Hyperboloidal Foliations in Relativity

• Hyperboloidal foliations are spacetime slicings by spacelike hypersurfaces that smoothly approach future null infinity, bridging the gap between Cauchy and characteristic methods.
• They use conformal compactification techniques and height functions to regularize wave equations, enabling accurate modeling in gravitational-wave and black-hole perturbation studies.
• This framework underpins both analytic and numerical approaches in general relativity, offering insights into quasinormal modes, energy flux conservation, and asymptotically hyperboloidal initial data.

Searching arXiv for recent and foundational work on hyperboloidal foliations to support the article. Hyperboloidal foliations are spacetime slicings by spacelike hypersurfaces that asymptotically approach null directions and intersect future null infinity $\mathscr{I}^+$, while retaining a standard spacelike initial-value formulation. In contemporary usage they occupy an intermediate position between standard Cauchy foliations, which reach spatial infinity $i^0$, and characteristic foliations, which are exactly null. The modern literature presents them both as a geometric device rooted in conformal methods and as a practical framework for high-accuracy gravitational-wave modeling, black-hole perturbation theory, and asymptotically hyperboloidal initial-value problems (Hilditch et al., 13 Sep 2025).

1. Geometric definition and conformal structure

A foliation of spacetime $(\mathcal{M},g_{\mu\nu})$ is a decomposition into non-intersecting hypersurfaces $\Sigma_t$ labeled by a time function $t$. In the standard $3+1$ Cauchy formulation, the $\Sigma_t$ are spacelike Cauchy surfaces with induced metric $\gamma_{ij}$ and extrinsic curvature $K_{ij}$. Hyperboloidal foliations preserve this spacelike structure, but their asymptotic endpoint is $\mathscr{I}^+$ rather than $i^0$0 (Hilditch et al., 13 Sep 2025).

Foliation type	Hypersurfaces	Asymptotic endpoint
Cauchy	spacelike	spatial infinity $i^0$1
Characteristic	null	null infinity $i^0$2
Hyperboloidal	spacelike, asymptotically null	future null infinity $i^0$3

This distinction is usually formulated through Penrose’s conformal compactification. One introduces an unphysical metric

$i^0$4

with $i^0$5 at $i^0$6 and $i^0$7 there. Hyperboloidal slices in the physical spacetime correspond to spacelike hypersurfaces in the unphysical spacetime that intersect the conformal boundary transversely. A recurring geometric requirement is smoothness of $i^0$8 at $i^0$9, because quantities such as Bondi mass, news, and peeling are defined in that conformal setting (Hilditch et al., 13 Sep 2025).

A persistent misconception is that hyperboloidal slices are null. They are not: they are strictly spacelike hypersurfaces whose asymptotics become null only at $(\mathcal{M},g_{\mu\nu})$0. This is precisely what allows one to combine radiative asymptotics with the standard machinery of spacelike evolution (Hilditch et al., 13 Sep 2025).

2. Height functions, compactification, and scri-fixing

The standard construction starts from a familiar coordinate system and deforms the time coordinate by a height function,

$(\mathcal{M},g_{\mu\nu})$1

chosen so that $(\mathcal{M},g_{\mu\nu})$2 is spacelike in the interior and asymptotic to outgoing null surfaces for large $(\mathcal{M},g_{\mu\nu})$3 (Hilditch et al., 13 Sep 2025). In flat spacetime a simple example is

$(\mathcal{M},g_{\mu\nu})$4

so that $(\mathcal{M},g_{\mu\nu})$5; the resulting slices are spacelike and asymptotically approach outgoing null cones (Hilditch et al., 13 Sep 2025).

To include infinity at finite coordinate location one combines the height-function transformation with radial compactification. In spherical symmetry a representative choice is

$(\mathcal{M},g_{\mu\nu})$6

so that $(\mathcal{M},g_{\mu\nu})$7 represents $(\mathcal{M},g_{\mu\nu})$8. This is the essence of scri-fixing: the spatial coordinate location of null infinity is kept fixed on the grid (0712.4333). The same paper constructs such coordinates explicitly on Minkowski, Schwarzschild, and Kerr spacetimes and emphasizes that the outer boundary is then the physical null boundary $(\mathcal{M},g_{\mu\nu})$9, not an artificial timelike cutoff (0712.4333).

