Infinitesimal Newton-Okounkov bodies and jet separation (Q2359431)

In the present paper, the authors provide a fundamental link between several positivity notions for divisors and objects coming from convex geometry. Let us recall some notions. For a projective variety \(X\) over the complex numbers, let \(L\) be a big line bundle, and let \(x \in X\) be a closed point. We say that \(L\) is \textit{locally positive} or \textit{locally ample} at \(X\) if there exists a neighborhood \(x \in \mathcal{U} \subset X\) such that the Kodaira map \(\phi_{mL}\) restricted to \(\mathcal{U}\) is an embedding for \(m\gg 0\). Once a line bundle \(L\) has been proven to be locally ample at a given point \(x \in X\), it is natural to try to measure the extent of its positivity there. The most traditional way to do this is using the Seshadri constant, but in the setting of the present paper the proper notion for these purposes is the moving Seshadri constant. The main idea of the paper is to apply the theory of infinitesimal Newton-Okounkov bodies to decode both local ampleness and moving Seshadri constants in a satisfactionary way. Let us recall what infinitesimal Newton-Okounkov bodies are. For a smooth projective variety \(X\) of dimension \(n\), let \(x \in X\) be a closed point, and consider \(\pi: X' \rightarrow X\) the blowing-up of \(X\) at \(x\) with the exceptional divisor \(E\). An \textit{infinitesimal flag} \(Y_{\bullet}\) over \(x\) is an admissible flag \[ Y_{\bullet} : Y_{0} = X' \supseteq Y_{1} = E \supseteq Y_{2} \supseteq \dots\supseteq Y_{n}, \] where each \(Y_{i}\) is a linear subspace of \(E \simeq \mathbb{P}^{n-1}\) of dimension \(n-i\) for each \(i=2, \dots,n\). The Newton-Okounkov body with respect to \(Y_{\bullet}\) on \(X'\) will be denoted by \(\widetilde{\triangle}_{Y_{\bullet}}(D)\). The first main result of the paper provides a complete characterization of nefness in terms of the shape of infinitesimal Newton-Okounkov bodies. {Theorem 1.} Let \(X\) be a smooth projective variety of dimension \(n\), and let \(D\) be a big \(\mathbb{R}\)-divisor on \(X\). The following conditions are equivalent: (i) \(D\) is nef; (ii) for every point \(x \in X\) there exists an infinitesimal flag \(Y_{\bullet}\) over \(x\) such that \(\mathbf{0} \in \widetilde{\triangle}_{Y_{\bullet}}(D)\); (iii) one has \(\mathbf{0} \in \widetilde{\triangle}_{Y_{\bullet}}(D)\) for every infinitesimal flag over \(X\). In the case of ampleness, we need the following definition. For \(\xi >0\) we define the convex body \[ \triangle_{\xi}^{-1} = \text{ convex hull of } \{\mathbf{0}, \xi e_{1}, \xi(e_{1} + e_{2}),\dots, \xi (e_{1} + e_{n}) \} \subset \mathbb{R}^{n}, \] where \(e_{1},\dots,e_{n}\) denote the standard basis vectors for \(\mathbb{R}^{n}\), and we call this body as the inverted standard simplex of size \(\xi\). The second result can be formulated as follows. Theorem 2. Let \(X\) be a smooth projective variety of dimension \(n\), and let \(D\) be a big \(\mathbb{R}\)-divisor on \(X\). Then the following are equivalent: (i) \(D\) is ample; (ii) for every point \(x \in X\) there exist an infinitesimal flag \(Y_{\bullet}\) over \(x\) and a real number \(\xi >0\) for which \(\triangle_{\xi}^{-1} \subseteq \widetilde{\triangle}_{Y_{\bullet}}(D)\); (iii) \(\widetilde{\triangle}_{Y_{\bullet}}(D)\) contains a non-trivial inverted standard simplex for every infinitesimal flag \(Y_{\bullet}\) over \(X\). It is worth pointing out that infinitesimal Newton-Okounkov bodies on projective varieties always contain inverted standard simplices at points where the divisor is locally ample. For a given infinitesimal flag \(Y_{\bullet}\), the supremum of sizes of all such simplices is called the inverted largest simplex constant, and will be denoted by \(\xi_{Y_{\bullet}}(D;x)\). It turns out that this constant does not depend on the choice of the infinitesimal flag, which leads to the common value \(\xi(D;x)\). The following result provides an interesting relation between the moving Seshadri constants \(\varepsilon(||D||;x)\) (see Definition \(1.9\) therein) and the inverted largest simplex constants. Theorem 3. Let \(D\) be a big \(\mathbb{R}\)-divisor on a smooth projective variety \(X\), and assume that \(x\) is not contained in the restricted base locus \(\mathbb{B}_{-}(D)\). Then \[ \varepsilon(||D||;x) = \xi(D;x). \] It is known that there are difficulties to detect those rational points of Newton-Okounkov bodies which actually arise as actual images of global sections. As an interesting by-product, the authors provide the following result. Theorem 4. Let \(D\) be a big \(\mathbb{Q}\)-divisor on \(X\) and let \(x \in X\) be a closed point. Denote by \(Y_{\bullet}\) an infinitesimal flag over \(x\). If \(\triangle_{\xi}^{-1} \subseteq \widetilde{\triangle}_{Y_{\bullet}}(\pi^{*}D)\) for some \(\xi > 0\), then all vectors in \(\triangle_{\xi}^{-1} \cap \mathbb{Q}^{n}\)not lying on the face generated by the points \(\lambda e_{1}, \lambda (e_{1} + e_{2}),\dots, \lambda (e_{1} +e_{n})\) are valuative. The paper is nicely written and contains several interesting ideas -- some of them were applied in the context of restrictions on Seshadri constants for algebraic surfaces [\textit{Ł. Farnik} et al., Taiwanese J. Math. 21, No. 1, 27--41 (2017; Zbl 1357.14011)]. Moreover, in Section \(5\) the authors define the extendend Seshadri functions, and it seems to be interesting to explore properties of those functions in detail.