Papers
Topics
Authors
Recent
Search
2000 character limit reached

Joyce Hypercomplex Manifolds

Updated 10 July 2026
  • Joyce hypercomplex manifolds are compact homogeneous spaces G/L endowed with invariant hypercomplex structures defined by integrable complex operators I, J, and K using strongly orthogonal roots.
  • They feature naturally reductive hyper-Hermitian metrics that classify invariant HKT metrics, where layerwise decomposition and representation theory govern the geometric framework.
  • These structures interconnect HKT-Einstein conditions, Obata holonomy reductions, and twistor constructions, linking Lie group geometry with moduli space and Donaldson–Thomas theory applications.

Searching arXiv for recent and foundational papers on Joyce hypercomplex manifolds. Joyce hypercomplex manifolds are compact homogeneous spaces M=G/LM=G/L endowed with GG-invariant hypercomplex structures arising from Joyce’s root-theoretic construction, together with the metric and holonomy structures naturally attached to them. In the homogeneous setting treated in recent work, they are quotients M=G/LM=G/L of a compact connected Lie group GG by a closed subgroup LL, equipped with integrable complex structures I,JI,J satisfying I2=J2=idI^2=J^2=-\mathrm{id} and IJ=JI=:KIJ=-JI=:K, so that one obtains the quaternionic $2$-sphere H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\} (Bedulli et al., 24 Apr 2026). Their significance lies in the fact that they furnish an exact class of compact homogeneous hypercomplex manifolds, admit a complete classification of invariant hypercomplex and HKT metrics, and connect representation theory, Obata and Bismut holonomy, strong HKT geometry, and twistor constructions in Donaldson–Thomas theory (Bedulli et al., 24 Apr 2026, Bedulli et al., 2010, Bridgeland, 2024).

1. Homogeneous definition and Joyce’s root-theoretic construction

In the homogeneous setting, the basic objects are compact homogeneous spaces

GG0

with GG1 a compact connected Lie group, GG2 a closed subgroup, and a GG3-invariant hypercomplex structure (Bedulli et al., 24 Apr 2026). A hypercomplex structure on a GG4-dimensional manifold is given by integrable complex structures GG5 with

GG6

which generate the quaternionic family GG7 of complex structures (Bedulli et al., 24 Apr 2026). A hyperhermitian metric GG8 is Hermitian with respect to every GG9, and the associated M=G/LM=G/L0-form with respect to M=G/LM=G/L1 is

M=G/LM=G/L2

(Bedulli et al., 24 Apr 2026).

The homogeneous hypercomplex structures considered are exactly those produced by Joyce’s construction and classified by Dimitrov–Tsanov (Bedulli et al., 24 Apr 2026). One fixes a maximal torus in M=G/LM=G/L3, a maximal set of strongly orthogonal roots M=G/LM=G/L4, and obtains a decomposition of the Lie algebra

M=G/LM=G/L5

where each M=G/LM=G/L6, each M=G/LM=G/L7 is built from root spaces in the corresponding layer, and M=G/LM=G/L8 centralizes all M=G/LM=G/L9 (Bedulli et al., 24 Apr 2026). Under the condition

GG0

the tangent module GG1 decomposes as

GG2

and the hypercomplex structure is defined layerwise (Bedulli et al., 24 Apr 2026).

On the GG3-plane spanned by GG4, with GG5 satisfying the GG6 relations

GG7

one sets

GG8

and on GG9,

LL0

(Bedulli et al., 24 Apr 2026). This algebraic construction yields the invariant hypercomplex structure, and when LL1 is connected it integrates to a LL2-invariant hypercomplex structure on LL3 (Bedulli et al., 24 Apr 2026).

A foundational classification result states that any invariant hypercomplex structure on a compact homogeneous space LL4 admitting a hyper-Hermitian naturally reductive invariant metric is obtained via Joyce’s construction (Bedulli et al., 2010). More precisely, if LL5 is compact homogeneous, LL6 is a LL7-invariant hypercomplex structure, and LL8 is a LL9-invariant hyper-Hermitian naturally reductive metric, then I,JI,J0 is exactly the one obtained by Joyce’s construction; if I,JI,J1 is semisimple, every simple factor of I,JI,J2 is of type I,JI,J3 (Bedulli et al., 2010). This identifies Joyce hypercomplex manifolds, under the natural reductiveness hypothesis, with the full compact homogeneous hypercomplex class in that setting.

