A NEW FORM OF THE KERR SOLUTION

CHRIS DORAN1
Astrophysics Group, Cavendish Laboratory, Madingley Road,
Cambridge CB3 0HE, UK.

Abstract
arXiv:gr-qc/9910099v3 31 Jan 2000

A new form of the Kerr solution is presented. The solution involves

a time coordinate which represents the local proper time for free-falling
observers on a set of simple trajectories. Many physical phenomena
are particularly clear when related to this time coordinate. The chosen
coordinates also ensure that the solution is well behaved at the horizon.
The solution is well suited to the tetrad formalism and a convenient null
tetrad is presented. The Dirac Hamiltonian in a Kerr background is
also given and, for one choice of tetrad, it takes on a simple, Hermitian
form.

PACS numbers: 04.20.Jb, 04.70.Bw

1 Introduction
The Kerr solution has been of central importance in astrophysics ever since
it was realised that accretion processes would tend to spin up a black hole
to near its critical rotation rate [1]. A number of forms of the Kerr solu-
tion currently exist in the literature. Most of these are contained in Chan-
drasekhar’s work [2], and useful summaries are contained in the books by
Kramer et al. [3] and d’Inverno [4]. The purpose of this paper is to present a
new form of the solution which has already proved to be useful in numerical
simulations of accretion processes. The form is a direct extension of the
Schwarzschild solution when written as
  2M 1/2 2
ds2 = dt2 − dr + dt − r 2 (dθ 2 + sin2 θ dφ2 ). (1)
r
(Natural units have been employed.) This is obtained from the Eddington-
Finkelstein form
 2M  2 4M  2M  2
ds2 = 1 − dt̄ − dt̄ dr − 1 + dr − r 2 (dθ 2 + sin2 θ dφ2 ).
r r r
(2)

by the coordinate transformation

  r 1/2 
t = t̄ + 2(2M r)1/2 − 4M ln 1 + . (3)
2M
1
e-mail: C.Doran@mrao.cam.ac.uk, http://www.mrao.cam.ac.uk/∼cjld1/

1
In both metrics r lies in the range 0 < r < ∞, and θ and φ take their usual
meaning.
The metric (1) has a number of nice features [5], many of which extend
to the Kerr case. The solution is well-behaved at the horizon, so can be
employed safely to analyse physical processes near the horizon, and indeed
inside it [5]. Another useful feature is that the time t coincides with the
proper time of observers free-falling along radial trajectories starting from
rest at infinity. This is possible because the velocity vector

ẋi = (1, −(2M/r)1/2 , 0, 0) ẋi = (1, 0, 0, 0) (4)

defines a radial geodesic with constant θ and φ. The proper time along these
paths coincides with t, and the geodesic equation is simply

r̈ = −M/r 2 . (5)

Physics as seen by these observers is almost entirely Newtonian, making this

gauge a very useful one for introducing some of the more difficult concepts
of black hole physics. The various gauge choices leading to this form of
the Schwarzschild solution also carry through in the presence of matter
and provide a simple system for the study of the formation of spherically
symmetric clusters [6] and black holes [5].
A further useful feature of the time coordinate in (1) is that it enables
the Dirac equation in a Schwarzschild background to be cast in a simple
Hamiltonian form [5]. Indeed, the full Dirac equation is obtained by adding
a single term ĤI to the free-particle Hamiltonian in Minkowski spacetime.
This additional term is

ĤI ψ = i(2M/r)1/2 (∂r ψ + 3/(4r)ψ) = i(2M/r)1/2 r −3/4 ∂r (r 3/4 ψ). (6)

A useful feature of this gauge is that the measure on surfaces of constant

t is the same as that of Minkowski spacetime, so one can employ standard
techniques from quantum theory with little modification. One subtlety is
that the Hamiltonian is not self-adjoint due to the presence of the singularity.
This manifests itself as a decay in the wavefunction as current density is
sucked onto the singularity [5].
The time coordinate t in the metric of equation (1) has many of the
properties of a global, Newtonian time. This suggests that an attempt to
find an analogue for the Kerr solution might fail due to its angular momen-
tum. The key to understanding how to achieve a suitable generalisation
is the realisation that it is only the local properties of t that make it so
convenient for describing the physics of the solution. The natural extension
for the Kerr solution is therefore to look for a convenient set of reference
observers which generalises the idea of a family of observers on radial tra-
jectories. In Sections 2 and 3 we present a new form of the Kerr solution

