/
glo.tex
1590 lines (1442 loc) · 77.2 KB
/
glo.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{The concept of black hole 3: The global view}
\label{s:glo}
\minitoc
\section{Introduction}
Having attempted in Chaps.~\ref{s:def} and \ref{s:neh} to characterize a black hole by the local
properties of its boundary, we turn now to the general definition of a black
hole. As it could have been anticipated from the naive ``definition'' given
in Sec.~\ref{s:def:first_defin}, the mathematically meaningful definition
of a black hole cannot be local: it has to take into account the full
spacetime structure, in particular its future asymptotics. Indeed, to conclude
firmly that a particle has escaped from a given region, one has to wait until the ``end
of time'' to make sure that the particle will never be back...
In this chapter, we therefore consider the global spacetime picture to
arrive at the general definition of a black hole in
Sec.~\ref{s:glo:BH}.
This amounts to focusing on the
spacetime asymptotics, which can be seen as
the region where the ``distant observers'' live and may, or may not, receive
light signal from some ``central region''. This far-away structure is best
described in terms of the so-called \emph{conformal completion}, which brings
the spacetime infinity(ies) to a finite distance in another manifold
--- a technique initiated by R.~Penrose\index[pers]{Penrose, R.} \cite{Penro63,Penro64} (see
Refs.~\cite{Fraue04} and \cite{Nicol17} for a review).
We start by investigating the conformal completion of the simplest
of all spacetimes: Minkowski spacetime.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Conformal completion of Minkowski spacetime} \label{s:glo:conf_Mink}
In this section $(\M,\w{g})$ is the 4-dimensional Minkowski spacetime\index{Minkowski!spacetime},
i.e. $\M$ is a smooth manifold diffeomorphic to $\mathbb{R}^4$ and $\w{g}$
is the metric tensor whose expression in terms of some global coordinates
$(t, x, y, z)$ implementing the diffeomorphism to $\mathbb{R}^4$
(i.e. \emph{Minkowskian coordinates}\index{Minkowskian coordinates})
is
\be \label{e:glo:Mink_metric}
\w{g} = - \dd t^2 + \dd x^2 + \dd y^2 + \dd z^2 .
\ee
\subsection{Finite-range coordinates on Minkowski spacetime} \label{s:glo:finite_range_Mink}
Since we would like to deal with the ``far'' region, it is natural to introduce
$r := \sqrt{x^2+y^2+z^2}$ and the associated spherical coordinates
$(t,r,\th,\ph)$, which are related to the Minkowskian ones by
\be \label{e:glo:spherical_coord}
\left\{ \begin{array}{l}
x = r\sin\th\cos\ph \\
y = r\sin\th\sin\ph \\
z = r\cos\th .
\end{array} \right.
\ee
The coordinates $(t, r,\th,\ph)$ span
$\mathbb{R}\times(0,+\infty)\times (0,\pi) \times (0,2\pi)$; they do not cover
the whole manifold $\M$ as a regular chart (cf. Sec.~\ref{s:bas:def_manif} of Appendix~\ref{s:bas}), but only $\M\setminus \Pi$, where $\Pi$ is the closed half hyperplane defined
by $y=0$ and $x\geq 0$. Once expressed in terms of the
spherical coordinates, the Minkowski metric (\ref{e:glo:Mink_metric}) takes the form
\be \label{e:glo:Mink_metric_spher}
\w{g} = - \dd t^2 + \dd r^2
+ r^2 \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right) .
\ee
\begin{figure}
\centerline{\includegraphics[width=0.5\textwidth]{glo_null_coord.pdf}}
\caption[]{\label{f:glo:null_coord} \footnotesize
Lines of constant null coordinates $u$ (solid) and $v$
(dashed) in terms of the coordinates $(t,r)$.}
\end{figure}
Let us introduce the null coordinate system $(u,v,\th,\ph)$ where $u$ and
$v$ are respectively the retarded\index{retarded!time} and advanced\index{advanced!time}
time defined by (cf. Fig.~\ref{f:glo:null_coord})
\be \label{e:glo:advanced_retarded}
\left\{ \begin{array}{l}
u = t - r\\
v = t + r
\end{array} \right.
