Boundary matching rules derived for broad classes of gravity theories

A new mathematical study has delivered the most general formulation yet of the rules that govern how two pieces of spacetime can be glued together in modified theories of gravity, revealing that General Relativity and its popular extensions occupy a surprisingly privileged place in the landscape of gravitational physics. The work, published open access in the journal General Relativity and Gravitation, derives junction conditions for essentially arbitrary gravitational theories — those built from actions depending on any function of curvature invariants, and even on covariant derivatives of the curvature tensor. The results settle long-standing ambiguities about when thin shells of matter, impulsive gravitational waves, and exotic objects called gravitational double layers can appear at the interface between two spacetime regions.

The question of how to join two spacetimes is far older than the current boom in alternative gravity theories. In General Relativity, matter concentrated on an idealized surface of zero thickness — a domain wall, a brane-world shell, or a thin layer of material — is described by the famous Israel equations, first written down in the 1960s and building on earlier work by Kornel Lanczos in the 1920s. These equations relate the surface energy-momentum tensor of the shell to the jump in the extrinsic curvature of the joining hypersurface, the geometric quantity that measures how the surface bends within the surrounding spacetime. Null, or lightlike, shells were later treated separately, with applications ranging from classical models of imploding dust clouds to the formation of black holes and the propagation of concentrated lightlike signals.

The new analysis, carried out by José M. M. Senovilla of the University of the Basque Country in Bilbao, extends this machinery to the plethora of gravitational theories now under consideration as alternatives to Einstein’s theory — F(R) theories, quadratic and cubic gravity, Lanczos-Lovelock models, generalized quasi-topological gravities, and theories whose Lagrangians involve covariant derivatives of the Riemann tensor to arbitrarily high order. The guiding principle is disarmingly simple: whatever the theory, its field equations must remain mathematically well-defined when curvature develops singular, Dirac-delta-like spikes on the joining hypersurface. Terms that would require multiplying such distributions by themselves — operations with no rigorous meaning in the standard distributional calculus — must be excluded, and the exclusion conditions turn out to be precisely the junction conditions.

The technical starting point is the geometry of matching. Two regions of an (n+1)-dimensional Lorentzian manifold, each with its own smooth metric and possibly different matter content, are separated by a timelike hypersurface Sigma. A continuous global metric exists if and only if the first fundamental forms — the metrics induced on the surface from each side — agree. Once that basic gluing is secured, the connection and curvature can be computed in the sense of tensor distributions, with step functions capturing the two-sided character of the fields and Dirac deltas concentrated on Sigma encoding any singular behavior. The jump of any tensor across the surface is simply the difference between its limits from the plus and minus sides.

From this framework, the study extracts a hierarchy of results. In the generic case of theories whose action includes quadratic or higher-degree curvature terms, the field equations contain products of the curvature with itself, and such products make distributional sense only if the Riemann tensor carries no singular part. That forces the second fundamental form to be continuous across the hypersurface: mathematically, the jump of the extrinsic curvature must vanish. A further requirement eliminates ill-defined products of curvature derivatives, demanding that the Riemann tensor itself be continuous. When these conditions hold but the normal derivative of the curvature still jumps, the field equations generate a genuine thin shell of matter, and the paper provides a general closed-form expression for the shell’s energy-momentum tensor, built from the jump in the covariant derivative of the Riemann tensor contracted with a tensor that encodes the specific Lagrangian.

Remarkably, this shell tensor is proven to be tangent to the hypersurface — a property forced by covariant conservation of the total energy-momentum — and it obeys generalized Israel equations relating its surface divergence to the jump of the bulk energy-momentum across the shell. These relations have exactly the same structure as Israel’s original equations, but with the shell tensor now depending on the detailed form of the gravitational action. For a “proper matching” with no shell at all, an additional condition is needed: the first covariant derivative of the Riemann tensor must also be continuous. And crucially, one result holds universally, independent of the field equations: the normal components of the energy-momentum tensor must be continuous across any proper matching surface. In particular, matching onto vacuum requires the vanishing of normal pressure, a condition that can even determine where the matching surface must lie.

