In this paper
AbstractIntroductionAcknowledgmentAI DeclarationRigidity when Counterexample to Conjecture 1.1ReferencesWorksPhysics and Astronomy
A note on Positivity and Rigidity for an Expanding Quasi-Local Energy
Abstract
We study positivity and rigidity for an expanding quasi-local energy associated with constantly umbilic initial data , introduced in earlier work [ALP26] as an analogue of the Brown-York energy adapted to an expanding flat de Sitter background. Under nonnegative scalar curvature and suitable boundary assumptions, our previous results established positivity for below a geometrically determined threshold , while vanishing at forced the boundary to be round. In this paper, we strengthen this endpoint rigidity statement by showing that, if when , then is isometric to a closed Euclidean 3-ball of radius at most . The proof combines the rigidity theorem of Shi-Tam for the Brown-York energy with a strict concavity argument. We also construct an explicit counterexample to the positivity and rigidity conjecture formulated in [ALP26], showing that the energy can become negative beyond this threshold and that vanishing energy need not imply rigidity.
Introduction
Let be a compact connected Riemannian 3-manifold with nonempty boundary of positive Gauss curvature. By the Weyl embedding theorem [7, 8], each connected component of admits an isometric embedding into Euclidean space , unique up to rigid motions, whose image bounds a strictly convex domain. Let denote the mean curvature of in with respect to the outward unit normal, and let denote the mean curvature of its Weyl embedding in . When has nonnegative scalar curvature, , and , the Brown–York energy
is nonnegative and vanishes only if is a compact domain in Euclidean space by the theorem of Shi and Tam [9]. This provides a quasi-local version of the positive mass theorem for compact Riemannian 3-manifolds with boundary, corresponding to the time-symmetric case in the initial-data setting.
In joint work with Avalos [2], motivated by cosmological observations of the expansion of the universe, we introduced an analogue of the Brown–York energy adapted to an expanding flat de Sitter background. Fix and, instead of time-symmetric initial data, consider the constantly umbilic condition , which models the expansion of the universe. In this expanding setting, we defined the following quasi-local energy
provided (i.e., the mean curvature vectors are not timelike). This definition was motivated in part by the Liu–Yau energy [5, 6]. It can also be derived from the free-energy Hamiltonian in general relativity, see Eq. (105) in [4]. Whereas the Brown–York energy uses a Euclidean slice of Minkowski space as its reference, the energy is adapted to a Euclidean slice of the flat-expanding patch of de Sitter space with cosmological constant . In flat-expanding coordinates, the metric for de Sitter space can be written as
The manifold is . The hypersurfaces are intrinsically Euclidean and constantly umbilic, with second fundamental form . Moreover, the Killing field is timelike precisely when , with corresponding to the cosmological horizon. Motivated by this geometry, we impose the requirement that the Weyl embedding of lie within the cosmological horizon of a -slice, equivalently, within a Euclidean ball of radius . For further details concerning the definition of and its motivation, we refer to Section 2 of [2].
If the initial data set arises in a spacetime satisfying the dominant energy condition with respect to the energy-momentum tensor, , where is the spacetime metric and , then has nonnegative scalar curvature—a consequence of the first constraint equation. Therefore, in this constantly umbilic setting, the initial-data dominant energy condition reduces to that of on .
The physical expectation that the dominant energy condition should give rise to nonnegative gravitational energy underlies both the positive energy theorem and positivity of the Brown–York energy. In the constantly umbilic setting considered here, the analogous expectation suggests that nonnegative scalar curvature should imply , provided the Weyl embedding lies within the cosmological horizon and both relevant mean curvature vectors are non-timelike. This motivates the following conjecture, originally formulated in [2], which would generalize Theorem 1 of [9] to the expanding setting.
Conjecture 1.1. Fix . Let be a compact connected Riemannian -manifold whose nonempty boundary has positive Gauss curvature. Let denote the mean curvature of in , and let denote the mean curvature of the Weyl embedding of into Euclidean space . Suppose that
and that the image of the Weyl embedding is contained in a closed Euclidean ball of radius . Then
Moreover, if , then is isometric to a compact domain in ; hence the constantly umbilic initial data set is a compact domain in a flat-expanding slice within the cosmological horizon of de Sitter space.