In $\Sigma_t$0 Minkowski space, a specific hyperboloidal chart used in later analysis is

$\Sigma_t$1

Here $\Sigma_t$2, and the endpoints $\Sigma_t$3 correspond to future null infinity in the compactified picture (Kroon et al., 2024). This compact domain with pure-outflow endpoints is one of the central analytic and numerical advantages of the hyperboloidal setting.

For Kerr, a generic hyperboloidal framework is built by introducing

$\Sigma_t$4

with conformal factor $\Sigma_t$5. The free functions $\Sigma_t$6 and the regular part of the height function encode the hyperboloidal gauge. The “minimal gauge” fixes these functions so that the metric and Teukolsky equation simplify while still foliating Kerr by slices running from $\Sigma_t$7 to $\Sigma_t$8 (Macedo, 2019).

3. Hyperboloidal wave equations, null boundaries, and energy flux

A major payoff of hyperboloidal foliations is that wave equations become regular up to null infinity after conformal rescaling. For a scalar field $\Sigma_t$9 satisfying $t$0, one introduces

$t$1

and obtains, in four dimensions,

$t$2

With a scri-fixing choice of slicing, the coefficients remain regular at $t$3, so the conformally rescaled equation can be solved directly on the compactified domain (Hilditch et al., 13 Sep 2025).

The de Sitter toy model makes the mechanism completely explicit. On the static patch

$t$4

the hyperboloidal time

$t$5

transforms the metric into

$t$6

which is regular on $t$7. The Klein–Gordon equation becomes

$t$8

The endpoints $t$9 are null boundaries, so no boundary conditions are imposed there; instead, radiation escapes through them (Bizoń et al., 2020).

In that model the energy density and flux satisfy a conservation law

$3+1$0

with Bondi-type energy

$3+1$1

and

$3+1$2

The energy decay is therefore exactly the flux of outgoing radiation across the horizons (Bizoń et al., 2020). This geometrically encoded dissipation is one of the defining features of hyperboloidal evolution.

In the $3+1$3 Minkowski setting, the hyperboloidal analogue of d’Alembert’s formula shows that finite propagation and outflow survive on the compactified hyperboloidal domain, but with a nontrivial redistribution of initial data. In particular, for compactly supported $3+1$4 and $3+1$5, the late-time limit may be a nonzero $3+1$6-independent constant. The paper identifies this as a “permanent displacement” intrinsic to hyperboloidal evolution rather than a numerical artifact (Kroon et al., 2024). This suggests that late-time limits are foliation-dependent even in flat-space toy models.

4. Quasinormal modes on hyperboloidal slices

Hyperboloidal foliations provide a natural formulation of quasinormal modes as eigenmodes of a non-self-adjoint evolution operator. In the de Sitter toy model, substituting

$3+1$7

into the hyperboloidal wave equation yields a regular ODE on the compact interval $3+1$8. In the massless case, the smoothness requirement at the endpoints quantizes the spectrum to

$3+1$9

with eigenfunctions

$\Sigma_t$0

In this formulation no explicit outgoing-wave condition is imposed; outgoing radiation is already encoded in the geometry of the foliation (Bizoń et al., 2020).

For black-hole perturbations the same principle extends to asymptotically flat spacetimes. In the Reissner–Nordström case, a hyperboloidal time $\Sigma_t$1 and a compactified radial coordinate $\Sigma_t$2 transform the master wave equation into a regular PDE on a compact interval whose endpoints are the event horizon and $\Sigma_t$3. After Laplace transform in $\Sigma_t$4, one obtains an ODE on $\Sigma_t$5, and the inverse Laplace transform yields a spectral decomposition into QNM poles plus branch-cut tails (Macedo et al., 2018). The same framework also clarifies the origin of the regularisation factors in Leaver’s Cauchy-based formalism: in the hyperboloidal picture they arise from the Jacobian factor relating regular hyperboloidal fields to Cauchy fields (Macedo et al., 2018).

For Kerr, the hyperboloidal framework reformulates the Teukolsky equation on slices extending from $\Sigma_t$6 to $\Sigma_t$7. The minimal gauge introduced for Kerr makes the metric and the radial-frequency Teukolsky equation sufficiently simple for both time-domain and frequency-domain analysis, and it provides the spacetime interpretation of Leaver’s regularisation factors. The same framework also exhibits two distinct extremal limits, namely standard extremal Kerr and the near-horizon geometry, depending on the hyperboloidal gauge (Macedo, 2019).

