A Colored Version of Tverberg's Theorem (Q4024898)

In what follows, \(n\), \(N\), \(r\), \(d\) signify positive integers. All points are in \(\mathbb{R}^ d\). Tverberg's theorem is: Any set of \(r(d+1)-d\) points can be divided into \(r\) disjoint sets, whose convex hulls have a non-empty intersection. Suppose a set \(P_ n\) of \(n\) points can be divided into \(d+1\) disjoint subsets \(C_ 1,\dots,C_{d+1}\) called colors, each having at least \(r\) points, then \(P_ n\) is called \(r\)-properly colored. For a point set \(X\), write conv \(X\) for its convex hull. If \(p_ 1,\dots,p_{d+1}\) are distinct points from \(P_ n\), \(r\)-properly colored, then that set and the simplex \(\text{conv}\{p_ 1,\dots,p_{d+1}\}\) are called multicolored precisely when \(p_ i \in C_ i\), \(i=1,\dots,d+1\), possibly after a relabeling. Here is the colored Tverberg problem: find the least \(N(r,d)\) such that, if \(n \geq N(r,d)\), then any properly \(r\)- colored \(P_ n\) has \(r\) disjoint, multicolored subsets, each with \(d+1\) members, whose simplicial convex hulls have a non-empty intersection. The authors conjecture that \(N(r,d) = r(d+1)\); they prove this for \(d = 1,2\) and remark that \(N(2,d) = 2(d+1)\) is known, as well as provide a proof due to L. Lovász (1989). The main result of the paper can be put this way: given sets of red, white and green points, each of cardinality \(r\), it is possible to form \(r\) triangles such that no two triangles have a common vertex, each triangle has a vertex of each color and the \(r\) triangles have a common point.