Sharp scaling for metric cotype in Banach spaces

The metric cotype inequality compares long coordinate jumps of a map from a discrete torus with its short diagonal jumps. The scale is the side length of the torus. For a Banach space with Rademacher cotype q, how small can this scale be as the dimension n grows?

Why it matters and what counts as progress

Why it matters: Smaller scales strengthen distortion bounds for finite grids. At the sharp scale, the inequality rules out certain coarse and uniform embeddings between spaces.

What counts as progress: A smaller exponent in the general upper bound is progress. The sharp conjecture asks for a scale of order n1/qn^{1/q} in every Banach space of Rademacher cotype q.

How the problem developed

  1. 2008 · Metric cotype and the sharp scale [4]

    Manor Mendel and Assaf Naor Related metric cotype to Rademacher cotype for Banach spaces and asked whether the general upper bound matches the scale forced by a two-point test.

  2. 2011 · Improved general scale [3]

    Ohad Giladi, Manor Mendel and Assaf Naor Gave a general scale of order at most n1+1/qn^{1+1/q} using smoothing and approximation.

  3. 8 Nov 2026 · Improved scale exponent [1][2]

    Manor Mendel with GPT-6 Astra Reported a continuous-torus kernel giving a scale of order at most n1/2+1/qn^{1/2+1/q}.

AI contribution

Sources

  1. Manor Mendel, Improved scaling in the metric cotype inequality
  2. Manor Mendel’s proof of the improved scaling bound
  3. Giladi, Mendel and Naor’s previous bound
  4. Mendel and Naor, Metric cotype
Publication history