Exploring Continuous Tensegrities

TL;DR

This paper extends the concept of tensegrity frameworks to continuous, possibly infinite, vertex sets in Euclidean space, establishing conditions for bar equivalence and analyzing motions of curves like circles.

Contribution

It introduces continuous tensegrities with infinite vertices, proves conditions for bar equivalence based on stress positivity, and examines specific cases like curves and circle motions.

Findings

Positive stress implies bar equivalence.

Semipositive stress corresponds to partial bar equivalence.

No local arclength preserving motion can increase antipodal distances on a circle.

Abstract

A discrete tensegrity framework can be thought of as a graph in Euclidean n-space where each edge is of one of three types: an edge with a fixed length (bar) or an edge with an upper (cable) or lower (strut) bound on its length. Roth and Whiteley, in their 1981 paper "Tensegrity Frameworks", showed that in certain cases, the struts and cables can be replaced with bars when analyzing the framework for infinitesimal rigidity. In that case we call the tensegrity "bar equivalent". In specific, they showed that if there exists a set of positive weights, called a positive "stress", on the edges such that the weighted sum of the edge vectors is zero at every vertex, then the tensegrity is bar equivalent. In this paper we consider an extended version of the tensegrity framework in which the vertex set is a (possibly infinite) set of points in Euclidean n-space and the edgeset is a compact set of unordered pairs of vertices. These are called "continuous tensegrities". We show that if a continuous tensegrity has a strictly positive stress, it is bar equivalent and that it has a semipositive stress if and only if it is partially bar equivalent. We also show that if a tensegrity is minimally bar equivalent (it is bar equivalent but removing any open set of edges makes it no longer so), then it has a strictly positive stress. In particular, we examine the case where the vertices form a rectifiable curve and the possible motions of the curve are limited to local isometries of it. Our methods provide an attractive proof of the following result: There is no locally arclength preserving motion of a circle that increases any antipodal distance without decreasing some other one.