---
title: Generalized Proca Theory
url: https://www.emergentmind.com/topics/generalized-proca-theory
type: topic
---

# Generalized Proca Theory

to=arxiv_search  手机版天天中彩票_code='print(arxiv_search("Generalized Proca theory vector disformal transformation generalized Proca 2018 1807.06048"))' code_execution_config='{}'
to=arxiv_search  eqqars_code='Generalized Proca theory vector disformal transformation 1807.06048' code_execution_config='{}'
to=arxiv_search  彩神争霸代理_json='{"query":"Generalized Proca theory vector disformal transformation generalized Proca", "max_results": 10}' code_execution_config='{}'
to=arxiv_search  เงินฟรี าคาร่json='{"query":"Generalized Proca theory", "max_results": 10, "sort_by":"relevance"}' code_execution_config='{}'
to=arxiv_search  թե_args='query: "Generalized Proca theory", max_results: 10' code_execution_config='{}'
Generalized Proca theory is a vector–tensor theory for a single massive spin-1 field \(A_\mu\) in which \(U(1)\) gauge invariance is explicitly broken, but the self-interactions are arranged so that only the three Proca polarizations propagate, together with the two tensor polarizations of gravity. Its defining feature is that derivative self-interactions and non-minimal curvature couplings are chosen so that the equations of motion remain second order, or equivalently the vector kinetic matrix is degenerate in the required way, thereby avoiding Ostrogradsky instabilities. In the longitudinal limit \(A_\mu=\nabla_\mu\phi\), the theory reduces to the Horndeski class, while operators built from the antisymmetric field strength vanish [1705.05387] [1807.06048] [1402.7026].

## 1. Origins and defining idea

The starting point is the standard Proca action,
\[
S_{\rm Proca}=\int d^4x\left[-\frac14 F_{\mu\nu}F^{\mu\nu}-\frac12 m^2 A_\mu A^\mu\right],
\qquad
F_{\mu\nu}\equiv \partial_\mu A_\nu-\partial_\nu A_\mu,
\]
which propagates three physical degrees of freedom because the time component \(A_0\) is non-dynamical and generates a primary constraint together with a secondary one [1402.7026]. Generalized Proca theory extends this structure by adding derivative self-interactions and curvature couplings that keep the same degree-of-freedom count. In four dimensions the healthy derivative family terminates at sextic order [1705.05387] [1402.7026].

The theory is the vector analogue of the covariant Galileon/Horndeski construction. Its longitudinal Stückelberg mode reproduces scalar Galileon interactions in the decoupling limit, but generalized Proca is not merely “Horndeski with \(A_\mu\) replacing \(\nabla_\mu\phi\)”: it also contains intrinsic vector interactions built from \(F_{\mu\nu}\), \(\tilde F_{\mu\nu}\), and related contractions, which vanish in the pure scalar limit and therefore have no scalar-tensor counterpart [1705.05387] [1402.7026].

