The Schwarzschild metric, with the simplification c=G=0,

ds2 =(1 -2Mr) dt2 -(1 -2Mr )1 dr2 -r2dθ2 -r2sin2θ dφ2

describes the spacetime around a spherically symmetric source outside of the actual source material. It was first generalized to an arbitrary number of spatial dimensions by Tangherlini, working in standard higher-dimensional spherical coordinates. In that reference he states that the Einstein equations arise “[a]fter a standard calculation”, and while this is true it glosses over the details of the actual evaluation.

In higher-dimensional spherical coordinates, most components of the metric tensor depend on almost all independent variables. While this occurs in a regular pattern and can be handled, it is still a nontrivial calculation and rather tedious. The use of the Cartan formalism does not make this process much simpler compared to employing a traditional coordinate basis.

What does simplify the process is switching first to a different set of coordinates that go by the name of Kerr-Schild. These are algebraically special coordinates that do not apply in general to gravitating systems, but do apply in the Schwarzschild case. A metric in these coordinates has the general form

gik =ηik -φ(xm) vi vk

where the first term is the diagonal Minkowski metric of flat spacetime,

ηik =[1,1, 1,1, ]

the coefficient φ is a scalar function of spacetime variables, and the vector with n+1 components is null with respect to both the full metric and the Minkowski metric:

glm vlvm =0 =ηlm vlvm

The left-hand equality defines nullity in a metric space, while the right-hand equality is additional information. By convention a repeated index implies summation.

Due to the null nature of the vector, its index is essentially raised and lowered with the Minkowski metric:

vi =gim vm =ηim vm -φvi vmvm =ηim vm

Then it is almost trivial to verify that the inverse of the metric tensor is

gik =ηik +φ(xm) vi vk

with a change of sign on the second term. Differentiation of the statement of nulllity implies

k [vmvm =0] vm kvm =vm kvm =0

since the constant Minkowski metric moves through the differentiation. The null condition thus holds over single derivatives.

This is one situation where the Cartan formalism is a hindrance compared to a coordinate basis. Diagonalizing the metric as one typically does in the Cartan formalism simply returns to spherical coordinates, thereby defeating the purpose of the exercise. It is simpler to use the traditional evaluation.

Combinations of the null vector are great aids in simplifying quantities. The Christoffel symbols in raw form are

Γkli =12 gim[ k glm +l gkm -m gkl] Γkl i =12 gim[ k(φ vlvm) +l(φ vkvm) -m(φ vkvl)]

Contracting these symbols on the lower indices with the null vector clearly gives zero, since nullity holds over single derivatives. If one assumes the null vector defines geodesics in these coordinates, then a single contraction over the geodesic equation adds another useful combination:

vk[ vi ;k =vi ,k +Γkmi vm =0] vkk vi =0

That the null vector is geodesic can be shown in more generality: for this presentation it is sufficient to assume it is and check explicitly. The final index can be lowered with the constant Minkowski metric.

With the two combinations of the null vector, the Christoffel symbols become

Γkli =12 (ηim +φvi vm) [k( φvl vm) +l( φvk vm) -m( φvk vl)] Γ kli =12 [k( φvi vl) +l( φvi vk) -ηim m( φvk vl) -φvi vkvl vmmφ ]

where the geodesic condition allows the null vector to move through summed derivatives like a constant. The Ricci tensor in terms of the Christoffel symbols is

Rik =m Γikm -k Γimm +Γlmm Γikl -Γiml Γklm

The two middle terms contain the ‘trace’ of the Christoffel symbols, which is

Γkll =12 [l( φvl vk) -m( φvk vm)] =0

so that these terms do not contribute to the Ricci tensor. The fourth term is messy at first, but the last terms in each bracket cancel out immediately:

Γiml Γklm =14 [i( φvl vm) +m( φvl vi) -ηlp p( φvi vm) -φvl vivm vppφ ] ×[k( φvm vl) +l( φvm vk) -ηmr r( φvk vl) -φvm vkvl vrrφ ] Γiml Γklm =14 [i( φvl vm) +m( φvl vi) -ηlp p( φvi vm) ] ×[k( φvm vl) +l( φvm vk) -ηmr r( φvk vl) ]

Expanding derivatives it is not difficult to see that the first terms in each bracket cancel against the remaining terms, leaving

Γiml Γklm =14 [m( φvl vi) -ηlp p( φvi vm) ] [l( φvm vk) -ηmr r( φvk vl) ] Γiml Γklm =12 [m( φvl vi) l( φvm vk) -ηlp p( φvi vm) l( φvm vk) ]

where again the constant Minkowski metric moves through derivatives in the second step, allowing symmetry in the two unsummed indices. The Ricci tensor is thus

