Lemaître-Tolman-Bondi (LTB) Cosmology

• LTB is the most general spherically symmetric inhomogeneous dust solution of Einstein’s equations, characterized by shell-dependent radius, mass, curvature, and bang-time functions.
• The model distinguishes three dynamical regimes—elliptic, parabolic, and hyperbolic—by the sign of the curvature function, linking the dynamics to local expansion rates and void formation.
• LTB is widely applied in cosmology for void modeling, gravitational collapse, and reconstructing light-cone observables, with recent studies extending its framework to modified gravity and quantum dynamics.

Searching arXiv for recent and relevant LTB papers to satisfy the requirement for fresh literature support. arxiv_search(query="Lemaître-Tolman-Bondi review cosmology perturbations", max_results=10) arxiv_search(query="Lemaître-Tolman-Bondi cosmology perturbations", max_results=10) Searching "Lemaître-Tolman-Bondi cosmology perturbations". Lemaître–Tolman–Bondi (LTB) denotes the most general spherically symmetric solution of Einstein’s equations with zero pressure in comoving synchronous gauge. In its standard form it describes a radially inhomogeneous dust spacetime through shell-dependent areal radius, mass, curvature, and bang-time functions, with one radial coordinate freedom reducing the three free functions to two physical degrees of freedom. Because the model is exact rather than perturbative, LTB has become a canonical framework for void cosmology, gravitational collapse, inverse reconstruction from light-cone observables, relativistic structure formation, and a wide range of extensions involving pressure, modified gravity, thermodynamics, and quantum dynamics (Yoo, 2010, Sussman, 2010).

1. Geometric definition and Einstein dynamics

In a common notation, the LTB line element is

$$ds^2=-dt^2+\frac{R'(t,r)^2}{1+2E(r)}\,dr^2+R(t,r)^2\,d\Omega^2,$$

where $R(t,r)$ is the areal radius and $E(r)$ is the energy or curvature function. Equivalent forms replace $R$ by $A$ and $2E(r)$ by $-k(r)$, so that

$$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$

For dust, Einstein’s equations reduce to a shellwise Friedmann-like equation and a density relation,

$$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$

The function $M(r)$ is the effective mass inside coordinate radius $R(t,r)$0, while the bang-time function $R(t,r)$1 is defined implicitly by $R(t,r)$2 and encodes inhomogeneity in the local Big Bang time (Yoo, 2010).

An alternative but equivalent notation introduces a Misner–Sharp mass function $R(t,r)$3 through

$R(t,r)$4

with field equations

$R(t,r)$5

and the first integral

$R(t,r)$6

This formulation is convenient when LTB is embedded in a more general perfect-fluid or thermodynamic analysis (Chakraborty et al., 2010).

For cosmological applications it is often useful to define two local expansion rates,

$R(t,r)$7

or equivalently $R(t,r)$8 and $R(t,r)$9. The shellwise dynamics then resembles an FRW evolution with explicitly radius-dependent matter and curvature content. This shellwise interpretation is central in void cosmology and in numerical tests of relativistic structure growth (Alonso et al., 2010).

2. Free functions, gauge choices, and exact solution classes

The standard LTB solution is determined by the radial functions $E(r)$0, $E(r)$1, and $E(r)$2, together with one coordinate freedom $E(r)$3. After gauge fixing, two physical functional degrees of freedom remain. A frequently used gauge is to normalize the present-day areal radius by $E(r)$4, which turns the model into an initial-value problem on a chosen time slice (Yoo, 2010).

The sign of the energy function separates the exact solutions into three classes. When $E(r)$5, the evolution is elliptic; when $E(r)$6, parabolic; and when $E(r)$7, hyperbolic. In parametric form these branches read

$E(r)$8

for the elliptic case with $E(r)$9,

$R$0

for the parabolic case, and

$R$1

for the hyperbolic case with $R$2. These exact branches are the basis for collapse studies, void models, and FRW limits (Bochicchio et al., 2011).

A complementary reformulation uses quasi-local integral scalars $R$3, $R$4, and $R$5, together with exact relative fluctuations $R$6. In that language, the LTB dynamics takes an FRW-like form,

$R$7

with $R$8. The same framework recasts shell-crossing avoidance as Hellaby–Lake inequalities on initial data, while regularity at the symmetry centre requires $R$9, $A$0, $A$1, and $A$2 (Sussman, 2010).

3. Void cosmology and reconstruction from observations

A major cosmological use of LTB is the modeling of a gigaparsec-scale underdensity embedded in an asymptotic Einstein–de Sitter background. In the Garcia-Bellido–Haugbølle parameterization one specifies a radial matter fraction

$A$3

and either an independent $A$4 or, in the constrained version, a homogeneous bang time $A$5, which fixes $A$6 as a function of $A$7. The resulting models are controlled by $A$8, $A$9, $2E(r)$0, and $2E(r)$1 (Alonso et al., 2010).

