-
Notifications
You must be signed in to change notification settings - Fork 18
/
sta.tex
4028 lines (3789 loc) · 222 KB
/
sta.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
\chapter{Stationary black holes}
\label{s:sta}\index{stationary!black hole}
\minitoc
\section{Introduction}
Having defined black holes in all generality in Chap.~\ref{s:glo}, we focus
here on black holes in steady state, i.e. black holes in stationary spacetimes.
We have already discussed
non-expanding horizons and Killing horizons in Chap.~\ref{s:neh} as
possible models for the event horizon of a
steady state black hole. Actually, we shall see here that (each connected
component of) the event horizon of a black hole in a stationary spacetime has to be
a Killing horizon.
We shall start by defining properly the concept of a stationary
spacetime and investigating the first properties of a black hole
in such a spacetime (Sec.~\ref{s:sta:sta_st}). Then, we extend the study of Killing horizons
initiated in Chap.~\ref{s:neh} to encompass bifurcate Killing horizons
(Sec.~\ref{s:sta:bifur_Killing_hor}). In Sec.~\ref{s:sta:mass_angul_mom},
we discuss the concepts
of mass and angular momentum in asymptotically flat spacetimes, which are useful to characterize black holes. In Sec.~\ref{s:sta:EH_KH}, we shall
see that
if the Killing vector $\w{\xi}$ generating stationarity is null on
a connected event horizon $\Hor$, the latter is a Killing horizon with
respect to $\w{\xi}$. If $\Hor$ is non-degenerate and the
electrovacuum Einstein equation holds, this can only occur in a static spacetime
(\emph{staticity theorem}, Sec.~\ref{s:sta:staticity_thm}).
If on the contrary, $\w{\xi}$ is spacelike on some parts of $\Hor$ (the timelike
case is excluded for the event horizon is a null hypersurface), then, modulo the
electrovacuum Einstein equation and some additional hypotheses,
$\Hor$ is still a Killing horizon, albeit with respect to a Killing vector
distinct from $\w{\xi}$ (\emph{strong rigidity theorem}, Sec.~\ref{s:sta:strong_rigidity}).
Section~\ref{s:sta:Smarr} is devoted to an important relation between various global
quantities characterizing a stationary black hole: the \emph{Smarr formula}.
This is the opportunity to investigate electromagnetic
fields on the horizon of a stationary black hole, in particular to define
the black hole's electric charge and electric potential, both of them being involved
in the Smarr formula. The culmination point of this chapter is
Sec.~\ref{s:sta:no-hair}, which presents the famous
\emph{no-hair theorem}. Modulo some hypotheses, this theorem stipulates that in 4-dimensional
general relativity, all isolated stationary electrovacuum black holes are necessarily
Kerr-Newman black holes; in the pure vacuum case (the most relevant one for astrophysics),
they are Kerr black holes, to be explored in Part.~\ref{P:ker}.
\section{Stationary spacetimes} \label{s:sta:sta_st}
\subsection{Definitions} \label{s:sta:def_station}
\begin{greybox}
A spacetime $(\M,\w{g})$ is called \defin{stationary}\index{stationary!spacetime}
iff (i) it is invariant under
the action of the translation group $(\R,+)$ and (ii) the orbits of
the group action (cf. Sec.~\ref{s:neh:symmetries})
are everywhere timelike curves or (ii') $(\M,\w{g})$
admits a conformal completion (cf. Sec.~\ref{s:glo:conf_compl})
and the orbits of the group action are timelike in the vicinity of
the conformal boundary $\scri$.
It is equivalent to say that there exists a Killing vector field
$\w{\xi}$ (the generator of the translation group, cf. Sec.~\ref{s:neh:symmetries}) that is
timelike everywhere or at least in the vicinity of $\scri$ when there exists a conformal
completion. We shall say that $(\M,\w{g})$ is \defin{strictly stationary}\index{strictly!stationary}\index{stationary!strictly --} iff the Killing vector field $\w{\xi}$ is timelike in all $\M$,
i.e. iff property (ii) above is fulfilled.
\end{greybox}
\begin{remark} \label{r:sta:pseudo-stationary}
Some authors (e.g. Carter \cite{Carte73b}) call
\emph{pseudo-stationary}\index{pseudo-stationary} the stationary spacetimes
that obey (ii'), keeping the qualifier
\emph{stationary} for the strictly stationary case.
As we are going to see, when $\M$
contains a black hole, $\w{\xi}$ cannot be timelike everywhere,
so only \emph{pseudo-stationarity} in Carter's sense is relevant for such spacetimes.
Our terminology, namely keeping the qualifier \emph{stationary} even for (ii'), follows that of
Choquet-Bruhat \cite{Choqu09},
Chru\'sciel, Lopes Costa \& Heusler \cite{ChrusLH12},
Heusler \cite{Heusl96}
and Wald \cite{Wald01}.
\end{remark}
The Killing vector field $\w{\xi}$ is a priori not unique: the translation group $(\R,+)$
admits the reparametrizations
$t\mapsto t' = \alpha t$, where $\alpha$ is a nonzero constant, which yield
the rescallings $\w{\xi} \mapsto \w{\xi'} = \alpha^{-1} \w{\xi}$
of the Killing vector fields [cf. Eq.~(\ref{e:neh:xi_dxdt})]. When the
spacetime admits a conformal boundary $\scri$ --- which is required if a black hole
is present, given the definition (\ref{e:glo:def_BH}) ---, then
one selects $\w{\xi}$ by demanding that
\be \label{e:sta:xi_scri}
\w{\xi}\cdot\w{\xi} \to -1 \quad \mbox{near\ } \scri .
\ee
This determines $\w{\xi}$ uniquely, up to a factor $\pm 1$, since this fixes
the rescaling constant $\alpha$ to $\pm 1$. If the spacetime is time-orientable, one
can get rid of the $\pm 1$ ambiguity by demanding that $\w{\xi}$
is future-oriented near $\scri$.
If the stationary spacetime is endowed with electromagnetic and/or
matter fields, they must respect the stationarity as well. For the electromagnetic
field $\w{F}$ (cf. Sec.~\ref{e:fra:electrovacuum}), this is expressed by
the vanishing of the Lie derivative along $\w{\xi}$:
\be
\Lie{\xi} \w{F} = 0 .
\ee
This condition is similar to Killing-vector property
$\Lie{\xi} \w{g} = 0$ [Eq.~(\ref{e:neh:Lie_xi_g})].
A notion stronger than stationarity is that of \emph{staticity}:
\begin{greybox}
A spacetime $(\M,\w{g})$ is called \defin{static}\index{static!spacetime}
iff (i) it is stationary and (ii) the Killing vector field $\w{\xi}$
generating the stationary action is orthogonal to a family of hypersurfaces
(one says that $\w{\xi}$ is \defin{hypersurface-orthogonal}\index{hypersurface-orthogonal}).
The spacetime $(\M,\w{g})$ is called \defin{strictly static}\index{strictly!static}\index{static!strictly --}
iff moreover $\w{\xi}$ is timelike in all $\M$.
\end{greybox}
\begin{remark}
The same comment as in Remark~\ref{r:sta:pseudo-stationary} can be made: some authors
would call \emph{static} only spacetimes that are
\emph{strictly static} according to the above definition.
\end{remark}
In loose terms, a spacetime is stationary if ``nothing changes with time'', while
it is static if, in addition, ``nothing moves''. A prototype of a stationary
spacetime that is not static is a spacetime containing a steadily rotating
body, be it a star or a black hole.
\subsection{Coordinates adapted to stationarity or staticity}
In a $n$-dimensional stationary spacetime $(\M,\w{g})$, a coordinate system $(x^\alpha) = (x^0, x^1, \ldots x^{n-1})$
is said \defin{adapted to stationarity}\index{adapted!coordinates} iff the coordinate vector
$\wpar_t$, where $t := x^0$, coincides with the stationary Killing vector $\w{\xi}$.
According to Property~\ref{p:neh:isometry_adapt_coord}, $t$ is then an
ignorable coordinate\index{ignorable coordinate}\index{coordinate!ignorable --}, i.e. the
components $g_{\alpha\beta}$ of the metric tensor with respect to $(x^\alpha)$ obey $\dert{g_{\alpha\beta}}{t} = 0$
[Eq.~(\ref{e:neh:dgabdt_zero})]. It follows that
\be
g_{\alpha\beta} = g_{\alpha\beta}(x^1, \ldots x^{n-1}) .
\ee
If the spacetime $(\M,\w{g})$ is static, a coordinate system $(x^\alpha) = (x^0=t, x^1, \ldots x^{n-1})$
is said \defin{adapted to staticity} iff it is adapted to stationarity and the hypersurfaces
$t=\mathrm{const}$ are orthogonal to the static Killing vector $\w{\xi}$.
