September 28, 2018 16:21 WSPC/INSTRUCTION FILE

kerrschild˙091113˙fin˙grqc

International Journal of Geometric Methods in Modern Physics

c World Scientific Publishing Company
arXiv:1408.4601v1 [gr-qc] 20 Aug 2014

THE KERR-SCHILD ANSATZ REVISED

DONATO BINI
Istituto per le Applicazioni del Calcolo “M. Picone,” CNR I-00161 Rome, Italy
ICRA, University of Rome “La Sapienza,” I-00185 Rome, Italy
ICRANet, I-65100 Pescara, Italy
binid@icra.it

ANDREA GERALICO
Physics Department, University of Rome “La Sapienza,” I-00185 Rome, Italy
ICRA, University of Rome “La Sapienza,” I-00185 Rome, Italy
ICRANet, I-65100 Pescara, Italy
geralico@icra.it

ROY P. KERR
University of Canterbury, Christchurch, New Zealand
ICRANet, I-65100 Pescara, Italy
roy.kerr@canterbury.ac.nz

Received (Day Month Year)

Revised (Day Month Year)

Kerr-Schild metrics have been introduced as a linear superposition of the flat spacetime
metric and a squared null vector field, say k, multiplied by some scalar function, say
H. The basic assumption which led to Kerr solution was that k be both geodesic and
shearfree. This condition is relaxed here and Kerr-Schild ansatz is revised by treating
Kerr-Schild metrics as exact linear perturbations of Minkowski spacetime. The scalar
function H is taken as the perturbing function, so that Einstein’s field equations are
solved order by order in powers of H. It turns out that the congruence must be geodesic
and shearfree as a consequence of third and second order equations, leading to an alter-
native derivation of Kerr solution.

Keywords: Kerr-Schild metrics

1. Introduction
Kerr-Schild metrics have the form [1,2]
ds2 = gαβ dxα dxβ ≡ (ηαβ − 2Hkα kβ )dxα dxβ , (1)
where ηαβ is the metric for Minkowski space and kα is a null vector
ηαβ k α k β = gαβ k α k β = 0, k α = η αβ kβ = g αβ kβ . (2)
The inverse metric is also linear in H
g αβ = η αβ + 2Hk α k β , (3)

1
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

2 Donato Bini, Andrea Geralico and Roy P. Kerr

and so the determinant of the metric is independent of H

(ηαγ − 2Hkα kγ )(η γβ + 2Hk γ k β ) = δαβ −→ |gαβ | = |ηαβ | . (4)
Within this class of general metrics the Kerr solution was obtained in 1963 by
a systematic study of algebraically special vacuum solutions [3]. If (x0 = t, x1 =
x, x2 = y, x3 = z) are the standard Cartesian coordinates for Minkowski spacetime
with ηαβ = diag[−1, 1, 1, 1], then for Kerr metric we have
(rx + ay)dx + (ry − ax)dy z
− kα dxα = dt + + dz , (5)
r 2 + a2 r
where r and H are defined implicitly by
x2 + y 2 z2 Mr3
+ =1, H=− . (6)
r 2 + a2 r2 r 4 + a2 z 2
Kerr solution is asymptotically flat and the constants M and a are the total
mass and specific angular momentum for a localized source. They both have the
dimension of a length in geometrized units. The vector k is geodesic and shearfree,
implying that Kerr metric is algebraically special according to the Goldberg-Sachs
theorem [4]. Moreover, k is independent of M and hence a function of a alone. Note
that the mass parameter M appears linearly in the metric, i.e. in H.
In this paper we consider Kerr-Schild metrics (1) as exact linear perturbations
of Minkowski space and solve Einstein’s field equations order by order in powers
of H. The results of this analysis will be that k must be geodesic and shearfree
as a consequence of third and second order equations, leading to an alternative
derivation of Kerr solution.