2
and show that it has many of the desired properties. In Section 4 we give
various tetrad forms of the solution, and present a Hermitian form of the
Dirac Hamiltonian in a Kerr background. Throughout we use Latin letters
for spacetime indices and Greek letters for tetrad indices, and use the sig-
nature ηαβ = diag (+ − − −). Natural units c = G = h̄ = 1 are employed
throughout.

2 The Kerr Solution

The new form of the Kerr solution can be written in Cartesian-type coor-
dinates (t, x, y, z) in a manner analogous to the Kerr-Schild form [2, 4]. In
this coordinate system our new form of the solution is
 2α 
i j
2
ds = ηij dx dx − ai vj + α vi vj dxi dxj
2
(7)
ρ
where ηij is the Minkowski metric,

(2M r)1/2
α= (8)
ρ
a2 z 2
ρ2 = r 2 + 2 , (9)
r
and a and M constants. The function r is given implicitly by
r 4 − r 2 (x2 + y 2 + z 2 − a2 ) − a2 z 2 = 0, (10)
and we restrict r to 0 < r < ∞, with r = 0 describing the disk z = 0,
x2 + y 2 ≤ a2 . The maximally extended Kerr solution (where r is allowed to
take negative values) will not be considered here.
The two vectors in the metric (7) are
 
ay −ax
vi = 1, 2 , , 0 (11)
a + r 2 a2 + r 2
and
 
2 2 1/2 rx ry z
ai = (r + a ) 0, 2 2
, 2 2
, . (12)
a +r a +r r
These two vectors play an important role in studying physics in a Kerr
background. They are related to the two principal null directions n± by
n± = (r 2 + a2 )1/2 vi ± (αρvi + ai ). (13)
For computations it is useful to note that the contravariant components of
the spacelike vector in brackets are the same as those of −ai ,
 
i i 2 2 1/2 rx ry z
αρv + a = −(r + a ) 0, 2 , , . (14)
a + r 2 a2 + r 2 r

3
The vector vi also plays a crucial role in separating the Dirac equation in
a Kerr background, and is the timelike eigenvector of the electromagnetic
stress-energy tensor for the Kerr-Newman analogue of our form.

3 Spheroidal Coordinates
The nature of the metric (7) is more clearly revealed if we introduce oblate
spheroidal coordinates (r, θ, φ), where
z
cosθ = 0≤θ≤π (15)
r
y
tanφ = 0 ≤ φ < 2π, (16)
x
so that ρ recovers its standard definition

ρ2 = r 2 + a2 cos2 θ. (17)

The use of the symbols r and θ here are standard, though one must be aware
that when M = 0 (flat space) these reduce to oblate spheroidal coordinates,
and not spherical polar coordinates. This is clear from the fact that r does
√
not equal (x2 + y 2 + z 2 ).
In terms of (t, r, θ, φ) coordinates our new form of the Kerr solution is
 ρ 2
ds2 =dt2 − dr + α(dt − a sin2
θ dφ)
(r 2 + a2 )1/2
− ρ2 dθ 2 − (r 2 + a2 ) sin2 θ dφ2 . (18)