\iff
\left\{ \begin{array}{l}
t = \frac{1}{2} (v+u)\\[1ex]
r = \frac{1}{2} (v-u) .
\end{array} \right.
\ee
One has then $\dd u \, \dd v = \dd t^2 - \dd r^2$ and the
metric tensor (\ref{e:glo:Mink_metric_spher}) takes the shape
\be \label{e:glo:Mink_metric_uv}
\w{g} = - \dd u \, \dd v
+ \frac{1}{4} (v-u)^2 \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right) .
\ee
The coordinates $(u,v)$ span the half part of $\mathbb{R}^2$ defined by
$u<v$. In order to have coordinates within a finite range, let us consider
their arctangents (cf. Fig.~\ref{f:glo:atan}):
\be \label{e:glo:UV_uv}
\left\{ \begin{array}{l}
U = \arctan u \\
V = \arctan v
\end{array} \right.
\iff
\left\{ \begin{array}{l}
u = \tan U \\
v = \tan V .
\end{array} \right.
\ee
Given that $\arctan$ is a monotonically increasing function (cf. Fig.~\ref{f:glo:atan}),
the coordinates $(U,V)$ span the half part of $(-\pi/2, \pi/2)\times (-\pi/2, \pi/2)$
defined by $U < V$:
\be \label{e:glo:span_UV}
-\frac{\pi}{2} < U < \frac{\pi}{2}, \quad
-\frac{\pi}{2} < V < \frac{\pi}{2}, \quad\mbox{and}\quad U < V.
\ee
\begin{figure}
\centerline{\includegraphics[width=0.8\textwidth]{glo_atan.pdf}}
\caption[]{\label{f:glo:atan} \footnotesize
The arctangent function mapping $\mathbb{R}$ to $(-\pi/2, \pi/2)$.}
\end{figure}
Since
\[
\D u = \frac{\D U}{\cos^2 U}, \quad \D v = \frac{\D V}{\cos^2 V}
\quad\mbox{and}\quad
\tan V - \tan U = \frac{\sin(V-U)}{\cos U \cos V},
\]
the Minkowski metric (\ref{e:glo:Mink_metric_uv})
is expressed in terms of the coordinates $(U,V,\th,\ph)$
as\footnote{See also the SageMath notebook~\ref{s:sam:conformal_Mink}.}
\be \label{e:glo:g_UV}
\w{g} = \frac{1}{4\cos^2 U \cos^2 V}
\left[ - 4 \, \dd U \, \dd V + \sin^2(V-U) \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right)
\right] .
\ee
\begin{remark}
The retarded/advanced times $u$ and $v$ have the dimension of a time, or of a
length in the $c=1$ units that we are using. Therefore, one should introduce
some length scale, $\ell_0$ say, before taking their arctangent and rewrite
(\ref{e:glo:UV_uv}) as
\[
\left\{ \begin{array}{l}
U = \arctan (u/\ell_0) \\
V = \arctan (v/\ell_0)
\end{array} \right.
\iff
\left\{ \begin{array}{l}
u = \ell_0 \tan U \\
v = \ell_0 \tan V .
\end{array} \right.
\]
The coordinates $(U,V)$ are dimensionless and a global factor $\ell_0^2$ should be
introduced in the right-hand side of Eq.~(\ref{e:glo:g_UV}).
However, the length scale $\ell_0$ plays no essential role,
so that, to keep simple notations,
it is omitted in what follows. In other words, we are using units for
which $\ell_0=1$.
\end{remark}
\subsection{Conformal metric} \label{s:glo:conf_metric}
In the right-hand side of (\ref{e:glo:g_UV}),
the terms in square brackets defines a metric
$\w{\tilde{g}}$ such that
\be \label{e:glo:tilde_g_Omega}
\encadre{ \w{\tilde{g}} = \Omega^2 \w{g} } ,
\ee
where $\Omega$ is the scalar field $\M \rightarrow \mathbb{R}$ obeying
\begin{subequations}
\begin{align}
\Omega & = 2 \cos U \cos V \label{e:glo:Omega_UV} \\
& = \frac{2}{\sqrt{u^2+1}\sqrt{v^2+1}} \label{e:glo:Omega_uv}\\
& = \frac{2}{\sqrt{(t-r)^2+1}\sqrt{(t+r)^2+1}} . \label{e:glo:Omega_tr}
\end{align}
\end{subequations}
We notice on (\ref{e:glo:Omega_uv}) and (\ref{e:glo:Omega_tr}) that the function
$\Omega$ never vanishes on $\M$, so that the bilinear form $\w{\tilde{g}}$ defined by
(\ref{e:glo:tilde_g_Omega}) constitutes a well-behaved metric on $\M$.
