ENERGY, MOMENTUM, AND CENTER OF MASS IN

GENERAL RELATIVITY
arXiv:1605.04968v1 [gr-qc] 16 May 2016

MU-TAO WANG

Abstract. These notions in the title are of fundamental importance

in any branch of physics. However, there have been great difficulties
in finding physically acceptable definitions of them in general relativity
since Einstein’s time. I shall explain these difficulties and progresses
that have been made. In particular, I shall introduce new definitions of
center of mass and angular momentum at both the quasi-local and total
levels, which are derived from first principles in general relativity and
by the method of geometric analysis. With these new definitions, the
classical formula p = mv is shown to be consistent with Einstein’s field
equation for the first time. This paper is based on joint work [14, 15]
with Po-Ning Chen and Shing-Tung Yau.

1. Introduction
Einstein’s theory stipulates that the spacetime is a 4 dimensional Lorentzian
manifold (N 3,1 , ḡ), where ḡ is a Lorentzian metric, or a symmetric 2-tensor
of signature (−, +, +, +). For the purpose of exposition, we specialize to the
vacuum case in this article and thus there is no presence of matter fields. The
spacetime metric ḡ then satisfies the vacuum Einstein equation Ric(ḡ) = 0.
The scenario is purely gravitational, or purely geometrical. The simplest
example is the Minkowski spacetime R3,1 where the metric is of the diago-
nal form ḡ = (−1, 1, 1, 1). This is the spacetime of special relativity where
there is no gravitation and the curvature tensor vanishes. Other important
examples include the Schwarzschild and the Kerr spacetimes. We refer to
the survey article [22] for basic materials in general relativity.
In order to solve the Einstein equation, one considers the initial value for-
mulation [17, 19] of the equation Ric(ḡ) = 0, and the spacetime is developed
from an initial data set. Suppose the spacetime is foliated by the level sets
of a time function t and let gt be the 3-metric on each constant time slice,
Ric(ḡ) = 0 is roughly
∂2
gt ∼ ∆t gt ,
∂t2
Date: May 18, 2016.
This work was partially supported by a grant from the Simons Foundation (#305519 to
Mu-Tao Wang). The author was partially supported by the National Science Foundation
under grants DMS-1105483 and DMS-1405152. He would like to thank his collaborators
Po-Ning Chen and Shing-Tung Yau, to whom he owes gratitude.
1
2 MU-TAO WANG

where ∆t is the Laplace operator of gt . This is a second order nonlinear

hyperbolic system. Thus we need to prescribe g0 on the initial time slice and
the first time derivative of g0 , which is essentially the second fundamental
∂
form of the initial time slice, k0 ∼ ∂t gt |t=0 . This interpretation implies that
there is a compatibility condition of hypersurfaces for the data (g0 , k0 ). The
condition is the so-called vacuum constrain equation for (g, k):
1
(1.1) (R + (trg k)2 − |k|2g ) = 0, divg (k − (trg k)g) = 0,
2
where R is the scalar curvature of g. Such an (M, g, k) is called a vacuum
initial data set. It can be shown that the vacuum constraint equation is
satisfied for each subsequent time slice under suitable boundary conditions.
A natural boundary condition for the initial value problem of the Ein-
stein equation is “asymptotically flatness”. This corresponds to an isolated
gravitating system around a planet, a star, or a black hole, which is far
away from any other celestial bodies. We say an initial data set (M, g, k) is
asymptotically flat, if there exists a compact subset K of M such that M \K
is diffeomorphic to a finite union ∪(R3 \B) of ball complements of R3 . Each
complement R3 \B is an end and the diffeomorphism on each end provides
a coordinate system (x1 , x2 , x3 ), such that in this coordinate system
(1.2) gij = δij + O(r −1 ), kij = O(r −2 ),
qP
3 i 2 3
where r = i=1 (x ) and δij is the Euclidean metric on R \B. Suitable
decay conditions for the derivatives of g and k need to be imposed as well,
but we will be implicit about these conditions in this paper.
From now on, we shall assume an initial data set (M, g, k) is asymptoti-
cally flat and satisfies the vacuum constraint equation. Both conditions are
preserved by the vacuum Einstein equation in the evolution [18].
It is now natural to ask the total mass and energy for such a system
(M, g, k). The ADM energy-momentum [1] (E, P1 , P2 , P3 ) on each end is
defined to be
1
Z
(1.3) E = lim (gij,j − gjj,i )ν i
r→∞ 16π S
r