## 2. Covariant action and operator content

A standard covariant presentation uses the invariants
\[
X\equiv -\frac12 A_\mu A^\mu,\qquad
F_{\mu\nu}\equiv \nabla_\mu A_\nu-\nabla_\nu A_\mu,\qquad
\tilde F^{\mu\nu}\equiv \frac12 \epsilon^{\mu\nu\alpha\beta}F_{\alpha\beta},
\]
\[
F\equiv -\frac14 F_{\mu\nu}F^{\mu\nu},\qquad
U\equiv -\frac14 F_{\mu\nu}\tilde F^{\mu\nu},\qquad
Y\equiv F_{\mu\rho}F_\nu{}^\rho A^\mu A^\nu,
\]
and the double dual of the Riemann tensor
\[
L^{\mu\nu\rho\sigma}\equiv \frac14 \epsilon^{\mu\nu\alpha\beta}\epsilon^{\rho\sigma\gamma\delta}R_{\alpha\beta\gamma\delta}.
\]
With these ingredients, the action is
\[
S_{\rm GP}=\sum_{i=2}^{6} S_i,\qquad
S_i=\int d^4x\,\sqrt{-g}\,\mathcal L_i,
\]
with
\[
\mathcal L_2=G_2(X,F,U,Y),
\]
\[
\mathcal L_3=G_3(X)\nabla_\mu A^\mu,
\]
\[
\mathcal L_4=G_4(X)R+G_{4,X}(X)\left[(\nabla_\mu A^\mu)^2-\nabla_\rho A_\sigma\nabla^\sigma A^\rho\right],
\]
\[
\mathcal L_5
=G_5(X)G_{\mu\nu}\nabla^\mu A^\nu
-\frac16 G_{5,X}(X)\Big[(\nabla_\mu A^\mu)^3-3\nabla_\rho A_\sigma\nabla^\sigma A^\rho \nabla_\mu A^\mu+2\nabla_\mu A_\nu\nabla_\rho A^\mu\nabla^\nu A^\rho\Big]
-g_5(X)\tilde F_{\mu\rho}\tilde F_\nu{}^\rho \nabla^\mu A^\nu
-\mathcal G_5(X)\tilde F_{\mu\rho}\tilde F_{\nu\sigma}A^\mu A^\nu \nabla^\rho A^\sigma,
\]
\[
\mathcal L_6
=G_6(X)L^{\mu\nu\rho\sigma}\nabla_\mu A_\nu \nabla_\rho A_\sigma
+\frac12 G_{6,X}(X)\tilde F_{\mu\rho}\tilde F_{\nu\sigma}\nabla^\mu A^\nu \nabla^\rho A^\sigma.
\]
The functions \(G_i\), \(g_5\), and \(\mathcal G_5\) are arbitrary functions of \(X\), except that \(G_2\) may depend on \(X,F,U,Y\); parity preservation requires \(G_2\) to be even in \(U\) [1807.06048].

This operator basis clarifies the distinction between scalar-like and intrinsic-vector structures. Terms involving only \(\nabla_\mu A_\nu\) collapse to Horndeski/Galileon combinations when \(A_\mu=\nabla_\mu\phi\). Terms involving \(F_{\mu\nu}\) or \(\tilde F_{\mu\nu}\) vanish in that limit and encode purely vectorial dynamics. The \(\mathcal G_5\) term is often omitted in abbreviated presentations, but it still yields second-order equations and becomes necessary when one studies disformal closure of the theory [1807.06048] [1705.05387].

## 3. Constraint structure and the longitudinal mode

The absence of the extra, ghostlike polarization is enforced by degeneracy of the Hessian with respect to vector velocities. For a Lagrangian \(L(A,\partial A,\ldots)\), the Hessian is
\[
H^{\mu\nu}\equiv \frac{\partial^2 L}{\partial(\partial_0 A_\mu)\partial(\partial_0 A_\nu)}.
\]
Generalized Proca interactions are constructed so that
\[
H^{00}=H^{0i}=H^{i0}=0,
\]
while the spatial block is non-degenerate, guaranteeing that \(A_0\) remains auxiliary and that a primary plus secondary second-class constraint removes the would-be fourth vector mode [1705.05387].

The quartic flat-space sector furnishes the canonical example. Starting from
\[
\mathcal L_4=f_4(A^2)\Big[c_1(\partial\!\cdot\!A)^2+c_2\,\partial_\rho A_\sigma\partial^\rho A^\sigma+c_3\,\partial_\rho A_\sigma\partial^\sigma A^\rho\Big],
\]
the Hessian determinant is
\[
\det(H_{\mathcal L_4})=2(c_1+c_2+c_3)(-2c_2)^3.
\]
The healthy choice is therefore \(c_1+c_2+c_3=0\); setting \(c_1=1\) gives \(c_3=-(1+c_2)\), which isolates the standard generalized Proca quartic combination, while the term proportional to \(F_{\rho\sigma}F^{\rho\sigma}\) can be absorbed into \(\mathcal L_2\) [1705.05387] [1402.7026].

A broader Faddeev–Jackiw analysis of diffeomorphism-invariant generalized Proca theories coupled to arbitrary backgrounds showed that, once the sharpened Hessian assumptions \(H_{00}=H_{0i}=0\) and \(\det H_{ij}\neq0\) hold, most consistency conditions are automatically trivialized by diffeomorphism invariance. The remaining requirement is simply that a particular combination entering \(\{\phi_1,\phi_2\}\) not vanish identically, so the existence of exactly three vector degrees of freedom is generically easy to maintain [1907.12794].