A $\Sigma_t$8 modal Teukolsky solver on hyperboloidal foliations of Kerr uses the compactified radial coordinate

$\Sigma_t$9

and hyperboloidal time

$\gamma_{ij}$0

so that $\gamma_{ij}$1 sits at $\gamma_{ij}$2 and the horizon at $\gamma_{ij}$3. In this formulation the characteristic structure shows pure outflow at both boundaries, and symmetric time integrators can be used without a Courant restriction for the semi-discrete linear system (Markakis et al., 2023). In the broader perturbative literature, this hyperboloidal formulation has become closely tied to QNM expansions, late-time tails, self-force calculations, and EMRI waveform production (Hilditch et al., 13 Sep 2025).

5. Asymptotically hyperboloidal initial data and geometric invariants

Hyperboloidal foliations also arise at the level of initial data. In Minkowski space, a hyperboloidal time function

$\gamma_{ij}$4

with

$\gamma_{ij}$5

produces slices asymptotic to the unit hyperboloid

$\gamma_{ij}$6

The induced background metric is

$\gamma_{ij}$7

with lapse and shift

$\gamma_{ij}$8

for the standard hyperboloid (Sancassani et al., 17 Apr 2025).

Within this setting, asymptotically hyperboloidal initial data sets $\gamma_{ij}$9 are those whose geometry tends to the hyperboloidal model $K_{ij}$0 at infinity. Using Michel’s geometric-invariant formalism, one obtains charges of energy and linear momentum associated with the KIDs corresponding to time and spatial translations. The paper introduces E–P chargeability, proves that it is preserved under Einstein evolution for the chosen hyperboloidal time function, and derives energy-loss and linear-momentum-loss formulas along the hyperboloidal foliation that coincide with the Bondi–Sachs–Metzner formulas while working under weaker asymptotic assumptions (Sancassani et al., 17 Apr 2025).

A different but related development concerns foliations of asymptotically hyperboloidal initial data sets by closed $K_{ij}$1-surfaces of constant spacetime mean curvature. For initial data close to the anti-de Sitter-Schwarzschild hyperboloid, an exhaustive family of STCMC surfaces is obtained as the long-time limit of the volume preserving spacetime mean curvature flow started from the Neves–Tian CMC foliation. This yields a foliation of the asymptotic end and, in balanced coordinates with diagonal mass aspect tensor $K_{ij}$2, the barycenters of the STCMC leaves converge to the hyperbolic origin (Tenan, 2 Jul 2026). These are not spacetime slicings themselves but foliations of a hyperboloidal initial-data hypersurface; the distinction is essential.

6. Subtleties, limitations, and current direction

The hyperboloidal picture does not remove every analytical difficulty. A particularly sharp example arises in the theory of QNM bilinear products. Although hyperboloidal QNM solutions are smooth and finite on future-directed hyperboloids, the integrand of the bilinear form with respect to which the modes are orthogonal can still diverge. The reason identified in recent work is the reflection, equivalently CPT, transformation appearing in the definition of the products, which changes the boundary behaviour of the integrand. Several regularisation procedures and an alternative flux-based definition are therefore introduced to obtain finite bilinear forms, excitation factors, and excitation coefficients (Minucci et al., 14 Apr 2026). This is a useful corrective to the widespread intuition that hyperboloidal regularity by itself makes every mode product finite.

A second limitation is structural rather than local. The topical survey of the field emphasizes that linear problems on fixed backgrounds are now comparatively mature, whereas fully nonlinear $K_{ij}$3-dimensional hyperboloidal evolutions of strong-field binaries remain an open frontier. The same survey places current work in a lineage running from Penrose’s compactification and Friedrich’s conformal field equations to scri-fixing, hyperboloidal layers, high-order numerical methods, and gravitational-wave applications (Hilditch et al., 13 Sep 2025).

This suggests a characteristic trajectory for the subject. Hyperboloidal foliations began as tools in mathematical relativity for treating null infinity as a regular boundary; they now function simultaneously as a conformal-geometric framework, a boundary-adapted PDE formulation, and a numerical strategy for placing “infinity on the grid.” Their contemporary significance lies precisely in that synthesis.