Hyperboloidal Foliation Technique

• Hyperboloidal foliation technique is a framework that defines spacetime slices extending to future null infinity, eliminating artificial outer boundaries.
• It unifies wave and Klein–Gordon equations by using adapted hyperboloidal coordinates, energy estimates, and conformal compactification to optimize decay rates.
• The method has wide-ranging applications in analytical proofs, numerical evolutions, black-hole physics, and cosmological simulations.

Searching arXiv for fresh supporting papers on hyperboloidal foliation technique. The hyperboloidal foliation technique is a framework in which spacetime is foliated by spacelike hypersurfaces that reach future null infinity rather than spatial infinity. In the formulation presented in the monograph “The hyperboloidal foliation method,” the method is based on a $(3+1)$-foliation of Minkowski spacetime by hyperboloidal hypersurfaces, is geometric in nature and invariant under Lorentz transformations, and is used to establish global-in-time existence results, uniform energy bounds, and optimal rates of decay in time for systems of nonlinear wave equations; it also places the wave equation and the Klein–Gordon equation in a unified framework and applies to nonlinear wave–Klein–Gordon systems, including systems involving a massive scalar field such as the Dirac–Klein–Gordon system (LeFloch et al., 2014). In numerical and geometric applications, the same basic idea is paired with conformal compactification or related compactifying maps so that future null infinity $\mathscr I^+$ is represented at a finite coordinate location, eliminating the artificial timelike outer boundary that appears in standard truncated Cauchy evolutions (0712.4333).

1. Geometric definition and asymptotic role

A hyperboloidal slice is a spacelike hypersurface in an asymptotically flat spacetime that extends to future null infinity $I^+$. Such slices are spacelike everywhere, do not approach spatial infinity $i^0$ like standard asymptotically Euclidean Cauchy slices, and extend through null infinity as spacelike hypersurfaces. A characteristic feature emphasized in the scri-fixing literature is that their mean extrinsic curvature $\tilde K$ tends to a finite nonzero limit asymptotically; in the sign convention used there, future hyperboloidal slices have positive asymptotic $\tilde K$ (0712.4333).

This geometric placement changes the asymptotic organization of the evolution problem. Standard constant-time slices intersect $i^0$, whereas hyperboloidal slices “bend upward” toward outgoing null directions so that radiation leaves the domain through $I^+$. This is why the method is repeatedly used for radiative problems: outgoing radiation is represented where it is geometrically defined, rather than reconstructed indirectly from large-radius asymptotics. The same geometric feature underlies the PDE-analytic version of the method, where hyperboloids inside the light cone are used to organize decay for both wave and Klein–Gordon fields (Ma, 2011).

A recurrent clarification in the literature is that hyperboloidal slices are not null. They are sometimes described informally as “asymptotically null,” but the scri-fixing analysis explicitly warns against that phrasing: their tangent vectors never become null, and a surface that actually became null would have qualitatively different geometric properties (0712.4333). This distinction matters because the technique relies on retaining a spacelike $3+1$ formulation while still reaching null infinity.

2. Canonical construction in Minkowski space

In Minkowski spacetime, a standard model family is

$\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$

or, in equivalent notation,

$\mathscr I^+$0

These are the future hyperboloids used in the PDE literature on wave–Klein–Gordon systems (LeFloch et al., 2022, Ma, 2011).

A closely related coordinate construction introduces a hyperboloidal time

$\mathscr I^+$1

so that the level sets $\mathscr I^+$2 are hyperboloids of constant mean curvature. They asymptote to outgoing null cones as $\mathscr I^+$3, since $\mathscr I^+$4, and therefore reach $\mathscr I^+$5 instead of spatial infinity. In the multidimensional nonlinear-wave setting, this is combined with a compactifying radial coordinate $\mathscr I^+$6 chosen so that the induced spatial conformal metric becomes flat, with explicit map

$\mathscr I^+$7

hence $\mathscr I^+$8 corresponds to $\mathscr I^+$9, that is, to $I^+$0. The conformal factor is

$I^+$1

and $I^+$2 is fixed at the finite coordinate value $I^+$3; this is a scri-fixing compactification (Rinne, 1 Jul 2025).

The same geometric pattern appears in the more general scri-fixing formulation, where one chooses a compactifying radius $I^+$4 as the area radius in the conformal geometry and writes

$I^+$5

With the explicit choice $I^+$6, one has $I^+$7, so $I^+$8 is the location of future null infinity (0712.4333). This suggests a common structural principle across analytic and numerical work: a hyperboloidal time transformation is paired with a radial compactification chosen so that the asymptotic region becomes a finite boundary without changing the causal meaning of outgoing propagation.