The Stückelberg decomposition,
\[
A_\mu\to \hat A_\mu+\frac{\nabla_\mu\pi}{M},
\]
makes the longitudinal sector explicit. In the decoupling limit, the allowed interactions reduce to scalar Galileon terms and mixed \(F\)-\(\partial\partial\pi\) operators. This is the vector origin of the statement that generalized Proca is the spin-1 analogue of Galileon/Horndeski, with the important qualification that intrinsic vector operators survive away from the pure-gradient limit [1705.05387] [1402.7026].

## 4. Cosmological dynamics and perturbations

On a spatially flat FLRW background,
\[
ds^2=-dt^2+a^2(t)\delta_{ij}dx^i dx^j,
\qquad
\langle A_\mu\rangle=-\phi(t)\delta^0_\mu,
\qquad
X=\frac{\phi^2}{2},
\]
the vector equation is algebraic in \(\phi\), so the temporal component remains non-dynamical even cosmologically. De Sitter solutions with \(\dot H=0\) and \(\dot\phi=0\) arise naturally and can be stable late-time attractors [1603.05806] [1705.05387].

The tensor sector is characterized by
\[
q_T=2G_4-2\phi^2 G_{4,X}+H\phi^3 G_{5,X},
\qquad
c_T^2=\frac{2G_4+\phi^2\dot\phi G_{5,X}}{q_T},
\]
with stability conditions \(q_T>0\) and \(c_T^2>0\). For vector perturbations,
\[
q_V=G_{2,F}+2G_{2,Y}\phi^2-4g_5H\phi+2G_6H^2+2G_{6,X}H^2\phi^2,
\qquad
c_V^2=\mu_V/q_V,
\]
where \(\mu_V\) depends on \(G_{2,F}\), \(g_5\), \(G_6\), \(\mathcal G_5\), and the tensor sector. In the scalar sector the quadratic action is written in terms of background functions \(w_1,\ldots,w_7\), with
\[
q_S=3w_1^2+4q_T w_4,\qquad
Q_S=\frac{H^2 q_T q_S}{\phi^2(w_1-2w_2)^2},\qquad
c_S^2=\frac{\mu_S}{8H^2\phi^2 q_T q_V q_S},
\]
and the corresponding no-ghost/no-gradient conditions are \(q_S>0\), \(Q_S>0\), and \(\mu_S\ge 0\) [1807.06048].

Intrinsic vector interactions play a distinctive cosmological role. They do not modify the background equations or the second-order tensor action, but they do alter the vector no-ghost condition and the scalar/vector propagation speeds. In the quasi-static regime deep inside the sound horizon, the effective gravitational coupling \(G_{\rm eff}\) can be smaller than Newton’s constant \(G\), and the growth rate \(f\sigma_8\) and slip parameter \(\eta\) deviate from \(\Lambda\)CDM in a model-dependent way [1605.05066].

A widely studied family adopts power laws
\[
G_2=b_2 X^{p_2},\qquad
G_3=b_3 X^{p_3},\qquad
G_4=\frac{M_{\rm Pl}^2}{2}+b_4 X^{p_4},\qquad
G_5=b_5 X^{p_5},
\]
with \(p_3=(p+2p_2-1)/2\), \(p_4=p+p_2\), \(p_5=(3p+2p_2-1)/2\). These models admit radiation, matter, and de Sitter fixed points, with
\[
w_{\rm DE}:
\quad
-1-\frac{4s}{3}\ \to\ -1-s\ \to\ -1,
\qquad
s\equiv p_2/p,
\]
along the cosmic sequence, and viable trackers satisfy \(0\le s\lesssim 0.36\) [1603.05806].

A cubic luminal subset was implemented in a Boltzmann code and fit to cosmological data. With Planck + HST, the analysis found
\[
h=0.7334^{+0.0246}_{-0.0269},
\]
while adding BAO and JLA gave
\[
h=0.7041^{+0.0094}_{-0.0087},
\]
so the early- and late-universe determinations of \(H_0\) are removed in the first case and reduced in the second [2002.06782].

