What is the effective action of quantum gravity in the Landau background gauge?

What the Landau background gauge actually fixes

The paper quantizes gravity coupled to a massless scalar using the background field method, splitting the metric and scalar into background and quantum parts. Gauge fixing uses a generalized de Donder condition with parameter ξ, and the Landau background gauge is the limit ξ → 0. Because the interaction vertices from that gauge-fixing term are individually divergent as ξ → 0, the authors keep ξ finite through intermediate steps and take the limit only after the different contributions to a given calculation are combined.

They also present an equivalent formulation that allows ξ → 0 from the start, by introducing a Nakanishi-Lautrup auxiliary field B_μ. Integrating over B_μ imposes the Landau background gauge condition as a delta function, and the resulting Faddeev-Popov ghost Lagrangian involves vector ghost fields η_μ and η*_μ together with the background Ricci tensor. The full effective Lagrangian is invariant under BRST transformations, which is what makes the Slavnov-Taylor identity available as a check on the whole construction.

BRST invariance and the Ward identities behind the result

The paper derives the Slavnov-Taylor identity reflecting the BRST symmetry of the gauge-fixed theory, introducing external sources G*_{μν} and C*_μ coupled to the BRST variations of the graviton and ghost fields. After subtracting the Landau gauge-fixing term, the identity takes a universal form that the authors argue holds to all orders for the generating functional of one-particle-irreducible Green functions.

From the invariance of the background effective action under background gauge transformations, the paper then extracts Ward identities in momentum space. One of them, k_μ Π^{μν,αβ}(k) = 0, states that the background graviton self-energy is transverse. Another relates the four-divergence of the background graviton three-point vertex to the graviton self-energy. These identities are the structural reason the counterterm Lagrangian can be written in terms of background-invariant structures at all, and they are also the tool the authors later use to argue for finiteness beyond one loop.

Six structures survive in the off-shell counterterm

Dimensional analysis, symmetry, and the form of the interaction terms allow nine independent invariant structures built from the background curvature and the background scalar. Using the identity that removes the square of the Riemann tensor as an independent quantity, and because the Landau gauge-fixing term eliminates three further structures, the one-loop background counterterm Lagrangian reduces to six independent terms with dimensionless coefficients c₁ through c₆.

The coefficients are fixed diagram by diagram. The background graviton self-energy diagrams give c₁ = 43/240 and c₂ = 43/120. The divergent part of the scalar self-energy vanishes, so c₃ = 0. The background three-point function ⟨h φ φ⟩ gives c₄ = 1/3 and c₅ = −5/6, and the four-point scalar function gives c₆ = 1/4. Substituting these values produces the explicit off-shell counterterm Lagrangian of Eq. (3.12). The authors note that this result coincides with their earlier general-background-gauge computation once all gauge parameters are set to zero at the end, but that the Landau-gauge route is much simpler and more direct.

The scalar self-energy does not diverge at one loop

The divergent part of the scalar self-energy diagram vanishes in the Landau background gauge, which is what sets c₃ = 0. The paper goes further: it exhibits a diagrammatic identity between the background graviton-scalar-scalar vertex and the graviton propagator in Landau gauge, and argues that because this identity holds in any spacetime dimension D in the Landau gauge, the scalar self-energy must vanish in any dimension D as well.

This is a statement about the divergent part of the off-shell scalar self-energy at one loop, not about the full self-energy. The authors also note that on symmetry and power-counting grounds the on-shell (k² = 0) massless scalar self-energy vanishes at all orders. The one-loop finiteness is therefore the explicit verification of a property that the paper then argues may persist at higher orders.

On shell the Lagrangian collapses to one term

Imposing Einstein's equation R_{μν} − ½ R g_{μν} = (κ²/2) T_{μν}, with the energy-momentum tensor of the massless scalar, gives the on-shell relations R_{μν} = −(κ²/2) ∂_μ φ ∂_ν φ and R = −(κ²/2) ∂_μ φ ∂^μ φ. Substituting these into the off-shell counterterm Lagrangian leaves a single term: a (∂_μ φ ∂^μ φ)² structure with coefficient 203/320, multiplied by √g κ⁴/(16π²ε).

The paper states that this on-shell result agrees with the one obtained by 't Hooft and Veltman in the Feynman background gauge. The authors describe this agreement as expected in view of the Kallosh-DeWitt theorem, which asserts that the on-shell effective action should be independent of the gauge-fixing parameters. The comparison is therefore a consistency check on the calculation rather than a new physical prediction: the off-shell Lagrangian is gauge-dependent and simpler in Landau gauge, while the on-shell answer is gauge-independent and matches the older Feynman-gauge computation.

Why the higher-loop finiteness claim is an argument

The abstract states that the matter self-energy, finite at one loop in the Landau background gauge, may be finite also at higher loops. The paper's support for this is the background Ward identity involving the background graviton-scalar-scalar vertex, together with the diagrammatic identity between that vertex and the graviton propagator in Landau gauge. The authors present this as an inference, not a demonstrated result, and they say explicitly that it could be tested at two-loop order by computing the contributions from the scalar self-energy diagrams shown in Fig. 4, an issue they are currently considering.

The distinction matters for how the paper should be read. The one-loop off-shell counterterm Lagrangian, the vanishing divergent part of the scalar self-energy at one loop, and the on-shell agreement with 't Hooft and Veltman are explicit calculations. The extension to higher loops is a symmetry-based expectation awaiting an explicit two-loop check. The paper also frames quantum gravity as non-renormalizable by power counting but renormalizable in the modern sense, where divergences are cancelled by all counterterms allowed by the gauge symmetry.

What to take from the Landau gauge choice

The paper's practical priority is the comparison between two routes to the same physics. Off shell, the Landau background gauge reduces the one-loop counterterm Lagrangian to six structures with coefficients 43/240, 43/120, 0, 1/3, −5/6, and 1/4, and the authors report that this route is much simpler and more direct than the general background gauge calculation. On shell, the same Lagrangian collapses to a single (∂φ)⁴ term with coefficient 203/320, matching the Feynman-gauge result.

For a reader deciding what to trust, the split is between what is computed and what is argued. The six-coefficient off-shell Lagrangian, the vanishing one-loop divergent scalar self-energy, and the on-shell agreement with 't Hooft and Veltman are explicit one-loop results. The possibility that the scalar self-energy stays finite at higher loops rests on the background Ward identity and remains to be confirmed by the two-loop diagrams the authors identify. The Landau gauge's value here is therefore calculational economy plus a clean setting for the finiteness question, not a change to the on-shell answer.

Sources used in this answer