Curve-flat functions and Lipschitz quotients

TL;DR

This paper constructs specific complete and compact metric spaces with prescribed curve-flat quotients, revealing limitations on universality of Lipschitz quotients among compact metric spaces.

Contribution

It introduces a new method for constructing metric spaces with controlled curve-flat quotients and demonstrates non-existence of a universal compact metric space for all compact metric spaces as Lipschitz quotients.

Findings

Existence of complete metric spaces with prescribed curve-flat quotients.

Construction of compact spaces with high-order curve-flat quotients bi-Lipschitz equivalent to given spaces.

Non-existence of a universal compact metric space for all compact metric spaces as Lipschitz quotients.

Abstract

We show that for every complete metric space $M$ there exists another complete metric space $N$ of the same density character such that the curve-flat quotient of $N$ is isometric to $M$. Moreover, we show that if $M$ is compact and $\alpha$ is any countable ordinal, there exists a compact $N$ such that its curve-flat quotient of order $\alpha$ is bi-Lipschitz equivalent to $M$, with arbitrarily small distortion. Our constructions rely on a new method for constructing (compact) metric spaces, which consists in attaching iteratively compact spaces at countably many pairs of points to a snowflake-like distortion of a given (compact) metric space. We apply our results on high-order curve-flat quotients to obtain a new result concerning universality of Lipschitz quotients. Specifically, we show that there cannot exist a compact metric space $K$ such that every compact metric space is a Lipschitz quotient of $K$. This result stands in contrast to a theorem of Johnson, Lindenstrauss, Preiss and Schechtman, who showed that any separable Banach space containing $\ell_1$ has every separable geodesic complete metric space as a Lipschitz quotient.