Hypercomplex Schemes in Twistor Geometry

• Hypercomplex schemes are singular analogues of hypercomplex manifolds defined through a P1-family of equivalence relations that capture quaternionic geometry beyond smooth loci.
• They employ twistor space techniques by replacing differential-geometric tangent actions with analytic and algebraic quotient constructions, ensuring compatibility across singularities.
• This framework unifies methods from gauge theory and singularity analysis, providing a robust language for understanding finite triholomorphic quotients and conical hypercomplex structures.

Searching arXiv for recent and foundational papers on hypercomplex schemes and closely related hypercomplex geometry. Hypercomplex schemes are singular analogues of hypercomplex manifolds in which the quaternionic geometry is encoded not by a smooth tangent-bundle action alone, but by a $\mathbb P^1$-family of equivalence relations compatible with the hypercomplex structure on the regular locus. In the formulation proposed in “Hypercomplex analytic spaces and schemes” (Bielawski, 22 Jul 2025), the basic objective is to construct a theory of singular hypercomplex objects that remains faithful to the twistor-space viewpoint, accommodates quotient constructions by finite triholomorphic group actions, and extends the hypercomplex structure of the smooth locus across singularities in a controlled way. The resulting notion is analytic first and algebraic second: hypercomplex analytic spaces are the primary objects, while hypercomplex schemes arise by replacing analytic spaces and relations with their algebraic counterparts (Bielawski, 22 Jul 2025).

1. From hypercomplex manifolds to singular objects

A hypercomplex manifold is a smooth manifold whose tangent bundle carries a fiberwise action of the quaternions $\mathbb H$, equivalently a triple of anticommuting complex structures $I,J,K$ satisfying

$I^2=J^2=K^2=IJK=-1.$

Its geometry is organized by the $S^2$-family of induced complex structures and by the associated twistor space $Z=M\times S^2$, which becomes a complex manifold with a holomorphic projection $\pi:Z\to\mathbb P^1$ and an antiholomorphic involution covering the antipodal map (Bielawski, 22 Jul 2025). This manifold picture is the smooth background from which hypercomplex schemes emerge.

The motivation for singular hypercomplex objects comes from spaces that are expected to retain hyperkähler or hypercomplex features despite singularities. The paper explicitly points to singular spaces in gauge theory, especially Coulomb branches of $3d\;N=4$ theories, as motivating examples, and argues that Verbitsky’s notion of hypercomplex variety is too rigid for this setting (Bielawski, 22 Jul 2025). This suggests that the singular theory should be organized around the regular locus $Reg(X)$, the canonical sheaf

$\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$

and a twistor description that survives beyond the smooth category (Bielawski, 22 Jul 2025).

A central shift therefore occurs: instead of defining a singular hypercomplex space by local quaternionic endomorphisms of a tangent sheaf, one defines it through a family of equivalence relations indexed by $\mathbb H$0. This replaces direct differential-geometric data by quotient-type data that still recovers the quaternionic geometry on the smooth part. A plausible implication is that the theory is designed less as a direct analogue of smooth hypercomplex differential geometry and more as a singular twistor geometry.

2. Hypercomplex analytic spaces

The paper first defines a weakly hypercomplex real analytic space and then a hypercomplex one. Let $\mathbb H$1 be a real analytic space of pure dimension $\mathbb H$2 with thin singular locus $\mathbb H$3. It is weakly hypercomplex if there exist a reduced complexification $\mathbb H$4 and an analytic $\mathbb H$5-family of equivalence relations $\mathbb H$6 on $\mathbb H$7 such that the following conditions hold (Bielawski, 22 Jul 2025):

1. on $\mathbb H$8, the relations $\mathbb H$9 come from a genuine hypercomplex structure, and $I,J,K$0 is the closure of its restriction from the regular locus;
2. every $I,J,K$1-equivalence class meets $I,J,K$2, and its dimension at such a point is $I,J,K$3;
3. $I,J,K$4;
4. the intersection of all relations is the diagonal,

$I,J,K$5

A weakly hypercomplex space is hypercomplex if, in addition, for every $I,J,K$6, the equivalence classes of $I,J,K$7 are discrete (Bielawski, 22 Jul 2025). The paper interprets this as the singular analogue of transversality between distinct complex structures.

This definition is explicitly twistor-theoretic. The relations $I,J,K$8 should be read as the leaf relations associated with the complex structure indexed by $I,J,K$9. On the regular locus they are induced by an honest hypercomplex structure; across singularities they are extended by closure. The diagonal condition ensures that the full family separates points, while the discreteness condition for pairwise intersections encodes the nondegeneracy of the quaternionic family.

The theory also imposes geometric restrictions. If $I^2=J^2=K^2=IJK=-1.$0 admits a normal complexification satisfying the definition, then

$I^2=J^2=K^2=IJK=-1.$1

This excludes very small singular loci and indicates that hypercomplex singularities are constrained by the twistor-family structure (Bielawski, 22 Jul 2025).

