Born Structure in Parapquaternionic Geometry

• Born structure is a geometric compatibility structure on 2n-dimensional spaces that unifies complex, para-complex, metric, and symplectic data into a para-quaternionic framework.
• It refines almost Hermitian and para-Hermitian geometries by employing three mutually compatible endomorphisms and bilinear forms, and is pivotal in doubled geometry, tangent bundles, and Lie algebra models.
• The integrability of Born structures, marked by a unique, torsionless connection, links generalized geometry with Hessian and pseudo-Kähler frameworks, offering a unified method to recover physical metrics.

Born structure is a geometric compatibility structure on a $2n$-dimensional manifold or, in algebraic models, on a Lie algebra, in which complex, para-complex, metric, and symplectic-type data are combined into a single para-quaternionic package. In the formulations developed for doubled geometry, tangent bundles, and Lie algebras, it is explicitly stronger than an almost Hermitian or almost para-Hermitian structure, because it requires three mutually compatible endomorphisms together with compatible bilinear forms rather than a single complex or para-complex datum (Sakamoto, 31 Jul 2025, Freidel et al., 2018). In string-theoretic applications, Born structure is the natural enhancement of para-Hermitian geometry on doubled target spaces; in differential geometry it appears canonically on tangent bundles built from an affine connection and a metric; and in Lie theory it organizes pseudo-Kähler, bi-Lagrangian, and complex-product structures within one framework (Svoboda et al., 2019, Gil-García et al., 2024).

1. Algebraic definition and equivalent formalisms

A standard manifold-level definition begins with an almost para-quaternionic triple of $(1,1)$-tensor fields

$(I,J,K)$

satisfying

$$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$

An almost Born structure on a $2n$-dimensional manifold $N$ is then a tuple

$(I,J,K,h,k,\omega)$

in which $h$ is a Riemannian metric, $k$ is a pseudo-Riemannian metric of signature $(n,n)$, and $(1,1)$0 is a non-degenerate $(1,1)$1-form, constrained by

$(1,1)$2

These identities mean that the three endomorphisms and the three bilinear objects determine one another rather than constituting independent fields (Sakamoto, 31 Jul 2025).

A second, widely used formulation, especially in doubled geometry, starts from a triple

$(1,1)$3

where $(1,1)$4 is a split-signature metric of type $(1,1)$5, $(1,1)$6 is an almost symplectic $(1,1)$7-form, and $(1,1)$8 is a positive-definite generalized metric. Compatibility is imposed by

$(1,1)$9

The associated endomorphisms are

$(I,J,K)$0

and satisfy the para-quaternionic relations

$(I,J,K)$1

This is the formulation used in Born geometry for doubled spaces (Freidel et al., 2018).

A Lie-algebraic formulation replaces the manifold tensors by bilinear data $(I,J,K)$2 together with recursion operators $(I,J,K)$3 satisfying

$(I,J,K)$4

In that setting, the $(I,J,K)$5-eigenspaces of $(I,J,K)$6 are Lagrangian for $(I,J,K)$7, null for $(I,J,K)$8, and orthogonal for $(I,J,K)$9, so the structure simultaneously encodes neutral-metric and pseudo-Kähler behavior (Gil-García et al., 2024).

Setting	Basic data	Characteristic relations
Manifold / tangent bundle	$$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$0	$$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$1, $$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$2, $$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$3, $$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$4, $$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$5
Doubled-space Born geometry	$$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$6	$$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$7, $$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$8
Lie-algebraic form	$$-I^2 = J^2 = K^2 = \mathrm{id}, \qquad IJK = -\mathrm{id}.$$9 with $2n$0	$2n$1, $2n$2

These notations are distinct but closely aligned. This suggests a common underlying pattern: one neutral metric, one positive or pseudo-Riemannian metric, one nondegenerate $2n$3-form, and an induced para-quaternionic triple.

2. Relation to Hermitian, para-Hermitian, and doubled geometry

Born structure strictly refines the more familiar Hermitian and para-Hermitian geometries. In the manifold formulation, $2n$4 is an almost Hermitian structure, while the remaining pairs encode para-complex analogues. In the doubled-geometry formulation, the structure decomposes into three mutually compatible viewpoints:

1. para-Hermitian $2n$5,
2. chiral $2n$6,
3. Hermitian $2n$7 (Sakamoto, 31 Jul 2025, Freidel et al., 2018).

In string-theoretic applications, the underlying space is a doubled target space $2n$8 of dimension $2n$9, with local coordinates

$N$0

The para-Hermitian datum is a split-signature metric $N$1 and an almost para-complex structure $N$2 satisfying

$N$3

Its $N$4 eigenspaces define a splitting

$N$5

and the canonical $N$6-form is

$N$7

When one adds the generalized metric $N$8, the para-Hermitian kinematics become a full Born geometry, and the doubled space acquires the data needed to encode physical background fields and T-duality-covariant dynamics (Svoboda et al., 2019).

