Discrete Spin Connection in Quantum Gravity

• Discrete Spin Connection is a framework assigning finite group elements to discrete geometries, enabling parallel transport and encoding curvature through holonomies.
• It appears in various formulations—including loop quantum gravity, twisted geometry, and lattice fermion systems—each reproducing continuum behavior under appropriate limits.
• These constructions offer practical insights for quantized gravity by linking discrete models to traditional continuum theories through clear methodologies for spin transport and curvature representation.

Searching arXiv for recent and foundational papers on discrete spin connection and related formulations. A discrete spin connection is the parallel transport structure assigned to the elementary adjacency relations of a discrete geometry. In the spinor formulation of loop gravity, the group element $g_e\in SU(2)$ built from endpoint spinors is the discrete spin connection along edge $e$: it is the parallel transporter that maps the spinorial data and the associated flux or face normal from the source node to the target node (Livine et al., 2011). In twisted geometry, the corresponding torsionless Levi-Civita connection is concentrated on faces and is determined by the orthogonal part of the triad mismatch (Haggard et al., 2012). In cell-based discrete gravity, the spin connection lives on links $U_\alpha(n)\in SO(d)$ or Spin(d), while in discrete fermion models it appears as edge phases or as diagonal connection data in lattice covariant derivatives (Chamseddine et al., 2021, Debbasch, 2019, Djordjević et al., 11 Jul 2025). Across these settings, curvature is encoded by holonomy around closed loops, plaquettes, or hinges.

1. Parallel transport, holonomy, and curvature

In all of these constructions, the discrete spin connection is a finite version of the continuum spin connection $\omega$, whose holonomy along a path $\gamma$ is

$$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$

The discrete replacements are ordered products of elementary transporters: $G_f=\prod_{e\in f} g_e$ on a graph, $U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$ around a hinge in a triangulation, and

$$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$

around a lattice cell in a fermionic model (Haggard et al., 2012, Livine et al., 2011, Djordjević et al., 11 Jul 2025). Flatness is characterized by trivial loop holonomy, such as $G_f=\mathbb{1}$ on a graph or $e$0 across a face.

Gauge covariance is intrinsic to the notion. In loop gravity, edge holonomies transform as $e$1 under local $e$2 rotations at nodes. In twisted geometry, the across-face holonomy satisfies

$e$3

which is the required transformation law under local $e$4 frame rotations on the two sides of a face. In the cell-based formulation, link variables transform as

$e$5

and in the two-dimensional hyperbolic-lattice setting the discrete construction is covariant under local $e$6 transformations (Livine et al., 2011, Haggard et al., 2012, Chamseddine et al., 2021, Djordjević et al., 11 Jul 2025).

2. Spinorial holonomy–flux variables in loop gravity

Livine–Tambornino, following the Freidel–Speziale spinorial reformulation, assign to each oriented edge $e$7 a source spinor $e$8 at $e$9 and a target spinor $U_\alpha(n)\in SO(d)$0 at $U_\alpha(n)\in SO(d)$1, with dual spinor $U_\alpha(n)\in SO(d)$2, canonical brackets

$U_\alpha(n)\in SO(d)$3

and flux vectors

$U_\alpha(n)\in SO(d)$4

The vector $U_\alpha(n)\in SO(d)$5 is the oriented face normal, or area vector, of the polyhedron at the node (Livine et al., 2011).

The edge constraint is the area-matching condition

$U_\alpha(n)\in SO(d)$6

which generates opposite $U_\alpha(n)\in SO(d)$7 phase rotations on the two spinors and ensures a well-defined normalization for the holonomy. The $U_\alpha(n)\in SO(d)$8 holonomy is the explicit bilinear function

$U_\alpha(n)\in SO(d)$9

With normalized spinors, it transports source data to target data: $\omega$0 and rotates the fluxes as

$\omega$1

The fluxes satisfy the $\omega$2 algebra,

$\omega$3

and symplectic reduction of $\omega$4 by $\omega$5 yields the standard phase space $\omega$6, with

$\omega$7

At each node, the closure condition

$\omega$8

implements local $\omega$9 gauge invariance and implies the existence of a convex polyhedron whose outward face normals are $\gamma$0 and whose face areas are $\gamma$1. Loop holonomies

$\gamma$2

then encode discrete curvature. In the twisted-geometry language referenced by Livine–Tambornino, $\gamma$3 encode face normals and areas, while relative phases between spinors across an edge encode twist or dihedral angles; when shape matching holds, the holonomy packages a Levi-Civita spin connection compatible with polyhedral geometry, and when shapes mismatch but areas match it packages a twisted connection. The same formalism is unitarily equivalent to $\gamma$4, but the generalized Bargmann representation replaces Haar integrals by Gaussian integrals on $\gamma$5, which reorganizes calculations of holonomies and fluxes (Livine et al., 2011).