and
1
Z
(1.4) Pj = lim 2(kij − δij trg k)ν i , j = 1, 2, 3.
r→∞ 16π Sr

Here Sr is a coordinate sphere of radius r with respect to the coordinate

system on an end and ν i is the unit outward normal of Sr .
The most important property of the ADM energy-momentum is the posi-
tivity, which is now known as the positive mass theorem of Schoen-Yau [43]
and Witten [50]. A special case of the theorem implies that:
If (M, g, k) is an asymptotically flat vacuum initial data set, then the
ADM energy-momentum of each end satisfies E ≥ 0 and E 2 − 3i=1 (Pi )2 ≥ 0.
P
 ENERGY, MOMENTUM, AND CENTER OF MASS IN GENERAL RELATIVITY 3
q
In particular, m = E 2 − 3i=1 (Pi )2 is the ADM mass of the system.
P

The rigidity statement of the theorem asserts that m = 0 if and only if

(M, g, k) is a Minkowskian data.
Before proceeding further to ask for the center of mass of this system, one
naturally wonders why the mass expression is so different from the usual
definition. Recall that in Newtonian gravity, the equation is ∆Φ = 4πρ,
where ρ is the mass density. For a domain Ω3 ⊂ R3 , the total mass is
Z
(1.5) ρ
Ω
and the center of mass integral is
Z
(1.6) xk ρ,
Ω

where = 1, 2, 3 are the standard coordinate functions on R3 .

xk , k
However, in general relativity, a fundamental observation of Einstein is
the so-called equivalence principle: THERE IS NO DENSITY FOR GRAV-
ITATION. Therefore the formula that
Z
mass = mass density
Ω
does NOT hold true for gravitation. Mathematically, this means at any
point in spacetime, one can choose a normal coordinate system so that all
first derivatives of the Lorentzian metric ḡ vanish at this point.
With this in mind, let us briefly review some existing definitions of center
of mass in general relativity. The list works for exposition only and is not
intended to be complete.
Huisken-Yau definition [29]: Suppose (M, g) is complete Riemannian 3-
manifold that is asymptotically Schwarzschildean in the sense that the met-
m
ric g = (1 + 2r )δ + O(r −2 ) and m > 0 (again with suitable decay conditions
on the derivatives of g), then there exists a unique CMC foliation by 2-
surfaces Σr of constant mean curvature in M when r is large enough. This
foliation thus defines a family of canonical “coordinate spheres” on each end.
The Huisken-Yau center of mass is defined to be
xk
R
lim Σr .
r→∞ |Σr |

The definition pioneered a very active area of research with an extensive

volume of literatures by now [51, 42, 40, 32, 35, 26, 24, 6, 7, 34, 8].
Another definition is given by Regge-Teitelboim [41] (see also [4]): For
each end of a vacuum asymptotically flat initial data set in the sense of
(1.2), the center of mass integral is defined to be:

1
Z
(1.7) lim xk (gij,j − gjj,i ) ν i − (gik − δik )ν i + (gii − δii )ν k
r→∞ 16π Sr
4 MU-TAO WANG

We note that although the definition is intended for an initial data set
(M, g, k), only the metric g appears in the expression. This is clearly closely
related to the ADM definition of mass. There are important gluing con-
structions to prescribe the center of mass integral by Corvino-Schoen [23],
Chrusciel-Delay [21], and Huang-Schoen-Wang [28].
The two definitions are shown to be equivalent in many cases [26]. How-
ever, each one has its advantage and its problems. In particular, there are is-
sues of well-definedness and finiteness [7, 34, 8, 6], in addition to the physical
validity of the definition [9]. For example, expression (1.7) diverges appar-
ently for standard asymptotics and a parity (Regge-Teitelboim) condition
needs to be imposed to guarantee finiteness. More importantly, these are a
priori abstract definitions and one naturally wonders the physical meaning
of them. In particular, how are they related to the dynamics of the Einstein
equation? At this point, let us discuss another definition pertaining to this
issue. In [18], Christodoulou considered a “strongly asymptotically flat” ini-
m
tial data set (M, g, k) on which g = (1 + 2r )δ + O(r −1−ǫ ) and k = O(r −2−ǫ )
for a small positive constant ǫ. The condition implies the ADM linear mo-
mentum (1.4) (P1 , P2 , P3 ) = (0, 0, 0). Under this condition, it is shown that
there exists a definition of center of mass that is conserved along the Einstein
equation.

2. The new definition

In joint work with Po-Ning Chen and Shing-Tung Yau [14, 15], we intro-
duced new definitions of center of mass C i , i = 1, 2, 3 and angular momen-
tum J i , i = 1, 2, 3 that satisfy rather remarkable properties. For example,
we prove a finiteness theorem [14, Section 7] under an order expansion con-
dition, and in particular, no parity assumption is needed. More importantly,
we show that
Pi = E · ∂t C i
when (M, g(t), k(t)) is a family of vacuum initial data sets evolving by the
Einstein equation where Pi is the ADM linear momentum and E is the ADM
energy. This is the relativistic version of the classical formula p = mv. As
fundamental as it is, this seems to be the first time when it is shown to be
consistent with the nonlinear Einstein evolution.
In addition, we prove an invariance of angular momentum theorem in
the Kerr spacetime [14, Section 8]. It is likely that the finiteness theorem
can be generalized to more general asymptotics (with the order expansion
condition) using the gravitational conservation law in [14, Section 5].
To motivate our definition, let’s look at conserved quantities in the the-
ory of special relativity for a moment. In special relativity, a matter field
in the Minkowksi spacetime R3,1 is equipped with the energy-momentum
tensor of matter density T . Noether’s principle asserts that each continuous
symmetry of the theory corresponds to a conserved quantity. There are 10
dimensional Killing fields on R3,1 that represent the infinitesimal isometry.
 ENERGY, MOMENTUM, AND CENTER OF MASS IN GENERAL RELATIVITY 5