2. Modified ansatz
Let ǫ be an arbitrary constant parameter, eventually to be set equal to 1, so that
the Kerr-Schild metric (1) reads
gαβ = ηαβ − 2ǫHkα kβ , (1)
with inverse
g αβ = η αβ + 2ǫHk α k β , (2)
and suppose that coordinates are chosen so that the components ηαβ are constants,
but not necessarily of the form ηαβ = diag[−1, 1, 1, 1]. The connection is then
quadratic in ǫ
Γγ αβ = ǫ Γ γ αβ + ǫ2 Γ γ αβ , (3)
1 2

where
Γ1 γ αβ = −(Hkα k γ ),β − (Hkβ k γ ),α + (Hkα kβ ),λ η λγ ,
Γ2 γ αβ = 2H[H(k̇α kβ + k̇β kα ) + Ḣkα kβ ]k γ ≡ 2Hk γ (Hkα k β )˙ , (4)
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

The Kerr-Schild ansatz revised 3

a “dot” denoting differentiation in the k direction, i.e. f˙ = k(f ) = f,α k α . Note

that only the indices of k can be raised and lowered with the Minkowski metric.
Hereafter we will use an “index” 0 to denote contraction with k, i.e.
Γ0 αβ = Γγ αβ kγ = ǫ(Hkα kβ )˙ ,
Γγ α0 = Γγ αβ k β = −ǫ(Hkα k γ )˙ ,
Γγ 00 = Γγ αβ k α k β = 0 ,
Γ0 α0 = Γγ αβ k β kγ = 0 . (5)
The determinant of the full metric is independent of ǫ
|gαβ | = |ηαβ − 2ǫHkα kβ | = |ηαβ | = const. −→ Γβ αβ = 0 , (6)
and the contracted Riemann tensor therefore reduces to
Rαβ = Rγ αγβ = Γγ αβ,γ − Γγ αδ Γδ βγ . (7)
The simplest component is
Rαβ k α k β = Γγ αβ,γ k α k β − Γγ δ0 Γδ γ0 = Γγ 00,γ − 2Γγ α0 k α ,γ
(8)
= 2ǫH||k̇||2 .
In vacuum the LHS is zero, then ||k̇|| = 0 and so k̇ is a null-vector orthogonal to
another null-vector, k. Hence k̇ must be parallel to k and therefore k is a geodesic
vector.
The Ricci tensor expanded as series in ǫ is given by
Rαβ = ǫ R αβ + ǫ2 R αβ + ǫ3 R αβ + ǫ4 R αβ . (9)
1 2 3 4

The vacuum Einstein’s equations Rαβ = 0 imply that contributions of all orders
must vanish. Let us evaluate all such components.
The highest components of the expansion for the Ricci tensor are
R = − Γ ρ ασ Γ σ βρ = 0 , (10)
4 αβ 2 2

R = − Γ ρ ασ Γ σ βρ − Γ ρ ασ Γ σ βρ = 4H 3 ||k̇||2 kα kβ . (11)
3 αβ 1 2 2 1

The next component of Rαβ is

R = Γ ρ αβ,ρ − Γ ρ ασ Γ σ βρ
2 αβ 2 1 1
h i
= 2H (Hkα kβ ) + k σ ,σ (Hkα kβ )˙ − H k̇α k̇β
¨ (12)

− H 2 Φkα kβ − 2Hk(α ψβ) ,

where
Φ = 4η γλ η δµ k[λ,δ] k[µ,γ] , ψα = 2k̇ γ (Hkα ),γ . (13)
Finally, the first component of the Ricci tensor expansion is
R = Γ γ αβ,γ
1 αβ 1
(14)
= Akα kβ + 2k(α Bβ) + Xαβ ,
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

4 Donato Bini, Andrea Geralico and Roy P. Kerr

where
A = η λγ H,λγ ,
1
Bβ = −(Hk γ ),γβ + η λγ (H 2 kβ,γ ),λ ,
H
Xαβ = −2H (k(α,β) k γ ),γ + k(α,|γ| k γ ,β) − η λγ kα,γ kβ,λ
 

− 2k γ H,(α kβ),γ + H,γ k(α,β)

 
h i
= −2H k̇(α,β) + k γ ,γ k(α,β) − η λγ kα,γ kβ,λ
− 2Ḣk(α,β) − 2H,(α k̇β) . (15)

2.1. Kinematical properties of the congruence k

Taking the covariant derivative of k gives
∇α kβ = kβ,α − ǫ(Hkα kβ )˙ , (16)
so that its 4-acceleration is simply
a(k)β = k µ ∇µ kβ = k̇β . (17)
The other optical scalars of interest are the expansion
1 α 1 1
θ= k ;α = η αβ kβ,α = k α ,α , (18)
2 2 2
the vorticity
1 1  
ω2 = k[α;β] k α;β = k[β,α] η αµ η βν kµ,ν − 2ǫH k̇ α k β , (19)
2 2
and the shear, implicitly defined by
1 1 1
θ2 + |σ|2 = k(α;β) k α;β = k(β,α) η αµ η βν kµ,ν − ǫH||k̇||2 . (20)
2 2 2

