2-symmetric transformations for 3-manifolds of genus 2 (Q1850472)

The \textit{Heegaard genus} \(g(M)\) of a closed \(3\)-manifold \(M\) is the minimal integer~\(g\geq 0\) such that \(M\)~admits a decomposition into two handlebodies bounding a surface of genus~\(g\) (a \textit{Heegaard splitting}). It is well known that the only \(3\)-manifold of Heegaard genus~\(0\) is the \(3\)-sphere, and that the \(3\)-manifolds of genus~\(1\) are \(S^1\times S^2\), \(S^1\tilde\times S^2\), and the lens spaces. In contrast, the classification problem for \(3\)-manifolds of genus \(g(M)\geq 2\) is still unsolved. However, partial progress has been made for~\(g(M)=2\): any genus~\(2\) manifold~\(M\) can be represented (in a non-unique way) by a highly symmetric colored (multi)graph, and two such manifolds turn out to be homeomorphic if and only if there is a finite sequence of \textit{dipole moves} relating each pair of colored graphs representing them. Moreover, each colored graph associated to~\(M\) can be uniquely characterized by a \(6\)-tuple of integers. The homeomorphism problem for \(3\)-manifolds of genus~\(2\) thus translates into determining when two \(6\)-tuples determine the same manifold. Additionally, \(g(M)\)~can be recovered from the colored graphs associated to~\(M\), in the sense that \(g(M)\)~is the minimal genus~\(g(C)\) of a surface into which a colored graph~\(C\) associated to~\(M\) can be regularly embedded. Unfortunately, a single dipole move can change~\(g(C)\), and hence cannot be used for defining an equivalence relation on the set of \(6\)-tuples representing colored graphs that would reflect the homeomorphism of the corresponding \(3\)-manifolds. In the present paper, the authors describe composite sequences of dipole moves that change the colored graph, but leave its genus and the manifold it represents invariant. Different equivalence classes of colored graphs under such composite sequences may still represent the same \(3\)-manifold, but the authors show that these equivalence classes ``almost always'' contain infinitely many elements, and characterize when this happens. Furthermore, they give criteria for a \(6\)-tuple associated to a colored graph to be minimal with respect to a suitable complexity measure. The authors further suggest that their findings may be useful in a computer enumeration of \(3\)-manifolds of genus~\(2\), but do not report on the results of such an enumeration.