Using equivariant obstruction theory in combinatorial geometry (Q2643064)

The paper under review begins with a terse but quite readable and very useful general introduction to equivariant obstruction theory. This includes an equivariant version of Poincaré duality. The main result (Theorem 1.8) goes as follows: Let \(M\) be a compact \((n+1)\)-dimensional manifold with a free \(G\)-action. Further let \(W\) be a \(d\)-dimensional submanifold, and let \(\Sigma\) be the union of a finite \(G\)-invariant arrangement of \((d-n-1)\)-dimensional smooth submanifolds. If the pair \((W,\Sigma)\) meets certain requirements of connectedness and general position then there exists a \(G\)-map from \(M\) to \(W\setminus\Sigma\) if and only if a certain obstruction element in \(H_n(W\setminus\Sigma,\mathbb Z)_G\) vanishes. This theorem is strong enough to derive short proofs of all of the following: Lovász' conjecture on lower bounds of certain graph coloring problems, initially solved by \textit{E. Babson} and \textit{D. N. Kozlov} [Ann. Math. (2) 165, No. 3, 965--1007 (2007; Zbl 1132.05019)], a result on the partition of sphere measures by hyperplanes due to \textit{V. V. Makeev} [J. Math. Sci., New York 104, No. 4, 1358--1363 (2001); translation from Zap. Nauchn. Semin. POMI 252, 165--174 (1998; Zbl 0989.52002)], and a further result on the measure partition by fans.