2.2. First result: k be geodesic

The third order field equations (11) imply that k be geodesic. Then it can be
normalized so that k̇ = 0. The optical scalars (19) and (20) thus become
1 αµ βν
ω2 = η η k[β,α] kµ,ν ,
2
1
θ2 + |σ|2 = η αµ η βν k(β,α) kµ,ν . (21)
2
The second order Ricci tensor (12) simplifies to

R = 2HDkα kβ , D = Ḧ + 2θḢ + 4Hω 2 , (22)

2 αβ

leading to the condition D = 0, which gives the following equation for H

0 = Ḧ + 2θḢ + 4Hω 2 . (23)
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

The Kerr-Schild ansatz revised 5

Finally, the first order Ricci tensor (14)–(15) becomes

R = η λγ H,λγ kα kβ + 2k(α Bβ)
1 αβ
h i (24)
− 2 (Ḣ + 2θH)k(α,β) − η λγ Hkα,γ kβ,λ ,

with
Bβ = −(Ḣ + 2θH),β + η λγ (2H,λ kβ,γ + Hkβ,γλ ) . (25)
The vector k is an eigenvalue of the Ricci tensor, i.e.
Rασ k σ = (Bσ k σ )kα . (26)
It proves easier to handle with the remaining set of first order field equations
by specifying a general field of real null direction in Minkowski space together with
an adapted tetrad frame, then setting to zero each individual frame component of
the first order Ricci tensor.

2.3. Simplified tetrad procedure

Following [5,6] introduce the set of null coordinates in Minkowski space (u, v, ζ, ζ̄),
which are related to the standard Cartesian coordinates (t, x, y, z) by
1 1
u = √ (t − z) , v = √ (t + z) ,
2 2
1 1
ζ = √ (x + iy) , ζ̄ = √ (x − iy) . (27)
2 2
The metric (1) becomes
ds2 = 2(dζdζ̄ − dudv) − 2ǫHkα kβ dxα dxβ . (28)
A general field of real null directions in Minkowski space is given by
k = −[du + Y Ȳ dv + Ȳ dζ + Y dζ̄] , k = Y Ȳ ∂u + ∂v − Y ∂ζ − Ȳ ∂ζ̄ , (29)
where Y is an arbitrary complex function of coordinates. In fact the independent
components of k reduce to two real functions of the coordinates, due to the two
conditions 1) k forms a lightlike world line and 2) k has an arbitrary parametriza-
tion. In Eq. (29) these two real functions of the coordinates collapsed in a single
complex function Y , namely k = k(Y, Ȳ ).
We introduce the following frame
ω 1 = dζ + Y dv , ω 2 = dζ̄ + Ȳ dv , ω 3 = −k , ω 4 = dv + ǫHω 3 , (30)
so that
ds2 = 2ω 1 ω 2 − 2ω 3 ω 4 . (31)
The dual frame is
e1 = ∂ζ − Ȳ ∂u , e2 = ∂ζ̄ − Y ∂u , e3 = ∂u − ǫHk , e4 = k . (32)
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

