Scalar-Tensor Representation

Updated 2 February 2026

Scalar-tensor representation is a framework where higher-order tensors are reformulated using one or more scalar fields to capture nonlinearity and additional degrees of freedom.
It employs techniques like Legendre transforms and auxiliary variables to reduce complex tensor equations to second-order scalar field equations, enhancing computational tractability.
The approach is pivotal in modified gravity and constitutive tensor modeling, enabling consistent reformulations in both high-energy physics and applied continuum mechanics.

A scalar-tensor representation refers to any mathematical or physical framework in which entities that originally possess higher-rank tensor structure, or encode nonlinear or nonminimal degrees of freedom, are equivalently reformulated in terms of one or more scalar fields coupled to a lower-rank background (typically the metric in gravitational theories). Such representations are fundamental both in mathematical physics—for example, in the reduction of higher-order gravity or Feynman integrals—and in applied tensor analysis, as in constitutive laws for anisotropic media. Scalar-tensor representations often exploit auxiliary variables, Legendre transforms, and invariant-theoretic constructions. Their technical foundations and implications are illustrated in tensor reduction strategies in high-energy theory (Kreimer et al., 2010), reformulated constitutive tensor function theory (Madadi et al., 15 Oct 2025), and crucially in modern modified gravity frameworks (Capozziello et al., 2011, Rosa et al., 2021, Gonçalves et al., 2021, Kaczmarek et al., 2023).

1. Foundations and Mathematical Structure

The essence of scalar-tensor representations is the replacement of a functional dependence on high-order tensors or nonlinear functionals (such as $f(R)$ in gravity, where $R$ is the Ricci scalar) by the introduction of one or more scalar fields through Legendre or related transforms. For a prototypical example, consider the $f(R,T)$ gravity action in $D$ -dimensional spacetime: $S = \frac{1}{2\kappa^2} \int d^Dx \sqrt{-g}\, f(R,T) + S_m[g_{MN},\chi]$ Auxiliary fields $\alpha$ , $\beta$ are introduced, yielding a dynamically equivalent action: $S = \frac{1}{2\kappa^2} \int d^Dx\, \sqrt{-g}\left[f(\alpha,\beta) + f_\alpha(\alpha,\beta)(R-\alpha) + f_\beta(\alpha,\beta)(T-\beta)\right] + S_m$ The scalar fields $\varphi = f_\alpha(R,T)$ and $\psi = f_\beta(R,T)$ , together with a Legendre potential

$V(\varphi, \psi) = \varphi R + \psi T - f(R,T)$

generate the scalar-tensor representation: $S[g, \varphi, \psi, \chi] = \frac{1}{2\kappa^2} \int d^Dx \sqrt{-g}\left[\varphi R + \psi T - V(\varphi,\psi)\right] + S_m$ This yields second-order dynamical equations for metric and scalar fields, replacing generically fourth-order equations in the original frame (Rosa et al., 2021, Gonçalves et al., 2021).

The procedure generalizes to any nonlinear function of geometric or physical invariants, e.g., $f(Q,T)$ with nonmetricity $Q$ (Kaczmarek et al., 2024), $f(\mathcal{G}, T)$ with Gauss-Bonnet term $\mathcal{G}$ (Kaczmarek et al., 2023), or to tensor function representation theory in anisotropic continua (Madadi et al., 15 Oct 2025).

2. Scalar-Tensor Duality in Gravity and Field Theory

Scalar-tensor representations are central in modern modified gravity. In $f(R)$ gravity, the scalar-tensor (O’Halon) form: $S_J[g, \phi, \Psi] = \int d^4x \sqrt{-g} [\phi R - U(\phi) + 2\kappa^2 \mathcal{L}_m(g,\Psi)]$ with $\phi = f'(R)$ and $U(\phi) = \phi R(\phi) - f(R(\phi))$ , is dynamically equivalent to the original $f(R)$ action (Capozziello et al., 2011). The Einstein-frame reformulation via $g_{\mu\nu} \to \tilde{g}_{\mu\nu} = \phi g_{\mu\nu}$ yields manifestly canonical kinetic terms and clarifies the nature (massive, nontrivial potential) of the scalar mode relative to Brans-Dicke gravity.

In bi-scalar-tensor representations, such as for $f(R, T)$ or $f(\mathcal{G}, T)$ gravity, two scalars $\varphi, \psi$ couple nonminimally to curvature, matter trace, or topological invariants. Each additional functional dependence in the geometric Lagrangian corresponds to a new scalar field in the dual representation. The Legendre potential $V(\varphi,\psi)$ (or $V(\phi, \psi)$ ) encodes the full nonlinear structure and, through its derivatives, enforces the on-shell equivalence with the original theory (Gonçalves et al., 2021, Kaczmarek et al., 2023).

