Quadratic Ricci Scalar Newton–Cartan Gravity

• Quadratic Ricci Scalar Newton–Cartan gravity is a non-relativistic higher-order gravity theory characterized by curvature-square corrections and a modified Poisson equation.
• It employs Newton–Cartan and torsional frameworks with Schrödinger-covariant methods to construct invariant curvature terms like R² in Hořava–Lifshitz models.
• The theory uses a higher-order relativistic limit to cancel divergences and yield boost-covariant field equations that enrich dynamical TNC formulations.

Quadratic Ricci Scalar Newton–Cartan gravity appears in the recent literature in two closely related settings. One is the curvature-squared sector of dynamical Newton–Cartan and torsional Newton–Cartan formulations of Hořava–Lifshitz gravity, where the spatial Ricci scalar enters the $z=2$ potential as an $R^2$ term together with other marginal operators. The other is the zero-torsion non-relativistic theory obtained as the Newton–Cartan limit of a higher-order relativistic model, where the quadratic Ricci-scalar coupling produces a corrected Poisson equation and a full boost-covariant multiplet of non-relativistic field equations (Afshar et al., 2015, Hartong et al., 2015, Cardona et al., 7 Jul 2025).

1. Geometric setting

Newton–Cartan and torsional Newton–Cartan geometry are formulated in terms of a clock one-form and spatial projectors. In one presentation, the basic fields are the time-like vielbein $\tau_\mu$, the spatial vielbein $e_\mu{}^a$, and their inverses $\tau^\mu$, $e^\mu{}_a$, satisfying

$$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$

with temporal and spatial metrics

$$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$

A parallel tensor formulation specifies $\tau_\mu$, a velocity field $v^\mu$ dual to $R^2$0, an inverse spatial metric $R^2$1, its covariant partner $R^2$2, and a Bargmann $R^2$3 gauge field $R^2$4, with

$R^2$5

The affine connection is chosen to satisfy

$R^2$6

These structures define the non-relativistic spacetime data used in both the Schrödinger-covariant and dynamical TNC approaches (Afshar et al., 2015, Hartong et al., 2015).

The torsion sector is organized by conditions on $R^2$7. Three regimes are distinguished:

• Newton–Cartan: $R^2$8.
• Twistless torsional Newton–Cartan: $R^2$9.
• Fully torsional Newton–Cartan: no constraint on $\tau_\mu$0.

For twistless torsion, the foliation is hypersurface orthogonal; equivalently, there exists a scalar khronon $\tau_\mu$1 such that $\tau_\mu$2. In the Schrödinger-gauged formulation, twistless torsion is encoded by

$\tau_\mu$3

The zero-torsion specialization is

$\tau_\mu$4

These distinctions are decisive for the status of projectable and non-projectable Hořava–Lifshitz gravity, and later for the zero-torsion higher-order limit (Afshar et al., 2015, Hartong et al., 2015).

2. Schrödinger-covariant construction of curvature-square terms

A systematic route to Newton–Cartan and Hořava–Lifshitz invariants is provided by the non-relativistic conformal method, defined as the non-relativistic version of the relativistic conformal method and based on the $\tau_\mu$5 Schrödinger algebra. Its generators are $\tau_\mu$6, $\tau_\mu$7, $\tau_\mu$8, $\tau_\mu$9, $e_\mu{}^a$0, $e_\mu{}^a$1, and $e_\mu{}^a$2, and the construction introduces a compensating complex scalar

$e_\mu{}^a$3

Gauge fixing

$e_\mu{}^a$4

removes local dilatations and central charge transformations; after this, $e_\mu{}^a$5 becomes the vector $e_\mu{}^a$6, while the spatial components $e_\mu{}^a$7 survive as torsion contributions. This is the mechanism by which Schrödinger-invariant scalar field theories are converted into Galilean or Newton–Cartan invariants (Afshar et al., 2015).