6 Donato Bini, Andrea Geralico and Roy P. Kerr

The connection coefficients are given by

Γcab = −ec µ ea µ;ν eb ν . (33)
Note that ωµ1 = −kµ,Ȳ and ωµ2 = −kµ,Y , trivially implying ω 1 (k) = 0 = ω 2 (k),
because
k · ω 1 = η αβ kα ωβ1 = −η αβ kα kβ ,Ȳ = −k α kα ,Ȳ = 0 . (34)
Similarly k · ω 2 = 0.
The derivative of k is quite simple
kµ,ν = kµ,Ȳ Ȳ,ν + kµ,Ȳ Ȳ,ν = −ωµ1 Ȳ,ν − ωµ2 Y,ν . (35)
Next introduce the following standard notation for the directional derivatives
along the frame vectors
D ≡ ∇k = ∂v + Y Ȳ ∂u − Y ∂ζ − Ȳ ∂ζ̄ ,
∆ ≡ ∇e3 = ∂u − ǫHD ,
δ ≡ ∇e1 = ∂ζ − Ȳ ∂u . (36)
The geodesic curvature κ, complex expansion ρ and shear σ of the null congruence
k are given by
κ ≡ −Γ414 = −k α De1α = DȲ ,
ρ ≡ −Γ412 = −k α δ̄e1α = δ̄ Ȳ ,
σ ≡ −Γ411 = −k α δe1α = δ Ȳ , (37)
respectively. It is also useful to introduce the quantity
τ ≡ −Γ413 = −k α ∆e1α = ∂u Ȳ . (38)
If the principal null vector k is geodesic, then κ = 0, i.e.
0 = DȲ = Ȳ,v + Y Ȳ Ȳ,u − Y Ȳ,ζ − Ȳ Ȳ,ζ̄ . (39)
If it is also shearfree, then σ = 0, i.e.
0 = δ Ȳ = Ȳ,ζ − Ȳ Ȳ,u , → (c.c.) 0 = Y,ζ̄ − Y Y,u , (40)
where “c.c.” stands for “complex conjugate.” Substituting it into Eq. (39) then
yields
0 = Ȳ,v − Ȳ Ȳ,ζ̄ , → (c.c.) 0 = Y,v − Y Y,ζ . (41)
The conditions (40) and (41) thus give
Y,ζ̄ = Y Y,u , Y,v = Y Y,ζ , (42)
whence if 0 is the flat-space wave operator, then
0 Y ≡ η αβ Y,αβ = 2Y,ζ̄ζ − 2Y,uv = (Y 2 ),uζ − (Y 2 ),ζu = 0 , (43)
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

The Kerr-Schild ansatz revised 7

and therefore Y is a solution of the wave equation in Minkowski space whenever

the congruence is geodesic and shearfree. They also show that the congruence k
must satisfy the Kerr Theorem, i.e. Y is a root of an analytic equation

0 = F (Y, ζ̄Y + u, vY + ζ) , (44)

where F is an arbitrary function analytic in the three complex variables Y , ζ̄Y + u

and vY + ζ.

2.4. Completion of the solution

In terms of the connection coefficients previously introduced the optical scalars
write as
1 1
θ = − (ρ + ρ̄) , ω 2 = − (ρ − ρ̄)2 , (45)
2 4
so that the single equation (23) coming from the vanishing of second order Ricci
tensor reads

0 = Ḧ − (ρ + ρ̄)Ḣ − (ρ − ρ̄)2 H . (46)

The nonvanishing relevant frame components of the first order Ricci tensor (24)
are given by

R = 2σ[Ḣ − (ρ̄ − ρ)H] , (47a)

1 11

R = (ρ + ρ̄)Ḣ − (ρ2 + ρ̄2 − 2σσ̄)H , (47b)

1 12

R = δ Ḣ + (ρ − ρ̄)δH + 2σ δ̄H − τ Ḣ − (δ ρ̄ + 2τ̄ σ + 2τ ρ − δ̄σ)H , (47c)

1 13
 
R 33 = 2 δ δ̄H − (ρ,u + ρ̄,u )H − τ δ̄H − τ̄ δH − ρH,u ,
1
(47d)
2
R = Ḧ − (ρ + ρ̄)Ḣ − (ρ − ρ̄) H , (47e)
1 34

since R 22 and R 23 are c.c. of R 11 and R 13 respectively. The identities

1 1 1 1

δ̄τ = ρ,u + τ τ̄ , δτ = σ,u + τ 2 ,

δρ = δ̄σ + τ (ρ − ρ̄) , Dρ = σσ̄ + ρ2 ,
Dτ = τ̄ σ + τ ρ , Dσ = σ(ρ + ρ̄) , (48)

as well as the commutation relations

∂u D − D∂u = −τ̄ δ − τ δ̄ , δD − Dδ = −ρ̄δ − σ δ̄ ,

δ∂u − ∂u δ = τ ∂u , δ̄δ − δ δ̄ = −(ρ − ρ̄)∂u , (49)

have been used here to simplify the expressions involving frame derivatives of H.
Setting to zero each component of Eqs. (47a)–(47e) gives a set of first order equa-
tions. Note that the condition coming from Eq. (47e) is the same as Eq. (46).
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

