G-Structures in Differential Geometry

• G-Structures are reductions of frame bundles that endow manifolds with geometric properties defined by a subgroup of GL(n,ℝ) or GL(n,ℂ).
• Intrinsic torsion quantifies the failure of G-structures to be integrable and classifies them into specific torsion classes with measurable geometric implications.
• Applications of G-structures span Riemannian, complex, and symplectic geometries, influencing theories in string theory, supergravity, and deformation studies.

A $G$-structure on a smooth or complex manifold encapsulates the reduction of its frame bundle to a subgroup $G \subset \mathrm{GL}(n, \mathbb{R})$ or $\mathrm{GL}(n, \mathbb{C})$, endowing the manifold with geometric structures modeled on the symmetries and tensor invariants of $G$. This concept provides a foundational, unifying language for diverse geometric conditions such as complex, symplectic, Riemannian, Hermitian, and exceptional holonomy geometries, as well as for formulating and analyzing their integrability, deformation theory, and applications in physics, notably string theory and supergravity.

1. Formal Definition and Examples

Let $M$ be an $n$-dimensional manifold. Its (oriented) frame bundle $F(M) \to M$ is a principal $\mathrm{GL}(n,\mathbb{R})$ (or $\mathrm{GL}(n,\mathbb{C})$) bundle. A $G$-structure is a principal $G \subset \mathrm{GL}(n, \mathbb{R})$0-subbundle $G \subset \mathrm{GL}(n, \mathbb{R})$1, where $G \subset \mathrm{GL}(n, \mathbb{R})$2 is a Lie subgroup of $G \subset \mathrm{GL}(n, \mathbb{R})$3 or $G \subset \mathrm{GL}(n, \mathbb{R})$4, typically closed and of positive dimension (Fiset, 2019). This reduction admits locally defined tensor fields whose pointwise stabilizer under $G \subset \mathrm{GL}(n, \mathbb{R})$5 is precisely $G \subset \mathrm{GL}(n, \mathbb{R})$6. In the holomorphic category, $G \subset \mathrm{GL}(n, \mathbb{R})$7-structures arise from holomorphic principal $G \subset \mathrm{GL}(n, \mathbb{R})$8-bundles $G \subset \mathrm{GL}(n, \mathbb{R})$9 together with $\mathrm{GL}(n, \mathbb{C})$0-equivariant soldering forms $\mathrm{GL}(n, \mathbb{C})$1, with $\mathrm{GL}(n, \mathbb{C})$2 (McKay, 2022).

Canonical geometric structures presented in this formalism include:

• Riemannian structures: $\mathrm{GL}(n, \mathbb{C})$3, determined by a Riemannian metric.
• Almost complex structures: $\mathrm{GL}(n, \mathbb{C})$4 embedded into $\mathrm{GL}(n, \mathbb{C})$5.
• (Almost) symplectic structures: $\mathrm{GL}(n, \mathbb{C})$6.
• Hermitian structures: $\mathrm{GL}(n, \mathbb{C})$7.
• Calabi-Yau structures: $\mathrm{GL}(n, \mathbb{C})$8, characterized by a covariantly constant holomorphic volume form.
• Exceptional holonomy: $\mathrm{GL}(n, \mathbb{C})$9 or $G$0 in dimensions 7 and 8 (Fiset, 2019, Legramandi et al., 2023).
• Null structures: reductions of the Lorentz frame bundle to subgroups fixing a null direction (Papadopoulos, 2018).

2. Intrinsic Torsion, Prolongation, and Integrability

The notion of intrinsic torsion quantifies the obstruction to the existence of a torsion-free (i.e., integrable) $G$1-structure. For $G$2-structures on oriented Riemannian manifolds, fix the Killing-orthogonal splitting $G$3. The Levi-Civita connection form $G$4 decomposes into a $G$5-connection $G$6 and intrinsic torsion $G$7 valued in $G$8: $G$9 where $M$0 is Levi-Civita and $M$1 is the minimal $M$2-connection (Niedzialomski, 2015). Integrability is equivalent to $M$3.

