Generalized Proca and Vector–Tensor Interactions

A massive spin-1 field should propagate three polarizations, yet a generic derivative self-interaction makes its temporal component dynamical and introduces a fourth mode. Generalized Proca theory avoids that mode by arranging a degenerate velocity Hessian and the associated constraint pair. The construction is background-sensitive: second-order field equations and a formal flat-space degeneracy are useful design criteria, but the constraints and kinetic signs must still be checked on the branch being used.

Required background. Frames, Tetrads, and Spinors supplies the local Lorentz decomposition into temporal, transverse, and longitudinal vector components.

Helpful background. Einstein Equations supplies the metric dynamics; Gravitational Actions and Boundary Terms supplies action variation; Formulations of General Relativity supplies constraint counting; and Gravity Beyond General Relativity supplies the wider consistency framework.

The Proca constraint before interactions

Use signature $(-,+,+,+)$ and

$$F_{\mu\nu}=\nabla_\mu A_\nu-\nabla_\nu A_\mu, \qquad X=-\frac12 A_\mu A^\mu.$$

The flat-space Proca Lagrangian is

$$\mathcal L_{\mathrm P}=-\frac14F_{\mu\nu}F^{\mu\nu} -\frac12m^2A_\mu A^\mu.$$

Because $F_{00}=0$, no $\dot A_0^2$ occurs and the conjugate momentum obeys $\pi^0=0$. With $A_i$ as canonical coordinates, $\pi^i=F_{0i}=\dot A_i-\partial_iA_0$. Preserving the primary constraint gives $\partial_i\pi^i+m^2A_0=0$. For $m\ne0$ the pair is second class. Starting with four fields and four momenta, two second-class constraints leave $(8-2)/2=3$ physical degrees of freedom: two transverse modes and one longitudinal mode. This canonical structure is reviewed in Heisenberg 2014, §1, around Eq. (1.1), PDF.

A generic term involving $\nabla_\mu A_\nu$ gives $A_0$ its own momentum. The design problem is therefore to retain a null direction of

$$\mathcal H^{\mu\nu} =\frac{\partial^2\mathcal L}{\partial\dot A_\mu\partial\dot A_\nu}, \qquad \det\mathcal H=0,$$

and then verify that preservation of the primary relation produces an independent secondary constraint rather than fixing a multiplier incorrectly.

The interactions L₂ through L₆

On curved spacetime, a convenient parity-even second-order generalized Proca basis is

$$S=\int d^4x\,\sqrt{-g}\sum_{n=2}^{6}\mathcal L_n,$$

where $G_{i,X}=\partial G_i/\partial X$ and

$$\begin{aligned} \mathcal L_2={}&G_2(X,F,Y),\\ \mathcal L_3={}&G_3(X)\nabla_\mu A^\mu,\\ \mathcal L_4={}&G_4(X)R+G_{4,X}\left[(\nabla\!\cdot\!A)^2 -\nabla_\rho A_\sigma\nabla^\sigma A^\rho\right], \end{aligned}$$

with $F=-F_{\mu\nu}F^{\mu\nu}/4$ and $Y=A^\mu A^\nu F_{\mu\alpha}F_\nu{}^\alpha$. The remaining terms are

$$\begin{aligned} \mathcal L_5={}&G_5G_{\mu\nu}\nabla^\mu A^\nu -\frac{G_{5,X}}6\big[(\nabla\!\cdot\!A)^3 -3(\nabla\!\cdot\!A)\nabla_\rho A_\sigma\nabla^\sigma A^\rho\\ &\qquad\qquad+2\nabla_\rho A_\sigma\nabla^\gamma A^\rho \nabla^\sigma A_\gamma\big] -g_5\widetilde F^{\alpha\mu}\widetilde F^\beta{}_{\mu} \nabla_\alpha A_\beta,\\ \mathcal L_6={}&G_6L^{\mu\nu\alpha\beta} \nabla_\mu A_\nu\nabla_\alpha A_\beta +\frac{G_{6,X}}2\widetilde F^{\alpha\beta}\widetilde F^{\mu\nu} \nabla_\alpha A_\mu\nabla_\beta A_\nu. \end{aligned}$$

All functions other than $G_2$ depend on $X$. The duals are defined explicitly by

$$\widetilde F^{\mu\nu}=\frac12\epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}, \qquad L^{\mu\nu\alpha\beta} =\frac14\epsilon^{\mu\nu\rho\sigma} \epsilon^{\alpha\beta\gamma\delta}R_{\rho\sigma\gamma\delta}.$$

They use the four-dimensional metric volume tensor and the site’s curvature convention. The Maxwell kinetic term is absorbed into $G_2$ here; it must not be added a second time. The Einstein–Hilbert term is included by a constant part $G_4\supset M_{\mathrm{Pl}}^2/2$, with $M_{\mathrm{Pl}}^2=(8\pi G)^{-1}$. The complete curved basis and its counterterms appear in Heisenberg 2019, §9.2, Eqs. (298)–(301), PDF; the flat-space construction is derived in Heisenberg 2014, §§2–3, Eqs. (2.1)–(3.23), PDF.