Each K corresponds to a conserved quantity

Z
T (K, u)
Ω3
where u is unit normal of Ω3 for Ω3 ⊂ R3,1 . In the standard coordinate
∂
system (t, x1 , x2 , x3 ) on R3,1 , the timelike translating Killing field ∂t corre-
∂
sponds to the energy, so does any Killing field in the orbit of ∂t under the
isometric action of SO(3, 1).
∂
Similarly, each spacelike translating Killing in the orbit of ∂x i , i = 1, 2, 3
corresponds to a component of the linear momentum. Each rotation Killing
field in the orbit of xi ∂x∂ j − xj ∂x∂
i , i < j corresponds to a component of the
∂ ∂
angular momentum. Each boost Killing field in the orbit of xi ∂t + t ∂x i
corresponds to a component of the center of mass.
However, in general relativity where the spacetime is in general curved,
we encounter two major difficulties: there is no density and there is (in a
generic spacetime) no symmetry.
The idea is to define conserved quantities quasi-locally on the boundary
∂Ω3 = Σ2 [38, 39, 45]. We note that this is possible in Newtonian gravity.
1
R ∂Φ
The total mass (1.5) equals to 4π ∂Ω ∂ν by integrating by parts. This holds
true for the center of mass integral (1.6) as well. Once a quasi-local definition
is available, we can take the limit along a foliation of surfaces to define total
conserved quantities. On an asymptotically flat initial data set (M, g, k),
we look at conserve quantities on coordinates spheres Sr and take the limit
as r → ∞. The same principle works for asymptotically hyperbolic and
asymptotically null cases [12, 16].
To define conserved quantities quasi-locally, we start by looking for the
“best match” of the 2-surface of interest, Σ ⊂ N (a physical spacetime) as a
reference surface Σ̃ ⊂ R3,1 . A canonical identification of the normal bundles
of the two surfaces is also needed to pull back symmetries (Killing fields) of
R3,1 from Σ̃ ⊂ R3,1 to Σ ⊂ N .
In the rest of this paper, we explain how to implement these ideas, which
goes back to the proposal of optimal isometric embedding equations in [48].
Given a spacelike 2-surface Σ with the topology of S 2 in a spacetime N .
Let σ be the induced metric on Σ and H be the mean curvature vector field,
which is the unique normal vector field on Σ such that the variation of the
area function |Σ| satisfies
Z
δV |Σ| = − hH, V i
Σ
for any variation field V . We assume H is spacelike and extract from it a
function |H| and a connection one-form αH of the normal bundle in the mean
curvature gauge (see [14] for the definition of αH ). (σ, |H|, αH ) determines
the “best match” surface Σ̃ ⊂ R3,1 .
Let X : Σ → R3,1 be an isometric embedding of σ, i.e hdX, dXi = σ.
This is an under-determined PDE system with 4 unknowns (the coordinate
6 MU-TAO WANG

functions of R3,1 ) and 3 equations (the components of σ). We need to

impose one more equation in order to make the system well-determined.
This is the optimal isometric embedding equation, which is obtained by a
variational method.
Let τ be the time-component of X. τ = 0 identically corresponds to an
isometric embedding into a totally geodesic R3 ⊂ R3,1 . This case was solved
by the work of Nirenberg [36] and Pogorelov [37], and applied to define
and study earlier proposals of quasi-local mass [5, 44, 31, 30]. The optimal
isometric embedding essentially imposes an equation on τ in the general case
in order to resolve some unwanted difficulty of previous proposals (see [48]).
Let H0 be the mean curvature vector of the image surface X(Σ) = Σ̃ and
again we extract a function |H0 | and a one-form αH0 . Consider the following
function ρ and 1-form ja on Σ:
q q
(∆τ )2 (∆τ )2
|H0 |2 + 1+|∇τ | 2 − |H|2 + 1+|∇τ |2
(2.1) ρ= p
1 + |∇τ | 2

and
ρ∆τ
(2.2) ja = ρ∇a τ − ∇a [sinh−1 ( )] − (αH0 )a + (αH )a .
|H0 ||H|
Here ∆ is the Laplace operator of the metric σ and a = 1, 2 denotes a
coordinate index on Σ.
The quasi-local energy with respect to the isometric embedding X with
time component τ is
1
Z
(2.3) E(Σ, τ ) = (ρ + ja ∇a τ ).
8π Σ
Minimizing among possible τ ’s, the Euler-Lagrangian equation for critical
points of E is
(2.4) ∇a ja = 0,
see [14] for the derivation. (2.4) is exactly the additional optimal embedding
equation imposed on an isometric embedding into R3,1 . The expression
(2.3) originates from the boundary term of Hamilton-Jacobi analysis of the
gravitational action [5, 25]. We refer to [46] for more detailed discussions on
this aspect.
For a critical point X that satisfies the optimal isometric embedding equa-
tion, by integration by parts, the quasi-local energy becomes
1
Z
ρ.
8π Σ
ρ is thus called the quasi-local mass density. The energy (2.3) satisfies the
important positivity and rigidity properties by the work of [47, 48], see also
[44, 31].
 ENERGY, MOMENTUM, AND CENTER OF MASS IN GENERAL RELATIVITY 7

For each optimal isometric embedding X : Σ → R3,1 , i.e. hdX, dXi = σ

and ∇σ j = 0, we obtain a pair ρ, ja that correspond to quasi-local energy-
momentum density and a reference surface X(Σ) = Σ̃ in R3,1 that is iso-
metric to Σ. We can thus proceed to define quasi-local conserved quantities
with respect to X.
Given any Killing field K in R3,1 , we define the corresponding quasi-local
conserved quantity to be:

1
Z
− (hK, T0 iρ + (K ⊤ )a ja ),
8π Σ

where T0 = (1, 0, 0, 0) and K⊤ corresponds to the tangential part of K to Σ.