Given that a normal 1-form to the hypersurfaces $t=\mathrm{const}$ is $\dd t$,
the coordinates $(x^\alpha)$ are adapted to staticity iff
\be \label{e:sta:xi_wpar_t}
\w{\xi} = \wpar_t \qquad\mbox{and}\qquad \uu{\xi} = W\, \dd t ,
\ee
where $\uu{\xi}$ is the metric dual of $\w{\xi}$ (cf. Sec.~\ref{s:bas:metric_dual})
and $W := \w{g}(\w{\xi},\w{\xi}) = \langle \uu{\xi}, \w{\xi} \rangle$.
The orthogonality of $\wpar_t$ and the hypersurfaces $t=\mathrm{const}$
translates to $g_{0i}=0$ for $i\in\{1,\ldots,n-1\}$,
the $g_{\alpha\beta}$'s being the metric components with respect to the coordinates
$(x^\alpha)$. Hence one may write the metric of a static spacetime of dimension $n$ as
\be \label{e:sta:static_metric}
\w{g} = W \dd t^2 + g_{ij} \dd x^i \dd x^j ,
\ee
where the indices $(i,j)$ range in $\{1,\ldots,n-1\}$
and $W$ and $g_{ij}$ are functions of $(x^1,\ldots,x^{n-1})$ only.
It is clear that the metric (\ref{e:sta:static_metric}) is invariant\footnote{Would
(\ref{e:sta:static_metric}) have contained a non-vanishing $g_{0i}\, \D t \, \D x^i$ term,
this would not have been the case.} in the
transformation $t\mapsto-t$. One says that a static spacetime is
\defin{time-reflection symmetric}\index{time!reflection symmetry}\index{reflection!time --}.
\begin{example}[staticity of anti-de Sitter spacetime]
The anti-de Sitter spacetime\index{anti-de Sitter spacetime} AdS$_{4}$ considered in Example~\ref{x:neh:AdS} of Chap.~\ref{s:neh}
is strictly static, with the coordinates $(\tau, r, \th,\ph)$ being
adapted to staticity (hence their name: \emph{global static}).
Indeed, the metric (\ref{e:neh:AdS4}) is of the form (\ref{e:sta:static_metric})
with $t = \tau$ and $W = -\ell^2(1+ r^2)$. The strict staticity follows from $W < 0$, which
shows that the Killing vector $\w{\xi} = \wpar_t$ is everywhere timelike.
\end{example}
\begin{example}[staticity of Schwarzschild spacetime]
The Schwarzschild spacetime\index{Schwarzschild!spacetime}, introduced in Example~\ref{x:def:Schw_hor} of Chap.~\ref{s:def}, is static, but not
strictly static. The staticity is however not obvious from the metric components
given by Eq.~(\ref{e:def:Schw_metric}) because the coordinates
$(t,r,\th,\ph)$ used there are not adapted to staticity: the Killing vector
$\w{\xi} = \wpar_t$ is not orthogonal to the hypersurfaces $t=\mathrm{const}$,
given that $g_{tr} \neq 0$. We shall introduce coordinates adapted to
staticity in Chap.~\ref{s:sch}, namely the \emph{Schwarzschild-Droste coordinates}.
That the Schwarzschild spacetime is not strictly static can be read directly on
Eq.~(\ref{e:def:Schw_metric}): $\w{\xi}\cdot\w{\xi} = g_{tt} = -1 + 2m/r$, so
that $\w{\xi}$ is timelike only in the region $r > 2 m$.
\end{example}
\subsection{Black holes in stationary spacetimes}
\label{s:sta:BH_stationary}
Let us consider a spacetime $(\M,\w{g})$ that \emph{contains a black hole}, as defined in
Sec.~\ref{s:glo:def_BH}. In particular, $(\M,\w{g})$ admits a future
infinity $\scri^+$.
Furthermore, we assume that $(\M,\w{g})$ is \emph{stationary},
as defined in Sec.~\ref{s:sta:def_station}.
Since $(\M,\w{g})$ is invariant under the action of the isometry group $(\mathbb{R},+)$,
so is $\scri^+$ (under some proper extension of $\w{\xi}$ to the conformal
completion $\tilde{\M}$)
and therefore its causal past $J^-(\scri^+)$. Since the event horizon $\Hor$
is the boundary of $J^-(\scri^+)$
inside $\M$ [Eq.~(\ref{e:glo:Hor_bound_past_scrip})], we get
\begin{prop}[stationary event horizon]
\label{p:sta:stationary_hor}
The event horizon $\Hor$ of a black hole in a stationary spacetime
is globally invariant under the action of the stationary group $(\R,+)$.
\end{prop}
The word \emph{globally} stresses that
$\Hor$ is invariant \emph{as a whole}, not that
each point of $\Hor$ is a fixed point of the group action.
Let us assume that $\Hor$ is smooth (this sounds likely in a stationary context;
a rigorous proof can be found in Ref.~\cite{ChrusDGH01});
it is then a null hypersurface (Property~\ref{p:glo:prop4} in Sec.~\ref{s:glo:properties_H}).
Now, $\Hor$ is globally invariant if, and only if, the
generator $\w{\xi}$ of the isometry group is tangent to $\Hor$.
Since a timelike vector cannot be tangent to a null hypersurface (cf.
Lemma~\ref{p:def:tangent_to_null_hyp} in Sec.~\ref{s:def:spacelike_sections}), we conclude:
\begin{prop}[stationary Killing vector tangent to the event horizon]
\label{p:sta:xi_tangent_H}
In a stationary spacetime containing a black hole,
the stationary Killing vector field $\w{\xi}$ is tangent to the event horizon
$\Hor$, which implies that $\w{\xi}$ is either null or spacelike on $\Hor$.
Let $\Hor_0$ be a connected component of $\Hor$
($\Hor_0 = \Hor$ if $\Hor$ is connected).
If $\w{\xi}$ is null on all $\Hor_0$, one says that $\Hor_0$
is \defin{non-rotating}\index{non-rotating horizon}, while if $\w{\xi}$ is
spacelike on some part of $\Hor_0$, one says that $\Hor_0$ is
\defin{rotating}\index{rotating horizon}.
\end{prop}
Since $\w{\xi}$ cannot be timelike on $\Hor$, it follows immediately
that a stationary spacetime containing a black hole cannot be strictly stationary,
according to the definition given in Sec.~\ref{s:sta:def_station}.
We shall discuss in detail the two cases --- null or spacelike --- allowed
for $\w{\xi}$ on $\Hor$ in Sec.~\ref{s:sta:EH_KH}.
It is rather intuitive that the event horizon $\Hor$ of a black hole in
a stationary spacetime must have a vanishing expansion $\theta_{(\wl)}$ along its null
normals $\wl$. Indeed, if $\theta_{(\wl)}$ were nonzero,
the area of cross-sections would vary when dragged
along $\wl$ and this would define some ``evolution'' along $\Hor$ (e.g. from small areas
to larger ones),
which would break the invariance of $\Hor$ under the action of the stationary group.
This of course results from $\Hor$ being part of the global spacetime structure,
since not any null hypersurface in a stationary spacetime has a vanishing
expansion: for instance, a future light cone in Minkowski spacetime
(which is stationary!) has $\theta_{(\wl)}>0$
(cf. Example~\ref{x:def:light_cone6} on p.~\pageref{x:def:light_cone6}),
but the light cone is not invariant by any time translation isometry.
The rigorous proof that $\theta_{(\wl)} = 0$
for stationary event horizons can be found in Hawking \& Ellis' textbook
(Proposition~9.3.1 of Ref.~\cite{HawkiE73}).
Here, we shall simply state:
\begin{prop}[non-expanding horizons for stationary black holes]
\label{p:sta:hor_non_expanding}
The event horizon $\Hor$ of a black hole in a stationary spacetime
is a null hypersurface of vanishing expansion:
\be
\theta_{(\wl)} = 0 .
\ee
Moreover, if the cross-sections $\Sp$ of $\Hor$
are closed manifolds such that $\Hor$ has the topology $\R\times \Sp$, $\Hor$ is a
\emph{non-expanding horizon}\index{non-expanding!horizon}\index{horizon!non-expanding --},
according to the definition given in Sec.~\ref{s:neh:neh}.
\end{prop}
In dimension 4, one can strongly constrain the topology of the horizon cross-sections:
\begin{prop}[topology theorem 1 \textnormal{(Hawking 1972 \cite{Hawki72})}]
\label{p:sta:topology1}
Let $(\M, \w{g})$ be a 4-dimensional stationary spacetime
containing a black hole of event horizon $\Hor$.
Let $\Sp$ be a connected component of a complete cross-section of $\Hor$.
Let us assume that (i) $\Sp$ is closed (compact without boundary) and orientable;
(ii) the null dominance condition\index{null!dominance condition}\index{dominance!null -- condition} (\ref{e:neh:null_dominant_cond})
is fulfilled on $\Hor$ for some scalar field $f \geq 0$
[for general relativity with $f=\Lambda$
this is equivalent to assuming $\Lambda \geq 0$ and
the dominant null energy condition\index{null!dominant energy condition}\index{energy!condition!null dominant --} (\ref{e:neh:null_dominant_cond_T})]
and (iii) when
displaced into the black hole exterior along $-\w{k}$ (the opposite of
the ingoing null normal $\w{k}$ to $\Sp$),
$\Sp$ becomes a surface with $\theta_{(\wl)} > 0$.