This neatly generalises the Schwarzschild form of equation (1), replacing

√ √
(2M/r) with (2M r)/ρ, and introducing a rotational component. The
line element can be simplified further by introducing the hyperbolic coor-
dinate η via a sinhη = r, though this can make some equations harder to
interpret and will not be employed here. The metric (18) is obtained from
the advanced Eddington-Finkelstein form of the Kerr solution,
 2M r  2M r
ds2 = 1 − 2 dv 2 − 2 dv dr + 2 (2a sin2 θ)dv dφ̄ + 2a sin2 θ dr dφ̄
ρ ρ
 2M r 
− ρ2 dθ 2 − (r 2 + a2 ) sin2 θ + 2 (a2 sin4 θ) dφ̄2 , (19)
ρ
via the coordinate transformation
dr
dt = dv − (20)
1 + (2M r/(r 2 + a2 ))1/2
a dr
dφ = dφ̄ − 2 . (21)
r + a + (2M r(r 2 + a2 ))1/2
2

4
This transformation is well-defined for all r, though the integrals involved
do not appear to have a simple closed form.
The velocity vector

ẋi = (1, −α(r 2 + a2 )1/2 /ρ, 0, 0) ẋi = (1, 0, 0, 0) (22)

defines an infalling geodesic with constant θ and φ, and zero velocity at

infinity. The existence of these geodesics is a key property of the solution.
The time coordinate t now has the simple interpretation of recording the
local proper time for observers in free-fall along trajectories of constant θ
and φ. As in the spherical case, many physical phenomena are simplest
to interpret when expressed in terms of this time coordinate. An example
of this is provided in the following section, where we show that the time
coordinate produces a Dirac Hamiltonian which is Hermitian in form. The
difference between this free-fall velocity and the velocity vi (defined by the
gravitational fields) also provides a local definition of the angular velocity
contained in the gravitational field.

4 Tetrads and the Dirac Equation

The metric (18) lends itself very naturally to the tetrad formalism. From
the principal null directions of equation (13) one can construct the following
null tetrad, expressed in (t, r, θ, φ) coordinates,
1 2 1/2
li = 2 2 2 2 2


(r + a , r + a − 2M r(r + a ) , 0, a) (23)
r 2 + a2
1
1/2
ni = 2 (r 2 + a2 , −(r 2 + a2 ) − 2M r(r 2 + a2 ) , 0, a) (24)
2ρ
1
mi = √ (ia sinθ, 0, 1, i cscθ). (25)
2(r + ia cosθ)
In this frame the Weyl scalars Ψ0 , Ψ1 , Ψ3 and Ψ4 all vanish, and
M
Ψ2 = − . (26)
(r − ia cosθ)3
A second tetrad, better suited to computations of matter geodesics, is
given by

e0 i = (1, 0, 0, 0)
e1 i = (α, ρ/(r 2 + a2 )1/2 , 0, −αa sin2 θ)
e2 i = (0, 0, ρ, 0)
e3 i = (0, 0, 0, (r 2 + a2 )1/2 sinθ). (27)

This defines a frame for all values of the coordinate r, so is valid inside
and outside the horizon. Combined with the techniques described in [5] this

5
tetrad provides a very powerful way of analysing and visualising motion in
a Kerr background.
A further tetrad is provided by reverting to the original Cartesian-type
coordinates of equation (7) and writing
α
eµ i = δiµ − vi aj η jµ , (28)
ρ

where vi and ai are as defined at equations (11) and (12). The inverse is
found to be
α ij
eµ i = δµi + η aj δµk vk . (29)
ρ
This final form of tetrad is the simplest to use when constructing the Dirac
equation in a Kerr background. We will not go through the details here
but will just present the final form of the equation in a Hamiltonian form.
Following the conventions of Itzykson and Zuber [7] we denote the Dirac-
Pauli matrix representation of the Dirac algebra by {γ µ } and write αi =
γ 0 γ i , i = 1 . . . 3. Since eµ 0 = δµ0 , premultiplying the Dirac equation by γ0 is
all that is required to bring it into Hamiltonian form. When this is done,
the Dirac equation in a Kerr background becomes

i∂t ψ = −iαi ∂i ψ + mγ0 ψ + ĤK ψ (30)

where
√ 
2M
(r 3 + a2 r)1/4 i∂r (r 3 + a2 r)1/4 ψ − a cosθ r 1/4 αφ i∂r r 1/4 ψ



ĤK ψ = 2
ρ

a cosθ 2 2 1/2
− (r + a ) γ5 ψ (31)
2

and

αφ = − sinφ α1 + cosφ α2 . (32)

The measure on hypersurfaces of constant t is again the same as that of

Minkowski spacetime, since the covariant volume element is simply

dx dy dz = ρ2 sinθ dr dθ dφ. (33)

As with the Schwarzschild case the interaction Hamiltonian ĤK is not self-
adjoint when integrated over these hypersurfaces. This is because the sin-
gularity causes a boundary term to be present when the Hamiltonian is
integrated.

