Kodaira-Iitaka dimension and multiplicity: an analytic perspective
Siarhei Finski · 2026
Read Kodaira-Iitaka dimension and multiplicity: an analytic perspective by Siarhei Finski (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-22194.txt.
Introduction
Finski gives analytic formulas for two invariants of graded linear series. The invariants are the Kodaira-Iitaka dimension and the multiplicity. Both are expressed through the intersection theory of a plurisubharmonic envelope. He refines the formulas at two levels, pointwise and metric. At the pointwise level, partial Bergman kernels converge weakly. At the metric level, he computes volume ratios for unit balls under sup norms. He also defines a non-pluripolar version of the dimension. It sits above the classical invariant and below earlier numerical versions.
Catalog record
| CREATOR | Siarhei Finski |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 127618 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.22194
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Kodaira-Iitaka dimension and multiplicity: an analytic perspective Siarhei Finski Abstract MnLargeSymbols’164 MnLargeSymbols’171 Abstract. We express the Kodaira-Iitaka dimension and the multiplicity of graded linear series in terms of the intersection theory of the plurisubharmonic envelope associated with the linear series, and obtain two refined versions of these formulas at the pointwise and at the metric levels. At the pointwise level, we focus on the weak convergence of the partial Bergman kernel associated with the linear series and a Bernstein-Markov measure. At the metric level, we compute the asymptotic ratio of the volumes of unit balls defined by the sup-norms on the linear series. Based on our findings, we introduce a non-pluripolar version of the numerical Kodaira-Iitaka dimension for a line bundle, show that this invariant dominates the classical Kodaira-Iitaka dimension and is, in turn, bounded above by the numerical versions proposed so far. Table of contents 1 Introduction We fix a connected complex projective manifold X , \dim_{\mathbb{C}}X=n , and a holomorphic line bundle L over X . The main object of the present study is a (graded) linear series W=\oplus_{k=0}^{\infty}W_{k} , W_{0}=\mathbb{C} , defined as a subring of the associated section ring R(X,L)\mathrel{\mathop{\ordinarycolon}}=\oplus_{k=0}^{\infty}H^{0}(X,L^{\otimes k}). (1.1) To exclude some trivial cases, we make an additional assumption that W\neq\mathbb{C} , implying in particular that L is effective , i.e., H^{0}(X,L^{\otimes k})\neq\{0\} for k\in\mathbb{N}^{*} large enough. Without loosing the generality (as otherwise one could restrict to a tensor power of L ), we further assume W_{k}\neq\{0\},\quad\text{for }k\in\mathbb{N}\text{ large enough}. (1.2) Using Okounkov bodies, Kaveh-Khovanskii in [ 45 , Theorem 3.4] established that there is \kappa(W)\in\{0,1,\ldots,n\} , called the Kodaira-Iitaka dimension of W , so that the following limit {\rm{vol}}_{\kappa}(W)\mathrel{\mathop{\ordinarycolon}}=\lim_{k\to\infty}\frac{\dim W_{k}}{k^{\kappa(W)}/\kappa(W)!} (1.3) exists and is positive; see also Lazarsfeld-Mustață [ 48 ] for the regime \kappa(W)=n , and Iitaka [ 43 ] , Fujita [ 36 ] for the previous works in realms of the complete linear series, that is, W=R(X,L) . We call the limit in ( 1.3 ) the multiplicity of W . When W is a complete linear series, we call the above quantity the multiplicity of L and denote it by {\rm{vol}}_{\kappa}(L) . The corresponding Kodaira-Iitaka dimension is denoted by \kappa(L) and is called the Kodaira-Iitaka dimension of L . This paper has three complementary aims: to establish analytic formulas for the Kodaira-Iitaka dimension and the multiplicity of graded linear series, and to refine these formulas at both the pointwise and metric levels. By the pointwise level, we refer to the asymptotic analysis of the corresponding partial Bergman kernels. By the metric level, we mean comparing the sequences of norms on W_{k} induced by the sup-norms associated with two continuous metrics on L . We begin with the former formula, which will be based on a notion of a plurisubharmonic (psh) envelope of a metric associated with the linear series. The latter definition can be traced back to Siciak [ 67 ] ; in the current context of linear series it was introduced by Hisamoto [ 41 ] . To recall it, for any norm N_{k} on W_{k} , we associate a singular metric FS(N_{k}) on L^{\otimes k} as |l|_{FS(N_{k}),x}=\inf_{\begin{subarray}{c}s\in H^{0}(X,L^{\otimes k})\\ s(x)=l\end{subarray}}\|s\|_{N_{k}},\quad\text{for any }x\in X,l\in L^{\otimes k}_{x}, (1.4) with the convention that the infimum equals +\infty if there is no s\in