Family 046 — Shafarevich’s conjecture: two claimed failures of holomorphic convexity

← All digests

Family 046 · Algebraic and complex geometry
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-12

1. The question

The universal cover of a space unwinds all its loops. For compact complex curves, this leads to the sphere, the complex plane, or the unit disc. These highly structured models invite a higher-dimensional question: how well behaved is the universal cover of a smooth projective variety?

Shafarevich’s holomorphic-convexity conjecture predicts a strong analytic answer. Holomorphic functions should be plentiful enough to prevent the hull of a compact set from spreading indefinitely through the universal cover.

Family 046 contains two manuscripts claiming counterexamples. M2 constructs a projective surface whose universal cover contains an infinite connected chain of compact rational curves. Holomorphic functions must be constant along the whole chain, defeating holomorphic convexity.

M1 gives a different and stronger obstruction under an additional geometric condition: a projective fourfold with large fundamental group, whose universal cover contains no compact positive-dimensional analytic subvariety, but is nevertheless not Stein. The two constructions address different ways the conjecture could fail and should not be merged into one example.

2. Background and essential definitions

Universal covers and fundamental-group images

For a connected manifold $X$, its universal cover $\widetilde X$ is a simply connected manifold mapping locally isomorphically to $X$. If $X$ is complex, the cover inherits a complex structure.

The fundamental group $\pi_1(X)$ records loops up to continuous deformation. An inclusion $Z\subset X$ induces a homomorphism from the loops in $Z$ to those in $X$. This map need not be injective: loops essential inside $Z$ may contract in the larger space. Its image determines how a connected lift of $Z$ sits over $Z$.

When that image is infinite, a connected lift can have infinitely many sheets. Keeping an image infinite after embedding into a projective variety is the main construction problem in both manuscripts.

Holomorphic hulls

For a compact nonempty $K\subset V$, its holomorphic hull is

$$\widehat K_V= \{x\in V: |f(x)|\le\sup_K|f|\text{ for every }f\in\mathcal O(V)\}.$$

The notation $\mathcal O(V)$ means all holomorphic complex-valued functions on $V$. A point lies in the hull when no such function separates its size from the values already seen on $K$.

The space $V$ is holomorphically convex if every compact set has compact hull. A Stein manifold is a complex manifold with particularly strong global function theory; it is the analytic analogue of an affine variety. In particular, Stein manifolds are holomorphically convex and have no compact positive-dimensional analytic subvarieties.

The converse needs care. Merely having no compact analytic curves does not, by definition, make a manifold Stein. For a holomorphically convex manifold, however, Remmert reduction contracts its compact connected analytic fibers to a Stein space. If there are no positive-dimensional compact analytic subvarieties, this reduction identifies the manifold itself as Stein.

“Large” has a precise meaning

A smooth projective $X$ has large fundamental group if every positive-dimensional irreducible subvariety $Z\subset X$ has infinite image

$$\operatorname{im}\bigl(\pi_1(Z^\nu)\to\pi_1(X)\bigr),$$

where $Z^\nu$ is its normalization, which separates the branches of singularities. In this setting, largeness is equivalent to the absence of compact positive-dimensional analytic subvarieties in $\widetilde X$.

This is not a statement that the group has many generators or grows quickly. It is a condition on every subvariety and its induced loops.

3. Development history and previous partial results

Shafarevich’s question extends the uniformization viewpoint for curves to higher-dimensional projective geometry. Campana and Kollár developed reduction maps that organize subvarieties whose normalized fundamental groups have finite image. These are powerful geometric reductions, but a map defined generically on the projective variety does not automatically yield a proper holomorphic reduction of its whole universal cover.

Several major affirmative results impose conditions on the fundamental group. Gurjar–Shastri handled smooth projective elliptic surfaces. Katzarkov treated virtually nilpotent fundamental groups. Work of Katzarkov–Ramachandran and Eyssidieux developed the theory for linear and reductive representations.

The linear Shafarevich theorem of Eyssidieux–Katzarkov–Pantev–Ramachandran proves holomorphic convexity for smooth projective varieties whose fundamental groups admit faithful finite-dimensional complex representations. Campana–Claudon–Eyssidieux extended the linear result to compact Kähler manifolds. More recent reductive-cover and normal-space results further broaden this affirmative theory while retaining specific representation-theoretic hypotheses.

Potential counterexamples also have a history. Bogomolov–Katzarkov proposed constructions using degenerating curves and group quotients. Their proposed failures depended on infinitude assertions for groups after all geometric relations had been imposed. Those conditional constructions did not themselves settle the conjecture.

The infinite-chain obstruction was already understood in work of Lasell–Ramachandran and Katzarkov–Ramachandran. The missing ingredient was an actual projective construction retaining the necessary infinite group image. M2 proposes to supply that ingredient through weighted group presentations. M1 instead uses a cover of an abelian surface and a local method for showing loops survive.

Thus the claimed novelty is not the observation that holomorphic functions are constant on compact curves. It is realizing the required configuration while preserving its global fundamental-group behavior.

4. What the new manuscripts claim

M2: failure already in complex dimension two

M2 claims a smooth connected projective complex surface $X$ containing a connected nodal curve $Z_0$, all of whose irreducible components are smooth rational curves, such that

$$\operatorname{im}\bigl(\pi_1(Z_0)\to\pi_1(X)\bigr) \text{ is infinite}.$$

A rational curve is isomorphic to $\mathbb P^1$, hence simply connected. Nevertheless, a union of such curves can have loops formed by its incidence pattern: moving through several components and back can create a loop even though each component separately has none.