In particular, condition (iii) holds if $\theta_{(\w{k})} < 0$ and there is no trapped
surface (cf. Sec.~\ref{s:neh:trapped_surfaces}) in the black hole exterior.
Then the cross-section $\Sp$
has \emph{generically} the topology of the 2-sphere (i.e. $\Sp$ is
\emph{homeomorphic}\index{homeomorphic} to $\SS^2$). The non-generic case
is that of $\Sp$ having the topology of the 2-torus $\mathbb{T}^2 = \SS^1\times\SS^1$; this can occur only under special circumstances,
among which the metric induced by $\w{g}$ on $\Sp$ must be flat.
\end{prop}
\begin{proof}
Let $\wl$ be a future-directed null normal to $\Hor$ and $\w{k}$ a complementary
future-directed null vector field
normal to $\Sp$, normalized as in Eq.~(\ref{e:def:k_el_minus_one}), i.e.
$\w{k}\cdot\wl = -1$. At each point $p\in \Sp$, the pair $(\w{k},\wl)$ is
a basis of the 2-plane $T_p^\perp\Sp$ orthogonal to $\Sp$ (cf. Fig.~\ref{f:def:TS_ortho}),
with $\wl$ tangent to $\Hor$ and $\w{k}$ transverse to it.
Morever $\w{k}$ points towards the black hole interior, otherwise null geodesics
leaving $\Hor$ along $\w{k}$ would enter $J^-({\scri^+})$.
As in Sec.~\ref{s:def:geom_null_hypsurf}, let us consider that $\Hor$
is the level set $u=0$ of 1-parameter family of hypersurfaces $(\Hor_u)_{u\in\R}$.
This extends $\wl$ in the vicinity of $\Hor$ via Eq.~(\ref{e:def:wl_rho_u})
as a null vector field normal to each $\Hor_u$.
By Property~\ref{p:sta:hor_non_expanding}, we have $\theta_{(\wl)} = 0$
on $\Sp$. Let us displace $\Sp$ by a small parameter $\eps>0$ along $-\w{k}$. The expansion along $\wl$ of the obtained surface is
positive by hypothesis (iii). By taking the limit $\eps\to 0$, we form the
derivative $\Lie{-k} \theta_{(\wl)}$, which must obey
$\Lie{-k} \theta_{(\wl)} \geq 0$ since $\theta_{(\wl)} = 0$ on $\Sp$ and
$\theta_{(\wl)} > 0$ on the displaced surface. Given that
$\Lie{-k} \theta_{(\wl)} = -\Lie{k} \theta_{(\wl)}$, we see that hypothesis
(iii) implies
\be \label{e:sta:Lie_k_l}
\Lie{k} \theta_{(\wl)} \leq 0 .
\ee
A standard identity (cf. e.g. Eq.~(3b) of \cite{Haywa94}, Eq.~(3.1) of \cite{BoothF07},
Eq.~(5.1) of \cite{Cao11} or Eq.~(36) of \cite{Jaram13})
expresses $\Lie{k}\theta_{(\wl)}$ as
\be \label{e:sta:Lie_k_theta_l}
\Lie{k} \theta_{(\wl)} = - \frac{1}{2} {}^\Sp\!\! R - \DSc_a \Omega^a
+ \Omega_a \Omega^a + \w{G}(\wl,\w{k}) ,
\ee
where ${}^\Sp\!\! R$ is the Ricci scalar of the Riemannian metric $\w{q}$ on
$\Sp$ induced by the spacetime metric $\w{g}$, $\DS$ is the Levi-civita
connection associated to $\w{q}$,
$\w{G}$ is the Einstein tensor
of the spacetime metric $\w{g}$ and $\w{\Omega}$ is the 1-form on $\Sp$
that is the pullback of the 1-form $\w{\omega}$ defined by Eq.~(\ref{e:def:def_omega}).
We may also view $\w{\Omega}$ as a spacetime 1-form, defined as
the composition of $\w{\omega}$ with the orthogonal projector $\vw{q}$ onto
$\Sp$: $\w{\Omega} = \w{\omega} \circ \vw{q}$.
In view of Eq.~(\ref{e:def:omega_restrict_H}), we may then write
$\langle\w{\Omega}, \w{v} \rangle = - \w{k} \cdot\wnab_{\vw{q}(\w{v})} \w{\ell}$,
or in index notation, $\Omega_\alpha = - k_\mu \nabla_\nu \el^\mu q^\nu_{\ \, \alpha}$.
Besides, using Eq.~(\ref{e:neh:null_dominant_cond}), we have
$\w{G}(\wl, \w{k}) = \vw{G}(\wl)\cdot\w{k} = - \w{W}\cdot\w{k} - f \wl\cdot \w{k} = - \w{W}\cdot\w{k} + f$.
Then, integrating (\ref{e:sta:Lie_k_theta_l}) over the compact manifold $\Sp$
and setting the integral of the divergence term $\DSc_a \Omega^a$ to zero since
$\Sp$ has no boundary, we get
\be \label{e:sta:Euler_char}
\underbrace{\frac{1}{2} \int_\Sp {}^\Sp\!\! R \, \sqrt{q} \, \D^2 x}_{2\pi\chi} = \int_\Sp \Big[
\underbrace{\Omega_a \Omega^a}_{\geq 0}
\, \underbrace{- \w{W}\cdot\w{k}}_{\geq 0}
+ \underbrace{f}_{\geq 0}
\, \underbrace{- \Lie{k} \theta_{(\wl)}}_{\geq 0} \Big]
\sqrt{q} \, \D^2 x .
\ee
In the left-hand side, we have invoked the Gauss-Bonnet theorem\index{Gauss-Bonnet theorem}
to express the integral of ${}^\Sp\!\! R$ in terms
of the Euler characteristic\index{Euler characteristic} $\chi$ of the
surface $\Sp$, using the fact that in dimension 2,
the Ricci scalar ${}^\Sp\!\! R$ is twice the
Gaussian curvature\index{Gaussian!curvature}\index{curvature!Gaussian --}.
The signs of the terms of the integrand in the right-hand side
are justified as follows:
$\Omega_a \Omega^a = q^{ab} \Omega_a \Omega_b \geq 0$ because $\w{q}$ is a Riemannian metric,
$- \w{W}\cdot\w{k} \geq 0$ by Lemma~\ref{p:fra:lem2} in Sec.~\ref{s:fra:time_orientation},
given that $\w{k}$ is a future-directed null vector and $\w{W}$ is a future-directed
causal vector thanks to the null dominance condition (\ref{e:neh:null_dominant_cond}),
$f\geq 0$ by hypothesis and $- \Lie{k} \theta_{(\wl)} \geq 0$
follows from Eq.~(\ref{e:sta:Lie_k_l}). Now, the Euler characteristic $\chi$ is a topological
invariant, which is $2$ for the sphere $\SS^2$, $0$ for the torus $\mathbb{T}^2$
and $-2(g-1)$
for a connected orientable surface of genus $g$, i.e. with $g$ ``holes''.
Equation~(\ref{e:sta:Euler_char}) yields $\chi \geq 0$.
The only possibilities for $\Sp$ compact, connected and orientable
are $\SS^2$ ($\chi = 2$) and $\mathbb{T}^2$ ($\chi = 0$).
However the torus case is very special. Indeed, setting $\chi=0$ in Eq.~(\ref{e:sta:Euler_char})
implies that each term in the integrand of the right-hand side vanishes separately:
$\Omega_a \Omega^a = 0$, $\w{W}\cdot\w{k} = 0$, $f=0$ and $\Lie{k} \theta_{(\wl)} = 0$.
The last condition is the marginal case in the inequality (\ref{e:sta:Lie_k_l}). Moreover,
since $\w{q}$ is positive definite, $\Omega_a \Omega^a = 0$ implies $\w{\Omega} = 0$,
which in turn implies $\DSc_a \Omega^a = 0$. In addition,
$\w{G}(\wl, \w{k}) = - \w{W}\cdot\w{k} + f = 0$. We then deduce immediately from
Eq.~(\ref{e:sta:Lie_k_theta_l}) that ${}^\Sp\!\! R = 0$, i.e.
the Ricci scalar of $\w{q}$ is identically zero. Given that the Riemann
curvature tensor of a 2-dimensional metric is proportional to its
Ricci scalar [cf. Eq.~(\ref{e:bas:Riem_n_2})], it follows that $\w{q}$
is a flat metric.
\end{proof}
\begin{remark}
The topology theorem 1, as stated above, is slightly different from the original
version given by Hawking \cite{Hawki72,Hawki73,HawkiE73}, for it adds the ``outermost'' hypothesis (iii). In Hawking's version, (iii) is deduced from the non-existence of
trapped surfaces in the black hole exterior $J^-(\scri^+)$, the latter property
being proved by assumming that the spacetime is globally hyperbolic (Proposition~9.2.8
in Ref.~\cite{HawkiE73}).
