The Torelli groups for genus 2 and 3 surfaces (Q1209386)

The mapping class group \(\Gamma_ g\) is the group of isotopy classes of orientation-preserving diffeomorphisms of the closed orientable surface of genus \(g\). The action of \(\Gamma_ g\) on the first homology of the surface determines a representation \(\Gamma_ g\to \text{GL}(2g,Z)\), whose image is the symplectic group \(\text{Sp}(g,Z)\), and whose kernel is by definition the Torelli group \(\ell_ g\). Around 1980, the Torelli groups were studied by \textit{Dennis Johnson}, who proved among many results that \(\ell_ g\) is finitely generated when \(g\geq 3\). The Torelli groups have been investigated by many other authors, including \textit{Birman}, \textit{Powell}, \textit{Chillingworth}, \textit{Wagoner}, \textit{Morita}, and the reviewer, who with \textit{A. Miller} gave a topological proof that \(\ell_ 2\) is not finitely generated. In the paper under review, the author uses sophisticated methods from complex algebraic geometry to probe much more deeply into \(\ell_ 2\) (and \(\ell_ 3\)); in particular, he shows that \(\ell_ 2\) is actually free, and that a set of free generators is the Dehn twists on separating curves which form a set of representatives for the homology splittings of the first homology of the genus 2 surface (that is, the unordered pairs \(\{A,B\}\) of rank two subgroups such that the first homology is \(A\oplus B\) and \(A\) and \(B\) are orthogonal with respect to the intersection pairing). The first section of the paper is a concise review of theta functions, the Siegel space \(Z_ g\), abelian varieties and principal polarizations, and the period map. The period map (also called the Torelli map) \(t: T_ g/\ell_ g\to Z_ g\) carries the quotient of Teichmüller space \(T_ g\) by the action of \(\ell_ g\) into the Siegel space. Since \(T_ g\) is contractible and the action of \(\ell_ g\) is known to be free, \(T_ g/\ell_ g\) is a \(K(\ell_ g,1)\)-space (in particular, \(\pi_ 1(T_ g/\ell_ g)=\ell_ g\) and the homology and cohomology of \(T_ g/\ell_ g\) agree with the group theoretic homology and cohomology of \(\ell_ g\)). In the second section, the author examines Jacobians of genus 2 curves, proving that \(t: T_ 2/\ell_ 2\to Z_ 2\) is a holomorphic injection whose complement is a disjoint union of properly imbedded copies of \(U\times U\), where \(U\) is the upper half plane. These imbeddings correspond to the homology splittings, so Alexander duality already shows that \(H_ 1(T_ 2/\ell_ 2)\) not finitely generated. In the third section, the structure of the image \(t(T_ 2/\ell_ 2)\) in \(Z_ 2\) is studied more closely by constructing a Morse function on it. This gives a handle decomposition for \(T_ 2/\ell_ 2\) which has one 0- handle and countably many 1-handles which correspond to the homology splitting of the first homology. This implies the freeness of \(\ell_ 2\) and leads to the identification of its generators. The group \(\ell_ 3\) is examined by a similar approach. The period mapping goes from \(T_ 3/\ell_ 3\) into \(Z_ 3\), but is no longer injective: by the Torelli theorem it has branching along the points corresponding to hyperelliptic genus 3 curves. The author proves (although giving credit to personal communication with \textit{D. Johnson} and \textit{J. Millson}) that \(H_ 3(\ell_ 3)\) contains a free abelian subgroup of infinite rank. A final section of the paper contains some applications to calculation of the homology of \(\text{Sp}(2,Z)\) and \(\text{Sp}(3,Z)\). For example, the Lyndon-Hochschild-Serre spectral sequence of the extension \(\ell_ 2\to\Gamma_ 2\to\text{Sp}(2,Z)\) degenerates, since \(\ell_ 2\) is free, and applying more precise statements of the results sketched above leads to homological information.