Moreover, since $\Omega^2 > 0$, $\w{\tilde{g}}$ has the same signature as
$\w{g}$, i.e. $(-,+,+,+)$.
The expression of $\w{\tilde{g}}$ is deduced from (\ref{e:glo:g_UV})
and (\ref{e:glo:Omega_UV}):
\be \label{e:glo:tg_UV}
\w{\tilde{g}} = - 4 \, \dd U \, \dd V
+ \sin^2(V-U) \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right) .
\ee
In view of (\ref{e:glo:tilde_g_Omega}), one says that the metric $\w{\tilde{g}}$
is \defin{conformal to}\index{conformal} the metric $\w{g}$, or equivalently,
that the metrics $\w{g}$ and $\w{\tilde{g}}$ are
\defin{conformally related}\index{conformally related metrics},
or that $\w{\tilde{g}}$ arises from $\w{g}$ via a
\defin{conformal transformation}\index{conformal!transformation}.
The scalar field $\Omega$ is called the \defin{conformal factor}\index{conformal!factor}.
A key property of a conformal transformation is to preserve orthogonality
relations, since (\ref{e:glo:tilde_g_Omega}) clearly
implies, at any point $p\in\M$,
\[
\forall (\w{u},\w{v})\in T_p\M\times T_p\M,\quad
\w{\tilde{g}}(\w{u},\w{v}) = 0 \iff \w{g}(\w{u},\w{v}) = 0 .
\]
In particular, null vectors for $\w{\tilde{g}}$ coincide with null vectors for $\w{g}$:
\[
\forall \wl \in T_p\M,\quad
\w{\tilde{g}}(\wl,\wl) = 0 \iff \w{g}(\wl,\wl) = 0 .
\]
Consequently the light cones of $(\M,\w{g})$ and $(\M,\w{\tilde{g}})$
are identical, which implies that $(\M,\w{g})$ and $(\M,\w{\tilde{g}})$
have the same causal structure.
Moreover, since $\Omega^2>0$, the spacelike and timelike characters of vectors
is preserved as well:
\be
\begin{array}{ll}
\forall \w{v} \in T_p\M,\ &
\w{v} \mbox{\ spacelike for\ } \w{\tilde{g}} \iff \w{v} \mbox{\ spacelike for\ } \w{g} \\
& \w{v} \mbox{\ timelike for\ } \w{\tilde{g}} \iff \w{v} \mbox{\ timelike for\ } \w{g} .
\end{array}
\ee
It follows that a curve $\Li$ is timelike (resp. null, spacelike) for $\w{\tilde{g}}$
iff $\Li$ is timelike (resp. null, spacelike) for $\w{g}$. Similarly,
a hypersurface $\Sigma$ is timelike (resp. null, spacelike) for $\w{\tilde{g}}$
iff $\Sigma$ is timelike (resp. null, spacelike) for $\w{g}$.
What about geodesics? Let us first recall that a null curve is not necessarily
a null geodesic (cf. Remark~\ref{r:def:null_curves} on p.~\pageref{r:def:null_curves}
and Appendix~\ref{s:geo}),
so that one cannot deduce from the above results that conformal transformations
preserve null geodesics. However, this turns out to be true:
\begin{prop}[null geodesics preserved by conformal transformations]
Let $\w{g}$ and $\w{\tilde{g}}$ be two Lorentzian metrics on a manifold
$\M$ that are conformally related: $\w{\tilde{g}} = \Omega^2 \w{g}$.
A smooth curve $\Li$ in $\M$ is a null geodesic for $\w{\tilde{g}}$ iff
$\Li$ is a null geodesic for $\w{g}$; any affine parameter $\tilde{\lambda}$
of $\Li$ as a $\w{\tilde{g}}$-geodesic is then related to any affine parameter
$\lambda$ of $\Li$ as a $\w{g}$-geodesic by
\be \label{e:glo:conf_transf_null_geod}
\derd{\tilde{\lambda}}{\lambda} = a\, \Omega^2 ,
\ee
where $a$ is a constant.