Another difference with Hawking's version,
as stated in Proposition~9.3.2 of Ref.~\cite{HawkiE73}, is that
the torus topology is not totally excluded by Property~\ref{p:sta:topology1}.
However, as discussed in
the historical note below, it seems that the exclusion of the torus in
Hawking's proof requires additional hypotheses, which are not explicitely stated
in Refs.~\cite{Hawki72,HawkiE73}.
\end{remark}
\begin{remark}
The topology theorem 1 is not specific to stationary black hole event horizons, for
it relies only on quasilocal properties. Indeed, the proof does not rely on
$\Hor$ being the boundary of a black hole region, if one agrees to define the
``interior'' region as that pointed towards by $\w{k}$. The only required property is
$\Hor$ being a hypersurface (not even a null one)
sliced by spacelike compact surfaces $\Sp$ with
$\theta_{(\wl)} = 0$ and $\Lie{k} \theta_{(\wl)} \leq 0$.
The theorem is therefore valid for \emph{outer trapping horizons} \cite{Haywa94}
and (tubes of) \emph{apparent horizons}, as noticed by Hawking himself \cite{Hawki73} (p.~34),
which are not necessarily null hypersurfaces and which exist in non-stationary spacetimes.
Both concepts of \emph{outer trapping horizon} and \emph{apparent horizon}
will be discussed in Chap.~\ref{s:loc}.
\end{remark}
Another version of the topology theorem relies on the null convergence
condition, which is weaker than the null dominance condition (hypothesis (ii) above),
and replaces hypothesis (iii) by other ones, which are more global. It also
fully excludes the 2-torus topology:
\begin{prop}[topology theorem 2 \textnormal{(Chru\'sciel\index[pers]{Chrusciel, P.T.@Chru\'sciel, P.T.} \& Wald\index[pers]{Wald, R.M.} 1994 \cite{ChrusW94b})}]
\label{p:sta:topology2}
Let $(\M, \w{g})$ be a 4-dimensional asymptotically flat stationary spacetime
containing a black hole of event horizon $\Hor$.
Let us assume that (i) the
null convergence condition\index{null!convergence condition}\index{convergence!condition!null --} (\ref{e:neh:null_energy_cond}) is fulfilled,
(ii) the domain of outer communications\index{domain of outer communications} $\langle\langle \M\rangle\rangle$
[Eq.~(\ref{e:glo:def_doc})]
is globally hyperbolic,
(iii) $\langle\langle \M\rangle\rangle$ contains an achronal asymptotically
flat hypersurface $\Sigma$ that intersects $\Hor$ in a compact cross-section $\Sp$,
and (iv) some technical condition is fulfilled (cf. \cite{ChrusW94b} for the details).
Then, $\langle\langle \M\rangle\rangle$ is simply connected and
any connected component of $\Sp$ is homeomorphic to the sphere $\SS^2$.
\end{prop}
A part $\mathscr{U}$ of spacetime is said
\defin{globally hyperbolic}\index{globally!hyperbolic}\label{d:sta:glob_hyperbol}
iff it admits a \defin{Cauchy surface}\index{Cauchy!surface}\label{d:sta:Cauchy_surface}, i.e. a spacelike
hypersurface $\Sigma$ such that every inextendible timelike curve of $\mathscr{U}$
intersects $\Sigma$ exactly once (cf. Sec.~8.3 of Wald's textbook~\cite{Wald84} for more
details). We shall not give the proof of Property~\ref{p:sta:topology2} here; it can of course
be found in the original article by Chru\'sciel \& Wald \cite{ChrusW94b}.
\begin{remark}
\label{r:sta:Majumdar_Papapetrou}
The topology theorems 1 and 2 regard only
a \emph{connected component} of a given horizon complete cross-section.
Generally, the latter is a connected 2-manifold, but there exist
4-dimensional stationary (actually static) spacetimes containing a black hole,
the event horizon of which has
disconnected complete cross-sections: the
\emph{Majumdar-Papapetrou spacetimes}\index{Majumdar-Papapetrou!black hole}
\cite{Majum47,Papap47,HartlH72}. They are solutions of the
electrovacuum Einstein equation (cf. Sec.~\ref{e:fra:electrovacuum})
representing an arbitrary number of charged black holes,
which form a static configuration thanks to
an exact balance between the gravitational
attraction and the electrostatic repulsion. The Majumdar-Papapetrou black holes will be discussed
further in Sec.~\ref{s:sta:uniqueness_static} [cf. Eq.~(\ref{e:sta:MajPap_n4})].
\end{remark}
A generalization of the topology theorem 1 to spacetimes of
dimension $n>4$ has been obtained in 2006 by Galloway \& Schoen \cite{GalloS06}
and R\'acz\index[pers]{Racz, I.@R\'acz, I.} provided a simplified
proof of it in 2008 \cite{Racz08}.
However, the theorem for $n>4$ only says that some invariant
of the smooth structure of $\Sp$, called the
\emph{Yamabe invariant}\index{Yamabe invariant}, must be positive for
$f \geq 0$ (\footnote{In brief, Eq.~(\ref{e:sta:Euler_char}) holds for $n>4$
as well, except that the integral of the Ricci scalar ${}^\Sp\!\!R$ is no longer
proportional to the Euler characteristic of $\Sp$ (no Gauss-Bonnet theorem for $\mathrm{dim}\, \Sp \neq 2$ !) but is related to the Yamabe constant of the conformal class of the metric $\w{q}$.}). This result implies that $\Sp$ must admit metrics of positive scalar curvature; it is
less stringent about the topology of $\Sp$ than the theorem for $n=4$.
Actually, the higher $n$, the less constraints on
the topology are provided by the Yamabe invariant. For instance,
for $n=5$, the cross-sections of the \emph{Myers-Perry black holes}\index{Myers-Perry black hole} \cite{MyersP86,EmparR08,Reall14}, which generalize
Kerr black holes to $n>4$, have the
topology of the sphere $\SS^3$, but the topology
$\SS^1\times\SS^2$ is allowed as well, as demonstrated
by the \emph{black ring}\index{black!ring} solution found by Emparan \& Reall \cite{EmparR02,EmparR08,Reall14} (see also Sec.~5.3 of Ref.~\cite{Chrus20}).
\begin{hist}
The topology theorem for the spacetime dimension $n=4$ has been formulated
first by Stephen Hawking\index[pers]{Hawking, S.W.} in 1992 \cite{Hawki72}
(see also p.~34 of Ref.~\cite{Hawki73} and Proposition~9.3.2 in Hawking \& Ellis' textbook
\cite{HawkiE73}).
However, in 1993, Gregory Galloway\index[pers]{Galloway, G.J.} \cite{Gallo94} (p.~119)
pointed out some limitation in Hawking's proof, namely that it cannot
exclude the torus topology ($\chi = 0$) for the horizon's cross-sections
without any extra hypothesis. For instance, for the proof given in Ref.~\cite{Hawki72},
one shall require that the spacetime is \emph{analytic} (cf. Remark~\ref{r:bas:analytic}
in Sec.~\ref{s:bas:def_manif}), which is a rather strong hypothesis.
The proof presented above is based on that given
by Sean Hayward\index[pers]{Hayward, S.} in 1994 \cite{Haywa94}
for outer trapping horizons (see also the proof of Theorem~6.3 in Ref.~\cite{Newma87}).
The theorem obtained by Piotr Chru\'sciel\index[pers]{Chrusciel, P.T.@Chru\'sciel, P.T.}
and Robert Wald\index[pers]{Wald, R.M.} in 1994 \cite{ChrusW94b}
(Property~\ref{p:sta:topology2})
relies on the \emph{topological censorship theorem}\index{topological!censorship theorem} established by
John Friedman\index[pers]{Friedman, J.L.}, Kristin Schleich\index[pers]{Schleich, K.}
and Donald Witt\index[pers]{Witt, D.M.} in 1993 \cite{FriedSW93}.
It leads directly to the spherical topology, excluding the toroidal one.
\end{hist}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Mass and angular momentum} \label{s:sta:mass_angul_mom}
For an asymptotically flat stationary spacetime, containing a
black hole or not, there is a well-defined concept of mass: the \emph{Komar mass},
which we introduce here (Secs.~\ref{s:sta:Komar_mass}-\ref{s:sta:Komar_ADM}).
For axisymmetric spacetimes, which are relevant
for stationary rotating black holes, there is in addition the concept
of \emph{Komar angular momentum}, which we shall introduce in Sec.~\ref{s:sta:Komar_angu_mom}.