3. Torsionless Levi-Civita transport in twisted geometry

In the twisted-geometry construction, each 3-simplex of a triangulation is a flat tetrahedron, and adjacent tetrahedra assign the same area to a shared face while generally disagreeing on the induced two-dimensional face metric. Regge geometry is the special case in which the face metrics match completely. The difficulty is that the triad $\gamma$6 is discontinuous across a face, so the torsionless Cartan equation

$\gamma$7

is ill-defined there. The resolution is to thicken the face into a slab and interpolate the triad smoothly across it (Haggard et al., 2012).

Let $\gamma$8 be the triad mismatch across the face. Its polar decomposition is

$\gamma$9

with $$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$0 and $$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$1 symmetric positive-definite. Using the interpolation $$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$2, where $$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$3, the slab-crossing holonomy is

$$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$4

The distributional torsionless spin connection concentrated on the face is

$$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$5

with $$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$6. In coordinates adapted to a face with normal $$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$7, the orthogonal polar part is a rotation in the face plane by an angle $$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$8, so that $$U_\gamma=\mathcal{P}\exp\left(-\int_\gamma \omega\right).$$9 and $G_f=\prod_{e\in f} g_e$0.

This construction recovers Regge curvature when the geometry is Regge. If face metrics match, then each across-face holonomy $G_f=\prod_{e\in f} g_e$1 is the identity, and the only contributions to a small loop around a hinge come from in-tetrahedron frame changes. The net holonomy is a rotation about the hinge direction by the deficit angle

$G_f=\prod_{e\in f} g_e$2

The paper stresses a conceptual distinction that is often blurred in informal discussion: twisting refers to mismatch of induced face metrics, whereas torsion is not determined by a metric alone. The interpolated construction defines a torsionless connection $G_f=\prod_{e\in f} g_e$3 in the presence of twisting, so twisting does not encode torsion. Within loop quantum gravity, $G_f=\prod_{e\in f} g_e$4 is the metric-compatible part of the Ashtekar–Barbero connection,

$G_f=\prod_{e\in f} g_e$5

and the Barbero–Immirzi parameter $G_f=\prod_{e\in f} g_e$6 enters only in the extrinsic sector $G_f=\prod_{e\in f} g_e$7 (Haggard et al., 2012).

4. Link variables, tetrads, and discrete Cartan structure

Chamseddine and Mukhanov formulate discrete gravity on a $G_f=\prod_{e\in f} g_e$8-dimensional manifold of elementary cells labeled by $G_f=\prod_{e\in f} g_e$9, with shift operators

$U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$0

and tangent operators

$U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$1

Each cell carries orthonormal frame operators $U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$2, tetrads $U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$3, inverse tetrads $U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$4, and metric

$U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$5

The discrete spin connection is a link variable

$U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$6

in $U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$7 or Spin$U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$8, and parallel transport acts on the vielbein by

$U_\gamma=\prod_i U_{\sigma_i}U_{\tau_i}$9

(Chamseddine et al., 2021).

The discrete Cartan structure is expressed using the Clifford-dressed shift operator

$$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$0

The torsion-free condition is

$$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$1

which is the discrete analogue of $$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$2. Curvature is defined by the plaquette holonomy

$$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$3

with Lie-algebra projection

$$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$4

and scalar curvature

$$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$5

Local rotations $$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$6 or Spin$$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$7 act on the frames and links as

$$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$8

under which torsion is invariant, curvature transforms adjointly, and the scalar curvature is gauge invariant. In the continuum limit $$W(\partial\Sigma)=\prod_{\langle ij\rangle\in\partial\Sigma}\Omega_{ij}$$9, one has

$G_f=\mathbb{1}$0

and the discrete plaquette curvature reduces to the standard continuum expression

$G_f=\mathbb{1}$1

The discrete Einstein–Hilbert–type action $G_f=\mathbb{1}$2 and the discrete Dirac action likewise reduce to their continuum counterparts (Chamseddine et al., 2021).