A central feature is recovery of physical spacetime from an integrable polarization. If $N$9 is integrable, its leaves define an $(I,J,K,h,k,\omega)$0-dimensional manifold $(I,J,K,h,k,\omega)$1, and the doubled tangent bundle reduces to the generalized tangent bundle of the physical leaf space: $(I,J,K,h,k,\omega)$2 In an untwisted splitting, the generalized metric has block form

$(I,J,K,h,k,\omega)$3

so the ordinary spacetime metric $(I,J,K,h,k,\omega)$4 is recovered on the physical sector. Under a $(I,J,K,h,k,\omega)$5-transformation,

$(I,J,K,h,k,\omega)$6

which packages the metric and Kalb–Ramond $(I,J,K,h,k,\omega)$7-form in the standard generalized-geometric fashion (Svoboda et al., 2019).

The compatibility of $(I,J,K,h,k,\omega)$8, $(I,J,K,h,k,\omega)$9, and $h$0 also reduces the structure group: $h$1 This is one reason Born geometry is presented as a unification of generalized geometry, doubled string geometry, and ordinary Riemannian geometry (Freidel et al., 2018).

3. Canonical connection and generalized torsion

A major structural result is that a Born structure carries a canonical connection analogous to the Levi-Civita connection of Riemannian geometry. In generalized or doubled geometry based only on $h$2, there are many generalized connections that preserve the available structures and are torsionless in the generalized sense. The addition of $h$3 removes this ambiguity: for a Born geometry $h$4, there exists a unique connection $h$5 such that

$h$6

and whose generalized torsion vanishes (Freidel et al., 2018).

The generalized torsion of a connection $h$7 is defined by

$h$8

where $h$9 is the canonical D-bracket of the para-Hermitian structure. The torsionless condition is therefore adapted to doubled geometry rather than to the ordinary Lie bracket (Freidel et al., 2018).

Using the chiral projectors

$k$0

the Born connection is given explicitly by

$k$1

This formula shows how the connection mixes opposite chiralities and corrects same-chirality sectors through the para-complex structure $k$2. In the doubled-space interpretation, that makes the Born connection the direct analogue of Levi-Civita for a geometry that is simultaneously metric, symplectic-type, and chiral (Freidel et al., 2018).

The connection preserves the full Born structure because it preserves $k$3, $k$4, and $k$5, and hence also $k$6 and $k$7. Moreover, when the physical subbundle $k$8 is integrable, the projection of the Born connection to $k$9 agrees with the Levi-Civita connection of the induced metric $(n,n)$0: $(n,n)$1 This establishes Born geometry as an extension of ordinary Riemannian geometry rather than a replacement for it (Freidel et al., 2018).

4. Integrability and the tangent bundle of a Hessian manifold

A distinct but closely related development concerns induced Born structures on tangent bundles. For an $(n,n)$2-dimensional manifold $(n,n)$3 with affine connection $(n,n)$4 and Riemannian metric $(n,n)$5, Marotta and Szabo showed that one can construct an almost Born structure on $(n,n)$6. The 2025 analysis of tangent bundles of Hessian manifolds sharpens that construction by identifying precise equivalences between the geometry of $(n,n)$7 on $(n,n)$8 and integrability of the induced Born structure on $(n,n)$9 (Sakamoto, 31 Jul 2025).

In local coordinates $(1,1)$00 adapted to a chart on $(1,1)$01, with Christoffel symbols $(1,1)$02, the tangent bundle carries the adapted frame

$(1,1)$03

with dual coframe

$(1,1)$04

The para-quaternionic endomorphisms are then

$(1,1)$05

and in the adapted frame take the standard matrices

$(1,1)$06

The lifted tensors determined by $(1,1)$07 are

$(1,1)$08

$(1,1)$09

$(1,1)$10

Thus $(1,1)$11 is an almost Born structure on $(1,1)$12 (Sakamoto, 31 Jul 2025).

The integrability notions are twofold. An almost Born structure is integrable if $(1,1)$13, $(1,1)$14, and $(1,1)$15 are individually integrable and $(1,1)$16. It is strongly integrable if the entire para-quaternionic triple is simultaneously integrable and $(1,1)$17. For general Born structures the second condition is strictly stronger, but for the tangent-bundle structures induced from $(1,1)$18 these notions coincide exactly with Hessian geometry (Sakamoto, 31 Jul 2025).

The fundamental equivalences are: $(1,1)$19

$(1,1)$20

$(1,1)$21

where the dual connection $(1,1)$22 is defined by

$(1,1)$23

A pair $(1,1)$24 is Hessian if $(1,1)$25 is flat, torsion-free, and $(1,1)$26 is totally symmetric. The central theorem is then

$(1,1)$27

This identifies Born geometry on $(1,1)$28 as the tangent-bundle manifestation of Hessian geometry (Sakamoto, 31 Jul 2025).

The flat model is $(1,1)$29 with its standard affine connection and Euclidean metric. Under

$(1,1)$30

the induced structure becomes the standard strongly integrable Born structure on $(1,1)$31 (Sakamoto, 31 Jul 2025).

