Higher convexity for complements of tropical varieties (Q289818)

\textit{M. Gromov} [Rend. Semin. Mat. Fis. Milano 61, 9--123 (1991; Zbl 0820.53035)] generalized the notion of convexity: a subset \(X \subseteq \mathbb{R}^n\) is \(k\)-convex if for every \((k+1)\)-plane \(L\) the natural map on \(k\)th reduced homology \(i_k : \widetilde{H}_k (X \cap L) \longrightarrow \widetilde{H}_k (X)\) is an injection. In particular, an ordinary convex subset is just a connected \(0\)-convex subset. \textit{A. Henriques} [Adv. Geom. 4, No. 1, 61--73 (2004; Zbl 1053.14065)] conjectured that if \(V \subset (\mathbb{C}^{\times})^n\) has pure codimension \((k+1)\), then the complement of its amoeba is \(k\)-convex, but he only proved weak k-convexity: the map \(i_k\) sends no positive class to zero. \textit{N. A. Bushueva} and \textit{A. K. Tsikh} [Proc. Steklov Inst. Math. 279, 52--63 (2012; Zbl 1312.14135); translation from Tr. Mat. Inst. Steklova 279, 59--71 (2012)] proved Henriques' conjecture when the variety \(V\) is a complete intersection. Since tropical varieties are limits of amoebas, the authors naturally conjecture that the complement of a tropical variety of pure codimension \((k+1)\) is \(k\)-convex. They prove the complement of a tropical curve in \(\mathbb{R}^n\) is \((n-2)\)-convex. Their proof is via direct argument and uses a basic property of tropical curves which they called weak balancing. When \(\mathbb{K}\) is the field of complex Puiseux series, Nisse and Sottile prove that the complement of the nonarchimedian amoeba of a variety \(V \subset (\mathbb{K}^{\times})^n\) of codimension \(k+1\) is weakly \(k\)-convex. Moreover, if \(V\) is a complete intersection, then the complement is \(k\)-convex. Their proof relies on \textit{M. Jonsson}'s limit theorem [Math. Ann. 364, No. 1--2, 293--311 (2016; Zbl 1375.14214)] that such a nonarchimedian amoeba is a limit of archimedian amoebas together with a result of Bushueva and Tsikh in [loc. cit.].