Euclidean distortion conjecture for finite subsets of LpL_p

From papers

For n\finityn\to\finity and p>2p>2, let c2n(Lp)\mathsf{c}_2^n(L_p) denote the maximum, over all nn-point metric spaces, of the least distortion with which the space embeds into LpL_p and then into Euclidean space. The known bounds give c2n(Lp)=o(logn)\mathsf{c}_2^n(L_p)=o(\log n) when p=o(logn6/loglogn3)p=o\big(\sqrt[6]{\log n}/\sqrt[3]{\log\log n}\big), while c2n(Lp)logn\mathsf{c}_2^n(L_p)\asymp\log n when plognp\gtrsim\log n. Euclidean distortion conjecture. One should have

c2n(Lp)=o(logn)\mathsf{c}_2^n(L_p)=o(\log n)

in the remaining range

logn6/loglogn3p=o(logn).\sqrt[6]{\log n}/\sqrt[3]{\log\log n}\lesssim p=o(\log n).

This would determine the qualitative Euclidean distortion growth throughout the currently unresolved intermediate range; the endpoint regimes are supplied by the theorem and by the embedding of n\ell_\infty^n into LpL_p.

Progress summary

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Sources & referencesView supporting material

Primary source

Assaf Naor and Kevin Ren, “Euclidean embedding, randomized clustering, and Lipschitz extension for finite and doubling subsets of L_p when p>2”, arXiv:2502.10543 (2026).

Solutions 0

No solutions have been posted yet.