3. Scalar-Tensor Representation in Constitutive and Tensor Function Theory

In the representation theory of anisotropic tensor-valued functions, a different but structurally analogous scalar-tensor reduction occurs. Traditional approaches (Boehler–Liu) require the invariance of high-order structural tensors under symmetry groups, necessitating fourth- or sixth-order tensors and rendering the construction unwieldy for most crystalline point groups. The Man–Goddard reformulation replaces these by sets of lower-order structural tensors (second-order or less) invariant under group generators $G^*$ . Scalar-valued or tensor-valued functions are then fully determined by traces (scalar invariants) constructed from the argument tensor and lower-order structural tensors: $f(A) = \widehat{f}(I_1, I_2, ..., I_r)$ where the $I_k$ are minimal integrity basis invariants composed of $A$ , $M_i$ , and their products (Madadi et al., 15 Oct 2025).

This dramatically reduces both tensor order and the cardinality of the invariant basis needed for constitutive modeling, with immediate impact on the mathematical tractability and compute efficiency in finite elasticity, plasticity, and related fields.

4. Scalarization in Quantum Field Theory: Tensor Reduction

In multi-loop Feynman diagram calculations, tensor integrals are systematically reduced to scalar integrals at the price of enlarging the set of propagators. Each contraction involving products like $k_e \cdot k_f$ invokes recurrence relations (Tarasov shifts) that generate scalar parametric integrals over a larger set (the Feynman matroid). The full tensor structure of any amplitude is then encoded as a single scalar parametric integral attached to a representable matroid (Kreimer et al., 2010). This approach unifies planar and non-planar, tensor and scalar, and master-integral topologies by combinatorial extension, aiding analytic continuation and integration-by-parts identity generation.

5. Scalar-Tensor Invariants and Frame Independence

Scalar-tensor representations naturally lead to invariant formulations across different conformal frames. For a general scalar-tensor action: $S = \frac{1}{2\kappa^2} \int d^4x \sqrt{-g} \left[ F(\phi) R - Z(\phi) (\nabla \phi)^2 - 2U(\phi) \right] + S_m[e^{2\alpha(\phi)}g,\Psi]$ the invariants $I_1 = 1/F(\phi)$ , $I_2 = U(\phi)/F(\phi)^2$ , $I_3(\phi)$ (canonical field), and $I_4(\phi) = e^{2\alpha(\phi)}/F(\phi)$ are unchanged under conformal transformations and field redefinitions. All physical observables, including hydrostatic equilibrium (Tolman–Oppenheimer–Volkoff) equations and mass-radius relations, are then expressed in terms of these invariants, guaranteeing frame independence (Kozak et al., 2021).

6. Physical Interpretation and Stability Implications

The scalar fields in scalar-tensor representations correspond to dynamical degrees of freedom that mediate or modulate the effective gravitational or constitutive response. In modified gravity, the original $f(R)$ scalar modifies the effective Newton constant, introduces higher-derivative dynamics, and, if nonminimally coupled to matter, can mediate energy-momentum exchange via the trace coupling scalar. The Legendre potential encodes all nonlinearities and mutual couplings. Stability of perturbations (tensor, scalar modes) is generally governed by the structure of $V$ and is ensured if it is separable, e.g., for $V = P(\varphi) + Q(\psi)$ , allowing for a factorization in the perturbation (graviton) operator, as shown explicitly in thick-brane and brane-with-Gauss-Bonnet examples (Rosa et al., 2021, Lobão et al., 2023).

In the context of cosmology and brane-worlds, extra scalar modes can generate accelerated expansion, modify history (emulate $\Lambda$ CDM), or create internal structure in the graviton zero-mode under appropriate parameter choices (Rosa et al., 2021, Kaczmarek et al., 2024, Kaczmarek et al., 2023). In field theory, scalarization gives access to parametric integration techniques and combinatorial symmetry exploitation.

7. Special Cases and Extensions

Specific functional choices in the original (non-scalarized) action reduce the number of necessary scalars. For $f(R,T) = F(R) + T$ or $f(R,T) = R + G(T)$ , the Legendre transform is degenerate, and a single auxiliary scalar suffices, yielding correspondingly simplified actions and equations of motion (Rosa et al., 2021, Pereira, 2024). In the context of quantum field calculations, such reductions correspond to cases where tensor structure can be eliminated with minimal propagator augmentation (Kreimer et al., 2010).

Scalar-tensor representation frameworks are extensible to generalized geometric contexts (nonmetricity, Gauss-Bonnet, hybrid cosmologies) and to symbolic tensor computation analysis (Tensor Evolution), where recurrences of tensor objects are abstracted to tensor-valued analogs of scalar recurrence chains (Absar et al., 5 Feb 2025).

In summary, the scalar-tensor representation is a mathematically precise and physically motivated tool for reducing, analyzing, and reinterpreting tensorial and nonlinear structures in gravity, field theory, and continuum mechanics via the introduction of appropriately coupled scalar fields and invariants. Its development underpins modern approaches to both analytic and computational challenges in theoretical physics and engineering (Rosa et al., 2021, Kaczmarek et al., 2024, Gonçalves et al., 2021, Madadi et al., 15 Oct 2025, Kreimer et al., 2010, Kozak et al., 2021, Lobão et al., 2023, Capozziello et al., 2011).