In [2], we established several positivity results providing partial support for Conjecture 1.1. In particular, Corollary 3.4 in [2] shows that when has nonnegative scalar curvature and is sufficiently small relative to the geometry of the boundary. (In fact, the observed cosmological constant is small in the sense that for a hypothetical boundary enclosing a massive astrophysical object like a planet, star, or black hole.) These results left open the question of whether positivity continues to hold in the full range permitted by the conjecture.
Our first result in this paper, Theorem 2.1, strengthens the rigidity statement obtained in [2]. Specifically, Theorem 3.3 in [2] introduces a geometrically determined threshold for which whenever , while at forces the boundary to be round. Theorem 2.1 strengthens the latter conclusion by showing that, in the constantly umbilic setting, if and , then the entire fill-in is isometric to a closed Euclidean -ball of radius at most . The proof combines the rigidity theorem of Shi–Tam for the Brown–York energy with a strict concavity argument.
Our second result shows that Conjecture 1.1 is false in general. In § 3, we construct an example of a nonflat Riemannian -manifold satisfying all the hypotheses of Conjecture 1.1, but for which . Moreover, by continuity, we show that there exists some such that even though is not a compact domain in Euclidean space. Thus, both the positivity and rigidity statements of Conjecture 1.1 fail, demonstrating that the smallness conditions imposed in [2] cannot be removed altogether.
There are several other approaches to quasi-local energy in the presence of a positive cosmological constant. Chen, Wang, and Yau [3] defined a quasi-local energy with respect to de Sitter and anti-de Sitter reference spacetimes using differences of surface Hamiltonians and optimal isometric embeddings into the reference spacetime. Their construction allows for general isometric embeddings into the reference spacetime and, in particular, is not tied to a preferred spacelike foliation. More recently, Agostiniani, Borghini, and Mazzieri [1] introduced a family of mass-type invariants for time-symmetric initial data with positive cosmological constant, obtaining new rigidity characterizations of de Sitter space. (It would be interesting to explore whether their approach can yield positivity and rigidity results for constantly umbilic initial data with suitable boundary conditions at the cosmological horizon.) The setting considered here is different from both of these approaches: rather than working in the time-symmetric case or optimizing over general isometric embeddings into the reference spacetime, we restrict to constantly umbilic initial data and use the preferred Weyl embedding into a Euclidean slice of the flat-expanding foliation. Thus, our construction is adapted specifically to initial data modeling cosmological expansion and to the geometry of the flat-expanding patch of de Sitter space.
Acknowledgment
Acknowledgment. E. Ling was supported in part by the Austrian Science Fund (FWF) [Grant DOI 10.55776/EFP6].
AI Declaration
AI Declaration. Theorem 2.1 and its proof were obtained by the authors independently of AI assistance. The counterexample underlying Theorem 3.2 was suggested by Claude Opus 5.5; the authors subsequently developed the construction and established the proof presented here. AI-based tools were also used for editorial assistance and mathematical exploration. The authors take full responsibility for the mathematical content.
Rigidity when
Let be a compact connected Riemannian 3-manifold with nonempty boundary . Fix a constant and assume the Gauss and mean curvatures of satisfy and . By the Weyl embedding theorem, each component of isometrically embeds into Euclidean space . This embedding is unique up to rigid motions in . Let denote the mean curvature of embedded within . Recall the Brown–York energy, Liu–Yau energy, and are given by
Let denote the area radius of . Let denote the minimum of the Gauss curvature on . Define
Theorem 2.1. Fix . Let be a compact connected Riemannian -manifold with nonempty boundary of positive Gauss curvature . Assume
If , then is well defined and . If and , then is isometric to a closed Euclidean -ball with a radius ; hence the constantly umbilic initial data set is a compact domain in a flat-expanding slice within the cosmological horizon of de Sitter space.