\subsection{Mass and angular momentum of weakly relativistic stationary systems}
\label{s:sta:weakly_relativistic}
We shall call \defin{weakly relativistic}\index{weakly!relativistic}
a $n$-dimensional spacetime $(\M,\w{g})$
such that $\M$ is diffeomorphic to $\R^n$ and the metric tensor can be expressed as
\be \label{e:sta:weak_g_f_h}
\w{g} = \w{f} + \w{h} ,
\ee
where $\w{f}$ is a flat Lorentzian metric on $\M$ and $\w{h}$ is small in the following
sense: the components of $\w{h}$ obey $|h_{\alpha\beta}| \ll 1$
in any \defin{Minkowskian coordinates for}\index{Minkowskian coordinates} $\w{f}$, i.e. coordinates
$(x^\alpha)$ on $\M$ such that\footnote{We keep the notation $\eta$ for the matrix $\mathrm{diag}(-1,1,\ldots,1)$, so that
$(\eta_{\alpha\beta})$ stands for the components of $\w{f}$ in Minkowskian coordinates only.}
$(f_{\alpha\beta}) = \eta := \mathrm{diag}(-1,1,\ldots,1)$.
In addition, we assume that $\w{g}$
is ruled by the Einstein equation (\ref{e:fra:Einstein_eq}) with $\Lambda=0$
and an energy momentum tensor $\w{T}$ of compact support.
Given the flat metric $\w{f}$, the Minkowskian coordinates $(x^\alpha)$
are not unique: any Poincaré transformation\index{Poincaré!transformation}\label{p:sta:Poincare_transf}
${\tilde x}^\alpha = \Lambda^\alpha_{\ \, \mu} x^\mu + c^\alpha$, where
$(\Lambda^\alpha_{\ \, \beta})$ is a Lorentz matrix and the $c^\alpha$'s are
constant, leads to a coordinate system $({\tilde x}^\alpha)$ on $\M$ that is Minkowskian for $\w{f}$.
Furthermore, the flat metric $\w{f}$ itself, and hence the decomposition (\ref{e:sta:weak_g_f_h}), is
highly non-unique.
Indeed any change of coordinates of the form ${x'}^\alpha = x^\alpha + \zeta^\alpha(x^0,\ldots,x^{n-1})$
where $(x^\alpha)$ are Minkowskian coordinates for $\w{f}$ and $\zeta^\alpha$ are infinitesimal functions,
leads to the following components of the metric tensor with respect to $({x'}^\alpha)$:
\[
{g'}_{\alpha\beta} = g_{\mu\nu} \der{x^\mu}{{x'}^\alpha} \der{x^\nu}{{x'}^\beta}
= \left( \eta_{\mu\nu} + h_{\mu\nu}\right) \left( \delta^\mu_{\ \, \alpha} - \partial_\alpha \zeta^\mu \right)
\left( \delta^\nu_{\ \, \beta} - \partial_\beta \zeta^\nu \right) ,
\]
where we have used $x^\alpha = {x'}^\alpha - \zeta^\alpha$
and $\dert{\zeta^\alpha}{{x'}^\beta} = \dert{\zeta^\alpha}{x^\sigma} \times \dert{x^\sigma}{{x'}^\beta} \simeq
\dert{\zeta^\alpha}{x^\beta} =: \partial_\beta \zeta^\alpha$ to the first order in $\zeta^\alpha$.
Expanding to the first order in $\zeta^\alpha$ and $\w{h}$, we get
\[
{g'}_{\alpha\beta} = \eta_{\alpha\beta} + h_{\alpha\beta}
- \partial_\alpha \zeta_\beta - \partial_\beta \zeta_\alpha ,
\]
where $\zeta_\alpha := \eta_{\alpha\mu} \zeta^\mu$.
We may recast the above expression as
$\w{g} = \w{f'} + \w{h'}$,
where $\w{f'}$ is the metric whose components in the coordinates $({x'}^\alpha)$ are
$\eta_{\alpha\beta}$ (hence $\w{f'}$ is flat, as\footnote{Note that
the components of $\w{f}$ with respect to the coordinates $({x'}^\alpha)$ are not $\eta_{\alpha\beta}$
but $\eta_{\alpha\beta} - \partial_\alpha \zeta_\beta - \partial_\beta \zeta_\alpha$.} $\w{f}$)
and $\w{h'}$ has the following components with respect to the coordinates $({x'}^\alpha)$:
\be \label{e:sta:hprime_h_zeta}
{h'}_{\alpha\beta} = h_{\alpha\beta} - \partial_\alpha \zeta_\beta - \partial_\beta \zeta_\alpha .
\ee
The above relation can be viewed as expressing some \emph{gauge freedom}\index{gauge!freedom} on
$\w{h}$. This freedom actually reflects the freedom in the choice of the flat background metric $\w{f}$ in the
decomposition (\ref{e:sta:weak_g_f_h}).
\begin{prop}[Lorenz gauge for the metric perturbation]
The gauge freedom (\ref{e:sta:hprime_h_zeta}) can be used
to ensure that, in terms of $\w{f}$-Minkowskian coordinates $(x^\alpha)$,
the metric perturbation $\w{h}$ fulfills
\be \label{e:sta:Lorenz_h}
\encadre{ \partial_\mu \left( \eta^{\mu\nu} \bar{h}_{\alpha\nu} \right) = 0 },
\ee
where
\be \label{e:sta:def_h_bar}
\w{\bar{h}} := \w{h} - \frac{1}{2} h \, \w{f} ,
\ee
$h$ standing for the trace of $\w{h}$ with respect
to $\w{f}$: $h := \eta^{\mu\nu} h_{\mu\nu}$.
The choice (\ref{e:sta:Lorenz_h}) is referred to as the
\defin{Lorenz gauge}\index{Lorenz gauge} (sometimes
\defin{Hilbert gauge}\index{Hilbert gauge} \cite{Strau13}).
\end{prop}
\begin{proof}
From Eq.~(\ref{e:sta:hprime_h_zeta}), we get $h' := \eta^{\mu\nu} {h'}_{\mu\nu} = h - 2 \partial_\mu \zeta^\mu$,
so that ${\bar{h}}'_{\alpha\beta} = \bar{h}_{\alpha\beta}
- \partial_\alpha \zeta_\beta - \partial_\beta \zeta_\alpha + \partial_\mu \zeta^\mu \eta_{\alpha\beta}$.
It follows then that $\partial_\mu \left( \eta^{\mu\nu} {\bar{h}'}_{\alpha\nu} \right) = \partial_\mu \left( \eta^{\mu\nu} \bar{h}_{\alpha\nu} \right) - \Box_{\w{f}} \zeta_\alpha$,
where $\Box_{\w{f}} := \eta^{\mu\nu} \partial_\mu \partial_\nu$ is the d'Alembertian\index{d'Alembertian} operator with respect to $\w{f}$. Accordingly, if the Lorenz gauge (\ref{e:sta:Lorenz_h}) is not fulfilled,
it suffices to solve the d'Alembert equation
\be \label{e:sta:Box_zeta}
\Box_{\w{f}} \zeta_\alpha = \partial_\mu \left( \eta^{\mu\nu} \bar{h}_{\alpha\nu} \right)
\ee
and plug the solution $\zeta_\alpha$ into Eq.~(\ref{e:sta:hprime_h_zeta})
to get a metric pertubation $\w{h'}$ that obeys the Lorenz gauge.
\end{proof}
\begin{remark}
The Lorenz gauge (\ref{e:sta:Lorenz_h}) is equivalent
to the first-order expansion in $\w{h}$ of
the relation defining \defin{harmonic coordinates}\index{harmonic coordinates}
on $\M$, i.e.
\be
\Box_{\w{g}} x^\alpha = 0 \iff \partial_\mu \left( \sqrt{-g} \, g^{\mu\alpha} \right) = 0 .
\ee
\end{remark}
\begin{remark} \label{r:sta:freedom_Lorenz}
The Lorenz gauge does not fully specify the pair $(\w{f},\w{h})$. Indeed, the solutions of the
d'Alembert equation (\ref{e:sta:Box_zeta}) are non-unique: they depend on initial and boundary data.
\end{remark}
A standard computation (see e.g. Chap.~18 of \cite{MisneTW73} or Chap.~5 of \cite{Strau13},
noticing that the computation is independent of the spacetime dimension $n$)
shows that, at first order in $\w{h}$ and in the Lorenz gauge,
the Einstein tensor\index{Einstein!tensor!linearized --} of $\w{g}$ is
\be \label{e:sta:G_box_hb}
\w{G} = - \frac{1}{2} \Box_{\w{f}} \w{\bar{h}} ,
\ee
where $\Box_{\w{f}}$ stands for the d'Alembertian operator relative to the metric $\w{f}$:
in $\w{f}$-Minkowskian coordinates $(x^\alpha)$,
$\Box_{\w{f}} \bar{h}_{\alpha\beta} = \eta^{\mu\nu} \partial_\mu \partial_\nu \bar{h}_{\alpha\beta}$.
Let us now assume that $(\M,\w{g})$ is a stationary spacetime, with Killing vector
$\w{\xi}$. We may then choose $\w{f}$ so that the $\w{f}$-Minkowskian
coordinates $(x^\alpha)$ are adapted to stationarity, i.e.