When confronted with Type Ia supernovae, CMB, and BAO data, asymptotically flat LTB void models were found to accommodate all observations within $2E(r)$2, with best-fit $2E(r)$3 very close to that of $2E(r)$4CDM. In the specific analysis of GBH and constrained GBH profiles, the total $2E(r)$5 values were $2E(r)$6 for both void models and $2E(r)$7 for $2E(r)$8CDM, while Bayesian evidence mildly preferred the homogeneous model because of the extra void parameters (0802.1523).

The inverse problem asks which LTB spacetime reproduces prescribed light-cone observables such as the angular-diameter distance $2E(r)$9 and redshift-space mass density $-k(r)$0. In a light-cone gauge, these data generate a closed ODE system for $-k(r)$1, $-k(r)$2, $-k(r)$3, and $-k(r)$4. A globally regular solution exists if and only if

$-k(r)$5

where $-k(r)$6 is the unique maximum of $-k(r)$7. For flat $-k(r)$8CDM observables with $-k(r)$9, this condition is satisfied, and the reconstructed LTB model exhibits a gentle central underdensity of order $$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$0 relative to the asymptotic value (Yoo, 2010).

Redshift drift adds an independent metric observable to distance fitting. In LTB, the drift contains both endpoint expansion terms and an explicit line-of-sight integral over the radial gradient of the longitudinal Hubble rate,

$$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$1

Even LTB models constructed to reproduce Hubble diagrams indistinguishable from flat $$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$2CDM typically predict different redshift-drift signals, and the drift can therefore reduce the functional degeneracy of spherically symmetric inhomogeneous cosmologies (Codur et al., 2021).

Recent work on the constrained GBH parameterization has also produced symbolic-regression surrogates for $$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$3, $$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$4, $$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$5, $$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$6, and $$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$7. Over the domain $$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$8, with $$ds^2=-dt^2+\frac{A'(r,t)^2}{1-k(r)}\,dr^2+A^2(r,t)\,d\Omega^2.$$9, $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$0, $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$1 Gpc, and $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$2 Gpc, the relative mean error remains below $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$3 percent for all quantities except the radial Hubble function, where it reaches $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$4 percent (Carvalho et al., 22 Mar 2026).

4. Structure formation, perturbations, and averaging

Direct N-body tests show that Newtonian simulations can reproduce the relativistic nonlinear shell dynamics of LTB voids. In simulations of a large underdense region embedded in an Einstein–de Sitter background, the radial density and velocity profiles agree with the exact LTB solution to better than a few percent over the entire void except for the most extreme cases, and the local matter density contrast at the void centre grows from $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$5 at $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$6 to $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$7 by $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$8. The same analysis shows that the local growth deep in the void follows that of an open universe with $$\dot R^2=\frac{2M(r)}{R}+2E(r),\qquad 4\pi G\,\rho=\frac{M'(r)}{R^2R'}.$$9, while large radii recover the Einstein–de Sitter scaling $M(r)$0 (Alonso et al., 2010).

Linear perturbations on an LTB background are substantially more intricate than in FLRW because background inhomogeneity couples gauge-invariant perturbations already at first order. In a Regge–Wheeler gauge and after spherical-harmonic decomposition, the perturbations satisfy a coupled system of linear PDEs for the metric variables $M(r)$1, $M(r)$2, and $M(r)$3, together with constraint equations for the fluid perturbations. Numerical evolution with finite element methods and Gaussian random initial conditions shows significant couplings up to $M(r)$4 for large and deep gigaparsec-scale voids of the type invoked to fit supernova distance-redshift relations (Meyer et al., 2014).

At the same time, LTB perturbation theory admits exact gauge-invariant constructions analogous to the conserved curvature perturbations of homogeneous cosmology. In particular, the spatial metric trace perturbation

$M(r)$5

can be combined into a gauge-invariant quantity $M(r)$6 whose evolution is

$M(r)$7

For a barotropic fluid, $M(r)$8, so $M(r)$9 is conserved exactly on all scales (Leithes et al., 2014).

Quasi-local scalars provide another exact restructuring of the problem. In this formulation the LTB spacetime becomes a nonlinear spherical perturbation of an FLRW background, and the same variables connect naturally with Buchert averaging. In an explicit small-curvature LTB model with inhomogeneous bang time, the averaged Hubble rate satisfies

$R(t,r)$00

where $R(t,r)$01 is determined by the curvature at the boundary of the averaging domain and $R(t,r)$02 depends on integrals over $R(t,r)$03 and $R(t,r)$04. In that toy model, backreaction vanishes for homogeneous bang time $R(t,r)$05 (Sussman, 2010, Isidro et al., 2016).

