Towards a classification of connected components of the strata of \(k\)-differentials (Q2152283)

For \(g \geq 0, k \geq 1\) and an integer partition \(\mu = (k_1, \dots, k_n)\) of \(k(g-2)\), the stratum of \(k\)-differentials \(\Omega^k \mathcal M_g (\mu)\) parametrizing sections of the \(k\)th power of the canonical bundle on genus \(g\) Riemann surfaces with \(n\) distinct zeros or poles of orders given by the signature \(\mu\) is a complex orbifold [\textit{M. Bainbridge} et al., Algebr. Geom. 6, No. 2, 196--233 (2019; Zbl 1440.14148)]. It is natural to ask when the space \(\Omega^k \mathcal M_g (\mu)\) is connected. When \(k=1\), the connected components were determined for holomorphic differentials (the case \(k_i \geq 0\)) by \textit{M. Kontsevich} and \textit{A. Zorich} [Invent. Math. 153, No. 3, 631--678 (2003; Zbl 1087.32010)] and for meromorphic differentials by \textit{C. Boissy} [Comment. Math. Helv. 90, 255--286 (2015; Zbl 1323.30060)]. Extra components arise due to hyperelliptic and spin structures. When \(k=2\) and \(k_i \geq -1\) (at worst simple poles), \textit{E. Lanneau} determined the connected components [Ann. Sci. Éc. Norm. Supér. (4) 41, No. 1, 1--56 (2008; Zbl 1161.30033)], where there are sometimes hyperelliptic components, but no components arising from spin structures. Building on the strategies of the pioneering works above, the authors develop a framework and new techniques towards studying the general case. They use the theory of multi-scale \(k\)-differentials due to \textit{M. Costantini} et al. [``The area is a good enough metric'', Ann. Inst. Fourier (to appear)] to extend to all \(k \geq 1\) the operations of \textit{breaking up a zero} and \textit{bubbling a handle at a zero} (this increases genus by one) seen in the earlier work. They then generalize hyperelliptic components from \(k \leq 2\) to all \(k\) and determine precisely for which \(\mu\) the space \(\Omega^k \mathcal M_g (\mu)\) has a hyperelliptic component. Connected components of strata of lower order differentials give rise to connected components of higher order differentials by taking powers; these loci can be removed, leaving the remaining \textit{primitive locus} \(\Omega^k \mathcal M_g (\mu)^{\text{prim}}\). The authors generalize the notion of spin parity from \(k \leq 2\) to all \(k\) and prove that for \(g \geq 1\), the parity is constant on \(\Omega^k \mathcal M_g (\mu)^{\text{prim}}\) for \(k\) even, but when \(k\) is odd there are components of \(\Omega^k \mathcal M_g (\mu)^{\text{prim}}\) with distinct parities with the exception of \(\Omega^3 \mathcal M_g (6)^{\text{prim}}\), which is connected. For \(k=2\) (quadratic differentials) with some \(k_i \leq -2\) and \(g \geq 2\) the authors give a complete answer for primitive differentials: there are four types of special values of \(\mu\) for which \(\Omega^2 \mathcal M_g (\mu)^{\text{prim}}\) has two connected components (one hyperelliptic, the other not) and \(\Omega^2 \mathcal M_g (\mu)^{\text{prim}}\) is connected for all other \(\mu\). The paper concludes with an appendix in which the authors give an approach towards computing parity of \(k\)-differentials for curves of genus \(g \leq 1\), which leads them to a number-theoretic conjecture.