A Note On Static Metrics: Robert Bartnik and Paul Tod

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

A Note on Static Metrics

Robert Bartnik∗
and
arXiv:gr-qc/0512097v1 16 Dec 2005

Paul Tod†

Abstract

Conditions are given which, subject to a genericity condition on the


Ricci tensor, are both necessary and sufficient for a 3-metric to arise from
a static space-time metric.

2000 Mathematics Subject Classification: 83C15, 53C50.


PACS numbers: 04.20 Jb
Keywords and Phrases: Einstein equations, static space-times.

The vacuum field equations of General Relativity reduce for a static solution
to a coupled system involving a (Riemannian) spatial metric gij and a potential
function V equal to the square-root of the norm of the Killing vector. The
system is over-determined and one can ask whether a given metric allows the
existence of a V for which the equations are satisfied. This is the question: when
is a 3-metric static? which in turn is the simplest case of a larger question: when
are Cauchy data for a vacuum solution actually data for a static solution?
In this note we give necessary and sufficient conditions on a 3-metric for
it to be static, in the case when the Ricci tensor viewed as an endomorphism
of vectors has distinct eigenvalues, which is the generic case. In this case, our
method gives an explicit candidate for the gradient dV in terms of the curvature.
If the Ricci tensor is degenerate, in the sense of having a repeated eigenvalue, the
method fails. One can then ask if it is possible to have two different potentials
with the same metric, a problem solved in [1]. Our methods can also be used to
give information on a number of related geometrical equations, which we shall
note below. The larger question above can also be solved, and will be considered
elsewhere.
∗ School of Mathematical Sciences, Monash University.
† Mathematical Institute, Oxford, OX1 3LB.

1
In terms of a metric gij with Ricci tensor Rij and a potential V , the static
vacuum field equations are as follows:

Rij = V −1 ∇i ∇j V (1)
∆g V = 0, (2)

where ∆g is the Laplacian for g (for these equations see, for example, [2]). It
follows from (1) and (2) that the scalar curvature vanishes. Define the Cotton-
York tensor, as usual, by

Cij = ǫjpq (∇q Rip − 41 gip ∇q R), (3)

then Cij is symmetric, trace-free and divergence-free in the sense that ∇j Cij =
0. Differentiating (1) and applying (3) and the Ricci identity appropriate for
dimension 3 leads to

V Cij = −ǫjpq (2Rip δqk + gip Rkq )∇k V, (4)

which gives a system of five equations (since both sides are symmetric and trace-
free) for the three components of Ui := V −1 ∇i V . For this system to be solvable
for Ui two linear constraints must hold on the five components of the tensor
Cij . These constraints are

Cij Rij = 0, Cij Rjk Rki = 0, (5)

which, for brevity, we may write as C · R = 0 and C · R2 = 0. The necessity of


these conditions follows readily from (4); sufficiency can be seen by calculating
in the Ricci eigenframe. In a frame which diagonalises the Ricci tensor, say
R = diag(λ, µ, ν) with λ + µ + ν = 0, we see from (4) that C is purely off-
diagonal,  
0 c b
C≃ c 0 a  (6)
b a 0
and for U , provided λ, µ and ν are distinct, we obtain the expression

U = [a/(µ − ν), b/(ν − λ), c/(λ − µ)]. (7)

Thus, provided the eigenvalues of the Ricci tensor are distinct, we have an
explicit expression for the gradient of the (logarithm of the) static potential.
All that remains is to see whether (1) and (2) are satisfied, which we do below.
Note that the eigenvalues of the Ricci tensor are distinct exactly when the
discriminant D satisfies

D := 4σ23 + 27σ32 < 0, (8)

where σ1 = λ + µ + ν = trg R = 0, σ2 = − 21 |R|2 = λµ + µν + νλ < 0 and


σ3 = det R = λµν are the elementary symmetric functions of the eigenvalues of

2
R. Note the alternative expressions

D = − 21 |R|6 + 3(trg R3 )2
= − 21 (Rij Rji )3 + 3(Rij Rjk Rki )2
= −(λ − µ)2 (µ − ν)2 (ν − λ)2 .
After some calculation, we may obtain the general form of (7) in a matrix
notation as