Intrinsic torsion decomposes into irreducible $M$4-modules ("torsion classes"), providing an effective classification scheme. For example, SU(3)-structures admit five torsion classes $M$5 appearing in the decompositions: $M$6 where the vanishing of all $M$7 yields a Calabi–Yau structure (Fiset, 2019). Similar decompositions govern G$M$8- and Spin(7)-structures, with their differential forms and torsion classes encoding integrability.

The prolongation theory for $M$9-structures identifies higher-order invariants and their obstructions to local flatness or geometric realization. The first prolongation is defined as $n$0 for the Spencer operator $n$1. Infinite prolongation signals non-rigidity and arises for certain non-parabolic $n$2-structures (McKay, 2022, Hwang, 2017).

3. Classifying Algebroids, Curvature Theory, and Integrability Criteria

Every regular $n$3-structure with connection is associated to a classifying transitive Lie algebroid $n$4, constructed via the invariants determined by the structure functions of the solder form and connection. The anchor and bracket encode the geometric data: $n$5 with structure equations $n$6 and $n$7 for torsion $n$8 and curvature $n$9. Local or global equivalence of $F(M) \to M$0-structures is characterized by isomorphism of the corresponding algebroids (Fernandes et al., 2021).

Generalized integrability requires the vanishing of a finite tower of generalized "essential curvatures" $F(M) \to M$1, with $F(M) \to M$2 being Spencer cohomology groups. For $F(M) \to M$3-reductive homogeneous models (i.e., models with a Lie algebra grading $F(M) \to M$4), a $F(M) \to M$5-structure $F(M) \to M$6 is locally immersible into the flat model $F(M) \to M$7 if and only if all essential curvatures vanish up to the required order (Santi, 2013).

In particular, classical geometric embedding problems, such as isometric or conformal immersions and the related Gauss–Codazzi–Ricci equations, are recast in this algebroid formalism as vanishing of calibrated Spencer classes.

4. Variants and Extensions: Homogeneous, Null, and Stratified G-Structures

Homogeneous $F(M) \to M$8-structures extend the classical notion by equipping the frame bundle with a compatible $F(M) \to M$9-action, parameterized by a degree morphism $\mathrm{GL}(n,\mathbb{R})$0. They capture contact structures (as homogeneous Sp($\mathrm{GL}(n,\mathbb{R})$1)-structures of degree identity), cosymplectic, almost contact, and contact-metric geometries, with integrability conditions phrased in terms of homogeneous forms or Atiyah algebroid curvature vanishing (Tortorella et al., 2019).

Null $\mathrm{GL}(n,\mathbb{R})$2-structures, relevant in Lorentzian geometry and general relativity, are reductions of the frame bundle to the stabilizer of a null line, generating geometric frameworks such as Robinson structures or supersymmetric null geometries in supergravity. Fundamental forms are built from forms satisfying $\mathrm{GL}(n,\mathbb{R})$3 and $\mathrm{GL}(n,\mathbb{R})$4, with their geometric content tied to null geodesic congruences, induced geometry on null hypersurfaces, and the symmetry algebras interpolating between BMS and Lorentz (Papadopoulos, 2018).

Stratified $\mathrm{GL}(n,\mathbb{R})$5-structures arise in compactifications of M-theory and related settings, where the structure group and associated forms vary across the base manifold, leading to rich decompositions (e.g., SU(4), G$\mathrm{GL}(n,\mathbb{R})$6, SU(3), SU(2) strata in eight or nine dimensions) and intricate correspondence between the algebraic and geometric data (Babalic et al., 2015).

5. Torsion, Connections with Skew Torsion, and Worldsheet/Worldline Physics

Given a $\mathrm{GL}(n,\mathbb{R})$7-structure, metric connections with skew-symmetric torsion compatible with the structure may exist and, depending on the group, may be unique or parameterized by auxiliary data. The Killing–Yano equation with torsion for a fundamental form $\mathrm{GL}(n,\mathbb{R})$8 (e.g., Hermitian 2-form, associative 3-form) requires that $\mathrm{GL}(n,\mathbb{R})$9 is parallel with respect to a metric connection with torsion $\mathrm{GL}(n,\mathbb{C})$0: $\mathrm{GL}(n,\mathbb{C})$1 subject to an extra algebraic-differential constraint on $\mathrm{GL}(n,\mathbb{C})$2. For $\mathrm{GL}(n,\mathbb{C})$3 and $\mathrm{GL}(n,\mathbb{C})$4 structures, there is frequently a finite-dimensional family of possible torsionful connections, classified by the symmetry and cohomology of fundamental forms (Papadopoulos, 2011). Uniqueness is restored in higher symmetry cases (e.g., Sp($\mathrm{GL}(n,\mathbb{C})$5), $\mathrm{GL}(n,\mathbb{C})$6).

