---
title: Spacetime Positive Mass Theorem
url: https://www.emergentmind.com/topics/spacetime-positive-mass-theorem
type: topic
---

# Spacetime Positive Mass Theorem

The spacetime positive mass theorem is the statement that, for an asymptotically flat initial data set \((M^n,g,k)\) satisfying the dominant energy condition, the total ADM energy–momentum is future causal, equivalently \(E\ge |P|\), so the associated mass is nonnegative. In the initial-data formulation this is a theorem about a Riemannian metric \(g\), a second fundamental form \(k\), and the Einstein constraint equations on a spacelike slice; in the rigidity case, vanishing mass characterizes flat spacetime data. Modern treatments extend the theorem beyond the classical \(3+1\)-dimensional setting to arbitrary dimensions, asymptotically hyperboloidal ends, boundary problems, singular or creased data, and several quantitative or quasi-local variants [2604.24746].

## 1. Initial-data formulation and classical statement

An initial data set for the Einstein equations is a triple \((M^n,g,k)\), where \(M^n\) is an \(n\)-dimensional manifold, \(g\) is a Riemannian metric, and \(k\) is a symmetric \(2\)-tensor interpreted as the second fundamental form of the embedding of \(M\) as a spacelike hypersurface in a spacetime \((\mathcal{M}^{1+n},\mathbf g)\). The constraint equations are written as
\[
16\pi\mu = R_g + (\operatorname{Tr}_g k)^2 - |k|_g^2,
\]
\[
8\pi J = \operatorname{div}_g\big(k - (\operatorname{Tr}_g k) g\big),
\]
where \(\mu\) is the energy density, \(J\) is the momentum density, and \(R_g\) is the scalar curvature. The dominant energy condition is
\[
\mu \ge |J|_g.
\]
For asymptotically flat data, the ADM energy and linear momentum are defined by flux integrals at infinity,
\[
E = \frac{1}{2(n-1)\omega_{n-1}} \lim_{r\to\infty} \int_{S_r} (g_{ij,i} - g_{ii,j}) \nu^j \, dS,
\]
\[
P_i = \frac{1}{2(n-1)\omega_{n-1}} \lim_{r\to\infty} \int_{S_r} (k_{ij} - (\operatorname{Tr}_g k) g_{ij}) \nu^j \, dS,
\]
and the ADM mass is
\[
m = \sqrt{E^2 - |P|^2}.
\]
In the time-symmetric case \(k=0\), the theorem reduces to the Riemannian positive mass theorem, since \(J=0\) and \(R_g\ge 0\) follows from the dominant energy condition [1906.11352].

In its standard spacetime form, the theorem asserts that for a complete asymptotically flat initial data set satisfying the dominant energy condition, one has
\[
E \ge |P|,
\]
and hence \(m\ge 0\). In the classical spinorial or low-dimensional formulations, equality corresponds to data arising from Minkowski spacetime. A convenient scalar quantity used in one geometric approach is the energy–momentum scalar curvature
\[
S_{\mathrm{em}}(g,h) := 2(\mu - |J|_g)
= S(g) - \left(|h|_g^2 - (\operatorname{tr}_g h)^2\right) - 2\,\big|\operatorname{div}_g h - d(\operatorname{tr}_g h)\big|_g,
\]
so that the dominant energy condition is equivalent to \(S_{\mathrm{em}}\ge 0\) [1612.07505].

## 2. Equality and rigidity

The rigidity problem asks what the null case \(E=|P|\) means geometrically. A decisive formulation is that, for asymptotically flat initial data satisfying the dominant energy condition, null ADM energy–momentum forces trivial total energy–momentum. In dimensions \(3\le n\le 7\), it was proved that if
\[
E = |P|,
\]
then in fact
\[
E = |P| = 0,
\]
so there are no nontrivial asymptotically flat data sets with null ADM energy–momentum vector [1706.03732].

A sharper rigidity statement identifies the zero-mass data as Minkowskian. In dimension \(3\), an elementary proof removes earlier extra decay assumptions on \(\mu\) and \(J\): if a complete asymptotically flat initial data set \((M,g,k)\) satisfies \(\rho\ge |J|\) and \(E=|P|\), then \(E=0\), \(P=0\), and \((M,g,k)\) arises as a spacelike slice of Minkowski spacetime. That argument uses spacetime harmonic functions, flatness of their level sets, Liouville’s theorem, and an alternative Killing-development construction [2203.01984].

A more general rigidity theorem is formulated in terms of local validity of the positive mass inequality near the given data. If the positive mass theorem is true for all complete asymptotically flat initial data sets in a neighborhood of \((g,k)\), then \(E=|P|\) implies \(E=|P|=0\); moreover, if \(E=0\), then \((M,g)\) isometrically embeds into Minkowski space with \(k\) as its second fundamental form. This formulation is independent of whether the inequality \(E\ge |P|\) is obtained by spinors, Jang-type arguments, or other methods [2302.06040].