## 5. Disformal transformations and the gravitational-wave constraint

A central development after GW170817/GRB170817A was the analysis of vector disformal transformations,
\[
\bar g_{\mu\nu}=\Omega^2\left(g_{\mu\nu}+B A_\mu A_\nu\right),
\qquad
\Omega={\rm const.},\quad B={\rm const.},
\]
with \(A_\mu\) held fixed as a one-form. Invertibility requires \(\Omega\neq0\) and \(1-2BX\neq0\). Under this map,
\[
\bar X=\frac{X}{\Omega^2(1-2BX)},
\qquad
\sqrt{-\bar g}=\Omega^4\sqrt{1-2BX}\sqrt{-g},
\]
and the transformed action remains within the generalized Proca class only if one includes the \(\mathcal G_5\) operator in \(\mathcal L_5\) and allows \(\mathcal L_2\) to contain the \(U^2\) and \(Y\) structures that had often been described as “beyond generalized Proca” [1807.06048].

On an FLRW background, the sound speeds transform uniformly:
\[
\bar c_T^2=\frac{c_T^2}{1-2BX},\qquad
\bar c_V^2=\frac{c_V^2}{1-2BX},\qquad
\bar c_S^2=\frac{c_S^2}{1-2BX},
\]
while the kinetic prefactors rescale accordingly. This shows that constant vector disformal transformations simply reshape the background light cone. Observable ratios such as \(c_T^2/c_\gamma^2\) remain frame-invariant once the matter coupling is fixed, so changing frame can shift the apparent burden between gravity and matter sectors but does not eliminate physical constraints [1807.06048].

The multimessenger bound,
\[
-3\times10^{-15}\lesssim \delta c_T\equiv \frac{c_T-c_\gamma}{c_\gamma}\lesssim 7\times10^{-16},
\]
severely restricts derivative couplings that enter \(c_T^2\), principally \(G_{4,X}\) and \(G_{5,X}\). On a self-accelerating background with \(\dot H=\dot\phi=0\), luminal propagation requires
\[
G_{4,X}=\frac12 H\phi\,G_{5,X}.
\]
Tracker solutions can also be tuned, but that tuning is background dependent and therefore fragile [1807.06048].

A major implication is that homogeneous tuning is generally destabilized by inhomogeneities. Expanding around a tuned FLRW solution gives
\[
\delta c_T^2\simeq -\epsilon\,C\,\delta\Omega_m,
\]
and a rough estimate using halos and subhalos yields
\[
\epsilon C \lesssim 2\times10^{-14}\,\Delta^{-1}(1\,{\rm Mpc}/L),
\]
with typical structures implying \(\epsilon\lesssim {\rm few}\times10^{-16}\). This suggests that viable models must either suppress higher-\(X\) derivatives independently of background or be further fine-tuned beyond the homogeneous level [1807.06048].

## 6. Quantum behavior and radiative stability

At the level of the pure generalized Proca effective field theory, explicit one-loop calculations up to four-point show nontrivial cancellations of the naively most dangerous terms. In a minimal flat-space model organized by the scales \(m\), \(\Lambda_2\), and \(\Lambda_3=(\Lambda_2^2 m)^{1/3}\), the surviving divergences either renormalize gauge-preserving structures such as \(F^2\) and \(F^4\), or are suppressed so that any induced ghost mass lies above the EFT cutoff. A decoupling-limit analysis explains this in terms of the scalar Galileon non-renormalization pattern combined with gauge-invariant \(\tilde F\tilde F\,\partial\partial\phi\) operators [2005.01639].

The situation changes once matter is coupled. One-loop scalar-matter corrections distinguish sharply between nonlinear couplings,
\[
\frac{1}{M_1^2}A^2 T+\frac{1}{M_2^2}A^\mu A^\nu T_{\mu\nu},
\]
and linear derivative couplings,
\[
\frac{1}{M_3^2}\partial_\mu A^\mu\,T+\frac{1}{M_4^2}\partial^{(\mu}A^{\nu)}T_{\mu\nu}.
\]
The former renormalize the generalized Proca interactions without activating ghosts within the EFT regime. The latter generate higher-derivative operators such as
\[
\frac{1}{M_3^4}\big(\Box(\partial\!\cdot\!A)\big)^2,
\]
which introduce ghost poles. The dominant ghost scale is \(M_{\rm ghost}\sim M_3\), and the EFT remains predictive only for external momenta below scales such as \(M_3\), \(\sqrt{M_3 M_i\alpha_3}\), \((M_i^2M_j^4\alpha_4)^{1/4}\), and \((M_i^8\alpha_4)^{1/6}\) [1612.06938].