In $\widetilde X$, a connected component above $Z_0$ is claimed to be a closed, noncompact, locally finite union of compact rational curves. Every global holomorphic function is constant on it. Consequently even the hull of one point is noncompact, so $\widetilde X$ is not holomorphically convex.

This surface does not have large fundamental group: its rational curves already violate that condition.

M1: failure with large fundamental group

M1 claims a smooth connected projective fourfold $X$ containing a simple abelian surface $A$, with

$$X\text{ having large fundamental group}, \qquad \operatorname{im}\bigl(\pi_1(A)\to\pi_1(X)\bigr)\simeq\mathbb Z.$$

An abelian surface is a projective complex torus of dimension two. It is simple if it contains no nonzero proper abelian subvariety. Its fundamental group is $\mathbb Z^4$.

The asserted infinite cyclic image means the kernel has rank three. A connected lift of $A$ is therefore diffeomorphic to

$$\mathbb R\times(S^1)^3,$$

and has nonzero third homology. A Stein surface has the homotopy type of a cell complex of real dimension at most two, so this lifted surface cannot be Stein. It is a closed complex submanifold of $\widetilde X$; closed complex submanifolds of Stein spaces are Stein. Thus $\widetilde X$ cannot be Stein either.

Largeness excludes compact positive-dimensional analytic subvarieties in this cover. Remmert reduction then makes non-Steinness a failure of holomorphic convexity as well. The contradiction uses the complex dimension of the lifted surface, not an incorrect claim that a Stein fourfold cannot have third homology.

5. Why the advance matters

Importance: 5/5. If correct, these constructions would disprove a central conjecture about universal covers of projective varieties, including its especially direct large-fundamental-group version.

The surface example locates failure in the smallest possible complex dimension beyond curves. The fourfold example would show that removing all compact positive-dimensional analytic subvarieties is still insufficient. These are complementary lessons about the role of topology in complex analysis.

The affirmative linear theory would remain intact. Instead, the examples would identify behavior that faithful finite-dimensional matrix representations cannot capture. In particular, the fundamental groups of the claimed examples cannot be linear over $\mathbb C$, by the established linear theorem.

6. The main idea

M2: leave enough group information after compactification

M2 begins with a family of spheres carrying 48 moving punctures. Near a distinguished parameter, the punctures collide in 24 pairs. After modifications, the special fiber consists of a main sphere with attached rational components. Suitable finite covers preserve the rational nature of those components.

The hard part is making the incidence loops survive in the final compact surface. Filling missing fibers and resolving singularities introduce group relations, and sufficiently many relations can collapse an initially infinite group to a finite one.

The manuscript uses a weighted Golod–Shafarevich method. In a presentation by generators and relations, different relations receive different costs according to their depth in a filtration. A strict inequality leaves room for additional high-depth relations while ensuring the resulting quotient stays infinite. The construction is carried out using inverse limits of finite groups of order a power of two.

After compactification, it chooses a finite quotient that detects each entire local boundary-group image. This is designed to preserve the infinite quotient through normalization and resolution, including the exceptional divisors. The final transfer places that infinite image on the compact rational special fiber.

Once the geometry exists, the analytic obstruction is short: holomorphic functions are constant on each compact rational component by the maximum principle, and intersections force all those constants to agree along the noncompact connected lift.

M1: compress an abelian surface’s loops to one infinite direction

M1 starts from a genus-two curve with simple Jacobian. Arithmetic complex-hyperbolic arrangements impose relations that reduce its group image to a cyclic subgroup. Showing the image is cyclic gives only an upper bound: it might still be finite or trivial.

To establish infinitude, the manuscript uses compatible local covers with path labels modeled on an infinite dihedral group. Its two involutions have a product of infinite order. A local criterion using hyperbolic path graphs is intended to detect every nonzero power after the global geometric identifications.

The labels need not form a representation of the full final fundamental group. That distinction matters because the construction is meant to lie outside the affirmative linear setting.

Two copies of the construction, followed by quotients and projective modifications, produce the required simple abelian surface with cyclic image. A complete intersection in a product with an additional abelian variety then gives a fourfold with large fundamental group, preserving that surface and its group map. The rank-three covering obstruction supplies the final analytic conclusion.

7. Scope, limitations, and open questions

These are existence claims over $\mathbb C$. They do not classify all projective varieties with nonconvex universal covers or provide a criterion that decides holomorphic convexity from arbitrary equations.

The two examples have different dimensions and properties. M2 contains rational curves and is not large; M1 is a fourfold and excludes compact positive-dimensional analytic subvarieties upstairs. The family does not claim a large-fundamental-group surface counterexample.

M2 also discusses consequences using separate claims about special compact Kähler manifolds and semialgebraic universal covers, covered in family 057 and family 058. Those comparisons are additional dependencies, not needed to formulate the basic infinite-chain obstruction.

Questions about which non-linear fundamental groups still force convexity, and what further geometric hypotheses ensure Steinness in the large case, remain meaningful after the stated counterexamples.

8. Sources and further reading

M1’s Theorem 1.1 states the fourfold and abelian-surface construction; its opening reduction explains the homological obstruction. M2’s Theorem 1.1 states the surface example, and Lemma 1.2 explains why its lifted nodal curve destroys holomorphic convexity.

The primary historical papers linked above distinguish established affirmative theorems from earlier conditional constructions. Background on Stein manifolds and covering spaces is useful, but the hull definition and the two concrete obstructions are enough to understand what the family claims.

Source manuscripts

• M1. OpenAI, A projective fourfold with large fundamental group and non-Stein universal cover (October 5, 2026).
• M2. OpenAI, A surface counterexample to Shafarevich holomorphic convexity (September 23, 2026).

Sources consulted on 12 October 2026 at repository revision fd4aeeb. Download the manuscript citations.