On uniqueness in Steiner problem

TL;DR

This paper investigates the uniqueness of solutions in the planar Steiner problem, establishing bounds on the size of non-unique configuration sets and extending the analysis to arbitrary Riemannian manifolds.

Contribution

It proves Hausdorff dimension bounds for non-unique Steiner configurations and develops a general framework for analyzing uniqueness in arbitrary Riemannian manifolds.

Findings

Non-unique Steiner configurations form a set of Hausdorff dimension at most 2n-1.

Configurations with multiple minimal trees of equal length also have Hausdorff dimension at most 2n-1.

The set of configurations with a unique Steiner solution in Euclidean space is path-connected.

Abstract

We prove that the set of $n$-point configurations for which the solution of the planar Steiner problem is not unique has the Hausdorff dimension at most $2n-1$ (as a subset of $\mathbb{R}^{2n}$). Moreover, we show that the Hausdorff dimension of the set of $n$-point configurations on which at least two locally minimal trees have the same length is also at most $2n-1$. Methods we use essentially require rely upon the theory of subanalytic sets developed in~\cite{bierstone1988semianalytic}. Motivated by this approach we develop a general setup for the similar problem of uniqueness of the Steiner tree where the Euclidean plane is replace by an arbitrary analytic Riemannian manifold $M$. In this setup we argue that the set of configurations possessing two locally-minimal trees of the same length either has the dimension $n\dim M-1$ or has a non-empty interior. We provide an example of a two-dimensional surface for which the last alternative holds. In addition to abovementioned results, we study the set of set of $n$-point configurations for which there is a unique solution of the Steiner problem in $\mathbb{R}^d$. We show that this set is path-connected.