A Riemann-Hurwitz-Plücker formula (Q2689538)

In the paper under review the authors provide an interesting generalization of the classical Riemann-Hurwitz and Pluecker formulas. Here is the setting of the paper under review. Let \(X\) be a smooth projective variety defined over an algebraically closed field \(\mathbb{K}\) of arbitrary characteristic, \(S\) a projective variety of dimension \(n\) and let \(\mathcal{D} \subset X \times S\) be a flat family of divisors on \(X\). The key definition that the authors introduce is a generalization of the well-known definition of inflection points. Definition. If we are given a smooth projective curve \(C\) and a morphism \(f : C \rightarrow X\), we say that a \(\mathbb{K}\)-point \(p \in C\) is an inflection point relative to \(\mathcal{D}\) if there exists a \(z \in S\) with the corresponding divisor \(D = \mathcal{D}_{z} \subseteq X\) such that \(f^{*}D\) has multiplicity at least \(n+1\) at \(p\). The formula for the total inflection of such a morphism will be expressed in terms of two quantities, namely \(N(X,\mathcal{D}) \in \mathbb{Z}_{\geq 0}\) and \(\mathcal{H}(X,\mathcal{D}) \in \mathrm{Pic}(X)\) which arise in that setting. More precisely, for any positive integer \(m\) there is a natural evaluation morphism on the space of \(m\)-tuples of points on divisors in \(\mathcal{D}\), which is denoted by \[ \mathrm{ev}_{m}\,:\, \mathcal{D}\times_{S}\mathcal{D} \times_{S} \cdots \times_{S} \mathcal{D} \rightarrow X^{m}, \] where we have \(m\) copies of \(\mathcal{D}\) on the lefthand side . Since both the source and the target of \(\mathrm{ev}_{n}\) have dimension \(n\cdot\dim X\) and the target is in addition irreducible, one defines \[ N(X,\mathcal{D})=\deg \, \mathrm{ev}_{n}. \] Thus \(N(X,\mathcal{D})\) is simply the number of divisors in \(\mathcal{D}\) passing through \(n\) general points in \(X\). Secondly, if \(\alpha \in A_{(n+1)\dim X - 1}(X^{n+1}) = \mathrm{Pic}(X^{n+1})\) is the pushforward of the fundamental class of the source of \(\mathrm{ev}_{m+1}\) to its target, and \(\delta : X \rightarrow X^{n+1}\) is the small diagonal immersion, one defines \[ \mathcal{H}(X,\mathcal{D}) = \delta^{*}\alpha. \] Thus \(\mathcal{H}(X,\mathcal{D})\) is the pullback to \(X\) under the small diagonal immersion of the class of the divisor consisting of \((n+1)\)-tuples of points on \(X\) contained in some divisor in \(\mathcal{D}\). The main result of the paper under review can be formulated as follows. Main Result. In the setting presented above, if we are given any smooth proper curve \(C\) and any morphism \(f : C \rightarrow X\) such that the set \(I\subset C\) of inflection points relative to \(\mathcal{D}\) is finite, then it is possible to assign non-negative integer multiplicities \(m_{p}\) for all \(p \in I\) such that \[ \sum_{p \in I} m_{p}p \sim f^{*}\mathcal{H}(X,\mathcal{D}) + \binom{n+1}{2}N(X,\mathcal{D})K_{C},\tag{\(\star\)} \] where \(\sim\) denotes linear/rational equivalence. In fact, one has \(m_{p} > 0\) for all \(p \in I\) if the induced morphism \(S \rightarrow \mathrm{Hilb}(X)\) to the Hilbert scheme is geometrically finite, respectively \(m_{p}=0\) for all \(p \in I\) otherwise. Note that in fact the lefthand side of \((\star)\) is a divisor while the righthand side is a divisor class, so it would be more precise to say that the lefthand side is an element of the righthand side. For the completeness of the picture, the authors show the following. Theorem. The righthand side of \((\star)\) specializes to the classical Riemann-Hurwitz and Pluecker formulas in the special cases that \(X\) is a curve, or that \(X\) is a projective space, respectively. Moreover, \(m_{p}\)'s constructed in the paper recover the usual multiplicities occurring in these formulas.