Rik =12 m [i( φvm vk) +k( φvm vi) -ηmp p( φvi vk) -φvm vivk vppφ ] -12 [m( φvl vi) l( φvm vk) -ηlp p( φvi vm) l( φvm vk) ]

This can be simplified greatly by raising one index. First write

Rik =gin Rnk Rik =12 m [ηin n( φvm vk) +k( φvm vi) -ηmp p( φvi vk) -φvi vpp( φvmvk) ] -12 [m( φvl vi) l( φvm vk) -ηlp p( φvi vm) l( φvm vk) ] -12 φvi vn m [n( φvm vk) +k( φvm vn) -ηmp p( φvn vk) -φvm vnvk vppφ ]

When the factors outside of the overall derivative of the third bracket are brought inside the derivative, they will zero out the final three terms inside the bracket, leaving a single term that cancels the final term inside the first bracket. One merely needs to add compensating contributions from the derivative:

Rik =12 m [ηin n( φvm vk) +k( φvm vi) -ηmp p( φvi vk) ] -12 [m( φvl vi) l( φvm vk) -ηlp p( φvi vm) l( φvm vk) ] +12 m( φvi vn) [n( φvm vk) +k( φvm vn) -ηmp p( φvn vk) -φvm vnvk vppφ ]

The second and fourth terms inside the third bracket are zero, and the first and third terms inside the same bracket cancel the second bracket. This just leaves

Rik =12 m [ηin n( φvm vk) +k( φvm vi) -ηmp p( φvi vk) ]

as the simplified form of the Ricci tensor in mixed indices.


Now consider transforming the Schwarzschild metric above in spherical coordinates to Kerr-Schild coordinates. In keeping with the foregoing, first write it more generally as

ds2 =(1-φ) dt2 -dr2 1-φ -r2dθ2 -r2sin2θ dφ2

If the temporal variable is replaced using

dt=du +dr 1-φ

the metric becomes

ds2 =(1-φ) du2 +2dudr -r2dθ2 -r2sin2θ dφ2 ds2 =(du +dr)2 -dr2 -r2dθ2 -r2sin2θ dφ2 -φdu2 ds2 =(du +dr)2 -dx2 -dy2 -dz2 -φdu2

where the completely spatial part of the metric has been rewritten in Cartesian coordinates. Now define a new temporal variable with

dτ=du+dr

so that the metric becomes

ds2 =dτ2 -dx2 -dy2 -dz2 -φ (dτ -dr )2 ds2 =dτ2 -dx2 -dy2 -dz2 -φ (dτ -xdx+ydy +zdzr )2

The first part of the metric is flat Minkowski spacetime. From the second term, the vector

vi =[1, xr, yr, zr]

is null with respect to the Minkowski metric. For the nonconstant components one can easily verify that is is geodesic (here indices run over spatial variables only):

vmm vi =xmr m xir =xmr [δim r -xi xm r3] =0

The metric tensor in Kerr-Schild coordinates is thus

gik =ηik -2Mr vi vk =ηik -φ vi vk

with the given null vector. The metric is clearly in a form easily extensible to an arbitrary number of dimensions: the problem is now to determine the dependence of the one metric function on the dimension.

The Ricci tensor as given above already holds in an arbitrary number of dimensions. One need merely pick a component that is particularly suitable for determining the single metric function and set it equal to zero. The simplest component is

R00 =12 mm φ =12 2φ

and the function is determined from the ordinary flat-space Laplacian! First evaluate the Laplacian (here indices run over spatial variables only),

mm φ(r) =m( xmr φ) =xm2 r2 φ +(1r -xm2 r2) φ

then sum over spatial variables and set equal to zero,

φ +(n-1) rφ =0 lnφ +(n-1) lnr=lnc φ=2M rn-2

where the constant of integration has been set to match the three-dimensional metric. Then one need merely reverse the transformation process for the final metric

ds2 =(1 -2M rn-2) dt2 -(1 -2M rn-2 )1 dr2 -r2d Ω(n)2

with an n-dimensional differential of solid angle.

This evaluation via Kerr-Schild coordinates is almost trivial compared to a direct evaluation in spherical coordinates, once one has the Ricci tensor in the same coordinates. There is admittedly a bit of work involved in obtaining that general result, but thanks the to the two combinations of the null vector it is basically a matter of watching terms go away.


Uploaded 2021.06.07 analyticphysics.com