8 Donato Bini, Andrea Geralico and Roy P. Kerr

Equation (47a) implies σ = 0, i.e. the congruence k must be shearfree. The

remaining first order equations thus simplify as
0 = (ρ + ρ̄)Ḣ − (ρ2 + ρ̄2 )H , (50a)
0 = δ Ḣ + (ρ − ρ̄)δH − τ Ḣ − (δ ρ̄ + 2τ ρ)H , (50b)
0 = δ δ̄H − (ρ,u + ρ̄,u )H − τ δ̄H − τ̄ δH − ρH,u , (50c)
and the identities (48) become
δ̄τ = ρ,u + τ τ̄ , δτ = τ 2 ,
δρ = τ (ρ − ρ̄) , ρ̇ = ρ2 ,
τ̇ = τ ρ . (51)
Taking the δ derivative of Eq. (50a) together with Eq. (50b) gives rise to the fol-
lowing compatibility condition
ρ(ρ + ρ̄)δH = [τ ρ(ρ + ρ̄) + τ ρ̄(ρ + 3ρ̄) + ρδ ρ̄]H . (52)
Take the complex conjugate of this equation and then its δ derivative; Eq. (50c)
thus gives rise to a second compatibility condition
   
ρτ̄ ρ̄τ ρ̄,u ρ,u τ τ̄
(ρ + ρ̄)H,u = 3 δ ρ̄ + 2 δ̄ρ + (ρ̄ + 3ρ) + (ρ − 3ρ̄) +6 H . (53)
ρ̄2 ρ ρ̄ ρ ρ̄
By using Eq. (50a), Eq. (46) rewrites as
Ḧ = 2(ρ + ρ̄)Ḣ − 2ρρ̄H . (54)
Let the complex expansion be nonzero, i.e. ρ 6= 0. It is easy to check that ρρ̄ and
ρ + ρ̄ are particular solutions, and therefore the general solution is
1
H= M (ρ + ρ̄) + Bρρ̄ , Ṁ = Ḃ = 0 , (55)
2
where M (Y, Ȳ ) and B(Y, Ȳ ) are real functions of Y and Ȳ . Substituting the general
solution (55) for H into Eq. (50a) one easily gets B = 0, by using the relation ρ̇ = ρ2 ,
so that
1
H = M (ρ + ρ̄) . (56)
2
Substituting now this solution for H into Eq. (52) leads to
3M
δM = τ ρ̄ . (57)
ρ
But M = M (Y, Ȳ ), so that δM = M,Y ρ̄, implying that
3M 3M
M,Y = τ , M,Ȳ = τ̄ . (58)
ρ ρ̄
The second compatibility condition (53) then yields
 
ρ̄ τ
δ̄τ − δ̄ρ − c.c. = 0 , (59)
ρ ρ
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

The Kerr-Schild ansatz revised 9

where the relation

 
1 1
M,u = 3M τ τ̄ + (60)
ρ ρ̄
has been used. Equation (59) is an additional equation for Y and Ȳ which we will
discuss later.
Following the original work [5] we now introduce P = (M/m)−1/3 , where m is
a real constant. The first equation of (58) thus becomes
τ
P −1 P,Y = − . (61)
ρ
By taking δ of both sides we then find
τ2
− ρ̄P −2 (P,Y )2 + ρ̄P −1 P,Y Y = − ρ̄ = −ρ̄P −2 (P,Y )2 , (62)
ρ2
since δP = ρ̄P,Y and δP,Y = ρ̄P,Y Y , and the identities (51) have been used to
replace δρ and δτ on the RHS. Equation (62) thus implies P,Y Y = 0, whose solution
is
P = pY Ȳ + qY + q̄ Ȳ + c , (63)
where p and c are real constants and q is a complex constant.
Let us turn to the remaining compatibility condition (59). First note that it can
be equivalently rewritten as
 
τ
ρ̄δ̄ − c.c. = 0 . (64)
ρ
By using Eq. (61) we have
 
τ
ρ̄δ̄ = ρρ̄P −1 [P −1 P,Y P,Ȳ − P,Y Ȳ ] . (65)
ρ
Take the complex conjugate of this expression taking into account that P is real;
substituting then into Eq. (64) we find that it is identically satisfied.
Finally, taking the exterior derivative of Y gives
dY = δY ω 1 + Y,u ω 3 = P −1 ρ̄[P ω 1 − P,Ȳ ω 3 ]
= P −1 ρ̄[(qY + c)(dζ + Y dv) − (pY + q̄)(du + Y dζ̄)] , (66)
whose general solution is
0 = F ≡ φ(Y ) + (qY + c)(ζ + Y v) − (pY + q̄)(u + Y ζ̄) , (67)
according to Eq. (44), with φ an arbitrary analytic function of the complex variable
Y . In fact, differentiating Eq. (67) leads to
F,Y dY = dF = F,α dxα = (qY + c)(dζ + Y dv) − (pY + q̄)(du + Y dζ̄) . (68)
Furthermore, taking the δ derivative of F , i.e.
ρ̄F,Y = δF = (∂ζ − Ȳ ∂u )F = P , (69)
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