$x^0$ is an ignorable coordinate, or equivalently
$\wpar_0=\w{\xi}$. Let then $\Sigma$ be a hypersurface $x^0 = \mathrm{const}$
and $\w{\gamma}$ the metric induced by $\w{f}$ on $\Sigma$.
The coordinates\footnote{Latin indices $i,j,k,\ldots$
range in $\{1,\ldots,n-1\}$, while Greek ones range in $\{0,\ldots,n-1\}$.}
$(x^i)_{1\leq i \leq n-1}$ form
a Cartesian coordinate system of $(\Sigma,\w{\gamma})$:
$\w{\gamma} = \delta_{ij} \, \dd x^i \otimes \dd x^j = (\dd x^1)^2 + \cdots + (\dd x^{n-1})^2$.
The operator $\Box_{\w{f}}$ reduces to the Laplace operator of $\w{\gamma}$,
$\Delta$ say, and, thanks to property (\ref{e:sta:G_box_hb}), the
Einstein equation (\ref{e:fra:Einstein_eq}) becomes a system of
$n(n+1)/2$ independent Poisson equations:
\be \label{e:sta:Delta_h_bar_T}
\Delta {\bar{h}}_{\alpha\beta} = - 16 \pi T_{\alpha\beta} .
\ee
The solutions are obtained via the Green function of the $(n-1)$-dimensional
Laplace operator:
\be \label{e:sta:hbar_Green_function}
{\bar{h}}_{\alpha\beta}(x) = \frac{16\pi}{(n-3)\Omega_{n-2}}
\int_\Sigma \frac{T_{\alpha\beta}({x'})}{|x - {x'}|^{n-3}} \, \D^{n-1} {x'},
\ee
where $x:=(x^1, \ldots, x^{n-1})$, ${x'} :=({x'}^1, \ldots, {x'}^{n-1})$,
$|x - {x'}|^2 := \sum_{i=1}^{n-1}(x^i - {x'}^i)^2$
and $\Omega_{n-2}$ is the area of the unit sphere $\SS^{n-2}$; the latter is given by the formula
\be \label{e:sta:area_p_sphere}
\Omega_{p} = \frac{2\pi^{(p+1)/2}}{\Gamma((p+1)/2)}, \qquad \mbox{with}\quad
\Gamma(u) := \int_0^{+\infty} t^{u-1} \mathrm{e}^{-t}\, \D t,
\ee
so that
\be \label{e:sta:area_p_sphere_examples}
\Omega_2 = 4\pi,\quad \Omega_3 = 2\pi^2,\quad \Omega_4 = \frac{8}{3} \pi^2, \quad
\Omega_5 = \pi^3,\ \ldots
\ee
\begin{prop}[asymptotic metric of a weakly relativistic system]
Let $(\M,\w{g})$ be a weakly relativistic stationary spacetime of dimension $n\geq 4$, with
$\w{g}$ obeying the Einstein equation (\ref{e:fra:Einstein_eq}) with $\Lambda=0$
and an energy momentum tensor $\w{T}$ of compact support (the ``source''),
where the energy-density dominates over the spatial stresses (weakly relativistic matter).
Within the Lorenz gauge, there exists
a coordinate system $(x^\alpha)$ such that the metric tensor has the following behavior when
$r:=\sqrt{(x^1)^2 + \cdots + (x^{n-1})^2} \to +\infty$:
\begin{subequations}
\label{e:sta:metric_weakly_relat}
\begin{align}
g_{00} = & - 1 + \frac{16\pi}{(n-2)\Omega_{n-2}} \frac{M}{r^{n-3}} + \bigO\left( \frac{1}{r^{n-1}} \right)
\label{e:sta:g00_weakly_relat} \\
g_{0i} = & \frac{8\pi}{\Omega_{n-2}} \frac{J_{ij} x^j}{r^{n-1}} + \bigO\left( \frac{1}{r^{n-1}} \right) \\
g_{ij} = & \left(1 + \frac{16\pi}{(n-2)(n-3)\Omega_{n-2}} \frac{M}{r^{n-3}} \right) \delta_{ij}
+ \bigO\left( \frac{1}{r^{n-2}} \right) ,
\end{align}
\end{subequations}
where
\be \label{e:sta:mass_weakly_relat}
M := \int_\Sigma T_{00}({x}) \, \D^{n-1} {x}
\ee
and
\be \label{e:sta:J_weakly_relat}
J_{ij} := \int_\Sigma \left( x^j T_{0i}(x) - x^i T_{0j}(x) \right) \, \D^{n-1} {x} ,
\ee
$\Sigma$ being any hypersurface $x^0 = \mathrm{const}$.
The quantities $M$ and $J_{ij}$ are independent of $\Sigma$ (i.e. of $x^0$)
and are called respectively the \defin{mass} and
the \defin{angular momentum} of the central source.
The coordinates $(x^\alpha)$ correspond to the \emph{central source rest-frame}
and take their origin at the \emph{center of mass}, in the sense
that
\be \label{e:sta:rest_frame_mass_center}
\int_\Sigma T_{0i}(x) \, \D^{n-1} {x} = 0
\qand
\int_\Sigma {x}^i T_{00}(x) \, \D^{n-1} {x} = 0
\ee
The dominance of the energy-density over
the spatial stresses is expressed in terms of the components of $\w{T}$ with respect to
the coordinates $(x^\alpha)$ as $T_{00} \gg |T_{ij}|$.
\end{prop}
\begin{proof}
Let $(\w{f},\w{h})$ obeys the Lorenz gauge and $(x^\alpha)$ be some corresponding
$\w{f}$-Minkowskian coordinates. By a Poincaré transformation (cf. p.~\pageref{p:sta:Poincare_transf}),
one can inforce (\ref{e:sta:rest_frame_mass_center}).
Far from the source, one may expand
the term $1/|x - {x'}|^{n-3}$ in Eq.~(\ref{e:sta:hbar_Green_function})
in powers of ${x'}^i/r$:
\[
\frac{1}{|x - {x'}|^{n-3}} = \frac{1}{r^{n-3}} \left( 1 + (n-3) \frac{x^j}{r} \frac{{x'}^j}{r}
+ \bigO\left( \frac{|x'|^2}{r^2} \right) \right) ,
\]
where Einstein's summation convention is assumed on the repeated index
$j\in\{1,\ldots,n-1\}$.
Accordingly, Eq.~(\ref{e:sta:hbar_Green_function}) yields
\be \label{e:sta:bar_h_expand}
{\bar{h}}_{\alpha\beta}(x) = \frac{16\pi}{\Omega_{n-2}}
\left( \frac{1}{(n-3)r^{n-3}} \int_\Sigma T_{\alpha\beta}({x'}) \, \D^{n-1} {x'}
+ \frac{x^j}{r^{n-1}} \int_\Sigma {x'}^j T_{\alpha\beta}({x'}) \, \D^{n-1} {x'}
\right) + \bigO\left( \frac{1}{r^{n-1}} \right) .
\ee
For the component $00$, the first integral in the right-hand side
is nothing but $M$,
as defined by (\ref{e:sta:mass_weakly_relat}), while the second integral
vanishes due to the second equation in~(\ref{e:sta:rest_frame_mass_center}).
Hence, we get
\be \label{e:sta:h_bar_00}
{\bar{h}}_{00}(x) = \frac{16\pi}{(n-3)\Omega_{n-2}} \frac{M}{r^{n-3}}
+ \bigO\left( \frac{1}{r^{n-1}} \right) .
\ee
Regarding the component $0i$ of Eq.~(\ref{e:sta:bar_h_expand}),
the first integral in the right-hand side vanishes due to the first
equation in~(\ref{e:sta:rest_frame_mass_center}). There remains then
\be \label{e:sta:h_bar_0i}
{\bar{h}}_{0i}(x) = \frac{16\pi}{\Omega_{n-2}} \frac{x^j}{r^{n-1}}
\int_\Sigma {x'}^j T_{0i}({x'}) \, \D^{n-1} {x'}
+ \bigO\left( \frac{1}{r^{n-1}} \right)
= \frac{8\pi}{\Omega_{n-2}} \frac{J_{ij} x^j}{r^{n-1}} + \bigO\left( \frac{1}{r^{n-1}} \right) ,
\ee
where the second equality follows from the identity
\be \label{e:sta:indent_J}
J_{ij} = 2 \int_\Sigma {x}^j T_{0i}({x}) \, \D^{n-1} {x} .
\ee
To prove it, consider
\[
\partial_k (x^i x^j T^{0k}) = \delta^i_{\ k} x^j T^{0k} + x^i \delta^j_{\ k} T^{0k}
+ x^i x^j \underbrace{\partial_k T^{0k}}_{0}
= x^j T^{0i} + x^i T^{0j} ,
\]
where $\partial_k T^{0k} = 0$ results from the equation of energy-momentum conservation
(\ref{e:fra:divT}) specialized to stationary spacetimes and expressed at 0th order in $\w{h}$.
