Local-to-global Urysohn width estimates (Q2054210)

In metric geometry, simplicial complexes are special and important spaces. One important method to understand spaces in geometry and topology is to approximate general spaces by simplicial complexes. In connection with this, we need to define and estimate the degree of approximation. For any compact metric space \(X\) and every positive integer \(d\), there is the Urysohn \(d\)-width, which measures the optimal distortion from \(X\) to all metric simplicial complexes of dimension up to \(d\). The basic problem under study in this paper is to understand relation between the local Urysohn \(d\)-width of compact Riemannian manifolds \(X\) and their global Urysohn \(d\)-width, where by the local Urysohn \(d\)-width, we mean the Urysohn \(d\)-width of all the unit balls in \(X\). This fits into the general problem from local to global. After proving that local volume estimates give bounds on the global Urysohn width, Larry Guth raised the question: For a compact Riemannian manifold \(X\), if the local Urysohn \(d\)-width for a certain \(d\) is uniformly and sufficiently small, does it imply that the (global) Urysohn \(d'\)-width of \(X\) is bounded by some quantity which depends on the local one, where \(d'\geq d\)? The main result of this paper is to give a negative answer to this question. This paper also contains several other important results. For example, another result gives a bound on the global Urysohn 1-width in terms of the first homology and the local Urysohn 1-width. This result is completed by a result on Riemannian manifolds of dimension \(n\) whose first homology is small, for example, whose underlying topological space is the ball, the sphere or the torus, for which the global Urysohn \((n-1)\)-width is at least one while the local Urysohn \(d\)-width can be arbitrarily small, for some dimension \(d< n\).