\end{prop}
\begin{proof}[Sketch of proof]
Write explicitly the geodesic equation [Eq.~(\ref{e:geo:eq_geod})]
and express the Christoffel symbols of $\w{\tilde{g}}$ in terms of those
of $\w{g}$ and the derivatives of $\Omega$ (see e.g. Appendix~D of Wald's
textbook \cite{Wald84} for details).
\end{proof}
On the contrary, conformal transformations preserve neither timelike
geodesics nor spacelike ones.
The coordinates $(U,V)$ are of null type; let us consider instead
the ``time+space'' coordinates $(\tau,\chi)$ defined by\footnote{Notice the
similarity with (\ref{e:glo:advanced_retarded}) up to some $1/2$ factors.}
\be \label{e:glo:tau_chi_U_V}
\left\{ \begin{array}{l}
\tau = V + U \\
\chi = V - U
\end{array} \right.
\iff
\left\{ \begin{array}{l}
U = \frac{1}{2} (\tau - \chi) \\[1ex]
V = \frac{1}{2} (\tau + \chi) .
\end{array} \right.
\ee
Given (\ref{e:glo:span_UV}), the range of these new coordinates is
\be \label{e:glo:range_tau_chi}
0 < \chi < \pi \quad\mbox{and}\quad
\chi - \pi < \tau < \pi - \chi .
\ee
In other words, if we draw the Minkowski spacetime in the $(\tau,\chi)$ plane,
it takes the shape of a half-diamond, as depicted in Fig.~\ref{f:glo:conf_diag_Mink}.
\begin{figure}
\centerline{\includegraphics[width=0.35\textwidth]{glo_conf_diag_Mink.pdf}}
\caption[]{\label{f:glo:conf_diag_Mink} \footnotesize
Conformal diagram of Minkowski spacetime. Constant-$r$ curves are drawn in
red, while constant-$t$ ones are drawn in grey.
\textsl{[Figure generated by the notebook \ref{s:sam:conformal_Mink}]}
}
\end{figure}
By combining (\ref{e:glo:advanced_retarded}) (\ref{e:glo:UV_uv}) and
(\ref{e:glo:tau_chi_U_V}), we get the link between $(t,r)$ and
$(\tau,\chi)$:
\be \label{e:glo:tau_chi_t_r}
\left\{ \begin{array}{l}
\tau = \arctan(t+r) + \arctan(t-r) \\
\chi = \arctan(t+r) - \arctan(t-r)
\end{array} \right.
\iff
\left\{ \begin{array}{l}
\displaystyle t = \frac{\sin\tau}{\cos\tau + \cos\chi}\\[2ex]
\displaystyle r = \frac{\sin\chi}{\cos\tau + \cos\chi} .
\end{array} \right.
\ee
We may use these relations to draw the lines $t=\mathrm{const}$ and
$r=\mathrm{const}$ in Fig.~\ref{f:glo:conf_diag_Mink}.
The expression of the conformal factor in the
coordinates $(\tau,\chi,\th,\ph)$ is easily deduced from
(\ref{e:glo:Omega_UV}) and
(\ref{e:glo:tau_chi_U_V}):
\be \label{e:glo:Omega_tau_chi}
\Omega = \cos\tau + \cos\chi .
\ee
\subsection{Conformal completion} \label{s:glo:conf_complet_Mink}
The expression of the conformal metric in terms of the coordinates
$(\tau,\chi,\th,\ph)$ is easily deduced from that in terms of
$(U,V,\th,\ph)$ as given by (\ref{e:glo:tg_UV}):
\be \label{e:glo:tg_Einstein}
\w{\tilde{g}} = - \dd\tau^2
+ \dd \chi^2
+ \sin^2\chi \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right) .
\ee
Restricting to a $\tau = \mathrm{const}$ hypersurface, i.e. setting $\dd\tau=0$,
we recognize the standard metric of the hypersphere
$\mathbb{S}^3$ in the hyperspherical coordinates $(\chi,\th,\ph)$.