Integrating over $\Sigma$ and invoking the Gauss-Ostrogradsky theorem to set the integral of
the divergence term $ \partial_k (x^i x^j T^{0k}) $ to zero (since $\w{T}$ is has compact support),
we get
\[
\int_\Sigma x^j T^{0i}(x) \, \D^{n-1} {x} + \int_\Sigma x^i T^{0j}(x) \, \D^{n-1} {x} = 0 .
\]
In view of the definition (\ref{e:sta:J_weakly_relat}) of $J_{ij}$, the
identity (\ref{e:sta:indent_J}) follows, since $T^{0i} = - T_{0i}$ at the 0th order in $\w{h}$.
Let us now reconstruct $\w{h}$ from $\w{\bar{h}}$. Taking the trace of Eq.~(\ref{e:sta:def_h_bar}) with respect to $\w{f}$
yields $\bar{h} := \eta^{\mu\nu} \bar{h}_{\mu\nu} = h - (h/2) \times n = (2 - n) h / 2$.
Hence we may invert Eq.~(\ref{e:sta:def_h_bar}) to
\be \label{e:sta:h_h_bar}
\w{h} = \w{\bar{h}} - \frac{\bar{h}}{n-2} \w{f}
\quad\mbox{with}\quad
\bar{h} = - \bar{h}_{00} + \bar{h}_{ii} .
\ee
The part\footnote{Note that Einstein's summation convention is used:
$\bar{h}_{ii} = \bar{h}_{11} + \cdots + \bar{h}_{n-1,n-1}$.}
$\bar{h}_{ii}$ of the trace $\bar{h}$ is deduced from
Eq.~(\ref{e:sta:bar_h_expand}):
\[
{\bar{h}}_{ii}(x) = \frac{16\pi}{\Omega_{n-2}}
\left( \frac{1}{(n-3)r^{n-3}} \int_\Sigma T_{ii}({x'}) \, \D^{n-1} {x'}
+ \frac{x^j}{r^{n-1}} \int_\Sigma {x'}^j T_{ii}({x'}) \, \D^{n-1} {x'}
\right) + \bigO\left( \frac{1}{r^{n-1}} \right) .
\]
The second integral vanishes identically, as it can be seen from the identity
\begin{align}
\partial_i \left( x^k x^j T^{ik} - \frac{1}{2} r^2 T^{ij} \right) & =
\delta^k_{\ i} x^j T^{ik} + x^k \delta^j_{\ i} T^{ik} +
x^k x^j \underbrace{\partial_i T^{ik}}_{0}
- r \underbrace{\partial_i r}_{x^i/r} T^{ij} - \frac{1}{2} r^2 \underbrace{\partial_i T^{ij}}_{0} \nonumber \\
& = x^j T^{ii} = x^j T_{ii} , \nonumber
\end{align}
where $\partial_i T^{ij}=0$ follows from the energy-momentum conservation law (\ref{e:fra:divT}). Hence the integral of $x^j T_{ii}$ over $\Sigma$ is that of the divergence of a vector
field that vanishes outside the source and the Gauss-Ostrogradsky theorem allows one to set it to zero.
We are thus left with
\be \label{e:sta:h_bar_ii}
{\bar{h}}_{ii}(x) = \frac{16\pi}{(n-3)\Omega_{n-2}}
\frac{1}{r^{n-3}} \int_\Sigma T_{ii}({x'}) \, \D^{n-1} {x'}
+ \bigO\left( \frac{1}{r^{n-1}} \right) .
\ee
Gathering Eqs.~(\ref{e:sta:h_bar_00}) and (\ref{e:sta:h_bar_ii}), we get
\[
\bar{h} = \frac{16\pi}{(n-3)\Omega_{n-2}}
\frac{1}{r^{n-3}} \left( - M + \int_\Sigma T_{ii}({x'}) \, \D^{n-1} {x'} \right)
+ \bigO\left( \frac{1}{r^{n-1}} \right) .
\]
However, given expression (\ref{e:sta:mass_weakly_relat}) for $M$, the weakly relativistic matter condition
$T_{00} \gg |T_{ij}|$ implies that the integral of $T_{ii}$ over $\Sigma$ is negligible in front of $M$. We conclude that
\[
\bar{h} = - \bar{h}_{00} + \bigO\left( \frac{1}{r^{n-1}} \right) .
\]
It follows then from Eq.~(\ref{e:sta:h_h_bar}) that, up to terms decaying at least as $1/r^{n-1}$,
\begin{align}
h_{00} & = \bar{h}_{00} + \frac{\bar{h}_{00}}{n-2} \times (-1) = \frac{n-3}{n-2}\, \bar{h}_{00} \nonumber \\
h_{0i} & = \bar{h}_{0i} \nonumber \\
h_{ij} & = \bar{h}_{ij} + \frac{\bar{h}_{00}}{n-2} \, \delta_{ij} =
\frac{\bar{h}_{00}}{n-2} \, \delta_{ij} + \bigO\left( \frac{1}{r^{n-2}} \right)
\nonumber .
\end{align}
The last equality holds because the $1/r^{n-3}$ term in $\bar{h}_{ij}$, which is proportional
to the integral of $T_{ij}$ over $\Sigma$ according to Eq.~(\ref{e:sta:bar_h_expand}),
is negligible in front of the $1/r^{n-3}$ term in $\bar{h}_{00}$, given
that $T_{00} \gg |T_{ij}|$. The above three equations, along with the values
(\ref{e:sta:h_bar_00}) and (\ref{e:sta:h_bar_0i}) for $\bar{h}_{00}$
and $\bar{h}_{0i}$, establish formulas (\ref{e:sta:metric_weakly_relat}) for
the components $g_{\alpha\beta} = \eta_{\alpha\beta} + h_{\alpha\beta}$
of the metric tensor.
\end{proof}
\subsection{Asymptotic expression of the metric in stationary asymptotically flat spacetimes}
\label{s:sta:asymptotic_metric}
Let us now consider a generic (not necessarily weakly relativistic) $n$-dimensional spacetime
$(\M,\w{g})$ that is stationary and asymptotically flat, with $\w{g}$
being ruled by the Einstein equation with $\Lambda=0$ and with $\w{T}$ vanishing in the
asymptotic region.
The decomposition $\w{g} = \w{f} + \w{h}$, with $\w{f}$ flat and $\w{h}$ ``small''
[Eq.~(\ref{e:sta:weak_g_f_h})] still holds in some neighborhood $\mathscr{U}$ of infinity.
Moreover, one can still choose the pair $(\w{f},\w{h})$ such that
the Lorenz gauge (\ref{e:sta:Lorenz_h}) is fulfilled and
the linearized Einstein equation in $\mathscr{U}$ reduces to $\Delta \w{\bar{h}} = 0$
[Eq.~(\ref{e:sta:Delta_h_bar_T}) with $\w{T} = 0$].
There exists then a $\w{f}$-Minkowskian coordinate system $(x^\alpha)$ on $\mathscr{U}$
such that the solution for $\w{g}$ is given
by Eq.~(\ref{e:sta:metric_weakly_relat}) to the lowest order in $1/r$,
the difference being that the constants
$M$ and $J_{ij}$ can no longer be expressed by integrals of the energy-momentum
tensor, as in Eqs.~(\ref{e:sta:mass_weakly_relat}) and (\ref{e:sta:J_weakly_relat}).
This can be shown rigorously by using the expansion in powers of $1/r$ of
the generic solution of $\Delta f = 0$ and the Lorenz
gauge (to set some terms to zero), cf. Exercice~19.3 of MTW \cite{MisneTW73}
or Sec.~5.7 of Straumann's textbook \cite{Strau13} for details.
For a configuration that is not assumed to be weakly relativistic, one
should not limit oneself to the linearized Einstein equation.
However, going beyong the first order expansion in $\w{h}$
does not spoil the first terms in the expansions (\ref{e:sta:metric_weakly_relat}) of
$\w{g}$'s components. This simply adds terms with a higher power
of $1/r$. For instance, for $n=4$, the second order expansion in $\w{h}$ introduces
the term $-2M^2/r^2$ in $g_{00}$ (see the above references for details, as well
as Refs.~\cite{Blanc14,PoissW14} for the nonlinear expansion within a post-Newtonian framework).
We may then assert:
\begin{prop}[asymptotic metric of a stationary asymptotically flat spacetime]
Let $(\M,\w{g})$ be a stationary asymptotically flat spacetime of dimension $n\geq 4$, with
$\w{g}$ obeying the Einstein equation (\ref{e:fra:Einstein_eq}) with $\Lambda=0$
and with the energy-momentum tensor $\w{T}$ vanishing in the
asymptotic region.
