0,1\}^n. Determine the optimal order of growth of CS1(m,n)C_{S_1}(m,n) as a function of mm and nn.

References

Source: arXiv

Progress summary

Refreshed
Open · solution reported

An unrefereed preprint claims to settle the optimal scaling question, but the result has not been independently verified.

The Mendel–Naor problem asks for the optimal scaling in the trace-class metric-cotype inequality. A preprint by Yong Jiao, Sijie Luo, and Lian Wu claims an answer with explicit dimension dependence.

October 2026 claimed settlement

The preprint claims the scaling 370m+6n3\sqrt{70m}+6\sqrt n and argues that the n\sqrt n term is necessary when the metric-cotype constants are fixed. This would settle the optimal-scaling question, but the source is an unrefereed preprint and the claim remains unverified.

Current status (as of October 2026): The optimal scaling is claimed to be 370m+6n3\sqrt{70m}+6\sqrt n, with n\sqrt n necessary, but this claimed resolution remains unverified.

Sources

Solutions 0

No solutions have been posted yet.