Family 041 — Hyperkähler fibrations: turning a boundary class into a map

← All digests

Family 041 · Algebraic and complex geometry
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-12

1. The question

A polynomial equation can define a geometric space; a collection of functions on that space can define a map to a simpler one. In complex geometry, line bundles organize the functions that may be used together. A basic difficulty is that a line bundle can have the right numerical properties for such a map while apparently supplying no functions at all.

For hyperkähler manifolds, a particularly important numerical condition predicts a map whose fibers have exactly half the dimension of the original space. The fibers should be Lagrangian: the ambient symplectic form becomes zero when restricted to them. This is the holomorphic version of a fibration picture closely related to mirror symmetry.

Family 041 contains two complementary claims. M2 says that every nonzero nef isotropic line bundle on a compact irreducible hyperkähler manifold actually produces the expected fibration. M1 says that, for a projective Lagrangian fibration in its stated setting, the base is ordinary complex projective space, even if it was initially allowed to be singular.

The first question is whether the map exists on the given manifold. The second is what its target can be.

2. Background and essential definitions

The geometry carried by a hyperkähler manifold

In these manuscripts, the relevant object is a compact irreducible holomorphic symplectic Kähler manifold $X$. It is simply connected and has a holomorphic two-form $\sigma$ that is nondegenerate everywhere; all its holomorphic two-forms are scalar multiples of $\sigma$. Its complex dimension is even, written $2n$.

“Kähler” means the complex structure has a compatible closed positive differential two-form. A hyperkähler metric provides a family of related complex structures on the same underlying real manifold. Changing the complex structure within this family is called hyperkähler rotation. A submanifold that is real and specially positioned in one complex structure can become a complex submanifold in another.

A complex $n$-dimensional submanifold is Lagrangian if the restriction of $\sigma$ is zero. A Lagrangian fibration is a surjective holomorphic map with connected fibers whose components have this half-dimensional property. Smooth fibers in the setting here are abelian varieties, hence complex tori with an algebraic polarization.

Positivity on the boundary

A line bundle $L$ has a first Chern class $c_1(L)$, recording its topological curvature class. It is nef when this class belongs to the closure of the Kähler cone. Thus it is a limit of positive classes, but need not itself be positive in every direction.

Hyperkähler manifolds carry a quadratic form $q_X$ on second cohomology, the Beauville–Bogomolov–Fujiki form. It controls top intersections through

$$\int_X a^{2n}=c_X q_X(a)^n, \qquad c_X>0.$$

A class is isotropic when $q_X(a)=0$. A nonzero nef isotropic class therefore has vanishing top self-intersection, but still has positive mixed intersections through dimension $n$. This is the numerical pattern expected of a class pulled back from an $n$-dimensional base.

A line bundle is semiample if some positive power is generated by global sections everywhere. Such sections give a holomorphic map with no unresolved points of definition. Nefness and isotropy are numerical conditions; semiampleness is a statement about actual holomorphic sections. Bridging that gap is the main existence problem.

For a K3 surface, where $n=1$, the model picture is an elliptic fibration over the projective line. The higher-dimensional question asks for an analogous structure with higher-dimensional torus fibers.

3. Development history and previous partial results

The special Lagrangian fibration picture arose in the mirror-symmetry proposal of Strominger, Yau, and Zaslow. Hyperkähler rotation relates that real-geometric picture to holomorphic Lagrangian fibrations. Hassett–Tschinkel and Sawon formulated early versions involving nef isotropic divisors.

Matsushita established the remarkable rigidity of nontrivial fibrations from projective irreducible symplectic manifolds: the base has half the dimension, and the fibers are Lagrangian. This identifies the geometry once a suitable map exists, but does not manufacture sections of an arbitrary nef isotropic bundle.

Two approaches produced substantial partial progress. Analytic positivity results gave sections or semiampleness under extra metric assumptions. Constructions using moduli spaces of sheaves, stability conditions, and monodromy handled the known deformation types, including manifolds related to K3 surfaces, generalized Kummer varieties, and the two O’Grady examples.

A particularly decisive reduction is the Soldatenkov–Verbitsky deformation theorem: for the fixed isotropic class in the appropriate deformation component, finding one semiample deformation suffices to obtain semiampleness at every nef point. This already establishes the strong conjecture for the known deformation classes. The remaining task is to construct the needed deformation without restricting to those classes.

The base question developed alongside the existence question. Hwang proved that a smooth base of a projective irreducible symplectic fibration is projective space. Greb–Lehn extended the smooth-base statement to compact Kähler total spaces. Other work treated known deformation types, four-dimensional total spaces, and particular classes of base singularities. The unresolved issue in the general projective setting was therefore the possible singularity of the base, rather than a shortage of candidates for a smooth base.

These earlier results are essential inputs, not conclusions first supplied by this family. M2 addresses the missing class-preserving construction; M1 addresses singular bases in arbitrary dimension and deformation type.

4. What the new manuscripts claim

M2: the strong hyperkähler SYZ statement

For every compact irreducible holomorphic symplectic Kähler $2n$-fold $X$, M2 claims

$$c_1(L)\ne0,\quad L\text{ nef},\quad q_X(c_1(L))=0 \quad\Longrightarrow\quad L\text{ semiample}.$$

It asserts this for the original line bundle on the original manifold. No birational replacement is required, and no restriction on deformation type or the second Betti number is added.