3. Twistor spaces and the section-space description

Given a complexification $I^2=J^2=K^2=IJK=-1.$2 and the family $I^2=J^2=K^2=IJK=-1.$3, the paper forms a relation $I^2=J^2=K^2=IJK=-1.$4 on $I^2=J^2=K^2=IJK=-1.$5 by setting $I^2=J^2=K^2=IJK=-1.$6. When the quotient ringed space

$I^2=J^2=K^2=IJK=-1.$7

is a reduced complex space, it is called a twistor space of $I^2=J^2=K^2=IJK=-1.$8 (Bielawski, 22 Jul 2025). This twistor space carries a holomorphic map

$I^2=J^2=K^2=IJK=-1.$9

and an antiholomorphic involution $S^2$0 covering the antipodal map.

A key structural statement is that if the complexification $S^2$1 is normal, then after shrinking if necessary, $S^2$2 is a complex space, and both $S^2$3 and its fibres are normal (Bielawski, 22 Jul 2025). The proof uses standard complex-analytic quotient theory, including normal equivalence relations and Remmert-type arguments.

The converse viewpoint is equally important. The paper constructs a map

$S^2$4

sending $S^2$5 to the $S^2$6-equivalence class of $S^2$7, and proves that $S^2$8 is surjective with discrete fibres and unbranched away from $S^2$9 (Bielawski, 22 Jul 2025). For each $Z=M\times S^2$0, the induced map

$Z=M\times S^2$1

is a holomorphic section of $Z=M\times S^2$2, and the assignment $Z=M\times S^2$3 embeds $Z=M\times S^2$4 analytically into the real locus $Z=M\times S^2$5 of the Douady space of $Z=M\times S^2$6-invariant sections (Bielawski, 22 Jul 2025).

This section-space realization is one of the paper’s central theorems. Hypercomplex spaces are characterized as real analytic section spaces of their twistor spaces, provided the fibrewise intersection maps satisfy the stated surjectivity, discreteness, and unbranchedness properties. In effect, the singular hypercomplex object is recovered not from a tangent sheaf, but from a real locus inside a space of twistor sections.

4. Proper hypercomplex spaces and the definition of hypercomplex schemes

The paper identifies a particularly well-behaved class. A hypercomplex space $Z=M\times S^2$7 is proper if the quotient map

$Z=M\times S^2$8

is finite (Bielawski, 22 Jul 2025). Locally, every hypercomplex space is proper. If $Z=M\times S^2$9 admits a twistor space and is proper, then $\pi:Z\to\mathbb P^1$0 is a real analytic covering, and $\pi:Z\to\mathbb P^1$1 is identified with the closure of a union of connected components of $\pi:Z\to\mathbb P^1$2 (Bielawski, 22 Jul 2025).

The algebraic definition is then obtained by direct translation. A pure $\pi:Z\to\mathbb P^1$3-dimensional complex space $\pi:Z\to\mathbb P^1$4 is called hypercomplex if it is equipped with a $\pi:Z\to\mathbb P^1$5-family of equivalence relations $\pi:Z\to\mathbb P^1$6 satisfying the following conditions (Bielawski, 22 Jul 2025):

1. for every $\pi:Z\to\mathbb P^1$7, the family

$\pi:Z\to\mathbb P^1$8

is a pure $\pi:Z\to\mathbb P^1$9-dimensional closed complex subspace near $3d\;N=4$0;

1. every $3d\;N=4$1-class is the closure of its intersection with $3d\;N=4$2;
2. $3d\;N=4$3 avoids $3d\;N=4$4, and on $3d\;N=4$5 the relations define a genuine hypercomplex structure;
3. $3d\;N=4$6;
4. for $3d\;N=4$7, equivalence classes of $3d\;N=4$8 are discrete.

For schemes, the prescription is explicit: replace analytic spaces by schemes of finite type over $3d\;N=4$9 or $Reg(X)$0, replace “discrete” by “quasifinite,” and replace analytic equivalence relations by algebraic ones (Bielawski, 22 Jul 2025). A hypercomplex scheme is therefore the algebraic incarnation of the same twistor-family formalism.

The paper also stresses that the algebraic category is more restrictive and does not cover certain important analytic examples, including Taub–NUT modifications (Bielawski, 22 Jul 2025). This suggests that “hypercomplex scheme” is not the universal ambient notion for singular hypercomplex geometry; rather, it is the algebraic subtheory inside a broader analytic framework.

5. Finite triholomorphic quotients and canonical singular hypercomplex spaces

One of the principal results is the construction of a canonical proper hypercomplex space from a finite quotient. Let $Reg(X)$1 be a connected hypercomplex manifold and $Reg(X)$2 a finite group acting triholomorphically on $Reg(X)$3. Then $Reg(X)$4 has a natural complexification $Reg(X)$5, defined as the connected component of the Douady space $Reg(X)$6 containing $Reg(X)$7, where $Reg(X)$8 is the twistor space of $Reg(X)$9 (Bielawski, 22 Jul 2025). The $\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$0-action extends to $\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$1 and commutes with the real structure $\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$2. By Cartan’s theorem, the quotient $\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$3 is a normal complex space, and one defines

$\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$4

The paper proves that $\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$5 is a canonical proper hypercomplex space associated to the finite quotient (Bielawski, 22 Jul 2025).