Strictly speaking the equasi-local energy (2.3) depends on the pair (X, T0 )
where T0 is a general future timelike translating Killing field in R3,1 and
τ = −hX, T0 i is the time component with respect to T0 . In this article, we
only consider the case when T0 = (1, 0, 0, 0) for simplicity. In general, we
need to take into account of the Lorentz group action on Killing fields as in
[14].
Therefore, a boost Killing field defines a component of the center of mass
and a rotation Killing field defines a component of the angular momentum.
For a given asymptotically flat vacuum initial data set (M, g, k), we con-
sider the quasi-local conserved quantities on coordinate sphere Sr . It is
shown [33, 12, 13] that if the ADM mass of (M, g, k) is positive, there is
a unique, energy-minimizing, optimal isometric embedding of Sr whose im-
age approaches a large round sphere in R3 . Take the limit as r → ∞ of
the quasi-local conserved quantities on Sr , we obtain (E, Pi , J i , C i ) where
(E, Pi ) are the same as the ADM energy-momentum vector [49]. In compari-
son to the Huisken-Yau and Regge-Teitelboim definitions, the new definition
does involve the second fundamental form k of an initial data set (M, g, k),
in which the dynamical information is encoded.
When (M, g(t), k(t)) is evolved by the vacuum Einstein equation, we de-
rived that ∂t J i = 0 and ∂t C i = PEi [14, 15]. The same result holds in the
presence of matter fields, as long as one assumed suitable decay conditions
of the energy momentum tensor at infinity, see also [34]. The proof is based
on several ingredients:
1. The Einstein evolution equation and the vacuum constraint equation
(1.1).
2. The optimal isometric embedding equation (2.4).
3. Conservation law of the quasi-local conserved quantity [14, Section 5].
We refer to the articles [14, 15], in particular [14, Section 9], for more
details.

References
[1] R. Arnowitt, S. Deser and C. W. Misner, The dynamics of general relativity, in
Gravitation: An introduction to current research, 227–265, Wiley, New York.
8 MU-TAO WANG