U = D−1 (σ2 I + 3R2 )2 [C, R]∗


= 14 D−1 (|R|2 I − 6R2 )2 [C, R]∗ , (9)

where |R|2 := Rij Rji and [C, R]∗ := ǫijk Cjp Rpk . More explicitly,

Ui = 41 D−1 (|R|2 δij − 6Rpi Rjp )(|R|2 δjk − 6Rqj Rkq )ǫkrs Crt Rts (10)

We shall say that a 3-metric g is Ricci-non-degenerate if its Ricci tensor has


distinct eigenvalues; equivalently, if D 6= 0.
Theorem 1 A Ricci-non-degenerate metric g, defined on a simply-connected
region, admits a static potential V exactly when the vector field U (defined in
terms of g, R, C by (9) or (10)) satisfies

∇i Uj + Ui Uj = Rij . (11)

Proof: If g is Ricci-non-degenerate and static with potential V , then Ui =


V −1 ∇i V satisfies (11) by (1), and the above calculations show (9) holds because
D < 0. Conversely, if Ui defined by (9) satisfies (11), then ∇[i Uj] = 0 and thus
Ui is a gradient, Ui = ∇i log V , where the potential V is unique up to an
arbitrary multiplicative constant. It then follows directly from (11) that V is a
static potential for g.
Equation (11) with Ui as in (10) is an equation directly on the Ricci tensor,
and it is of interest to find the form of the derivatives of highest degree. We
have the identities
2 2
∇i Cjk = ∇(i Cjk) + ∇[i Cj]k + ∇[i Ck]j
3 3
ǫijm ∇i Cjk = ∆Rm k − 3R
mn
Rnk + δkm Rpq Rpq
Thus the highest derivatives in (11) are ∇(i Cjk) and ∆Rij . However, given a
solution of (1) and (2), ∆Rij can be expressed as

∆Rij = 6Rki Rjk − 2gij Rkm Rkm + U m (∇i Rjm + ∇j Rim − 3∇m Rij )
in terms of Rij , ∇i Rjk and Ui , and we obtain a necessary condition on just
∇(i Cjk) from (11) and (10).

The conditions found here give an algorithm for testing a metric for staticity.
Assuming Ricci-nondegeneracy, one proceeds by asking the questions:

3
1. does the scalar curvature vanish?
2. if so, is Cij purely off-diagonal in the Ricci eigenframe?
3. if so, is Ui from (10) a gradient?
4. if so, is (11) satisfied?
This algorithm can be tested on, for example, spatially homogeneous 3-metrics
when it finds all the static cases.
Finally, we note the following two equations which have arisen in the litera-
ture:

∇i ∇j F = Rij − λgij (12)


2∇i ∇j F = F Rij . (13)

The first is the gradient Ricci soliton equation (see e.g. [3]) while the second
has arisen in the study of black holes (see e.g. [4]). For both equations, the
interest is typically in global solutions on compact manifolds, but our methods
in the 3-dimensional case again deduce candidate dF given the metric.

Acknowledgments: We are grateful to the Isaac Newton Institute, Cam-


bridge, where part of this work was done, for hospitality and financial support.
The first author (RAB) also thanks the Clay Institute for its support.

References
[1] K.P.Tod, Spatial metrics which are static in many ways, Gen.Rel.Grav. 32
(2000) 2079
[2] H.Stephani, D.Kramer, M.MacCallum, C.Hoenselaers and E.Herlt, Exact
solutions of Einstein’s field equations Cambridge University Press, Cam-
bridge, 2003.
[3] T.Ivey, New examples of complete Ricci solitons, Proc. Amer. Math. Soc,
122 (1994) 241–245
[4] J.Lewandowski and T.Pawlowski, Extremal Isolated Horizons: A Local
Uniqueness Theorem, Class.Quant.Grav. 20 (2003) 587-606; gr-qc/0208032

You might also like