2. Classification of invariant HKT metrics

For a hyperhermitian structure I,JI,J4, the relevant metric condition is the HKT condition. Let I,JI,J5 be the Bismut connection of I,JI,J6. The structure is HKT when the three Bismut connections of I,JI,J7, I,JI,J8, and I,JI,J9 coincide (Bedulli et al., 24 Apr 2026). A fundamental characterization used in the homogeneous theory is

I2=J2=idI^2=J^2=-\mathrm{id}0

where I2=J2=idI^2=J^2=-\mathrm{id}1 is the Dolbeault operator with respect to I2=J2=idI^2=J^2=-\mathrm{id}2 (Bedulli et al., 24 Apr 2026). In the Lie algebra formulation this becomes an algebraic condition on I2=J2=idI^2=J^2=-\mathrm{id}3 (Bedulli et al., 24 Apr 2026).

The decisive metric classification theorem for Joyce hypercomplex manifolds gives a complete description of all invariant HKT metrics (Bedulli et al., 24 Apr 2026). There is a distinguished I2=J2=idI^2=J^2=-\mathrm{id}4-invariant hyperhermitian scalar product I2=J2=idI^2=J^2=-\mathrm{id}5 on

I2=J2=idI^2=J^2=-\mathrm{id}6

such that each layer I2=J2=idI^2=J^2=-\mathrm{id}7 is orthogonal, with

I2=J2=idI^2=J^2=-\mathrm{id}8

(Bedulli et al., 24 Apr 2026). The HKT condition forces strong restrictions: on each layer I2=J2=idI^2=J^2=-\mathrm{id}9, the restriction of any invariant HKT metric must be a scalar multiple of IJ=JI=:KIJ=-JI=:K0, and different layers are orthogonal (Bedulli et al., 24 Apr 2026).

Accordingly, every invariant HKT metric has the form

IJ=JI=:KIJ=-JI=:K1

and, conversely, if IJ=JI=:KIJ=-JI=:K2 is connected, every such choice defines a IJ=JI=:KIJ=-JI=:K3-invariant HKT metric (Bedulli et al., 24 Apr 2026). Thus the invariant HKT cone is an IJ=JI=:KIJ=-JI=:K4-dimensional positive cone parametrized by the Joyce layers. This result is representation-theoretic: the root decomposition and the strongly orthogonal roots completely control the HKT geometry (Bedulli et al., 24 Apr 2026).

The connection with naturally reductive geometry is also structural. On a homogeneous Hermitian manifold, if the invariant metric is naturally reductive,

IJ=JI=:KIJ=-JI=:K5

then the canonical homogeneous connection has torsion

IJ=JI=:KIJ=-JI=:K6

its torsion IJ=JI=:KIJ=-JI=:K7-form is

IJ=JI=:KIJ=-JI=:K8

and this canonical connection coincides with the Bismut connection (Bedulli et al., 24 Apr 2026). However, on Joyce hypercomplex manifolds the canonical connection is not necessarily HKT for an arbitrary hyperhermitian metric; the classification above isolates exactly those invariant metrics for which the HKT condition holds (Bedulli et al., 24 Apr 2026).

3. HKT-Einstein geometry

The HKT-Einstein condition used in the homogeneous theory is defined via the Chern connection rather than the Levi-Civita or Bismut Ricci tensor (Bedulli et al., 24 Apr 2026). For a Hermitian structure IJ=JI=:KIJ=-JI=:K9, let $2$0 be its Chern connection and $2$1 the Chern-Ricci form. An HKT metric is HKT-Einstein if there exists a smooth function $2$2 such that

$2$3

On compact manifolds, $2$4 must be a constant $2$5, and $2$6 is equivalent to the structure being balanced and the canonical bundle $2$7 being holomorphically trivial (Bedulli et al., 24 Apr 2026).

For a Joyce hypercomplex manifold, the Chern-Ricci form can be computed explicitly from the root system. Writing

$2$8

one has

$2$9

and H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}0 vanishes on H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}1 (Bedulli et al., 24 Apr 2026). Evaluating the HKT-Einstein condition on the Joyce layers shows that it is equivalent to the existence of H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}2 with

H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}3

Up to overall scaling, this forces

H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}4

(Bedulli et al., 24 Apr 2026).

These coefficients are positive and admit the root-theoretic expression

H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}5

(Bedulli et al., 24 Apr 2026). Consequently, every Joyce hypercomplex manifold admits a unique invariant HKT-Einstein metric up to scale, at least in the simply connected case (Bedulli et al., 24 Apr 2026). This metric is the canonical representative in the invariant HKT cone.

