Deformation families of open Calabi-Yau manifolds and the Steinness boundary

From complete metrics to deformation families of open Calabi-Yau manifolds

The starting point is a rational elliptic surface S obtained by blowing up CP2 at nine points lying on the image of an elliptic curve C0, with C the strict transform of that curve [1]. Hans-Joachim Hein had constructed complete Calabi-Yau metrics on the quasi-projective variety S\C, so these complements are complete open Calabi-Yau manifolds [1]. Fan Xu's Theorem 2.6 upgrades this from a single space to a compactifiable deformation family: over a suitably chosen base T, the family (F-tilde, S-tilde, T, pi-tilde, o, j) has fibers that are all complete open Calabi-Yau manifolds, and S-tilde is a closed analytic subset whose complement is flat over T [1]. Corollary 2.7 then shows the family is nontrivial by choosing T so that the square-zero elliptic curves over distinct parameters are not isomorphic, which prevents an algebraic isomorphism between the corresponding open fibers [1].

This places the paper in a lineage that treats compactifications as a tool for probing the complex-analytic nature of open manifolds. Van Why's work on non-affine Stein manifolds shows that a Stein manifold can admit a normal-crossing divisor compactification and still fail to be biholomorphic to any affine variety, using contact-boundary and plumbing arguments in the symplectic setting [3]. The new paper asks a related but distinct question: when a compactifiable deformation family is built from rational elliptic surfaces, are its fibers Stein? The answer it gives is no for the families it constructs, and the mechanism is a holomorphic involution rather than a symplectic growth-rate obstruction [1].

The holomorphic involution and the non-Steinness verdict

The non-Steinness argument rests on a holomorphic involution on the central fiber and on the family. Xu defines an involution l on CP2 by l([x:y:z]) = [x:-y:x], and arranges the nine blow-up points so that the center point o = (tau0, q1, ..., q8) satisfies qi = -q_{10-i} for i = 2, ..., 5, with q9 = -q1 [1]. Under these conditions, Corollary 1.2 produces a holomorphic involution on the deformation family (F-tilde, S-tilde, T, pi-tilde, o, j) [1]. The involution is induced by an involution on the lattice C0(tau) and is connected to mirror symmetry through the Borcea-Voisin construction, and it also connects to the hypersurface involution on the rational elliptic surface with global sections [1].

The presence of this involution forces the family to have two distinct fibers that are algebraically isomorphic to each other no matter how T is chosen [1]. Xu then states that S\C is not Stein in this case, so all fibers of the constructed compactifiable deformation families are not Stein [1]. This is a structural obstruction, not a metric one: it does not rely on the absence of a complete Calabi-Yau metric, since Hein's metrics already exist on these complements [1]. The result is also deliberately scoped. The non-Steinness conclusion is claimed for all the compactifiable deformation families constructed in the paper, not for every open Calabi-Yau manifold, and the construction is limited to the specific compactifiable families built from rational elliptic surfaces [1].

The Brunella conjecture and the nine-point blow-up of CP2

A second strand of the paper addresses the quasi-projective variety S\C coming from the blow-up of CP2 at nine points, in connection with a conjecture of Marco Brunella [1]. Xu classifies the value of the sum of the inverse images of the nine blow-up points on the elliptic curve. If that sum is 0 mod <1, tau>, or has order m >= 2, or equals m1 + i m2 mod <1, tau> with (m1, m2) a Diophantine number pair, then S\C is not Stein [1]. In the same cases, the anticanonical line bundle of S admits a smooth Hermitian metric whose curvature form is semipositive [1]. The paper also constructs a one-dimensional compactifiable deformation family (F1-tilde, S4-tilde, Vi, pi-F1-tilde, x_origin, j_origin) whose fibers are almost all not Stein, while the anticanonical line bundle is semipositive for almost all fibers [1].

The remaining cases are framed as open. Xu notes that the question of whether the anticanonical line bundle is semipositive for every fiber of the complex analytic family (F1-tilde, Vi, pi-F1-tilde) is related to a conjecture of Takayuki Koike, and that the answer also depends on the Ueda type of the strict transform of the embedded elliptic curve in each fiber [1]. This is an explicit boundary of the result: the non-Steinness verdict covers the listed arithmetic cases, but the full Brunella conjecture and the Koike-related question are not settled by the construction [1]. The paper's own framing treats these as left work rather than as consequences of the main theorem [1].

Where Steinness survives: canonical extensions with nef tangent bundles

The paper's final strand constructs a different kind of deformation family, in the sense of Elizabeth Gasparim and Francisco Rubilar, whose fibers are Stein [1]. Proposition 4.5 shows that if (M, pi_M, B) is a nontrivial complex analytic family with B a sufficiently small polydisc centered at 0, and if there is a polydisc B1 compactly contained in B such that M_B1 = pi_M^{-1}(B1) is Kähler, then one can construct a deformation family of the canonical extension W_M of the compact complex manifold M over B1 [1]. The construction uses the extension 0 -> T*M -> W -> O_M -> 0 associated to a Kähler form, and defines W_M = P(W) \ P(T*M) [1].

Corollary 4.6 then states that if the tangent bundle for every fiber of (M_B1, pi_M_B1, B1) is nef, the fibers of the canonical-extension deformation family are all Stein [1]. This is presented as following directly from a theorem that holds unconditionally as stated in the cited work [1]. The paper also notes that Andreas Höring and Thomas Peternell conjectured that the fibers are all Stein if and only if the tangent bundle for each fiber is numerically effective [1]. A concrete example is the deformation family of the canonical extension of a torus, where all fibers are Stein by a cited proposition [1]. This Stein-fiber family is not a counterexample to the non-Steinness result; it is a separate construction with different hypotheses, and it shows that the paper's conclusions are conditional on the family being considered [1].

What earlier work established and what remains open

The broader deformation-theoretic background is well established. Keller's work on deformed Calabi-Yau completions defines and investigates deformed n-Calabi-Yau completions of homologically smooth dg categories, shows that they have the Calabi-Yau property, and proves compatibility with derived equivalences and localizations [2]. That is a different notion of deformation from Xu's compactifiable deformation families, but it establishes that Calabi-Yau properties can be stable under deformation in a controlled algebraic setting [2]. Wilson's boundedness results for Calabi-Yau threefolds address a separate question about whether the cubic cup-product form and the linear form given by c2 determine the threefold up to finitely many families, with boundedness proved for Picard number 2 and partial results for higher Picard number [5]. These are not direct inputs to Xu's construction, but they show that the moduli-theoretic behavior of Calabi-Yau objects is itself an active and only partially resolved area [5].

The limits of the new paper are therefore twofold. First, the non-Steinness conclusion is established for the compactifiable deformation families constructed in the paper, not for all open Calabi-Yau manifolds [1]. Second, the Brunella-related discussion leaves open the cases not covered by the arithmetic conditions on the sum of the nine points, and the Koike-related question about semipositivity of the anticanonical line bundle for every fiber remains dependent on the Ueda type [1]. The Stein-fiber construction is likewise conditional on the Kähler hypothesis and the nefness of the tangent bundle [1]. What the paper does establish is a clear contrast: in one construction, a holomorphic involution forces non-Steinness across the family, while in another, a nefness hypothesis yields Stein fibers [1].

About These Sources

This research page is built on 5 studies (2 peer-reviewed, 3 preprints) — published from 2021 to 2026, 3 from 2024 or later, 1 in Q1 journals, collectively cited 53 times — selected as the most relevant from 13 studies that passed quality screening, drawn from 121 papers retrieved from a database of over 500 million.

Sources used in this answer