On generalized canonical bundle formula and boundedness of complements in complex analytic setting
Kenta Hashizume · 2026
Read On generalized canonical bundle formula and boundedness of complements in complex analytic setting by Kenta Hashizume (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-2603-17485.txt.
Introduction
Hashizume works in complex analytic geometry, where spaces are built from complex numbers and maps between them. He proves a generalized canonical bundle formula for generalized lc-trivial fibrations. The bases in his setup need not be compact, and the coefficients need not be rational. He shows two objects, the discriminant b-divisor and the moduli b-divisor, behave well when cut down to open subsets. That match lets local and global steps line up. He also treats when complements stay bounded in this setting.
Catalog record
| CREATOR | Kenta Hashizume |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 122731 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.17485
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On generalized canonical bundle formula and boundedness of complements in complex analytic setting Kenta Hashizume Abstract . We establish the generalized canonical bundle formula for generalized lc-trivial fibrations with irrational coefficients over non-compact bases in the complex analytic setting, and we show that the discriminant b-divisor and moduli b-divisor are compatible with restriction to arbitrary open subsets. We also discuss the boundedness of complements in this setting. 2020 Mathematics Subject Classification Primary: 14D06, Secondary: 14J40, 14C20, 32C15 1. Introduction In this paper, we establish the generalized canonical bundle formula for generalized lc-trivial fibrations over non-compact bases and prove the boundedness of complements in the complex analytic setting. We work with the notion of generalized pairs, introduced by Birkar–Zhang [ BiZh16 ] . Generalized pairs have played a crucial role in Birkar’s work on the boundedness of complements [ Bir19 ] and the boundedness of Fano varieties of fixed dimension with mild singularities [ Bir21 ] . These developments have led to significant progress in birational geometry, and the language of generalized pairs is now standard in this area. In the complex analytic setting, there are two extensions of this notion depending on the framework adopted for the nef part. One is the notion of generalized pairs whose nef part is given by b-nef \mathbb{R} -b-divisors. Another is the notion of generalized pairs whose nef part is given by nef closed (1,1) -currents. The former is very natural in the context of projective morphisms between complex analytic spaces. The latter, on the other hand, is useful in the context of compact but non-projective Kähler varieties because of the lack of ample divisors. We adopt the former definition in this paper. The canonical bundle formula plays a central role in birational geometry, especially in inductive arguments on the dimension of varieties. The canonical bundle formula for klt-trivial fibrations was studied by Kawamata [ Kaw98 ] and Ambro [ Am04 ] , [ Am05 ] , and Fujino–Gongyo [ FG14 ] established the canonical bundle formula for lc-trivial fibrations. Their theorems were extended to the framework of generalized pairs by Filipazzi [ Fi20 ] . For recent developments on the b-semiampleness conjecture, see Bakker–Filipazzi–Mauri–Tsimerman [ BFMT25 ] . For the canonical bundle formula in the Kähler context, see Hacon–Păun [ HaPa24 ] . We extend Filipazzi’s theorem to generalized lc-trivial fibrations in Fujino’s setting of complex analytic varieties [ Fu22 ] . Theorem 1.1 (= Theorem 3.10 ) . Let \pi\colon(X,B+M)\to S be a generalized sub-pair with the nef part \boldsymbol{\rm M} , and let f\colon(X,B+M)\to Z be a generalized lc-trivial fibration over S . Let W\subset S be a compact subset such that \pi\colon X\to S and W satisfy (P). Suppose that • \boldsymbol{\rm M} is a finite \mathbb{R}_{>0} -linear combination of b-nef /S \mathbb{Q} -b-Cartier \mathbb{Q} -b-divisors on X , and • there exists a Zariski open dense subset U\subset Z such that B|_{f^{-1}(U)} is effective. Then, after replacing S with a suitable open neighborhood of W , we have the following properties. (i) Let \boldsymbol{\rm G} (resp. \boldsymbol{\rm N} ) be the discriminant \mathbb{R} -b-divisor (resp. the moduli \mathbb{R} -b-divisor) associated to f . Then \boldsymbol{\rm N} is a finite \mathbb{R}_{>0} -linear combination of b-nef /S \mathbb{Q} -b-Cartier \mathbb{Q} -b-divisors on Z . In particular, (Z,\boldsymbol{\rm G}_{Z}+\boldsymbol{\rm N}_{Z})\to S is a generalized sub-pair with the nef part \boldsymbol{\rm N} . Furthermore, if (X,B+M)\to S is generalized lc (resp. generalized klt), then (Z,\boldsymbol{\rm G}_{Z}+\boldsymbol{\rm N}_{Z})\to S is also generalized lc (resp. generalized klt). (ii) For any open subset \tilde{S}\subset S that does not necessarily contain W , if we put \tilde{X}\subset X and \tilde{Z}\subset Z as the inverse images of \tilde{S} , then the generalized lc-trivial fibration (\tilde{X},B|_{\tilde{X}}+M|_{\tilde{X}})\to\tilde{Z} over \tilde{S} satisfies ( i ). In particular, the discriminant \mathbb{R} -b-divisor and the moduli \mathbb{R} -b-divisor associated to f are compatible with the restriction over any open subset (Lemma 3.3 ). For the definition of property (P), see Definition 2.1 . We explain the differences between Theorem 1.1 and [ HaPa24 , Theorem 2.3] . The main difference lies in the compactness assumptions in generalized lc-trivial fibrations. Theorem 1.1 applies to generalized lc-trivial fibrations over non-compact bases. Hence, the nef part of the generalized sub-pair is not necessarily globally nef. In particular, if S=Z in Theorem 1.1 , then the nef part is only fiberwise semi-positive. Moreover, Theorem 1.1 allows divisors with irrational coefficients. We remark that Theorem 1.1 is new even in the case of lc-trivial fibrations and the theorem slightly improves [ Fu22 , Theorem 21.4] in the case of klt-trivial fibrations. However, the starting point of the proof of Theorem 1.1 is [ Fu22 , Theorem 21.4] . Using Filipazzi’s idea [ Fi20 ] , we first prove Theorem 1.1 for generalized klt-trivial fibrations. We then extend the framework to generalized lc-trivial fibrations. In the proof, weak semistable reduction and the minimal model program (MMP, for short) in the complex analytic setting play important roles. We refer to [ EH25 ] and [ EH24 ] for those key ingredients, respectively. We note that we do not use the variations of mixed Hodge structures in the proof (cf. [ FG14 , Theorem 3.6] ). To the best of our knowledge, the strategy of the proof does not appear in the algebraic setting. The first and most fundamental difficulty goes back to the definition of b-divisors. In the complex analytic setting, it is not clear whether there is a reasonable definition of the restriction of b-divisors to open subsets. This is because the number of bimeromorphic morphisms increases under restriction, and we do not have a compactification of open analytic varieties. Hence, it is not clear whether the discriminant \mathbb{R} -b-divisor and the moduli \mathbb{R} -b-divisor associated to generalized lc-trivial fibrations are compatible with restriction over the base. However, to run a relative MMP or take a weak semistable reduction, restriction over the base is unavoidable. Therefore, the first step toward Theorem 1.1 is to prove a suitable compatibility of the discriminant \mathbb{R} -b-divisor and the moduli \mathbb{R} -b-divisor with respect to restriction. This is achieved in Lemma 3.3 . The second difficulty is the \mathbb{R} -b-Cartier property of the moduli \mathbb{R} -b-divisor. Note that [ Fu22 , Theorem 21.4] does not explicitly assert the \mathbb{Q} -b-Cartier property of the moduli \mathbb{Q} -b-divisor. This is treated in Remark 3.4 and Theorem 3.5 , which is Theorem 1.1 in the case of klt-trivial fibrations. After extending Theorem 3.5 to the case of generalized klt-trivial fibrations using Filipazzi’s idea [ Fi20 ] , we encounter the final difficulty: the extension to the generalized lc case. We overcome this difficulty by using the techniques of the MMP [ EH24 ] together with Theorem 1.1 in the case of generalized klt-trivial fibrations. For details, see the proof of Proposition 3.9 . Once we establish the canonical bundle formula and the MMP [ EH26 ] , we expect that many results for projective varieties can be extended to projective morphisms between complex analytic varieties. Although we do not explore such applications in depth in this paper, we prove the boundedness of complements in the complex analytic setting. Roughly speaking, complements provide a way to measure singularities