Some Ramsey-type results on intrinsic linking of \(n\)-complexes (Q1954154)

The complete \(n\)-complex on \(N\) vertices, \(K_N^n\), is the \(n\)-skeleton of an \((N-1)\)-simplex. In this paper, the author establishes several Ramsey-type results for embeddings of complete \(n\)-complexes in \(\mathbb{R}^{2n+1}\), extending known results of the 1-dimensional case \(n=1\), i.e., on intrinsic linking of embeddings of complete graphs in \(\mathbb{R}^3\), such as shown by \textit{E. Flapan} [Algebr. Geom. Topol. 2, 371-381 (2002; Zbl 0998.57010), Fundam. Math. 201, No. 2, 131--148 (2008; Zbl 1170.57001), J. Knot Theory Ramifications 10, No. 8, 1143--1154 (2001; Zbl 0998.57008)], \textit{T. Fleming} [J. Knot Theory Ramifications 21, No. 7, 1250065, 10 p. (2012; Zbl 1239.57005)], and \textit{T. Fleming} and \textit{A. Diesl} [Algebr. Geom. Topol. 5, 1419--1432 (2005; Zbl 1099.57006)]. The main results are as follows, where a \(k\)-component link is \(k\) disjoint \(n\)-spheres embedded in \(\mathbb{R}^{2n+1}\). For \(N\) sufficiently large, every embedding of a complete \(n\)-complex \(K_N^n\) in \(\mathbb{R}^{2n+1}\) contains a nonsplit \(r\)-component link, where the lower bound of \(N\), the number of vertices required, grows at most linearly with respect to each of \(r\) and \(n\). For \(N\) determined from \(r\) and \(n\), every embedding of \(K_{N}^n\) in \(\mathbb{R}^{2n+1}\) contains an \(r\)-component link called a generalized keyring, which is a link such that the linking number of one fixed component and every other component is one modulo two, where \(N\) grows quadratically in \(r\) and linearly in \(n\). The other results concern linking numbers of 2-component links. For \(N\) determined from a given integer \(\lambda\) and \(n\), every embedding of \(K_N^n\) in \(\mathbb{R}^{2n+1}\) contains a 2-component link whose linking number is at least \(\lambda\) in absolute value, where \(N\) grows quadratically in \(\lambda\) and linearly in \(n\). For a sufficiently large \(N\), every embedding of \(K_N^n\) in \(\mathbb{R}^{2n+1}\) contains a 2-component link whose linking number is a nonzero multiple of a given positive integer \(q\), where the lower bound of \(N\) has growth \(O(q^{n+2})\); further, when \(q\) is prime, then this conclusion holds true for \(N\) with a lower bound with growth \(O(q^2)\); note that this result gives a better bound for \(n=1\) than that previously known. The proofs are based on the techniques from the corresponding 1-dimensional results, with modifications for higher dimensional case.