5. Discrete spin connection in lattice fermion systems

In the discrete-time quantum walk construction for fermions in two-dimensional curved spacetime, the walk is rewritten in terms of discrete covariant derivatives

$G_f=\mathbb{1}$3

a local eigenbasis of $G_f=\mathbb{1}$4, and a discrete zweibein extracted from the eigenvalues $G_f=\mathbb{1}$5. The discrete spin connection is encoded in the diagonal zero-components $G_f=\mathbb{1}$6 and $G_f=\mathbb{1}$7 in that local basis, together with fixed components $G_f=\mathbb{1}$8, $G_f=\mathbb{1}$9, and $e$00. Local Lorentz transformations act as

$e$01

and the transformation laws of $e$02 and $e$03 provide the discrete analogue of the abelian $e$04-dimensional spin-connection gauge law. Two discrete Riemann curvatures are then defined from the mixed action of the covariant differences; one of them, $e$05, tends in the continuum limit to the usual Riemann curvature component $e$06 (Debbasch, 2019).

In the symmetry-based theory of Dirac fermions on two-dimensional hyperbolic lattices, the continuum background is the Poincaré disk

$e$07

with nonzero spin connection components

$e$08

The discrete spin connection is constructed by parallel transport along geodesic edges,

$e$09

where $e$10 is determined by the difference in tangent directions at the endpoints, and for a central $e$11-gon one finds

$e$12

The Wilson loop

$e$13

encodes curvature, and the discrete tetrad hypothesis

$e$14

ensures compatibility between discrete frames and spin transport. The resulting lattice Dirac action includes the $e$15 directly in its nearest-neighbor kinetic term, so curvature modifies hopping through bond-dependent spinor phases. In this framework, coupling to the spin connection produces a finite density of states at zero energy for any finite curvature in hyperbolic spaces with $e$16, with the scaling heuristic $e$17, and this is tied to enhanced susceptibility to interaction-driven instabilities at weak coupling (Djordjević et al., 11 Jul 2025).

6. Spin foams, group averaging, and framed polyhedra

In the spin-foam representation of discrete $e$18 gauge theory, the discrete spin connection is the assignment of group elements $e$19 to the edges of a 2-complex. Face holonomies are

$e$20

the Wilson action is $e$21, and in the BF limit one imposes $e$22 on every face. Under local gauge transformations at vertices, $e$23 transforms by conjugation with the group elements at the source and target. The edge holonomies are therefore the discrete $e$24 spin connection in the standard lattice-gauge sense (Hnybida, 2015).

The coherent-state reformulation replaces explicit spin sums by spinor integrals and contour integrals. Local $e$25 gauge invariance at a node is implemented by the projector

$e$26

which encodes the intertwiner space. Summing over total spin yields

$e$27

so the sum over spins is traded for a residue at a simple pole. After Gaussian spinor integration, the vertex amplitude becomes a meromorphic function of contour variables, and the dependence on the connection is reorganized into cycle invariants on the boundary graph.

The same paper makes the boundary data geometrically explicit by reducing $e$28 spinors at a node modulo $e$29 to the Grassmannian $e$30 through the closure constraint

$e$31

In three-vector form this gives $e$32, so the node data describe a convex polyhedron of fixed total area together with face frames, hence a framed polyhedron. Matching constraints on links generate the residual $e$33 phases, and the resulting configurations are closed twisted geometries. In that language, adjacent framed polyhedra are related by $e$34 holonomies, and after integrating over the $e$35 those holonomies survive as cycle invariants and contour poles in the vertex generating functions (Hnybida, 2015).

Taken together, these formulations show that “discrete spin connection” is not a single construction but a family of closely related ones. The common content is the same: a link-, edge-, or face-based transporter for local frame or spinor data; a gauge-covariant notion of curvature through loop holonomy; and a discrete counterpart of Levi-Civita, Cartan, or Dirac transport that reproduces the standard continuum structures in the appropriate limit (Livine et al., 2011, Haggard et al., 2012, Chamseddine et al., 2021, Debbasch, 2019, Djordjević et al., 11 Jul 2025, Hnybida, 2015).