The relation to other Einstein notions is limited. The HKT-Einstein condition is imposed on the H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}6-antiinvariant part of the Chern-Ricci form, not on the Levi-Civita Ricci tensor (Bedulli et al., 24 Apr 2026). Hyperkähler metrics form a special case where the Chern-Ricci form vanishes and H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}7, but in general HKT-Einstein metrics need not be Levi-Civita Einstein (Bedulli et al., 24 Apr 2026).

4. Strong HKT metrics, Bismut parallelism, and rigidity

A HKT structure is called strong HKT if one, hence all, of the Hermitian structures H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}8 is SKT: H={aI+bJ+cK:a2+b2+c2=1}\mathsf H=\{aI+bJ+cK:a^2+b^2+c^2=1\}9 (Bedulli et al., 24 Apr 2026). On Joyce hypercomplex manifolds, strong HKT geometry is highly restrictive.

For invariant HKT metrics

GG00

the condition GG01 holds if and only if

GG02

(Bedulli et al., 24 Apr 2026). Equivalently, the restriction of GG03 to the root part GG04 must come from an GG05-invariant inner product on GG06 (Bedulli et al., 24 Apr 2026). Under the same hypothesis, the Bismut curvature is also parallel: GG07 Such manifolds are Bismut–Ambrose–Singer manifolds (Bedulli et al., 24 Apr 2026).

Strong HKT implies these parallelism properties automatically. If GG08 is a Joyce hypercomplex manifold with an invariant strong HKT metric GG09, then

GG10

(Bedulli et al., 24 Apr 2026). Moreover, if

GG11

with each GG12 compact, connected, simple and GG13 closed and connected, then each GG14 must be trivial, so GG15 itself is a compact Lie group and GG16 is left-invariant (Bedulli et al., 24 Apr 2026).

This leads to a sharp classification statement: if a simply connected Joyce hypercomplex manifold carries an invariant strong HKT metric, then it is isomorphic to a product

GG17

equipped with a bi-invariant hyperhermitian metric (Bedulli et al., 24 Apr 2026). In particular, no nontrivial fibrations GG18 with GG19 admit strong invariant HKT metrics (Bedulli et al., 24 Apr 2026).

A related group-manifold phenomenon appears at the level of explicit examples. When GG20, so that GG21 is a compact Lie group with a Joyce hypercomplex structure, the reference metric GG22 is bi-invariant and naturally reductive for simple factors of types GG23; in these cases it is strong HKT and Bismut-parallel (Bedulli et al., 24 Apr 2026). For GG24, the invariant HKT-Einstein metric coincides with the one previously constructed by modifying the Killing form (Bedulli et al., 24 Apr 2026).

5. Obata holonomy and twisted Calabi–Yau phenomena

The Obata connection is the unique torsion-free connection preserving the hypercomplex structure (Brienza et al., 9 Sep 2025). On a Lie group with a left-invariant hypercomplex structure, it is given by

GG25

for left-invariant vector fields (Brienza et al., 9 Sep 2025). Its holonomy lies in GG26, and the restricted holonomy lies in GG27 precisely under special geometric conditions (Brienza et al., 9 Sep 2025).

For Joyce hypercomplex group manifolds, a recent holonomy analysis shows that maximal Obata holonomy is not generic (Brienza et al., 9 Sep 2025). If in the Joyce decomposition

GG28

there exists an index GG29 with GG30, then the quaternionic subspace GG31 is parallel under the Obata connection, and the holonomy is strictly smaller than GG32 (Brienza et al., 9 Sep 2025). This yields a general reduction theorem: for all compact Lie groups in Joyce’s list except GG33 and the flat Hopf surface case, every left-invariant Joyce hypercomplex structure has

GG34

(Brienza et al., 9 Sep 2025).

The case GG35 is exceptional. For GG36, Soldatenkov proved

GG37

and there is now a higher-dimensional full-holonomy example on GG38 with

GG39

for a specific Joyce hypercomplex structure (Brienza et al., 9 Sep 2025). At the same time, for every GG40 there exist infinitely many Joyce hypercomplex structures on GG41 with reduced Obata holonomy, produced by choosing the Joyce basis so that a large GG42 is preserved by the hypercomplex structure and by the Obata connection (Brienza et al., 9 Sep 2025).