There exists a coordinate system $(x^\alpha)$ in the asymptotic region
such that the metric tensor has the following behavior when
$r:=\sqrt{(x^1)^2 + \cdots + (x^{n-1})^2} \to +\infty$:
\begin{subequations}
\label{e:sta:asymptotic_metric}
\begin{align}
g_{00} = & - 1 + \frac{16\pi}{(n-2)\Omega_{n-2}} \frac{M}{r^{n-3}} + \bigO\left( \frac{1}{r^{n-2}} \right)
\label{e:sta:asymptotic_g00} \\
g_{0i} = & \frac{8\pi}{\Omega_{n-2}} \frac{J_{ij} x^j}{r^{n-1}} + \bigO\left( \frac{1}{r^{n-1}} \right)
\label{e:sta:asymptotic_g0i} \\
g_{ij} = & \left(1 + \frac{16\pi}{(n-2)(n-3)\Omega_{n-2}} \frac{M}{r^{n-3}} \right) \delta_{ij}
+ \bigO\left( \frac{1}{r^{n-2}} \right) ,
\end{align}
\end{subequations}
where $\Omega_{n-2}$ is the area of the sphere $\SS^{n-2}$
[Eqs.~(\ref{e:sta:area_p_sphere})-(\ref{e:sta:area_p_sphere_examples})],
$M$ is a constant and $(J_{ij})$ is a constant antisymmetric $(n-1)\times(n-1)$
matrix. Furthermore, by a suitable $\mathrm{SO}(n-1)$ transformation of the coordinates
$(x^i)_{1\leq i\leq n-1}$, $(J_{ij})$ can be brought to the following block diagonal form,
depending only on $p:=[(n-1)/2]$ numbers $J_{(1)}$, $\ldots$, $J_{(p)}$ (possibly
equal to zero):
\be \label{s:sta:J_block_diagonal}
J_{ij} = \left(\begin{array}{ccccc}
0 & J_{(1)} & & & \\
-J_{(1)} & 0 & & & \\
& & 0 & J_{(2)} & \\
& & -J_{(2)} & 0 & \\
& & & & \ddots
\end{array}\right) ,
\ee
with the last raw and the last column
containing only zeros if $n$ is even.
\end{prop}
The last assertion follows from the fact that any real antisymmetric matrix $J$
is similar to a matrix $J'$ of the type (\ref{s:sta:J_block_diagonal})
via $J' = P J P^{-1}$, where $P$ is a special orthogonal matrix.
Note that the difference between (\ref{e:sta:asymptotic_metric})
and the asymptotic expansion (\ref{e:sta:metric_weakly_relat}) for
a weakly relativistic system is
$\bigO\left( {1}/{r^{n-2}} \right)$ in Eq.~(\ref{e:sta:asymptotic_g00})
versus $\bigO\left( {1}/{r^{n-1}} \right)$ in Eq.~(\ref{e:sta:g00_weakly_relat}).
As discussed above, this results from nonlinear terms in the expansion of the Einstein equation.
\subsection{Komar mass} \label{s:sta:Komar_mass}
In Newtonian gravity, the mass $M$ of an isolated body is defined
similarly to Eq.~(\ref{e:sta:mass_weakly_relat}), namely by the integral
of the mass density $\rho \sim T_{00}$ over the body. An alternative formula
for $M$ is provided by \defin{Gauss's law}\index{Gauss's law}: $M$ is $-1/(4\pi)$ times the flux of the gravitational field $\w{\mathfrak{g}} = - \wnab\Phi$
through any closed surface $\Sp$ surrounding the body, namely
\be \label{e:sta:Newtonian_mass}
M = \frac{1}{4\pi} \int_{\Sp} \vw{\nabla}{\Phi}\cdot \D \w{S} ,
\ee
where $\Phi$ is the Newtonian gravitational and
$\D \w{S}$ is the area element vector normal to $\Sp$. The above formula
is easily derived when $\Sp$ is a sphere of large radius, using spherical coordinates
$(r,\th,\ph)$. Indeed, for large $r$,
one has $\Phi \sim - M/r$, so that $\vw{\nabla}\Phi \sim (M/r^2) \, \wpar_r$. Given that
$\D \w{S} = r^2 \sin\th \, \D\th\, \D\ph\, \wpar_r$ and $\wpar_r\cdot\wpar_r = 1$,
formula~(\ref{e:sta:Newtonian_mass}) follows. This formula is
actually not restricted to a remote sphere, but is valid for any closed surface
$\Sp$ surrounding the body (this follows from the Gauss-Ostrogradsky theorem and Laplace's equation, $\Delta \Phi = 0$, which holds outside the central body).
Gauss's law (\ref{e:sta:Newtonian_mass}) reflects the ``gravitating aspect'' of the mass, while the
volume integral (\ref{e:sta:mass_weakly_relat}) identifies the mass with the ``amount of matter'' constituting
the central body. The latter definition would yield $M=0$ for any vacuum spacetime
(set $\w{T}=0$ in Eq.~(\ref{e:sta:mass_weakly_relat})) and therefore cannot be used to generalize
the concept of mass to strongly relativistic systems. In particular, one would certainly
demand $M > 0$ for the gravitating mass of Schwarzschild and Kerr black holes, which are pure vacuum solutions of the Einstein equation.
Accordingly, Gauss's law (\ref{e:sta:Newtonian_mass}), rather than the volume integral
(\ref{e:sta:mass_weakly_relat}), is the good basis for any attempt to
generalize the concept of mass to strongly relativistic spacetimes.
Taking a look at the asymptotic expression (\ref{e:sta:metric_weakly_relat}) of the metric tensor
of a weakly relativistic statrionary system, we notice that
the mass $M$ appears as the coefficient of the dominant $1/r^{n-3}$ term
in the expansion of $g_{00}$, so that one could
recover it by considering the flux integral of $\partial g_{00}/\partial r$ over
a $(n-2)$-dimensional surface at large $r$. In order to have a coordinate-invariant
definition, it is more appropriate to consider the 1-form $\uu{\xi}$ metric-dual
to the stationary Killing vector $\w{\xi}$. Indeed, in any coordinate system
adapted to stationarity, i.e. such that the components of
$\w{\xi}$ are $\xi^\alpha=(1,0,\ldots,0)$, the components of $\uu{\xi}$
are $\xi_\alpha = g_{0\alpha}$, so that in particular $\xi_0 = g_{00}$.
Instead of the partial derivative $\partial g_{00}/\partial r$, a natural coordinate-independent
quantity in then the exterior derivative\index{exterior!derivative}\index{derivative!exterior --} $\dd \uu{\xi}$ (cf. Sec.~\ref{s:bas:ext_deriv}). Since the concept of flux
is naturally conveyed by the integral of the Hodge dual (see e.g. Sec.~16.4.7 of Ref.~\cite{Gourg13}), one arrives at the
definition of mass given below [Eq.~(\ref{e:sta:def_Komar_mass})], named \emph{Komar mass}.
In addition to be coordinate-invariant, it shares the same property as the Newtonian mass (\ref{e:sta:Newtonian_mass}),
namely to be independent from the integration surface outside the central body,
at least within general relativity
(Property~\ref{p:sta:Komar_mass_invariant} below).
\begin{greybox}
Let $(\M, \w{g})$ be an asymptotically flat spacetime of dimension $n \geq 4$
that is stationary,
with stationary Killing vector $\w{\xi}$, normalized such that $\w{\xi}\cdot\w{\xi} \to -1$
near the asymptotic boundary [Eq.~(\ref{e:sta:xi_scri})].
Given a spacelike closed $(n-2)$-surface $\Sp\subset\M$,
the \defin{Komar mass over}\index{Komar!mass}\index{mass!Komar --} $\Sp$ is
defined by
\be \label{e:sta:def_Komar_mass}
\encadre{M_{\Sp} := - \frac{n-2}{16\pi(n-3)} \int_{\Sp} \star(\dd \uu{\xi}) },
\ee
where
(i) $\uu{\xi}$ is the 1-form associated to $\w{\xi}$
by metric duality (cf. Sec.~\ref{s:bas:metric_dual}), i.e. the 1-form
of components $\xi_\alpha = g_{\alpha\mu} \xi^\mu$, (ii) $\dd \uu{\xi}$ is
the exterior derivative of $\uu{\xi}$ (cf. Sec.~\ref{s:bas:ext_deriv}, especially Eqs.~(\ref{e:bas:def_ext_1f}) and (\ref{e:bas:def_ext_1f_nab})), namely the 2-form whose components
are
\be \label{e:sta:duxi_nab}
(\dd \uu{\xi})_{\alpha\beta} =
\partial_\alpha \xi_\beta - \partial_\beta \xi_\alpha =
\nabla_\alpha \xi_\beta - \nabla_\beta \xi_\alpha
= 2 \nabla_\alpha \xi_\beta ,
\ee
the last equality following from the Killing equation (\ref{e:neh:Killing_equation}),
and (iii) $\star(\dd \uu{\xi})$ is the $(n-2)$-form that is the
Hodge dual of the 2-form $\dd \uu{\xi}$. The
\defin{Hodge dual}\index{Hodge dual}\index{dual!Hodge --} of
any $p$-form $\w{A}$ is defined\footnote{See e.g.
Sec.~14.6 of Ref.~\cite{Strau13} or
Sec.~14.5 of Ref.~\cite{Gourg13} for an introduction to Hodge duality.}
as the $(n-p)$-form $\star\w{A}$ given by