Total curvature and packing of knots (Q857049)

Let \(K\) be a smooth knot embedded in \(\mathbb{R}^3\). For \(r>0\), consider the disks of radius \(r\) normal to \(K\), centered at points of \(K\). For \(r\) sufficiently small, these disks are pairwise disjoint and combine to form a tubular neighborhood of \(K\). Let \(R(K)\), the \textit{thickness radius} of \(K\), denote the supremum of such radii. The \textit{ropelength} of \(K\), denoted \(E_L(K)\), is the ratio of the total arclength of \(K\) over \(R(K)\). The \textit{average crossing number} of \(K\), denoted \(acn(K)\), is the average of the number of crossings taken over all projection directions that lead to a regular projection of \(K\). Let \(cr(K)\) be the minimum crossing number of \(K\) and \(\tau(K)\) denote the total curvature of \(K\). The main theorem of this article is the following: For a smooth knot \(K\) in \(\mathbb{R}^3\), \[ cr(K)\leq acn(K)< 4 E_L(K)\tau(K). \] The same authors showed in [Topology Appl. 91, No. 3, 245--257 (1999; Zbl 0924.57012)], that \(acn(K)\leq E_L(K)^{4/3}\) using similar techniques. The new arguments here are about the curvature of curves that are packed in certain ways. The bounds on the \(acn(K)\) imply that knot families whose crossing number grows as the \(4/3\)-power of their ropelength must have unbounded total curvature. In the final section of the article the authors conjecture that in fact \(acn(K)\leq 4 E_L(K)\tau(K)^{1/2}\) and give a heuristic argument to support this.