Compactifications of moduli spaces in real algebraic geometry (Q1187482): Difference between revisions

This interesting paper is the continuation of previous work of the author and \textit{M. Seppälä} on moduli spaces in real algebraic geometry [see for instance Math. Z. 201, No. 2, 151-165 (1989; Zbl 0645.14012)]. It is devoted mainly to the compactification of moduli spaces of real principally polarized abelian varieties. First is explained why the real points of the classical moduli spaces are not a good answer to the moduli problems in real algebraic geometry. The construction of the moduli space of real principally polarized abelian varieties is recalled, in the line of Comessatti theory. It is the quotient of the space of symmetric \(n\times n\) period matrices \({1\over 2}M+iT\) with \(T\) real positive definite and \(M\) with coefficient in \(\mathbb{Z}\), under the action of a subgroup of \(Sp_{2n}(\mathbb{Z})\). It comes with a canonical involution \(\Sigma\), which sends the antiholomorphic involution \(S\) defining the real structure to \(-S\), where \(-\) is from the group structure. The compactification is in two steps. The first one follows the construction of the Satake compactification. A second identification is imposed by the compatibility with the involution \(\Sigma\). There are here a lot of calculations with matrices. The case of real elliptic curves is explained in details. The moduli space is here the disjoint union of two lines, corresponding to real elliptic curves with one or two real components, respectively. The compactification is the circle obtained by adding two points, corresponding to the two real forms \(\mathbb{R}^*\) and \(S^ 1\) of the multiplicative group. These are also the generalized jacobians of the real cubics with a real node (resp. with a real isolated double point). This geometric interpretation of points in the boundary is extended to moduli spaces for higher dimensions, using real semi-abelian varieties, and generalized jacobians of semistable real curves.