Simple closed curves, finite covers of surfaces, and power subgroups of \(\mathrm{Out}(F_n)\) (Q2336101)

In this paper, the authors prove that for any punctured surface \(\Sigma\), there is a finite regular cover \(\tilde{\Sigma}\to \Sigma\), such that \(H_1(\tilde{\Sigma};\mathbb{Q})\) is not spanned by lifts of simple closed curves in \(\Sigma\) (Theorem A). This result promotes a result of \textit{T. Koberda} and \textit{R. Santharoubane} [Invent. Math. 206, No. 2, 269--292 (2016; Zbl 1400.57020)] from \(\mathbb{Z}\)-coefficients to \(\mathbb{Q}\)-coefficients. On the other hand, Koberda-Santharoubane's result works for closed surfaces but the result in this paper only works for punctured surfaces. To prove Theorem A, the authors prove a more general theorem in free groups (Theorem C). Some notations: let \(\mathcal{O}\subset F_n\) be a subset of the free group on \(n\)-generators, let \(R\lhd F_n\) be a finite index subgroup and let \(A\) be an abelian group, then define \(H_1^{\mathcal{O}}(R;A)\) to be the span of the following set in \(H_1(R;A)\): \[ \{[x^k]\in H_1(R;A)\ |\ x\in \mathcal{O},\ k\geq 1\text{\ such\ that\ }x^k\in R\}. \] Then Theorem C is stated as: let \(\mathcal{O}\subset F_n\) be a subset contained in the union of finitely many \(\text{Aut}(F_n)\)-orbits, then there is a finite-index normal subgroup \(R\lhd F_n\) such that \(H_1^{\mathcal{O}}(R;\mathbb{Q})\ne H_1(R;\mathbb{Q})\). The technical heart of this paper is the following Theorem D. For a prime number \(p\), let \(\mathfrak{D}_p\subset F_n\) be the subset consisting of elements whose image in \(H_1(F_n;\mathbb{F}_p)\) is nontrivial, then there is a finite-index normal subgroup \(R\lhd F_n\), such that \(H_1^{\mathfrak{D}_p}(R;\mathbb{Q})\ne H_1(R;\mathbb{Q})\). To prove Theorem D, the authors first prove a criterion for certifying that \(H_1^{\mathfrak{D}_p}(R;\mathbb{Q})\ne H_1(R;\mathbb{Q})\), in terms of the existence of certain representations of the finite group \(G=F_n/R\) (Theorem 2.1). The existence of the desired representation relies on the existence of certain finite \(p\)-groups (Proposition 2.3), while the desired finite \(p\)-group is constructed by the theory of \(p\)-restricted Lie algebras. Besides Theorem C, Theorem D also implies the existence of certain infinite quotients of \(\text{Out}(F_n)\) and \(F_n\) that map certain given elements (powers of transvections and elements in \(\mathfrak{D}_p\), respectively) to finite-order elements, which are Theorems E, F and G in the paper.