On real generalized Jacobian varieties (Q2573493)

\textit{C. Scheiderer} solved in the affirmative 17th Hilbert's problem for smooth real algebraic curves in his beautiful work [Trans. Am. Math. Soc. 352, No. 3, 1039--1069 (2000; Zbl 0941.14024)]. To that end he proved that every divisor on a connected projective smooth curve over the reals (with at least one real point) is linearly equivalent to a divisor supported on the real points of the curve, and he asked whether this result can be extended to singular projective curves. This seems to be the starting point of the paper under review. In what follows, \(X\) is a real algebraic curve, i.e. a proper, one dimensional and geometrically integral scheme over \(\mathbb R\). Let \(S\) be its singular locus, \(U=X\setminus S\) and let \(p_a\) be its arithmetic genus. Let \(l\) be the number of loops (connected components that are not singletons) of the set \(X(\mathbb R)\) of \(\mathbb R\)-rational points of \(X\). Scheiderer in the article cited above and the author of the present paper in [Adv. Geom. 3, 339--360 (2003; Zbl 1079.14013)], introduced two numerical invariants: i) \(N(X)=\) the smallest integer \(n\geq 1\) such that any invertible sheaf of degree \(n\) on \(X\) is isomorphic to a Cartier divisor \(\mathcal O(D),\) with \(D\) a totally real effective divisor on \(U\). Of course, \(N(X)=\infty\) if such an integer does not exist. ii) \(M(X)=\) the smallest integer \(m\geq 1\) such that any invertible sheaf \(\mathcal L\) of degree \(2m\) on \(X\) with the property that the degree of \(\mathcal L\) on each loop of \(X(\mathbb R)\) is even, is isomorphic to a Cartier divisor \(\mathcal O(D),\) with \(D\) a totally nonreal effective divisor on \(U\). If such an integer does not exist, then \(M(X)=\infty.\) The goal of this paper is to study the finiteness of these invariants and, in case they are finite, to estimate them in terms of the genus \(g\) of a desingularization \(\hat {X}\) of \(X\), the number \(s\) of connected components of \(\widehat {X}(\mathbb R)\) and the number \(l\) of loops of the set \(X(\mathbb R).\) These problems were raised by Scheiderer, who also proved the finiteness of \(N(X)\) in the smooth case. Monnier proved the finiteness of \(M(X)\) in the smooth case in the article cited above, where also some lower and upper bounds of \(N(X)\) and \(M(X)\) for smooth \(X\) are given. The main results of the present work can be summarized as follows. Theorem. i) \(g\leq M(X) \leq 2p_a\). ii) If the singularities of \(X\) are either real isolated and worse than nodes or nonreal, then \(N(X)=\infty.\) iii) If \(X\) is a nodal curve with only real nodes, then \(N(X)\) is finite. iv) \(l\leq 1+p_a.\) Moreover, if \(l\geq p_a,\) then \(N(X)\leq 2p_a.\) v) Suppose \(p_a\geq 2.\) Then \(N(X)\geq 1+ p_a.\) vi) Suppose \(g\geq 2.\) Then \(M(X) \geq g.\) This article, which is clearly written and uses both clever arguments in classical algebraic geometry and specific results of the real case, contains also many examples and consequences of the above theorem in some particular situations.