5. Born Lie algebras and low-dimensional classification

The Lie-algebraic theory of Born structures translates the differential-geometric compatibility conditions into purely algebraic form. In the formulation adopted from Hamilton–Kotschick–Pilatus, an integrable Born structure on a Lie algebra $(1,1)$32 is specified by $(1,1)$33 together with operators $(1,1)$34 satisfying the para-quaternionic relations, with integrability defined by three conditions:

• $(1,1)$35 in the Chevalley–Eilenberg complex,
• $(1,1)$36 has vanishing Nijenhuis tensor,
• the eigenspaces $(1,1)$37 of $(1,1)$38 are Lie subalgebras (Gil-García et al., 2024).

An equivalent characterization is especially useful: a Born structure on a Lie algebra is integrable if and only if the $(1,1)$39-eigenspaces are subalgebras and $(1,1)$40 is parallel with respect to the Levi-Civita connection of $(1,1)$41. Accordingly, an integrable Born Lie algebra may be viewed as a pseudo-Kähler Lie algebra $(1,1)$42 equipped with complementary $(1,1)$43-orthogonal subalgebras $(1,1)$44 exchanged by $(1,1)$45. This makes explicit its simultaneous relation to pseudo-Kähler geometry, bi-Lagrangian geometry, and complex-product geometry (Gil-García et al., 2024).

A central structural theorem states that every Born Lie algebra arises from a bicross product. One starts with Lie algebras $(1,1)$46 and $(1,1)$47 of the same dimension, representations

$(1,1)$48

satisfying the matched-pair compatibility equations, and an isometry

$(1,1)$49

The resulting bicross product $(1,1)$50 carries an associated Born structure, and conversely every Born Lie algebra is isomorphic to one obtained in this way from its two $(1,1)$51-eigensubalgebras (Gil-García et al., 2024).

The classification results presently available are explicit. In dimension $(1,1)$52, every $(1,1)$53-dimensional Lie algebra is Born, including $(1,1)$54 and $(1,1)$55. In dimension $(1,1)$56, the non-abelian Lie algebras admitting an integrable Born structure are

$(1,1)$57

A corollary is that a $(1,1)$58-dimensional Lie algebra admits an integrable Born structure if and only if it admits both a pseudo-Kähler structure and a bi-Lagrangian structure (Gil-García et al., 2024).

For non-abelian $(1,1)$59-dimensional nilpotent Lie algebras, the admissible cases are

$(1,1)$60

with shorthand

$(1,1)$61

$(1,1)$62

$(1,1)$63

The classifications are derived by reducing to matched pairs of type $(1,1)$64 or $(1,1)$65 (Gil-García et al., 2024).

The curvature analysis shows a striking asymmetry between the two Born metrics. In all classified $(1,1)$66-dimensional examples, the neutral metric $(1,1)$67 is flat, while the pseudo-Kähler metric $(1,1)$68 may be flat, Einstein, Ricci-soliton, or non-flat. In the $(1,1)$69-dimensional nilpotent case, most examples again have flat $(1,1)$70, but $(1,1)$71 allows both $(1,1)$72 and $(1,1)$73 to be non-flat for generic parameters (Gil-García et al., 2024).

6. Integrability, scope, and conceptual significance

Several recurring themes organize the modern theory of Born structures. First, Born structure is not merely an alternative notation for almost Hermitian or para-Hermitian geometry. The defining feature is the simultaneous presence of three compatible endomorphisms and three compatible bilinear objects; each of the familiar geometries appears only as a partial face of the full structure (Sakamoto, 31 Jul 2025, Freidel et al., 2018).

Second, integrability is genuinely multi-layered. In general, one may distinguish integrability of the individual tensors $(1,1)$74, $(1,1)$75, and $(1,1)$76 from simultaneous integrability of the entire para-quaternionic triple. The tangent-bundle theorem for Hessian manifolds is therefore structurally significant because it identifies a setting in which these notions collapse to a single condition. In that case, Hessian geometry on the base manifold is exactly the condition required for Born integrability upstairs on the tangent bundle (Sakamoto, 31 Jul 2025).

Third, the structure has a clear unifying role across otherwise separate domains. In doubled string geometry it provides the minimal package compatible with T-duality, physical spacetime recovery, generalized metric data, and flux twisting. In tangent-bundle geometry it is induced canonically from $(1,1)$77 and detects Hessianity. In Lie theory it organizes pseudo-Kähler and bi-Lagrangian data and admits constructive classification through bicross products (Svoboda et al., 2019, Gil-García et al., 2024).

A plausible implication is that Born structure should be viewed less as a single isolated object than as a compatibility principle. The recurring pattern is that one begins with two complementary geometric sectors—horizontal and vertical, physical and dual, or complementary Lie subalgebras—and then imposes a third datum that locks them into a para-quaternionic whole. In the doubled-space setting this extra datum resolves the connection ambiguity of double field theory by selecting the unique torsionless compatible Born connection; in the tangent-bundle setting it turns Hessian geometry into an integrable Born structure; and in the Lie-algebraic setting it turns matched pairs into explicit algebraic models (Freidel et al., 2018, Sakamoto, 31 Jul 2025, Gil-García et al., 2024).