3. Energy method, decay theory, and wave–Klein–Gordon unification

In the analytic literature, the central use of hyperboloids is to recast energy estimates in a geometry adapted simultaneously to wave and Klein–Gordon propagation. For the linear equation

$I^+$9

the hyperboloidal energy on $i^0$0 is defined by

$i^0$1

For $i^0$2 this controls tangential derivatives $i^0$3 and the weighted normal part $i^0$4; for $i^0$5 it also contains the coercive $i^0$6 term needed for Klein–Gordon analysis. The corresponding energy estimate is

$i^0$7

This is one of the core analytic mechanisms behind the method’s unification of the massless and massive cases (Ma, 2011).

The adapted derivatives are equally important. In the semi-hyperboloidal frame,

$i^0$8

the $i^0$9 are tangent to $\tilde K$0 and are regarded as good derivatives. Lorentz boosts are tangent to the hyperboloids and commute with $\tilde K$1, so the commutation algebra

$\tilde K$2

propagates high-order control without using the scaling vector field, which is incompatible with the Klein–Gordon operator $\tilde K$3 (LeFloch et al., 2022). In this setting, hyperboloidal Sobolev inequalities convert commuted $\tilde K$4 energies into pointwise decay.

The 2022 revisitation of the method shows how this framework can be strengthened when standard hyperboloidal energy is insufficient for almost bounded energy growth in coupled systems. For the model

$\tilde K$5

posed on an initial hyperboloid $\tilde K$6, the new ingredient is a hierarchy of fractional Morawetz energy estimates for the wave component, defined from two conformal transformations, with the optimal case corresponding to using the scaling vector field as a multiplier for the wave component only. The resulting “conformal hyperboloidal energy” keeps the original hyperboloidal machinery for the coupled system while adding stronger weighted control for the wave field (LeFloch et al., 2022).

A further extension appears in the Euclidian–Hyperboloidal Foliation Method, which glues hyperboloidal leaves in the interior to Euclidean leaves in the far exterior. The leaves are hyperboloidal in $\tilde K$7, Euclidian in $\tilde K$8, and smoothly interpolated in a transition region near the light cone. This is designed for non-compactly supported asymptotically flat data and for Einstein–massive field systems, precisely because it preserves the hyperboloidal advantages for Klein–Gordon decay while restoring Euclidean control in spacelike directions (LeFloch et al., 2017).

4. Conformal compactification, scri-fixing, and numerical hyperboloidal evolution

In numerical work, hyperboloidal foliations are typically paired with compactification so that future null infinity is included in the computational domain. The scri-fixing formulation states the motivation directly: one avoids an artificial timelike outer boundary in asymptotically flat problems by including $\tilde K$9 in the domain and fixing its coordinate location in time (0712.4333). This addresses both the outer-boundary problem and the radiation-extraction problem.

For multidimensional nonlinear wave equations,

$\tilde K$0

the hyperboloidal numerical method replaces usual Cauchy slicing with a hyperboloidal foliation of Minkowski spacetime and conformal compactification, so that future null infinity $\tilde K$1 sits at a finite grid location. The practical benefits stated in that work are: no artificial outer boundary, direct access to null infinity, and avoidance of the asymptotic “blueshift” problem that appears if one compactifies the radial coordinate without changing the time slicing (Rinne, 1 Jul 2025). The field is rescaled by

$\tilde K$2

which yields a regular conformally transformed wave equation on the compactified domain.

A particularly influential implementation is the hyperboloidal layer method for Regge–Wheeler–Zerilli evolution in Schwarzschild. There the inner region is left unchanged in standard Schwarzschild/tortoise coordinates, while an outer shell is deformed into hyperboloidal slices and compactified so that $\tilde K$3 lies at a finite coordinate value. The time transformation is

$\tilde K$4

and the compactifying coordinate $\tilde K$5 is chosen so that outgoing null rays keep the form

$\tilde K$6

With

$\tilde K$7

the point $\tilde K$8 represents $\tilde K$9. The radial characteristic speeds become

$i^0$0

so $i^0$1 at $i^0$2, and no outer boundary condition is needed there (Bernuzzi et al., 2011).

A complementary line of work combines hyperboloidal coordinates with the dual-foliation formalism. In that framework, one keeps the evolution variables in an asymptotically Minkowskian tensor basis and uses hyperboloidal coordinates only for the computational covering. The basic transformation is

$i^0$3

and generalized compactifications with $i^0$4 are used so that formally singular terms in the transformed generalized harmonic system admit regular or trivial limits at null infinity. A further generalization replaces the fixed height function by a dynamically evolved optical function $i^0$5, with

$i^0$6

so that the outgoing coordinate speed can be maintained at $i^0$7 exactly (Hilditch et al., 2016). In spherical symmetry, the generalized harmonic implementation uses either a height function or an eikonal equation, includes hyperboloidal layers, and evolves compactified hyperboloidal initial value problems with a massless scalar field, regular center, or black-hole excision, while extracting Bondi quantities at $i^0$8 (Peterson et al., 2024).