Remark. The converse of the rigidity statement in Theorem 2.1 is true as well. Specifically, if is isometric to the closed -ball of radius in Euclidean space, then for all we have and . The proof is a straightforward computation.
Proof. The case was already dealt with in the proof of Lemma 5.2 in [2], whose estimates remain valid under the weaker assumption .
Now assume with . Since , the equality argument in the proof of Lemma 5.2 in [2] applies. In particular, the inequalities corresponding to (34) and (35) there are equalities, and hence each connected component of is isometric to a round sphere. Consequently, and the Gauss curvature are constant on each component. Moreover, (34) shows that so is well defined.
Claim: .
We first use the claim to establish the theorem and then prove the claim at the end of the proof. Since , we have . By the rigidity part of Theorem 1 in [9], it follows that is isometric to a compact domain in Euclidean space (hence consists of a single component). Since is round, is a closed -ball in . If denotes the radius, then and hence .
Now we prove the claim. Fix a component of , and let denote the constant on . The result of Shi–Tam gives (omitting the measure for notational simplicity)
Consider the function for . Since , is strictly concave on . By Jensen’s inequality and (5), we have
Therefore
Since , (7) implies for each component . Thus
Therefore the inequalities in (6) are actually equalities:
The first equality in (9) is precisely the equality case of Jensen’s inequality. Since is strictly concave, this implies that is constant on . Then the second equality in (9) forces on since is strictly increasing on . The component was arbitrary; therefore on .
Counterexample to Conjecture 1.1
Before constructing the counterexample, we note that Conjecture 1.1 holds under the additional assumption that the induced metric on is round:
Proposition 3.1. Fix . Let be a compact connected Riemannian -manifold, whose nonempty boundary is isometric to a round sphere of radius . If and , then with equality if and only if is isometric to the closed Euclidean -ball of radius .
Proof. Since is constant, the arguments involving the Shi–Tam theorem and Jensen’s inequality in the proof of Theorem 2.1 apply. For example, (7) shows that . If , then (9) holds and the rigidity claim follows. □
Thus, to obtain a counterexample, one must depart from the round boundary geometry; the construction below does so by perturbing the round metric in an axially symmetric manner.
Theorem 3.2. There exists a compact connected nonflat Riemannian -manifold and constants such that, for every , the hypotheses of Conjecture 1.1 are satisfied. Moreover,
Consequently, both the positivity and rigidity statements of Conjecture 1.1 fail.
Proof. Let denote the closed unit ball in with boundary given by the unit two-sphere. Let denote the Euclidean metric on .
For sufficiently small , set
where is the second Legendre polynomial. Since is -harmonic, is scalar flat. However, is not flat—the Ricci tensor is nonvanishing.
The mean curvature of within in terms of the mean curvature of within can be calculated using conformal transformation laws,
Similarly, a calculation for the Gauss curvature gives
Clearly uniformly on for sufficiently small .
Set . Since is axially symmetric (i.e., invariant under rotations in ), the isometric Weyl embedding can be explicitly calculated as a surface of revolution in Euclidean space:
We have
Comparing the and coefficients gives
The minus sign was chosen for so that it agrees with the round sphere.
Since both the induced metric and the second fundamental form are diagonal in these coordinates, the meridian and parallel directions are principal directions of the surface of revolution. The principal curvature in the parallel direction with respect to the outward unit normal is
A computation shows that (15) implies
Using (15) and (16) this yields
By the Gauss equation, the other principal curvature satisfies . Using (12) and (18), the mean curvature of within Euclidean space is given by
Set . Define and via and . From equations (19) and (11), we have
We wish to expand out
in powers of . Taylor expanding both integrand terms gives
Therefore
To obtain a negative , the second term must dominate the positive Brown–York term. We will examine each term separately starting with the second.
Let denote the area form for the round metric on the unit two-sphere. Then
Therefore
The Brown–York energy is
Substituting (27) and (28) along with
into (25) yields
Take . Then for sufficiently small . The map is continuous on and . Define so that is the connected component of the set containing . Continuity implies ; moreover since .