6
5 Conclusions
The Kerr solution is of central importance in astrophysics as ever more com-
pelling evidence points to the existence of black holes rotating at near their
critical rate [8]. Any form of the solution which aids physical understanding
of rotating black holes is clearly beneficial. The form of the solution pre-
sented here has a number of features which achieve this aim. The solution is
well suited for studying processes near the horizon, and the compact form of
the spin connection for the tetrad of equation (28) makes it particularly good
for numerical computation. It should also be noted that this gauge admits
a simple generalisation to a time-dependent form which looks well-suited to
the study of accretion and the formation of rotating black holes.
A more complete exposition of the features of this gauge, including the
derivation of the Dirac Hamiltonian will be presented elsewhere. One reason
for not highlighting more of the advantages here is that many of the the-
oretical manipulations which exploit these properties have been performed
utilising Hestenes’ spacetime algebra [5, 9]. This language fully exposes
much of the intricate algebraic structure of the Kerr solution and brings
with it a number of insights. These are hard to describe without employ-
ing spacetime algebra and so will be presented unadulterated in a separate
paper.
The fact that the time coordinate measured by a family of free-falling
observers brings the Dirac equation into Hamiltonian form is suggestive of a
deeper principle. This form of the equations also permits many techniques
from quantum field theory to be carried over to a gravitational background
with little modification. The lack of self-adjointness due to the source itself
is also natural in this framework, as the singularity is a natural sink for
the current. In the non-rotating case the physical processes resulting from
the presence of this sink are quite simple to analyse [5]. The Kerr case is
considerably more complicated, due both to the nature of the fields inside
the inner horizon, and to the structure of the singularity. One interesting
point to note is that the sink region is described by r = 0, and so represents
a disk, rather than just a ring of matter. This in part supports the results
of earlier calculations described in [8], though much work remains on this
issue.

Acknowledgements
CD is supported by an EPSRC Fellowship. The author is grateful to An-
thony Lasenby and Anthony Challinor for helpful discussions.

References

7
[1] J.M. Bardeen. Kerr metric black holes. Nature, 226:64, 1970.

[2] S. Chandrasekhar. The Mathematical Theory of Black Holes. Oxford

University Press, 1983.

[3] D. Kramer, H. Stephani, M. MacCallum, and E. Herlt. Exact Solutions

of Einstein’s Field Equations. Cambridge University Press, 1980.

[4] R. d’Inverno. Introducing Einstein’s Relativity. Oxford University Press,

1992.

[5] A.N. Lasenby, C.J.L. Doran, and S.F. Gull. Gravity, gauge theories and
geometric algebra. Phil. Trans. R. Soc. Lond. A, 356:487–582, 1998.

[6] A.N. Lasenby, C.J.L. Doran, M.P. Hobson, Y. Dabrowski, and S.F. Gull.
Microwave background anisotropies and nonlinear structures I. Improved
theoretical models. Mon. Not. R. Astron. Soc., 302:748, 1999.

[7] C. Itzykson and J-B. Zuber. Quantum Field Theory. McGraw-Hill, New
York, 1980.

[8] A.N. Lasenby, C.J.L. Doran, Y. Dabrowski, and A.D. Challinor. Ro-
tating astrophysical systems and a gauge theory approach to gravity. In
N. Sánchez and A. Zichichi, editors, Current Topics in Astrofundamental
Physics, Erice 1996, page 380. World Scientific, Singapore, 1997.

[9] D. Hestenes. Space-Time Algebra. Gordon and Breach, New York, 1966.