These rigidity statements clarify the distinction between the inequality and its equality case. The inequality only says that the ADM energy–momentum is future causal; rigidity says that the null case is not merely borderline but forces flat spacetime geometry.

## 3. Proof mechanisms

There are several structurally distinct proofs of the spacetime positive mass theorem. The classical Schoen–Yau approach uses geometric analysis based on minimal hypersurfaces and, in the spacetime case, marginally outer trapped surfaces and the Jang equation. The classical Witten proof is spinorial and uses a modified Dirac operator; it requires a spin structure. A third geometric strategy replaces both with singular hypersurface analysis based on skin structures and surgeries on minimal and marginally outer trapped hypersurfaces, yielding an all-dimensional, topology-free proof scheme. In that framework, one reduces the theorem to nonexistence of an \(S_{\mathrm{em}}>0\)-island and then derives a contradiction by compactification, stable MOTS, conformal surgery, and heredity of positive scalar curvature [1612.07505].

A recent all-dimensional proof for asymptotically flat and asymptotically hyperboloidal data combines Jang’s equation with the Brendle–Wang Riemannian positive mass theorem. The argument constructs geometric Jang graphs \(\Sigma\subset M\times\mathbb R\), proves almost-minimizing regularity with singular set of Minkowski dimension at most \(n-7\), applies the Schoen–Yau scalar curvature identity
\[
\bar R = 2(\mu - J(w)) + |h-k|_{\bar g}^2 + 2|X|_{\bar g}^2 - 2\operatorname{div}_{\bar g} X,
\]
and then performs conformal modifications so that Brendle–Wang’s all-dimensional Riemannian theorem applies to the resulting asymptotically flat metric. This yields \(E_{\mathrm{ADM}}\ge |P_{\mathrm{ADM}}|\) in all dimensions, together with rigidity to pp-wave spacetimes in the asymptotically flat equality case and to Minkowski space in the asymptotically hyperboloidal spin or Riemannian equality case [2604.24746].

There is also a non-PDE Lorentzian perspective for stationary vacuum spacetimes near infinity. In that approach one constructs an intermediate diagonal metric
\[
\tilde h = A(r)\,dt^2 - B(r)\,\delta_{ij}\,dx^i dx^j,
\qquad
A(r)=1-\frac{2m}{r}+\frac{Gm^2}{r^2},
\qquad
B(r)=1+\frac{2m}{r}+\frac{m^2}{r^2},
\]
compares causal cones after conformal compactification, and rules out negative mass by combining failure of the Penrose property with focusing theorems. This yields a Lorentzian, causality-based positive mass argument for asymptotically flat solutions that are vacuum and stationary near infinity [2101.02081].

## 4. Extensions and variants

One extension replaces asymptotic flatness by asymptotically hyperboloidal asymptotics. In that setting the end approaches hyperbolic space rather than Euclidean space, and the mass functional is defined against hyperbolic static potentials. For complete asymptotically hyperboloidal initial data satisfying the dominant energy condition, the hyperboloidal energy–momentum vector satisfies
\[
E \ge |P|.
\]
Under additional hypotheses, equality implies that the data embed as a hyperboloidal slice of Minkowski space [2604.24746].

A second extension concerns noncompact boundaries and arbitrary ends. For asymptotically flat spin initial data sets with a noncompact boundary \(\Sigma\subset\mathcal E\), the tilted boundary dominant energy condition
\[
H_\Sigma \pm \cos\alpha\,\operatorname{tr}_\Sigma k
\ge
\sin\alpha\,|k(n,\cdot)^\top|
\]
leads to the tilted spacetime positive mass inequality
\[
E_\mathcal{E} \pm \cos\alpha\,(P_\mathcal{E})_n \ge \sin\alpha\,|P_\mathcal{E}|.
\]
The special cases \(\alpha=0\) and \(\alpha=\pi/2\) recover, respectively, normal and tangential boundary dominant energy conditions. A further refinement allows arbitrary other ends and proves a quantitative shielding theorem using a Dirac–Witten operator with a Callias potential and a mixed boundary value problem [2311.15252]. The earlier tilted formulation develops the corresponding mass-type invariant and spinorial proof in the noncompact-boundary half-space setting [2304.05208].

Another extension allows codimension-one singularities. For asymptotically flat spin initial data sets creased along a closed hypersurface \(\Sigma^{n-1}\), with a matching Bartnik-data condition involving spacetime rotations, one still obtains
\[
E \ge |P|.
\]
Under stronger weighted Hölder decay assumptions, vanishing mass implies a Lipschitz embedding into Minkowski spacetime, smooth away from the crease, whose induced metric and second fundamental form agree with the glued data [2508.17585].