Beyond EFT control, a 2026 functional-renormalization-group study investigated ultraviolet completion in a truncation with up to two derivatives and four powers of the vector. It found a triplet of nearby non-Gaussian fixed points; only one, termed the “Proca fixed point,” has a non-tachyonic mass. In the full \(A^4\) truncation it has five relevant directions, while the Gaussian and Reuter fixed points lie on singular hypersurfaces of the flow and act only as quasi-fixed points in certain regimes [2601.20944].

## 7. Compact objects and extensions

Generalized Proca theory supports black-hole solutions with vector hair. On a static, spherically symmetric background,
\[
ds^2=-f(r)dt^2+h(r)^{-1}dr^2+r^2 d\Omega^2,
\qquad
A_\mu=(A_0(r),A_1(r),0,0),
\]
exact stealth Schwarzschild, Reissner–Nordström, and extremal Reissner–Nordström branches arise for specific choices of \(G_4\), \(G_5\), and \(G_6\). Power-law models, including vector Galileons, also admit regular numerical black holes. The longitudinal mode can generate a primary hair, whereas intrinsic vector derivative interactions typically induce secondary hair. In these solutions the largest deviations from GR occur near the horizon, which is why strong-gravity gravitational-wave measurements were identified as particularly promising probes [1705.09662].

A major non-Abelian extension replaces the single Abelian field by an \(SU(2)\) triplet \(A_\mu^a\). The generalized \(SU(2)\) Proca theory was first built by an exhaustive classification of healthy operators up to six contracted Lorentz indices, requiring both the correct degree-of-freedom count and a healthy multi-Galileon longitudinal limit [1609.05870]. Subsequent work showed that, unlike the Abelian case, a nontrivial secondary constraint condition must also be enforced, which led to a reconstruction of the theory and the identification of “beyond GSU2P” operators [2009.03241].

On cosmological triad backgrounds, the \(SU(2)\) tensor sector contains coupled gravitational-wave and vector transverse-traceless modes. For a quartic self-interaction \(\mathcal L_4^1\) plus a double-dual curvature term, the background-independent no-ghost condition is
\[
\alpha_1>0,
\qquad
-0.366\lesssim \alpha_{\rm Curv}/\alpha_1 \lesssim 2.241,
\]
and in the subcase \(\alpha_{\rm Curv}=0\), one tensor eigenmode is exactly luminal [1907.07961]. Astrophysically, neutron-star solutions in generalized \(SU(2)\) Proca theory were found to be more compact than their GR counterparts for the realistic equations of state studied, and some branches exceed \(\sim2.5\,M_\odot\), offering an alternative route into the compact-object mass gap [2408.07674].

Related extensions include teleparallel generalized Proca theory, where curvature-based GP is transplanted into torsion-based gravity and supplemented by a new sector
\[
\mathcal L_{\rm TP}=G_{\rm TP}(X,F,Y,I_1,\ldots,I_{54}),
\]
made possible by the first-order nature of torsion. On flat FLRW with a purely temporal vector, this reduces effectively to \(G_{\rm TP}(X,I_1)\) with \(I_1=3AH\), enlarging the available background cosmologies [2012.11959]. A later dynamical-systems study of one perturbatively safe, \(c_T=1\), generalized \(SU(2)\) Proca sector found the absence of stable attractors and smooth cosmological transitions, ruling out that specific implementation as a complete description of the Universe’s expansion [2502.03483].

Generalized Proca theory therefore occupies a distinctive position among modified-gravity models. It is simultaneously a constrained vector EFT with a precise Hamiltonian structure, a cosmological dark-energy framework with nontrivial scalar, vector, and tensor phenomenology, and a source of strong-gravity solutions and non-Abelian generalizations. Its modern development has been shaped by three recurring themes: closure under field redefinitions, especially disformal ones; compatibility with high-precision gravitational-wave measurements; and the tension between classical richness and quantum consistency [1807.06048] [2005.01639] [2601.20944].

Source: https://www.emergentmind.com/topics/generalized-proca-theory