5. Black-hole, Kerr, self-force, and cosmological extensions

Hyperboloidal foliations have been extended far beyond Minkowski-space model problems. In the Kerr framework, generic compact hyperboloidal coordinates $i^0$9 are introduced by

$I^+$0

so that $I^+$1 is future null infinity and the slices connect the future horizon $I^+$2 to $I^+$3. For the Teukolsky equation, the regularized hyperboloidal master field is

$I^+$4

with $I^+$5 regularizing the horizon and $I^+$6 regularizing null infinity. In the frequency domain, the minimal gauge yields a formulation identified as the spacetime counterpart of Leaver’s formalism (Macedo, 2019).

A related Schwarzschild frequency-domain development for gravitational self-force adopts hyperboloidal slicing in ingoing Eddington–Finkelstein coordinates,

$I^+$7

with minimal gauge

$I^+$8

and explicit hyperboloidal height function

$I^+$9

for $3+1$0. The RWZ master functions are rescaled by

$3+1$1

so that retarded boundary behavior is encoded as regularity of the conformal field $3+1$2 on the compact interval $3+1$3, where $3+1$4 and $3+1$5 (Leather, 2024).

Hyperboloidal ideas have also been adapted to black-hole scattering and integrability. In one such formulation, Schwarzschild perturbations are rewritten in compactified hyperboloidal coordinates

$3+1$6

mapping the full tortoise line onto a finite interval $3+1$7 whose endpoints represent the horizon and future null infinity. The resulting time generator is non-selfadjoint, and its adjoint contains boundary delta terms supported at $3+1$8, reflecting the open nature of scattering through the horizon and null infinity (Vitel, 8 Dec 2025).

In cosmology, the same geometric strategy is used on expanding FLRW backgrounds. For spatially flat FLRW in conformal time,

$3+1$9

one introduces

$\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$0

Then $\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$1 at $\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$2 places $\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$3 at a finite numerical boundary. The numerical work uses this compactified hyperboloidal foliation to study linear decay and semilinear long-time dynamics in expanding universes, with no physical outer boundary condition imposed at $\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$4 (Rossetti et al., 27 Feb 2025).

6. Variants, misconceptions, and related constructions

A persistent misconception is that all hyperboloidal effects are purely numerical artifacts. The exact $\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$5-dimensional analysis of the wave equation on a hyperboloidal slice shows otherwise. With

$\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$6

the transformed wave equation has pure-outflow boundaries at $\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$7, since there it reduces to

$\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$8

The corresponding hyperboloidal analogue of d’Alembert’s formula shows that even when the initial hyperboloidal time derivative vanishes, compactly supported bump data in $\mathcal H_s := \{(t,x): t>0,\ t^2-r^2=s^2\},$9 can leave a late-time constant offset. The paper interprets this “permanent displacement” as a genuine feature of hyperboloidal slicing rather than numerical contamination (Kroon et al., 2024).

Another important distinction concerns what counts as a hyperboloidal foliation. In PDE analysis and numerical relativity, the term usually refers to a foliation of spacetime by spacelike hypersurfaces reaching $\mathscr I^+$00. In differential geometry, however, closely related asymptotic structures can appear as foliations by $\mathscr I^+$01-surfaces inside a hyperboloidal initial data set. The 2026 construction of constant spacetime mean curvature surfaces for asymptotically hyperboloidal initial data is of that second kind: it builds an exhaustive family of asymptotically round STCMC leaves near infinity, using the volume preserving spacetime mean curvature flow, and uses them to study center of mass. This is directly relevant to hyperboloidal asymptotics, but it is not the same object as the spacetime hyperboloidal foliations used in wave estimates and numerical evolution (Tenan, 2 Jul 2026).

Finally, the technique continues to acquire more structured numerical realizations. A recent fully three-dimensional SBP scheme for linear wave equations on hyperboloidal slices works in spherical polar coordinates, keeps grid points at the origin, on the axis, and at $\mathscr I^+$02, and uses compactification together with rescaling. One of its key observations is the relation

$\mathscr I^+$03

which simplifies the equations and energy density, while conditions on constraint-addition terms are identified to maintain symmetric hyperbolicity (Reddy et al., 1 Jun 2026). This suggests that the hyperboloidal foliation technique is no longer only a geometric or analytic device; it is also becoming a mature discretization framework for stable computations extending all the way to null infinity.