The analysis singles out several classes of theories as extraordinary. General Relativity is unique in having a Lagrangian linear in curvature, so its field equations are linear too; only the continuity of the extrinsic curvature is required for a proper matching, and shells with the familiar Israel surface stress tensor arise whenever it jumps. F(R) theories, which depend on the scalar curvature alone, turn out to be the only theories that permit shells of pure curvature — in plain terms, impulsive gravitational waves — because their equations involve only the scalar curvature and its derivatives, and these remain well defined provided only the trace of the extrinsic curvature is continuous. Impulsive gravitational waves, a concept going back to Roger Penrose’s geometric work in the 1970s, therefore live naturally in GR and F(R) gravity but are generically forbidden in other theories, where any jump of the full second fundamental form would render the equations ill-defined.

Even stranger behavior emerges in purely quadratic theories — those whose Lagrangian is a polynomial containing at most quadratic curvature invariants, such as combinations of the square of the scalar curvature, the square of the Ricci tensor, and the square of the full Riemann tensor. Because the derivative of the Lagrangian with respect to the Riemann tensor is then linear in the curvature, the problematic products of distributions never appear, and the Riemann tensor is allowed to jump. The second covariant derivative of the curvature distribution then contains derivatives of Dirac deltas on the hypersurface, forcing the energy-momentum tensor to include a structure mathematically analogous to the electric double layer at a charged surface. These gravitational double layers, which this research program has explored in earlier work, introduce new surface quantities: an external flux momentum and an external pressure or tension, alongside the ordinary tangential shell tensor, and the Israel equations acquire extra terms that measure the double-layer strength through the jumps of the scalar curvature and Einstein tensor.

The paper also tackles the most general setting of all: Lagrangians depending on covariant derivatives of the Riemann tensor up to some maximum order m, a class that includes infinite-derivative-inspired constructions truncated at finite order, though some such theories have been questioned on causality grounds in the quantum regime. Here the logic extends naturally. The Riemann tensor and all of its covariant derivatives up to order m must be continuous for the field equations to make distributional sense, and a thin shell necessarily appears whenever the (m+1)-th derivative jumps. A proper, shell-free matching requires continuity of the Riemann tensor through order m+1 — meaning the m-th derivative can jump only at the price of generating a shell. The general formula for the shell energy-momentum tensor in terms of that highest-order jump is provided, along with proofs that it is tangent to the surface and satisfies the same generalized Israel relations.

Senovilla is careful to delineate the scope of the results. The analysis is restricted to metric theories of gravity, leaving aside formulations in which the connection is independent — such as Palatini variants, whose junction conditions have been studied separately — and it assumes minimal coupling to matter, though the author notes that scalar-tensor theories with more general couplings can be attacked with the same techniques. Infinite-derivative Lagrangians, which have no maximum derivative order, remain harder: in principle they demand an infinitely differentiable Riemann tensor, and previous studies in that setting have had to add assumptions by hand. The author also flags that recent results obtained via the boundary-term, or Gibbons-Hawking-York, approach agree with the distributional analysis for quadratic theories except on the question of double layers, which the boundary method misses — raising doubts about the generality of that approach, which has underpinned some other recent claims in the literature.

The practical stakes are considerable. Thin-shell models are the standard tool for studying regular black holes formed by shell collapse, gravastars, thin-shell wormholes, brane-world cosmologies, and domain walls, and several recent constructions of regular black holes in higher-curvature gravity rest on junction technology. Using the wrong junction conditions in such models can silently introduce unphysical surface stresses or miss legitimate ones. The new work supplies a definitive checklist: continuity of the first fundamental form as the prerequisite, continuity of the extrinsic curvature in generic higher-curvature theories, continuity of successively higher derivatives of the Riemann tensor depending on the order of the action, the universal continuity of normal energy-momentum components for shell-free matchings, and special dispensations for the two exceptional families — GR with F(R), which alone host curvature shells and impulsive waves, and purely quadratic gravity, which alone can host gravitational double layers. For a field saturated with proposed modifications of Einstein’s theory, having the gluing rules for all of them in one mathematically rigorous package is a substantial step toward order in the generalized-gravity zoo.