[2] A. Ashtekar and R. O. Hansen, A unified treatment of null and spatial infinity in
general relativity. I. Universal structure, asymptotic symmetries, and conserved quan-
tities at spatial infinity, J. Math. Phys. 19 (1978), no. 7, 1542–1566.
[3] R. Bartnik, Quasi-spherical metrics and prescribed scalar curvature, J. Differential
Geom. 37 (1993), no. 1, 31–71.
[4] R. Beig and N. Ó Murchadha, The Poincaré group as the symmetry group of canonical
general relativity, Ann. Physics 174 (1987), no. 2, 463–498.
[5] J. D. Brown and J. W. York, Jr., Quasilocal energy and conserved charges derived
from the gravitational action, Phys. Rev. D (3) 47 (1993), no. 4, 1407–1419.
[6] S. Brendle and M. Eichmair, Large outlying stable constant mean curvature spheres
in initial data sets, Invent. Math. 197 (2014), no. 3, 663–682. arXiv:1312.6391
[7] C. Cederbaum and C. Nerz, Explicit Riemannian manifolds with unexpectedly be-
having center of mass, arXiv:1312.6391
[8] P.-Y. Chan and L. -F. Tam, A note on center of mass, arXiv:1402.1220
[9] P.-N. Chen, L.-H. Huang, M.-T. Wang, and S.-T. Yau, On the validity of the definition
of angular momentum in general relativity, arXiv:1401.0597.
[10] P.-N. Chen and M.-T. Wang Conserved Quantities of harmonic asymptotic initial
data sets, arXiv:1409.5105
[11] P.-N. Chen and M.-T. Wang Rigidity and minimizing properties of quasi-local mass,
arXiv: 1411.6251
[12] P.-N. Chen, M.-T. Wang, and S.-T. Yau, Evaluating quasilocal energy and solving
optimal embedding equation at null infinity, Comm. Math. Phys. 308 (2011), no.3,
845–863
[13] P.-N. Chen, M.-T. Wang, and S.-T. Yau, Minimizing properties of critical points of
quasi-local energy, Comm. Math. Phys. 329 (2014), no. 3 919–935
[14] P.-N. Chen, M.-T. Wang, and S.-T. Yau, Conserved quantities in general relativ-
ity: from the quasi-local level to spatial infinity, to appear in Comm. Math. Phys,
arXiv:1312.0985
[15] P.-N. Chen, M.-T. Wang, and S.-T. Yau, Quasilocal angular momentum and center
of mass in general relativity, arXiv:1312.0990
[16] P.-N. Chen, M.-T. Wang, and S.-T. Yau, Conserved quantities on asymptotically
hyperbolic initial data set, arXiv:1409.1812
[17] Y. Fourès-Bruhat, Théorème d’existence pour certains systèmes d’équations aux
dérivé es partielles non linéaires, Acta Math. 88, (1952). 141– 225.
[18] D. Christodoulou, Mathematical problems of general relativity. I, Zürich Lectures in
Advanced Mathematics. European Mathematical Society (EMS), Zürich, 2008.
[19] D. Christodoulou and S. Klainerman, The global nonlinear stability of the Minkowski
space. Princeton Mathematical Series, 41. Princeton University Press, Princeton, NJ,
1993.
[20] P. -T. Chruściel, On angular momentum at spatial infinity, Classical Quantum Grav.
4 (1987), no. 6, L205–L210.
[21] P.-T. Chruściel and E. Delay, On mapping properties of the general relativistic con-
straints operator in weighted function spaces, with applications, Mém. Soc. Math. Fr.
(N.S.) 94 (2003).
[22] P.-T. Chruściel, G. J. Galloway, and D. Pollack, Mathematical general relativity: a
sampler, Bull. Amer. Math. Soc. (N.S.) 47 (2010), no. 4, 567–638.
[23] J. Corvino and R. Schoen, On the asymptotics for the vacuum Einstein constraint
equations, J. Diff. Geom. 73(2), 185–217 (2006).
[24] M. Eichmair and J. Metzger, Large isoperimetric surfaces in initial data sets, J.
Differential Geom. 94 (2013), no. 1, 159–186.
[25] S. W. Hawking and G. T. Horowitz, The gravitational Hamiltonian, action, entropy
and surface terms, Classical Quantum Gravity 13 (1996), no. 6, 1487–1498.
 ENERGY, MOMENTUM, AND CENTER OF MASS IN GENERAL RELATIVITY 9