Lastly, we show that there is an such that for all the hypotheses of Conjecture 1.1 are met for . From (12), the Gauss curvature is positive uniformly on for sufficiently small . Moreover, since and and , we also obtain the uniform bounds
for all .
It remains to show that the image of the Weyl embedding is contained in a Euclidean 3-ball of radius . Indeed, we will show that converges uniformly to the unit sphere as . From (15),
Uniformly in , (17) yields
Therefore, choosing ,
Plugging (31) and (33) into (13) gives
uniformly on . Therefore
for some . Thus, for sufficiently small , is contained in the Euclidean 3-ball of radius centered at the origin.
Since only finitely many smallness conditions on have been imposed above, we may choose sufficiently small so that all of them hold simultaneously. □
We conclude with a few remarks.
The counterexample in the proof results from a competition between the positive Brown–York energy and a negative correction caused by the concavity of . Indeed, rewriting (25) gives
Although is not pointwise greater than , its first-order variation is twice the size of that of , which accounts for the coefficient multiplying . The concavity of therefore produces a negative second-order contribution, which outweighs the positive Brown–York term when is large and is small.
A similar calculation shows that the choice in equation (10) also yields a counterexample, although the negative leading term in the energy expansion occurs at order .
Since the metrics constructed in the proof of Theorem 3.2 are scalar flat, the data , where is an open neighborhood of on which , satisfy the vacuum Einstein constraint equations with cosmological constant . Their globally hyperbolic developments therefore realize these counterexamples within -vacuum spacetimes.
The counterexamples above can be rescaled to any expansion parameter . Indeed, under the scaling , the hypotheses of Conjecture 1.1 are preserved upon replacing by , and . Consequently, for every , there exist counterexamples to positivity, and separately, to rigidity at .
(Eric Ling) Department of Mathematics, Florida State University, Tallahassee, Florida
Email address: [email protected]
(Annachiara Piubello) Mathematics Department, Eberhard Karls Universität Tübingen
Email address: [email protected]
References
- [1][ABM26] V. Agostiniani, S. Borghini, and L. Mazzieri. Mass-type invariants in the presence of a cosmological constant, 2026. arXiv:2603.01543.arxiv.org/abs/2603.01543
- [2][ALP26] Rodrigo Avalos, Eric Ling, and Annachiara Piubello. On energy and its positivity in spacetimes with an expanding flat de sitter background. Classical and Quantum Gravity, 43(9):095006, 2026.arxiv.org/abs/2511.01713
- [3][CWY20] P.-N. Chen, M.-T. Wang, and S.-T. Yau. Quasi-local energy with respect to de Sitter/anti-de Sitter reference. Communications in Analysis and Geometry, 28(7):1489–1531, 2020.arxiv.org/abs/1603.02975
- [4][Kij97] J. Kijowski. A simple derivation of canonical structure and quasi-local Hamiltonians in general relativity. General Relativity and Gravitation, 29(3):307–343, 1997.DOI
- [5][LY03] C.-C. M. Liu and S.-T. Yau. Positivity of quasilocal mass. Physical Review Letters, 90(23):231102, 2003.DOI
- [6][LY06] C.-C. M. Liu and S.-T. Yau. Positivity of quasi-local mass II. Journal of the American Mathematical Society, 19(1):181–204, 2006.DOI
- [7][Nir53] L. Nirenberg. The Weyl and Minkowski problems in differential geometry in the large. Communications on Pure and Applied Mathematics, 6(3):337–394, 1953.DOI
- [8][Pog52] A. V. Pogorelov. Regularity of a convex surface with given gaussian curvature. Matematicheskii Sbornik (N.S.), 31(73)(1):88–103, 1952.
- [9][ST02] Y. Shi and L.-F. Tam. Positive mass theorem and the boundary behaviors of compact manifolds with nonnegative scalar curvature. Journal of Differential Geometry, 62(1):79–125, 2002.DOI