The spatial curvature sector is built from the Schrödinger rotation curvature $e_\mu{}^a$8 and its contractions. The linear and quadratic curvature densities include

$e_\mu{}^a$9

$\tau^\mu$0

After gauge fixing, these descend to Galilean invariants built from the Newton–Cartan curvature scalar and its quadratic contractions. In particular, the quadratic Ricci scalar term of the Hořava–Lifshitz potential originates from the Schrödinger-invariant progenitor $\tau^\mu$1, while the full classification also contains $\tau^\mu$2 and $\tau^\mu$3 (Afshar et al., 2015).

The same sector can be characterized in scale-covariant language. At $\tau^\mu$4, purely spatial curvature invariants such as $\tau^\mu$5, $\tau^\mu$6, and $\tau^\mu$7 contain no $\tau^\mu$8, and therefore are automatically Schrödinger-invariant once they are made locally scale-invariant with the appropriate compensator power. For $\tau^\mu$9, the compensator weight satisfies

$e^\mu{}_a$0

For $e^\mu{}_a$1, the same spatial curvature invariants remain locally scale-invariant with an appropriate compensator, but there is no $e^\mu{}_a$2-symmetry enhancement to the full Schrödinger group (Devecioglu et al., 2018).

3. Dynamical TNC geometry and the Hořava–Lifshitz potential

When torsional Newton–Cartan geometry is made dynamical, it gives rise to Hořava–Lifshitz gravity. In adapted coordinates $e^\mu{}_a$3, the standard ADM variables are recovered from

$e^\mu{}_a$4

and the extrinsic curvature is

$e^\mu{}_a$5

The torsion vector is

$e^\mu{}_a$6

which in adapted coordinates reduces to

$e^\mu{}_a$7

Projectable Hořava–Lifshitz gravity corresponds to torsionless Newton–Cartan geometry, in which $e^\mu{}_a$8 and $e^\mu{}_a$9; non-projectable Hořava–Lifshitz gravity corresponds to twistless torsional Newton–Cartan geometry, in which $$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$0 and $$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$1 (Hartong et al., 2015).

In $$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$2 dimensions and for $$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$3, TNC covariance organizes the most general Hořava–Lifshitz action into

$$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$4

The potential is the sum of all relevant and marginal scalars built from $$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$5, $$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$6, and spatial covariant derivatives. Excluding time-reversal-odd terms,

$$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$7

In two spatial dimensions, the only independent curvature invariant from the spatial Riemann tensor is $$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$8 itself; at $$\tau^\mu \tau_\mu = 1,\quad \tau^\mu e_\mu{}^a = 0,\quad \tau_\mu e^\mu{}_a = 0,\quad e_\mu{}^a e^\mu{}_b = \delta^a{}_b,$$9, the marginal, four-derivative curvature invariant is $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$0. In the projectable case all acceleration terms drop out, and the potential reduces to curvature invariants and a cosmological constant. In the non-projectable case, the $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$1-dependent operators are allowed by TNC symmetry and must be included for completeness (Hartong et al., 2015).

This framework also explains the Bargmann origin of the Hořava–Melby-Thompson $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$2. The central extension introduces $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$3 with $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$4, and a Stückelberg completion $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$5 makes the theory $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$6-invariant in torsional settings. In the projectable case with $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$7, one can realize a genuine local $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$8; otherwise the Stückelberg completion turns the $$\tau_{\mu\nu} = \tau_\mu \tau_\nu,\qquad h^{\mu\nu} = e^\mu{}_a e^\nu{}_b\,\delta^{ab}.$$9 into a redundancy of field variables (Hartong et al., 2015).

4. Higher-order relativistic parent and the non-relativistic limit

A distinct construction of quadratic Ricci Scalar Newton–Cartan gravity starts from a relativistic parent theory in $\tau_\mu$0 spacetime dimensions and performs a magnetic $\tau_\mu$1 limit. For the Ricci-scalar-squared model, the parent action is

$\tau_\mu$2

with $\tau_\mu$3. The quartic Maxwell term replaces the $\tau_\mu$4 piece in a way that cancels divergent orders in the $\tau_\mu$5-expansion. The paper states that this is the unique choice that cancels the $\tau_\mu$6 and $\tau_\mu$7 divergences in the magnetic limit (Cardona et al., 7 Jul 2025).