This construction is subtle because the naive topological quotient need not be the correct singular object. The paper emphasizes that $\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$6 need not coincide with the ordinary quotient as a real analytic space. In the special case where $\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$7 has no elements of order $\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$8, however,

$\Omega_X^{[1]} := i_*\Omega^1_{Reg(X)},$9

as real analytic spaces (Bielawski, 22 Jul 2025). Thus the obstruction is specifically tied to the structure of the finite group action.

The mechanism passes through the quotient twistor space $\mathbb H$00, equipped with its induced projection to $\mathbb H$01 and involution. The associated map

$\mathbb H$02

is a real analytic covering, unbranched away from the singular locus, so the section-space criterion applies and yields the hypercomplex structure (Bielawski, 22 Jul 2025). A plausible implication is that the category of hypercomplex spaces is designed precisely to contain finite triholomorphic quotients that would otherwise fall outside ordinary real-analytic quotient theory.

6. Model examples, cones, and scope of the theory

The basic example is $\mathbb H$03. The paper identifies the semialgebraic quotient with the space of real symmetric $\mathbb H$04 matrices of rank $\mathbb H$05 and nonnegative trace, and shows that its Zariski closure is the variety of all real symmetric $\mathbb H$06 rank $\mathbb H$07 matrices (Bielawski, 22 Jul 2025). This closure is the union of two copies of $\mathbb H$08 meeting at the origin. The associated twistor space is obtained from the twistor space of $\mathbb H$09, namely the total space of $\mathbb H$10, by the fibrewise $\mathbb H$11-action, and becomes the hypersurface

$\mathbb H$12

inside $\mathbb H$13 (Bielawski, 22 Jul 2025).

The real sections are described by explicit quadratic equations, including

$\mathbb H$14

together with

$\mathbb H$15

The resulting real analytic space $\mathbb H$16 is hypercomplex, and the paper shows that topologically it decomposes into two pieces $\mathbb H$17, corresponding to the two quaternionic structures $\mathbb H$18 and $\mathbb H$19, glued at the origin (Bielawski, 22 Jul 2025). This example demonstrates why the naive quotient is not the correct singular hypercomplex space.

The theory also extends to conical constructions. For certain conical complex spaces $\mathbb H$20 with a holomorphic $\mathbb H$21-action and a compatible stratification by hypercomplex manifolds, the paper constructs a hypercomplex space $\mathbb H$22 homeomorphic to two copies of $\mathbb H$23 glued at the cone vertex (Bielawski, 22 Jul 2025). This includes deformations of quaternionic cones and nilpotent cones of complex semisimple Lie algebras. In the nilpotent cone case, the resulting object is an integral proper hypercomplex scheme (Bielawski, 22 Jul 2025).

These examples clarify both the reach and the limits of the theory. Hypercomplex schemes are not merely singular hypercomplex manifolds in the differential-geometric sense. They are twistor-controlled singular spaces, often best understood via closures, section spaces, and quotient relations rather than via local tensor fields. This suggests that the subject sits at the intersection of quaternionic geometry, singularity theory, analytic and algebraic quotients, and gauge-theoretic moduli problems.

7. Position within hypercomplex geometry

The proposed notion is firmly rooted in classical hypercomplex geometry but changes the point of entry. Smooth hypercomplex manifolds are organized by their $\mathbb H$24-family of complex structures and twistor spaces (Bielawski, 22 Jul 2025), and related recent work continues to treat twistor constructions and quaternionic compatibility as central structural tools, whether in the study of ordinary hypercomplex manifolds (Federico et al., 22 Jun 2025) or in generalized hypercomplex geometry (Fino et al., 2024). Hypercomplex schemes inherit this emphasis on the $\mathbb H$25-family, but they do so through equivalence relations rather than smooth endomorphism bundles.

The principal conceptual novelty is therefore not a new smooth hypercomplex structure, but a singular extension of the twistor principle. The regular locus carries the genuine hypercomplex geometry; the singular locus is controlled by closure conditions, transversality encoded as discreteness or quasifiniteness, and section-space realizations inside a twistor quotient (Bielawski, 22 Jul 2025). In this framework, finite triholomorphic quotients and conical singularities become canonical examples rather than pathologies.

A plausible implication is that hypercomplex schemes provide a candidate foundational language for singular spaces expected to carry quaternionic geometry but not smooth hypercomplex structures in the ordinary sense. In the paper’s formulation, the decisive data are the $\mathbb H$26-indexed equivalence relations and the associated twistor space. Hypercomplex schemes are thus best understood as algebraic singular twistor geometries of quaternionic type (Bielawski, 22 Jul 2025).