In string theory and supersymmetric compactifications, $\mathrm{GL}(n,\mathbb{C})$7-structures determine supersymmetry preservation via their torsion classes, leading to modifications of worldsheet chiral algebras or worldline symmetry operators, and entering the anomaly cancellation and moduli theory for flux backgrounds (Fiset, 2019, Legramandi et al., 2023).

6. Holomorphic G-Structures, Characteristic Classes, and Universal Relations

For complex manifolds, holomorphic $\mathrm{GL}(n,\mathbb{C})$8-structures are equivalent to holomorphic Cartan geometries of type $\mathrm{GL}(n,\mathbb{C})$9 and yield universal constraints on the Dolbeault cohomology classes (e.g., Chern classes) of the manifold. The main result is that the classifying relations among characteristic forms $G$0---obtained via representation theory and prolongation algebras---directly induce universal polynomial relations among the Dolbeault representatives of characteristic classes for manifolds admitting a given holomorphic $G$1-structure. This approach generalizes classical vanishing and identity theorems (e.g., Baum–Bott theorem for holomorphic foliations) without recourse to specific metrics or connections, relying purely on the algebraic invariants of $G$2 (McKay, 2022).

For $G$3-structures of infinite type (i.e., whose prolongations do not terminate), only the finite stages up to given degree are relevant for characteristic classes of bounded degree.

7. Deformations, Moduli, and Generic G-Model Theory

Deformation theory for $G$4-structures is governed by bundle automorphisms and base-point dependent right actions. A deformation via $G$5 produces a valid $G$6-structure if and only if $G$7, the normalizer, for all $G$8 (Bunk, 2014). Under these transformations, one obtains explicit formulae for the induced connections and the change in intrinsic torsion. This framework accommodates conformal rescalings, central deformations, and moduli computation for specific geometric structures, including nearly Kähler and Sasaki–Einstein geometries.

In categorical logic and model theory, a $G$9-structure may be construed as a first-order structure with a $G \subset \mathrm{GL}(n, \mathbb{R})$00-action compatible with the signature, leading to the construction of generic $G \subset \mathrm{GL}(n, \mathbb{R})$01-models via colimits of presheaves of $G \subset \mathrm{GL}(n, \mathbb{R})$02-structures over topological spaces. These models ensure local forcing semantics and genericity for first-order formulas, providing a model-theoretic foundation for equivariant and geometric logic (Padilla et al., 2013).

---

References:

• (Fiset, 2019): "G-structures and Superstrings from the Worldsheet"
• (Niedzialomski, 2015): "Geometry of $G \subset \mathrm{GL}(n, \mathbb{R})$03-Structures via the Intrinsic Torsion"
• (Papadopoulos, 2011): "Killing-Yano equations with torsion, worldline actions and G-structures"
• (Santi, 2013): "A generalized integrability problem for G-Structures"
• (Babalic et al., 2015): "Internal circle uplifts, transversality and stratified G-structures"
• (McKay, 2022): "Characteristic forms of complex Cartan geometries III: G-structures"
• (Tortorella et al., 2019): "Homogeneous G-structures"
• (Fernandes et al., 2021): "The Classifying Lie Algebroid of a Geometric Structure II: G-structures with connection"
• (Bunk, 2014): "A method of deforming G-structures"
• (Legramandi et al., 2023): "G-structures for black hole near-horizon geometries"
• (Padilla et al., 2013): "Sheaves of G-structures and generic G-models"
• (Papadopoulos, 2018): "Geometry and symmetries of null G-structures"
• (Hwang, 2017): "Rational curves and prolongations of G-structures"