W_{k} such that s(x)=l . It is easy to see, see Lemma 3.9 , that for any continuous metric h^{L} on L , the sequence of Fubini-Study (singular) metrics associated with the sup-norms \textrm{Ban}_{k}^{\infty}[W](h^{L}) on W_{k} , k\in\mathbb{N}^{*} , is submultiplicative. In particular, by the Fekete’s lemma, the following limit P[W,h^{L}]=\Big(\lim_{k\to\infty}FS({\textrm{Ban}}_{k}^{\infty}[W](h^{L}))^{\frac{1}{k}}\Big)_{*} (1.5) exists as a (singular) metric on L , where (\cdot)_{*} is the lower semicontinuous regularization. Standard results imply that P[W,h^{L}] has a psh potential, [ 29 , Proposition I.4.24] . Define the closed positive (1,1) -current c_{1}(L,P[W,h^{L}]) as described after ( 1.29 ). On X , fix a Kähler form \omega and consider the following measures, defined in the non-pluripolar, see [ 11 ] , sense c_{1}(L,P[W,h^{L}])^{i}\wedge\omega^{n-i},\qquad i\in\{0,1,\ldots,n\}. (1.6) Although the above measures in general depend on \omega , many of the associated quantities turn out to be independent of it, as shown below. Theorem 1.1 . For any graded linear series W\subset R(X,L) , we have \kappa(W)=\max\Big\{i\in\{0,1,\ldots,n\}\mathrel{\mathop{\ordinarycolon}}c_{1}(L,P[W,h^{L}])^{i}\wedge\omega^{n-i}\neq 0\Big\}. (1.7) Remark 1.2 . At first sight, the condition in ( 1.7 ) appears to depend on the choice of the Kähler form \omega and of the metric h^{L} , but this is not the case; see the end of Section 3.3 for details. Our next result gives an analytic expression for the multiplicity of a graded linear series. It will involve another important invariant of the linear series, which we now recall. Recall that any linear series defines the rational maps X\dashrightarrow\mathbb{P}(W_{m}^{*}) that send a point x\in X to the hyperplane H_{m,x}\subset W_{m} consisting of all sections in W_{m} that vanish at x (and are undefined whenever all sections in W_{m} vanish at x ). We denote by Y_{m} the closure of the image of this map. A theorem of Chang-Jow [ 16 , Theorem 1.1] states that the rational maps \varphi_{m}\mathrel{\mathop{\ordinarycolon}}X\dashrightarrow Y_{m} (1.8) are birationally equivalent for m large enough. It is well-known, cf. [ 13 , Theorem 3.15] , that \kappa(W)=\dim(Y_{m}),\quad\text{for }m\in\mathbb{N}\text{ large enough}. (1.9) We say that W is birational if the maps ( 1.8 ) are birational for m large enough, cf. [ 48 , Definition 2.5] . Taking into account ( 1.9 ), we see that birational linear series only exist when L is big, i.e., such that the Kodaira-Iitaka dimension of L is n . We denote by [\omega]\in H^{1,1}(X)\cap H^{2}(X,\real) the cohomology class of \omega , and by \varphi_{*}[\omega]^{n-\kappa(W)} the fiber integral of \omega along a general fiber of \varphi_{m} for m large enough. By the result recalled after ( 1.8 ), this quantity depends only on the cohomology class [\omega] , as indicated by the notation. Theorem 1.3 . For any graded linear series W\subset R(X,L) , we have {\rm{vol}}_{\kappa}(W)=\frac{1}{\varphi_{*}[\omega]^{n-\kappa(W)}}\int_{X}c_{1}(L,P[W,h^{L}])^{\kappa(W)}\wedge\omega^{n-\kappa(W)}. (1.10) Remark 1.4 . a) When W=R(X,L) and L is big, by [ 68 ] and [ 34 , Theorem 2.16] , we have P[W,h^{L}]=P[h^{L}], (1.11) where P[h^{L}] is the psh envelope defined by P[h^{L}]=\inf\Big\{h^{L}_{0}\mathrel{\mathop{\ordinarycolon}}h^{L}_{0}\text{ has a psh potential and }h^{L}_{0}\geq h^{L}\Big\}. (1.12) Note that immediately from the definition, P[h^{L}] has minimal singularities in the sense described in Section 3.1 . Theorem 1.3 , Lemma 3.1 , ( 1.11 ) and the monotonicity formula, cf. Theorem 3.2 , then recover a theorem of Boucksom [ 12 ] which states that for a big line bundle L and any positive closed (1,1) -current T_{\min} with a potential of minimal singularities in the class [c_{1}(L)] , we have {\rm{vol}}_{\kappa}(L)=\int_{X}T_{\min}^{n}. (1.13) b) When W is birational, ( 1.10 ) is due to Hisamoto [ 41 ] . One may ask whether an analogue of ( 1.13 ) remains valid for line bundles that are not big. To discuss this in details, we introduce the non-pluripolar numerical Kodaira-Iitaka dimension, \kappa_{{\rm{np}}}(L) , and the non-pluripolar numerical multiplicity, {\rm{vol}}_{\kappa,{\rm{np}}}(L) , as follows \displaystyle\kappa_{{\rm{np}}}(L)=\max\Big\{i\in\{0,1,\ldots,n\}\mathrel{\mathop{\ordinarycolon}}T_{\min}^{i}\wedge\omega^{n-i}\neq 0\Big\}. (1.14) \displaystyle{\rm{\displaystyle vol}}_{\kappa,{\rm{np}}}(L)=\frac{1}{\varphi_{*}[\omega]^{n-\kappa(L)}}\int_{X}T_{\min}^{\kappa(L)}\wedge\omega^{n-\kappa(L)}, where T_{\min} is a positive closed (1,1) -current with a potential of minimal singularities in the class [c_{1}(L)] and \varphi_{*}[\omega]^{n-\kappa(L)} is defined for the complete linear