A more speculative extension replaces a single timelike normal by \(m\) timelike normal directions. For generalized initial data
\[
(M^n,g,k^1,\dots,k^m)
\]
on a pseudo-Riemannian manifold of signature \((n,m)\), with pairwise commuting \(k^\alpha\) and generalized dominant energy condition
\[
\mu \ge \|\mathcal J\|_{\mathrm{tr}},
\]
the main inequality becomes
\[
E \ge \|\mathcal P\|_{\mathrm{tr}},
\]
where \(\mathcal P\) is the \(m\times n\) matrix of ADM momenta and \(\|\cdot\|_{\mathrm{tr}}\) is the trace norm. Equality implies a foliation by flat codimension-\(m\) submanifolds, and under an umbilicity condition \(k^\alpha=f^\alpha g\), the data embed into a generalized pp-wave [2602.20081].

## 5. Stability, quasi-local positivity, and related inequalities

Beyond rigidity, one can ask for stability: what geometric control is implied by small mass rather than zero mass? In spherical symmetry, a sequence of asymptotically flat initial data sets satisfying the dominant energy condition, with no horizons except possibly an inner boundary, and with ADM masses tending to zero, admits graphical isometric embeddings into static spacetimes converging to Minkowski space in the pointed volume-preserving intrinsic flat sense. On tubular regions around fixed symmetry spheres, the difference between the second fundamental form of the graph and the prescribed initial-data tensor converges to zero in \(L^p\), \(1\le p<2\), assuming a uniform \(L^2\) bound on \(k_j\). The same work also proves a spacetime Penrose inequality with rigidity in spherical symmetry and analyzes examples showing why smooth or Gromov–Hausdorff convergence fails in general [1906.11352].

A quasi-local direction asks for positivity of mass attached to a bounded spacelike \(2\)-surface rather than to spatial infinity. A purely quasi-local spinor proof has been given for Wang–Yau-type masses, without Bartnik gluing or asymptotic extension. On a compact Riemannian manifold with boundary, one solves the Dirac equation with MIT Bag or APS boundary conditions and proves positivity of a spinor-defined quasi-local mass. For a spacelike topological \(2\)-sphere in a generic spacetime verifying the dominant energy condition, the resulting gravitational mass is nonnegative and vanishes only if the surface is embedded in Minkowski space [2401.13909].

Spacetime harmonic functions provide another scalar approach to mass inequalities. The spacetime harmonic equation
\[
\Delta u + (\operatorname{Tr}_g k)\,|\nabla u| = 0
\]
and the associated spacetime Hessian
\[
\bar\nabla_{ij}u = \nabla_{ij}u + k_{ij}|\nabla u|
\]
lead to integral inequalities of the form
\[
E+\langle\vec a,P\rangle
\ge
\frac{1}{16\pi}\int\Big(\frac{|\bar\nabla^2u|^2}{|\nabla u|}+2(\mu-|J|)|\nabla u|\Big),
\]
which recover the asymptotically flat and asymptotically hyperboloidal spacetime positive mass theorems, furnish a new interpolation-based notion of total mass, and yield an elementary proof of the positive mass theorem with charge [2102.11421].

## 6. Low-dimensional analogues and conceptual issues

In \(1+2\)-dimensional gravity, the natural asymptotic geometry is conical rather than asymptotically flat. For a complete, asymptotically conical, orientable two-dimensional Riemannian manifold \((\Sigma,g)\) with nonnegative scalar curvature, one defines the mass of the end by the angle defect
\[
m := 2\pi(1-P),
\]
where the asymptotic cone metric is
\[
ds^2_{\text{cone}} = dr^2 + P^2 r^2\,d\theta^2.
\]
The main theorem is that \(\Sigma\) is diffeomorphic to \(\mathbb R^2\), the angle defect satisfies \(m\ge 0\), and \(m=0\) only for the Euclidean plane. Moreover,
\[
m = \frac12\int_\Sigma S\,d\mathrm{vol}_g.
\]
This is an exact analogue of the time-symmetric positive mass theorem in \(2+1\) dimensions, with Gauss–Bonnet replacing the higher-dimensional machinery of minimal surfaces or spinors [1202.6279].

A separate conceptual issue concerns how one interprets global energy–momentum. A critical paper argues that a finite-volume integral
\[
P^\mu=\int T^{0\mu}\,d^3x
\]
of a divergence-free tensor need not define a Lorentz four-vector, using a finite-volume construction based on a “wave four-tensor.” The same discussion emphasizes that this critique is directed at a heuristic finite-volume interpretation and does not reconstruct the detailed asymptotically flat initial-data framework in which ADM energy–momentum is defined by surface integrals at infinity and used in the rigorous proofs of Schoen–Yau or Witten [2111.08497].

Taken together, these developments place the spacetime positive mass theorem at the center of a broad structure: a global causal or elliptic inequality for asymptotically flat and asymptotically hyperboloidal initial data, a rigidity principle characterizing Minkowski or pp-wave geometries in equality cases, a source of stability and Penrose-type estimates, and a template for quasi-local, boundary, singular, and even multi-time generalizations.

Source: https://www.emergentmind.com/topics/spacetime-positive-mass-theorem