10 Donato Bini, Andrea Geralico and Roy P. Kerr

implies that the complex expansion of the null vector k is given by

ρ̄ = P F,Y −1 . (70)
Equation (66) then immediately follows.
Summarizing, the solution is given by
m
ds2 = 2(dζdζ̄ − dudv) − 3 (ρ + ρ̄)[du + Y Ȳ dv + Ȳ dζ + Y dζ̄]2 , (71)
P
with
P = pY Ȳ + qY + q̄ Ȳ + c , ρ̄ = P F,Y −1 . (72)
The main properties of such a family of solutions are listed below (see e.g. [7]):
1. They are all algebraically special, with k shearfree and geodesic.
2. They all admit at least a one-parameter group of motions with Killing
vector
ξ = c∂u + p∂v + q̄∂ζ + q∂ζ̄ , (73)
which is simultaneously a Killing vector of flat spacetime. The solutions
can be simplified by performing a Lorentz transformation. One can thus
assume that if
√
a) ηαβ ξ α ξ β < 0, then P = (1 + Y Ȳ )/
√ 2, i.e. with ξ pointing along the
u + v (or t) direction (p = c = 1/ 2, √ q = 0);
b) ηαβ ξ α ξ β > 0, then P = (1 − Y Ȳ )/ √2, i.e. with ξ pointing along the
v − u (or z) direction (−p = c = 1/ 2, q = 0);
c) ηαβ ξ α ξ β = 0, then P = 1, i.e. with ξ pointing along the u direction
(p = q = 0, c = 1).
3. For a timelike Killing vector ξ, the particular case φ = −iaY , with m = M,
leads to the Kerr solution (1)–(6), once written in Kerr-Schild coordinates.

3. Concluding remarks
We have presented an alternative derivation of Kerr solution by treating Kerr-Schild
metrics as exact linear perturbations of Minkowski spacetime. In fact they have been
introduced as a linear superposition of the flat spacetime metric and a squared null
vector field k multiplied by a scalar function H.
In the case of Kerr solution the vector k is geodesic and shearfree and it is
independent of the mass parameter M, which enters instead the definition of H
linearly. This linearity property allows one to solve the field equations order by
order in powers of H in complete generality, i.e. without any assumption on the null
congruence k. The Ricci tensor turns out to consist of three different contributions.
Third order equations all imply that k must be geodesic; it must be also shearfree
as a consequence of first order equations, whereas the solution for H comes from
second order equations too.
 September 28, 2018 16:21 WSPC/INSTRUCTION FILE
kerrschild˙091113˙fin˙grqc

The Kerr-Schild ansatz revised 11

The present treatment can be generalized to include also the electromagnetic

field, i.e. to the case of Kerr-Newman. In fact, even in the charged Kerr solution
the congruence of k-lines depend only on the rotation parameter a and not on the
mass M or charge Q. Furthermore, the electromagnetic field is linear in Q and the
metric is linear in M and Q2 , since the function H is obtained simply by replacing
M → M − Q2 /(2r).

Acknowledgements
We thank Prof. R. Ruffini and ICRANet for support.

References
[1] R. P. Kerr and A. Schild, A new class of vacuum solutions of the Einstein field
equations, in Atti del convegno sulla relatività generale; problemi dell’energia e onde
gravitazionali, ed. G. Barbera (Firenze, 1965) p. 173.
[2] R. P. Kerr and A. Schild, Proc. Symp. Appl. Math. 17 (1965), 199.
[3] R. P. Kerr, Phys. Rev. Lett. 11 (1963), 237.
[4] J. N. Goldberg and R. K. Sachs, Acta Phys. Polon. 22 Suppl. (1962), 13.
[5] G. Debney, R. P. Kerr and A. Schild, J. Math. Phys. 10 (1969), 1842.
[6] R. P. Kerr and W. B. Wilson, Gen. Rel. Grav. 10 (1979), 273.
[7] H. Stephani, D. Kramer, M. A. H. MacCallum, C. Hoenselaers and E. Herlt, Exact
Solutions of Einstein’s Field Equations, 2nd ed. (Cambridge Univ. Press, Cambridge,
2003).