Pseudo-Anosov mapping classes not arising from Penner's construction (Q906193)

The mapping class group \(\mathrm{Mod} (S_{g,n})\) is the group of isotopy classes of orientation preserving homeomorphisms of \(S_{g,n},\) the orientable surface of genus \(g\) and \(n\) punctures. An element \(f \in\mathrm{Mod} (S_{g,n})\) is pseudo-Anosov if there exists a representative homeomorphism \(\psi\), a number \(\lambda >1\), and a pair of transverse invariant singular measured foliations \(\mathcal{F}^u\) and \(\mathcal{F}^s\) such that \(\psi (\mathcal{F}^u) = \lambda \mathcal{F}^u\) and \(\psi (\mathcal{F}^s) = \lambda^{-1} \mathcal{F}^s.\) The number \(\lambda\) is called the stretch factor of \(f\). A stretch factor is said to be coronal if the constant \(\lambda\) has a Galois conjugate on the unit circle. The authors give a brief history with references of several constructions of pseudo-Anosov mapping classes using Dehn twists. This paper is focused on a construction due to Penner. Penner's construction: let \(A=\{a_1, \ldots, a_n\}\) and \(B=\{b_1, \ldots, b_m\}\) be a pair of multicurves on a surface. If \(A\) and \(B\) are in minimal position with the complement of their union equal to the union of open disks and single punctured ones, then any product of positive Dehn twists about curves in \(A\) with negatives Dehn twists about curves in \(B\) is pseudo-Anosov provided that all \(n+m\) twists appear in the product at least once. This construction leads to pseudo-Anosov mapping classes with small stretch factors. Penner conjectured that every pseudo-Anosov mapping class has a power that arises from this construction. The authors prove that this is not the case for mapping classes associated with coronal stretch factors. The main theorem reads as follows: a coronal pseudo-Anosov mapping class has no power coming from Penner's construction. Moreover, there exists a coronal pseudo-Anosov mapping class on \(S_{g,n}\) when \(3g+n\geq 5\). It is also observed that if the stretch factor of \(f\) has a Galois conjugate on the unit circle then no power of \(f\) can arise from Penner's construction. This leads to the open question: let \(f\) be a pseudo-Anosov mapping class whose stretch factor does not have Galois conjugates on the unit circle. Does some power of \(f\) arise from Penner's construction? Section 2 in the paper is devoted to the proof of the main theorem. Section 3 starts with examples of coronal pseudo-Anosov mapping classes and, based on computer experiments by other researchers, the authors claim that pseudo-Anosov classes obtained through this construction may be rare. In the last part of this section, the authors raise a question on whether the Galois conjugated of Penner stretch factors from a given pair of multicurves can approach the unit circle. They conjecture the answer to be negative. In the appendix, the authors give a proof that Penner's conjecture is true for \(S_{1,0}\), and \(S_{1,1}\) but false for \(S_{0,4}\). Then they consider a modification of the original conditions, by allowing half twists in addition to Dehn twists, and they show the conjecture to be also true for \(S_{0,4}\).