[26] L.-H. Huang, On the center of mass of isolated systems with general asymptotics,
Classical Quantum Gravity 26 (2009), no. 1, 015012, 25 pp.
[27] L.-H. Huang, Private communication.
[28] L.-H. Huang, R. Schoen and M.-T. Wang, Specifying angular momentum and center
of mass for vacuum initial data sets, Comm. Math. Phys. 306 (2011), no. 3, 785–803.
[29] G. Huisken and S.-T. Yau, Definition of center of mass for isolated physical systems
and unique foliations by stable spheres with constant mean curvature, Invent. Math.
124, 281–311 (1996).
[30] J. Kijowski, A simple derivation of canonical structure and quasi-local Hamiltonians
in general relativity, Gen. Relativity Gravitation 29 (1997), no. 3, 307–343.
[31] C.-C. M. Liu and S.-T. Yau, Positivity of quasi-local mass. II, J. Amer. Math. Soc.
19 (2006), no. 1, 181–204
[32] J. Metzger, Foliations of asymptotically flat 3-manifolds by 2-surfaces of prescribed
mean curvature, J. Differential Geom. 77 (2007), no. 2, 201–236.
[33] P. Miao and L.-F. Tam, On second variation of Wang-Yau quasi-local energy, Ann.
Henri Poincaré 15 (2014), no. 7, 1367–1402.
[34] C. Nerz, Time evolution of ADM and CMC center of mass in general relativity,
arXiv:1312.6274
[35] A. Neves and G. Tian, Existence and uniqueness of constant mean curvature foliation
of asymptotically hyperbolic 3-manifolds, Geom. Funct. Anal. 19 (2009), no. 3, 910–
942.
[36] L. Nirenberg, The Weyl and Minkowski problems in differential geometry in the large,
Comm. Pure Appl. Math. 6, (1953). 337–394.
[37] A. V. Pogorelov, Regularity of a convex surface with given Gaussian curvature, (Rus-
sian) Mat. Sbornik N.S. 31(73), (1952). 88–103.
[38] R. Penrose, Some unsolved problems in classical general relativity, in Seminar on
Differential Geometry, 631–668, Ann. of Math. Stud., 102, Princeton Univ. Press,
Princeton, NJ.
[39] R. Penrose, Quasilocal mass and angular momentum in general relativity, Proc. Roy.
Soc. London Ser. A 381 (1982), no. 1780,
[40] J. Qing and G. Tian, On the uniqueness of the foliation of spheres of constant mean
curvature in asymptotically flat 3-manifolds. J. Amer. Math. Soc. 20 (2007), no. 4,
1091–1110.
[41] T. Regge and C. Teitelboim, Role of Surface Integrals in the Hamiltonian Formulation
of General Relativity, Ann. Phys. 88, 286–318 (1974).
[42] R. Rigger, The foliation of asymptotically hyperbolic manifolds by surfaces of constant
mean curvature (including the evolution equations and estimates), Manuscripta Math.
113 (2004), no. 4, 403–421.
[43] R. Schoen and S.-T. Yau, On the proof of the positive mass conjecture in general
relativity. Comm. Math. Phys. 65 (1979), no. 1, 45–76.
[44] Y. Shi and L.-F. Tam, Positive mass theorem and the boundary behaviors of compact
manifolds with nonnegative scalar curvature, J. Differential Geom. 62 (2002), no. 1,
79–125.
[45] L. B. Szabados, Quasi-Local Energy-Momentum and Angular Momentum in General
Relativity, Living Rev. Relativity 12, (2009), 4.
[46] M.-T. Wang, Quasilocal mass and surface Hamiltonian in spacetime, XVIIth Inter-
national Congress on Mathematical Physics, 229–238, World Sci. Publ., Hackensack,
NJ, 2014.
[47] M.-T. Wang and S.-T. Yau, Quasilocal mass in general relativity, Phys. Rev. Lett.
102 (2009), no. 2, no. 021101, 4 pp.
[48] M.-T. Wang and S.-T. Yau, Isometric embeddings into the Minkowski space and new
quasi-local mass. Comm. Math. Phys. 288 (2009), no. 3, 919-942.
10 MU-TAO WANG

[49] M.-T. Wang and S.-T. Yau, Limit of quasilocal mass at spatial infinity. Comm. Math.
Phys. 296 (2010), no.1, 271–283.
[50] E. Witten, A new proof of the positive energy theorem. Comm. Math. Phys. 80 (1981),
no. 3, 381–402.
[51] R. Ye, Foliation by constant mean curvature spheres on asymptotically flat manifolds.
Geometric analysis and the calculus of variations, 369–383, Int. Press, Cambridge,
MA, 1996.

Department of Mathematics, Columbia University, 2990 Broadway, New

York, NY 10027