More explicitly, there are a positive integer $m$, a normal projective variety $B$, an ample line bundle $A$ on $B$, and a map

$$f:X\longrightarrow B, \qquad L^{\otimes m}\simeq f^*A, \qquad \dim_{\mathbb C}B=n.$$

The map is surjective with connected fibers. Every component of every fiber has dimension $n$, and $\sigma$ vanishes on its smooth part. This includes singular fibers; it does not say all fibers are smooth tori.

M1: the base in the projective setting

M1 starts with a projective surjective holomorphic map from such an $X$ to a normal projective $n$-fold $B$, with connected Lagrangian fibers as specified above. It claims

$$B\simeq\mathbb P^n_{\mathbb C}.$$

Here projective map means a relatively ample line bundle exists. Together with a projective base, this makes the original total space projective. The theorem should therefore not be silently extended to every nonprojective total space covered by M2.

After identifying the base, existing theorems exclude codimension-one multiple fibers: for a prime divisor $D\subset B$, the multiplicities of the components of $f^*D$ have greatest common divisor one. Individual components may still occur with multiplicity greater than one.

A finiteness consequence with an extra hypothesis

M2 combines its semiampleness claim with Engel–Filipazzi–Greer–Mauri–Svaldi’s boundedness theorem to claim finitely many smooth deformation types in each complex dimension among manifolds with

$$b_2(X)\ge5.$$

Here $b_2$ is the dimension of second real cohomology. This restriction belongs to the finiteness deduction, not to the semiampleness theorem. The deduction uses a theorem on indefinite rational quadratic forms to produce an isotropic vector when at least five variables are available.

5. Why the advance matters

Importance: 5/5. The family claims solutions to two central structural problems in hyperkähler geometry, with no limitation to currently familiar examples.

The existence theorem would turn a computable boundary condition in cohomology into a global geometric decomposition. Instead of merely suggesting a hidden family of tori, the line bundle would define that family everywhere. The base theorem would then say that the projective versions all organize their fibers over the simplest possible compact base of the right dimension.

The finiteness consequence is also substantial: constructing examples is very different from knowing that only finitely many deformation types can occur in a fixed dimension. It remains a finiteness statement with its stated Betti-number condition, not an explicit classification or a list of all types.

6. The main idea

Produce one torus while remembering the desired class

M2 follows the deformation reduction, but must preserve the particular isotropic class supplied at the start. It constructs a degeneration whose monodromy remembers that class as a labelled integral vector. Monodromy is the change in cohomological coordinates obtained by moving once around the degeneration parameter.

Near the degeneration, the manuscript rescales Ricci-flat metrics and uses estimates to obtain a smooth flat cylinder in logarithmic coordinates. The cylinder contains an angular torus. A perturbation produces a special Lagrangian torus, on which the Kähler form vanishes and the holomorphic volume form has constant phase.

The torus’s position in the coordinate system matters: it connects the analytic construction back to the prescribed monodromy class. Producing an unrelated torus somewhere on a deformation would not suffice.

Hyperkähler rotation then makes this torus holomorphic and Lagrangian on a nonprojective deformation. An existing Greb–Lehn–Rollenske theorem supplies a fibration there. The manuscript arranges that the relevant rational isotropic direction is the original one, and the Soldatenkov–Verbitsky theorem carries semiampleness back to $(X,L)$.

Remove the base singularities in two stages

M1 first excludes singularities that are not finite quotients of smooth local spaces. Its local tensor analysis associates a hypothetical bad locus with an Euler-characteristic obstruction. Deformations that keep the same base supply incompatible global restrictions. The argument treats small second Betti number separately, including a case requiring a second Lagrangian fibration.

It then eliminates the remaining finite-quotient singularities. The second fibration supplies an abelian variety mapping finitely onto the base. Combining this cover with the original fibration produces translation and ramification constraints incompatible with nontrivial local stabilizers. Once the base is smooth, Hwang’s theorem identifies it with projective space.

M1 uses M2’s semiampleness theorem in constructing the second system. M2’s additional projective-base corollary uses M1 only after its own semiampleness argument. Their primary claims and their subsequent consequences therefore have distinct roles in the exposition.

7. Scope, limitations, and open questions

The line bundle in M2 must have a nonzero integral first Chern class, be nef, and be isotropic. An arbitrary real boundary class need not come from a line bundle, so the theorem does not assign a holomorphic map to every such real class.

The conclusion is semiampleness on a compact irreducible holomorphic symplectic Kähler manifold. It is not a general semiampleness theorem for every variety with trivial canonical bundle. Nor is it the full metric SYZ conjecture for all Calabi–Yau manifolds.

The projective-space conclusion here uses M1’s projectivity hypotheses. The finiteness corollary retains $b_2\ge5$, even though the main existence theorem does not. Neither paper supplies an effective universal power $m$, an explicit list of deformation types, or a classification of singular fibers.

These distinctions leave substantial geometric questions after the stated claims: how to describe the resulting fibrations concretely, determine their degenerations, and turn abstract finiteness into an explicit classification.

8. Sources and further reading

M2’s Theorem 1.1 is the semiampleness claim; Corollaries 1.2 and 1.3 separate the finiteness and projective-base consequences. M1’s Theorem 1.1 states the base theorem and spells out projectivity. Their introductions describe the earlier deformation, fibration, and analytic results used in the arguments.

For background, hyperkähler manifolds connect differential geometry to the holomorphic symplectic language used here. The basic distinction throughout is between a numerical cohomology class and the global sections needed to turn it into a map.

Source manuscripts

• M1. OpenAI, Projective-space bases of Lagrangian fibrations (September 23, 2026).
• M2. OpenAI, The strong hyperkähler SYZ conjecture (September 23, 2026).

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