The restricted holonomy condition GG43 has a different geometric meaning. On Joyce hypercomplex manifolds in the list

GG44

the strong HKT metric built from the Killing form has closed Lee form, and one obtains

GG45

(Brienza et al., 9 Sep 2025). Moreover, the pair GG46, with GG47, solves the twisted Calabi–Yau system on the complex manifold GG48 (Brienza et al., 9 Sep 2025). This produces compact twisted Calabi–Yau manifolds from Joyce hypercomplex geometry.

6. Cohomological and twistor-theoretic extensions of the Joyce framework

The term “Joyce” also appears in a second, distinct but related framework: Joyce structures on complex manifolds arising from Donaldson–Thomas theory. In that setting, a Joyce structure on a complex manifold GG49 consists of a period structure with skew form and a compatible pre-Joyce structure, and it induces a complex hyperkähler structure on the total space GG50 (Bridgeland, 2024, Bridgeland et al., 2020). Here the relevant objects are not compact homogeneous spaces GG51, but tangent bundles of moduli spaces such as spaces of stability conditions or moduli of quadratic differentials (Bridgeland, 2024, Bridgeland, 2022).

The local form of such a Joyce structure is governed by a holomorphic potential GG52 satisfying a heavenly-type equation (Bridgeland et al., 2020). Under the affine symplectic fibration condition, one obtains a complex hyperkähler structure on GG53 with

GG54

and the geometry is characterized locally by a nonlinear PDE for GG55 (Bridgeland et al., 2020). Strong Joyce structures include further conditions: oddness under fibre inversion, homogeneity under an Euler vector field, and lattice periodicity (Bridgeland et al., 2020). This setting provides a twistor-space interpretation of Joyce structures and their associated hyperkähler metrics (Bridgeland, 2024).

Concrete nontrivial examples arise from spaces of quadratic differentials. For the moduli space GG56 of genus GG57 curves with a quadratic differential with simple zeroes, one obtains a meromorphic Joyce structure and hence a complex hyperkähler structure on GG58 (Bridgeland, 2022). The period structure is built from the anti-invariant homology of the spectral curve GG59, and the non-linear connection comes from isomonodromic deformation theory via spectral correspondence and flat connections (Bridgeland, 2022). This suggests a geometric realization of Joyce structures expected from Donaldson–Thomas theory.

A further development constructs Joyce structures from moduli spaces of meromorphic quadratic differentials on the sphere, especially in cases with odd pole orders, again producing complex hyper-Kähler metrics with homothetic symmetry (Moy, 5 Sep 2025). Another explicit family is obtained from deformed polynomial oscillators of odd degree GG60, where the base GG61 is the unfolding of the GG62-singularity and the total space GG63 carries a complex hyper-Kähler metric compatible with an affine symplectic fibration (Dunajski et al., 2024). These are Joyce-type hyper-Kähler geometries rather than homogeneous Joyce manifolds in the Lie-theoretic sense.

A broader cohomological extension concerns hypercomplex manifolds in the sense of Joyce, where quaternionic Dolbeault and Bott–Chern cohomologies admit GG64-invariant and GG65-anti-invariant decompositions (Lejmi et al., 2023). On compact hypercomplex manifolds satisfying the GG66-Lemma, the structure is GG67-pure-and-full (Lejmi et al., 2023). In real dimension GG68, HKT existence on compact GG69-manifolds can be characterized in terms of dimensions of GG70-invariant and Bott–Chern cohomology groups (Lejmi et al., 2023). This cohomological viewpoint is not the homogeneous metric classification of Joyce hypercomplex manifolds, but it extends the Joyce framework into HKT detection, deformation theory, and almost abelian solvmanifolds (Lejmi et al., 2023).

These later developments suggest a bifurcation of terminology. In one usage, Joyce hypercomplex manifolds are the compact homogeneous manifolds GG71 obtained from strongly orthogonal roots and Joyce decompositions (Bedulli et al., 24 Apr 2026, Bedulli et al., 2010). In another, Joyce structures are geometric structures on moduli spaces whose tangent bundles carry complex hyperkähler metrics of “Joyce type” (Bridgeland, 2024, Bridgeland et al., 2020). A plausible implication is that the shared terminology reflects a common structural core: quaternionic or hyperkähler geometry controlled by period data, symmetry, and nonlinear flatness conditions.

Topic to Video (Beta)

No one has generated a video about this topic yet.

Whiteboard

No one has generated a whiteboard explanation for this topic yet.

Follow Topic

Get notified by email when new papers are published related to Joyce Hypercomplex Manifolds.