5. Horizons, thermodynamics, and local-system effects

For a perfect-fluid LTB spacetime written as

$R(t,r)$06

the trapping horizon coincides with the apparent horizon and is given by

$R(t,r)$07

where $R(t,r)$08 is the Misner–Sharp mass function. On this background the unified first law,

$R(t,r)$09

with $R(t,r)$10, is equivalent to the Einstein equations. Assuming the Clausius relation at the horizon and thermal equilibrium between the horizon and the enclosed matter, one obtains explicit conditions for the generalized second law on the apparent horizon for a perfect fluid and on the event horizon for holographic dark energy (Chakraborty et al., 2010).

LTB has also been used to quantify how cosmological inhomogeneity enters local Fermi-frame dynamics. The leading local tidal potential can be written as

$R(t,r)$11

with

$R(t,r)$12

Solar-system phenomenology gives $R(t,r)$13 and $R(t,r)$14 from inner-planet perihelia, $R(t,r)$15 from the Cassini Earth–Saturn range, and $R(t,r)$16–$R(t,r)$17 from Oort-cloud arguments. These bounds remain many orders of magnitude above the cosmological scale $R(t,r)$18 inferred from LTB cosmological fits (Iorio, 2010).

A separate local-dynamics construction uses an exact parabolic LTB model with two density peaks in a flat dust background. In that solution, the proper separation of the two peaks grows like $R(t,r)$19 at late times, so the system is strongly bound yet exactly comoving with the cosmic substratum (Bochicchio et al., 2011).

6. Extensions beyond dust and contemporary developments

Although standard LTB is a dust solution, several extensions retain much of the same structure. If the pressure is uniform in space at each comoving time, the spherically symmetric equations still admit an LTB-like form with

$R(t,r)$20

where $R(t,r)$21 evolves when $R(t,r)$22. The homogeneous reduction recovers the usual Friedmann equations with matter pressure and $R(t,r)$23, while explicit inhomogeneous models exhibit growing condensations and black-hole formation (Lynden-Bell et al., 2016).

In Palatini $R(t,r)$24 gravity, a generalized LTB spacetime for dissipative dust keeps the same basic metric ansatz but replaces the GR source by an effective tensor involving $R(t,r)$25 and its derivatives. The generalized Misner–Sharp mass,

$R(t,r)$26

still exists, but the expansion, shear, and structure scalars acquire explicit $R(t,r)$27-derivative corrections. The resulting generalized LTB spacetime retains properties comparable with LTB and yields structure scalars with a similar dependence on the material profile even in Palatini gravity (Bhatti et al., 2021).

Other extensions treat mixtures of baryons, cold dark matter, and dark energy. In one such formulation the quasi-local Einstein equations reduce to a first-order 7-dimensional autonomous dynamical system for $R(t,r)$28, $R(t,r)$29, $R(t,r)$30, and the corresponding inhomogeneous fluctuations. The homogeneous subsystem has four fixed points, including a future attractor $R(t,r)$31 corresponding to a homogeneous CDM plus dark-energy phase and a past repeller $R(t,r)$32 corresponding to a baryon-dominated era. Shells with $R(t,r)$33 can undergo a turn-around and bounce (Blanquet-Jaramillo et al., 2022).

The same exact framework can be specialized to nonstandard geometries. A particular choice of $R(t,r)$34, $R(t,r)$35, and $R(t,r)$36 produces an LTB wormhole embedded in a closed Friedmann universe, with a throat at a local minimum of the areal radius. Null geodesics in this dynamic background define a wormhole shadow whose angular diameter exhibits a non-monotonic dependence on observation time: an initial decrease is followed by growth as cosmic expansion dominates (Bronnikov et al., 11 Sep 2025).

At the quantum level, the marginally bound LTB dust model has been loop-quantized as a collection of non-interacting shells, each with an individual single-shell loop quantum dynamics. The loop Hamiltonian is a difference operator on superselection sectors, is essentially self-adjoint, and yields a universal density bound $R(t,r)$37. Wave packets initially peaked on collapsing trajectories undergo a bounce at Planckian energy densities and then follow an expanding classical branch, whereas the corresponding Wheeler–DeWitt quantization admits unbounded $R(t,r)$38 and depends on self-adjoint extension data. Near the bounce, interference patterns can suppress the accuracy of the effective theory close to the centre of the dust cloud (Cafaro et al., 5 Mar 2026).