The large-$\tau_\mu$8 field decomposition is

$\tau_\mu$9

so that

$v^\mu$0

The gauge field is expanded as

$v^\mu$1

and therefore

$v^\mu$2

The connection used to organize the non-relativistic limit is

$v^\mu$3

which obeys

$v^\mu$4

Zero torsion is enforced by the on-shell relativistic constraint $v^\mu$5, implying $v^\mu$6 after the limit. In this way the higher-order theory defines a zero-torsion Newton–Cartan system rather than a twistless torsional one (Cardona et al., 7 Jul 2025).

5. Non-relativistic equations and the corrected Poisson sector

The zero-torsion quadratic Ricci-scalar Newton–Cartan theory is defined by a constrained system of non-relativistic equations. Writing $v^\mu$7 and $v^\mu$8 for the Ricci tensor and scalar built from the Newton–Cartan connection, the constraints are

$v^\mu$9

and

$R^2$00

The paper emphasizes that the second condition is specific to the quadratic Ricci-scalar model and removes the remaining $R^2$01 obstruction in the Poisson combination (Cardona et al., 7 Jul 2025).

The leading scalar equation is the corrected Poisson equation

$R^2$02

The remaining equations in the multiplet are

$R^2$03

and

$R^2$04

supplemented by a further scalar equation $R^2$05 at order $R^2$06. These equations define zero-torsion quadratic Ricci-scalar Newton–Cartan gravity (Cardona et al., 7 Jul 2025).

The $R^2$07-dependence has two effects. First, the factor $R^2$08 dresses $R^2$09. Second, the term $R^2$10 adds a time-like second-derivative correction along the clock direction. In the $R^2$11 limit, or when curvature corrections are parametrically small, the standard Newton–Cartan Poisson sector is recovered. Unlike the Einstein–Gauss–Bonnet case discussed in the same work, these corrections do not vanish in $R^2$12; they persist in any dimension (Cardona et al., 7 Jul 2025).

6. Symmetry structure, projectability, and scope

The higher-order non-relativistic equations form a closed multiplet under Galilean boosts. In the zero-torsion sector,

$R^2$13

$R^2$14

The paper describes this as a reducible indecomposable representation of the Bargmann algebra. Thus the corrected Poisson equation is not an isolated scalar equation; it is one component of a boost-covariant set of Newton–Cartan field equations (Cardona et al., 7 Jul 2025).

A persistent distinction in the literature concerns action-level classification versus equations of motion. The Schrödinger-based construction classifies all locally Schrödinger-invariant curvature terms up to the stated derivative order and maps them to $R^2$15, $R^2$16, $R^2$17, $R^2$18, and mixed torsion-curvature terms in the Hořava–Lifshitz potential, but it does not compute the equations of motion arising from varying $R^2$19, $R^2$20, or higher-curvature potential terms. By contrast, the 2025 higher-order limit derives a full set of non-relativistic equations, but only in the zero-torsion sector (Afshar et al., 2015, Cardona et al., 7 Jul 2025).

A second distinction concerns dimensionality and independent invariants. In $R^2$21-dimensional dynamical TNC geometry, $R^2$22 is the unique independent curvature-squared scalar in the spatial metric sector; in broader $R^2$23 Schrödinger and scale-invariant classifications, the list also contains $R^2$24 and $R^2$25. A third distinction concerns projectability: projectable Hořava–Lifshitz gravity corresponds to torsionless Newton–Cartan geometry, while non-projectable Hořava–Lifshitz gravity corresponds to twistless torsional Newton–Cartan geometry with acceleration or torsion vector terms. Purely spatial curvature invariants such as $R^2$26 are compatible with local scale invariance for any $R^2$27, and at $R^2$28 they are automatically Schrödinger-invariant because they do not depend on $R^2$29 (Hartong et al., 2015, Devecioglu et al., 2018).

In this combined sense, Quadratic Ricci Scalar Newton–Cartan gravity denotes both a specific higher-derivative potential term in dynamical TNC/Hořava–Lifshitz gravity and a non-relativistic higher-order gravity theory with its own corrected Poisson sector. The former provides the diffeomorphism-covariant organization of $R^2$30 and related operators; the latter supplies an explicit non-relativistic field-equation realization of a quadratic Ricci-scalar theory.