On the nonexistence of certain branched covers (Q441117)

The main result in the paper under review is the following: every branched covering \(T^n \to N\) from the \(n\)-torus \(T^n\) to a closed oriented \(n\)-manifold \(N\) is an ordinary covering, provided that \(H^r(N; \mathbb R) \cong H^r(T^n; \mathbb R)\) for some \(1 \leq r < n\). In particular, for \(n = 4\) and \(N \cong \#_3(S^2 \times S^2)\), the connected sum of three copies of \(S^2 \times S^2\), the authors can conclude that there is no branched cover \(T^4 \to \#_3(S^2 \times S^2)\), while it is known that there are PL maps \(T^4 \to \#_3(S^2 \times S^2)\) of arbitrarily large degree [\textit{H. Duan} and \textit{S. Wang}, Acta Math. Sin., Engl. Ser. 20, No. 1, 1--14 (2004; Zbl 1060.57018)]. This shows that the picture changes completely when passing from dimension 3 to 4. In fact, every \(\pi_1\)-surjective map \(M \to N\) between closed, orientable 3-manifolds of degree \(> 3\) is homotopic to a branched cover [\textit{A. L. Edmonds}, Math. Ann. 245, 273--279 (1979; Zbl 0406.57003)]. [Editorial remark: This article was retracted by the authors due to a gap in the proof of Proposition~3.1. For the retraction note (received 14 February 2019), see \url{https://msp.org/gt/2012/16-3/gt-v16-n3-x02-Branched_covers_and_cohomology_retraction.pdf}.]