The curvature couplings are essential. Naively replacing $\partial$ by $\nabla$ in the quartic and quintic flat interactions produces higher derivatives through commutators of covariant derivatives. The $G_4R$, $G_5G_{\mu\nu}\nabla^\mu A^\nu$, and double-dual terms cancel those contributions while maintaining the constraint structure. This is a translation of the entire $(-,+,+,+)$ package: $X$, the Maxwell sign, the dual tensor, and the curvature convention are fixed together.

How the temporal mode is removed

The quartic flat-space combination exhibits the mechanism directly:

$$\mathcal L_4^{\mathrm{flat}} =f_4(X)\left[(\partial\!\cdot\!A)^2 -\partial_\rho A_\sigma\partial^\sigma A^\rho\right].$$

Since $\partial\!\cdot\!A=-\dot A_0+\partial_iA_i$, the first square contains $+f_4\dot A_0^2$. The $\rho=\sigma=0$ term in the second contraction is also $+f_4\dot A_0^2$ and is subtracted. The coefficient of $\dot A_0^2$ therefore cancels exactly; the remaining $\dot A_0$ dependence is at most linear. The Hessian has a null temporal direction, producing a primary constraint. At higher orders the antisymmetric Levi-Civita contractions enforce the analogous degeneracy, and in four dimensions the finite index supply makes the nontrivial series stop at $\mathcal L_6$. See Heisenberg 2019, §8.4, Eqs. (244)–(267), PDF.

This argument establishes neither ghost freedom on every background nor positivity. On a chosen background one must compute the full metric–vector Hessian, preserve the constraint under time evolution, and then inspect the reduced tensor, vector, and scalar kinetic matrices. A branch at which the rank changes can gain strong coupling even though the covariant formula still looks degenerate.

Gauge and decoupling limits

If $G_2$ reduces to the Maxwell term, all $X$-dependent symmetry-breaking interactions vanish, and $m\to0$, the two second-class Proca constraints become first class and generate $U(1)$ gauge symmetry. The physical count drops from three to two. Merely sending the mass parameter to zero while retaining $G_3(X)$ or $G_{4,X}$ does not restore gauge invariance.

The Stückelberg replacement

$$A_\mu=B_\mu+\frac1m\partial_\mu\pi$$

isolates the longitudinal polarization. With a correlated scaling of interaction coefficients, the $m\to0$ decoupling limit yields scalar Galileon combinations for $\pi$ plus a Maxwell field $B_\mu$. Without that scaling, inverse powers of $m$ expose a vanishing strong-coupling scale. Thus the gauge-invariant and scalar decoupling limits are controlled limits only after their coefficient scaling and cutoff are declared.

Common pitfalls

Equating second-order equations with three modes. A second-order equation for $A_0$ would make it dynamical. The decisive condition is a constraint-producing Hessian degeneracy and its preserved secondary relation.

Checking only the vector Hessian on a gravitational background. Metric velocities mix with vector velocities. The coupled kinetic matrix and actual background constraint rank are the relevant objects.

Calling $m\to0$ the Maxwell limit. $X$-dependent derivative interactions still break $U(1)$. State which functions vanish or scale with $m$.

Exercises

Verify the cancellation of $\dot A_0^2$ in $\mathcal L_4^{\mathrm{flat}}$ and use it to explain why one still needs a secondary constraint.

Solution

With $A^0=-A_0$,

$$(\partial\!\cdot\!A)^2=(-\dot A_0+\partial_iA_i)^2 =\dot A_0^2-2\dot A_0\partial_iA_i+(\partial_iA_i)^2.$$

For the second invariant, its $(\rho,\sigma)=(0,0)$ contribution is

$$\partial_0A_0\,\partial^0A^0 =\dot A_0(-\partial_0)(-A_0)=\dot A_0^2.$$

It enters with a minus sign, so the two quadratic terms cancel. Hence the momentum conjugate to $A_0$ is a relation among fields and the other momenta rather than an equation that determines $\dot A_0$; this is the primary constraint. One primary constraint alone would remove only half a configuration-space degree of freedom. Its time derivative must yield an independent secondary constraint, as in ordinary Proca theory, so the second-class pair removes one full unwanted polarization.

References

• Heisenberg, L. “A Systematic Approach to Generalisations of General Relativity and Their Cosmological Implications.” Physics Reports 796 (2019) 1–113. DOI. Open PDF.
• Heisenberg, L. “Generalization of the Proca Action.” Journal of Physics A: Mathematical and Theoretical 47 (2014) 335401. DOI. Open PDF.