Moreover, we notice that the full metric (\ref{e:glo:tg_Einstein})
is perfectly regular even if we relax
the condition on $\tau$ in (\ref{e:glo:range_tau_chi}), i.e. if we
let $\tau$ span the
entire $\mathbb{R}$. We may then consider the manifold
\be
\mathscr{E} = \mathbb{R}\times \mathbb{S}^3
\ee
and $\w{\tilde{g}}$ as the Lorentzian metric on $\mathscr{E}$ given by
(\ref{e:glo:tg_Einstein}).
The Lorentzian manifold
$(\mathscr{E},\w{\tilde{g}})$ is nothing but the
\defin{Einstein static universe}\index{Einstein!static universe}\index{static!universe (Einstein)}, also called the \defin{Einstein cylinder}\index{Einstein!cylinder}\index{cylinder!Einstein --},
a static solution of the Einstein equation (\ref{e:fra:Einstein_eq})
with $\Lambda > 0$ and some pressureless matter of uniform density
$\rho = \Lambda/(4\pi)$.
We have thus an embedding\footnote{Cf. Sec.~\ref{s:bas:embed} of Appendix~A} of Minkowski spacetime into the Einstein cylinder (cf. Fig.~\ref{f:glo:Einstcyl_Mink}):
\be \label{e:glo:embed_Mink_Einst}
\Phi:\ \M \longrightarrow \mathscr{E}
\ee
and this embedding is a conformal isometry from
$(\M,\w{g})$ to $(\Phi(\M),\w{\tilde{g}})$.
In the following, we shall identify $\Phi(\M)$ and $\M$, i.e. use the same
symbol $\M$ to denote the subset of $\mathscr{E}$ that is the image of $\M$ via the
embedding (\ref{e:glo:embed_Mink_Einst}).
\begin{figure}
\centerline{\includegraphics[width=0.6\textwidth]{glo_Einstcyl_Mink.pdf}}
\caption[]{\label{f:glo:Einstcyl_Mink}\footnotesize
Two views of the Einstein cylinder $\mathscr{E}$, with the conformal embedding of
Minkowski spacetime $\M$ in it. Each view represents a 2-dimensional cut of $\mathscr{E}$
at $\th=\pi/2$ and $\ph=0$ (one half of the plotted cylinder) or $\ph=\pi$ (the other half).
The $\mathbb{S}^3$ sections of $\mathscr{E}$ are then depicted as horizontal circles,
which are made of two half-circles corresponding to $\ph=0$ and $\ph=\pi$
respectively, with $\chi$ running from $0$ to $\pi$ on both on them.
The plotted piece of $\M$ is then the
2-dimensional cut $(y,z) = (0,0)$ of $\M$, which is a timelike plane spanned
by the coordinates $(t,x)$, with $x\geq 0$ on the $\ph=0$ half cylinder
and $x\leq 0$ on the $\ph=\pi$ half one.
The red curves are the same constant-$r$ curves
as in Fig.~\ref{f:glo:conf_diag_Mink}, while the black curves are
the same constant-$t$ curves as those drawn in grey in Fig.~\ref{f:glo:conf_diag_Mink}.
\textsl{[Figure generated by the notebook \ref{s:sam:conformal_Mink}]}
}
\end{figure}
Since $\mathscr{E}$ and $\M$ have the same dimension, $\M$ is an open subset of $\mathscr{E}$.
Its (topological) closure $\overline{\M}$ in $\mathscr{E}$ is (cf.
Figs.~\ref{f:glo:conf_diag_Mink} and \ref{f:glo:Einstcyl_Mink})
\be
\overline{\M} = \M \cup \scri^+ \cup \scri^- \cup \left\{ i^0 \right\} \cup
\left\{ i^+ \right\} \cup \left\{ i^- \right\} ,
\ee
where
\begin{itemize}
\item $\scri^+$ is the hypersurface of $\mathscr{E}$ defined by
$\tau = \pi - \chi$ and $0 < \tau < \pi$ $\iff$
$V=\pi/2$ and $-\pi/2< U < \pi/2$;
\item $\scri^-$ is the hypersurface of $\mathscr{E}$ defined by
$\tau = \chi - \pi $ and $-\pi < \tau < 0$ $\iff$
$U=-\pi/2$ and $-\pi/2< V < \pi/2$;
\item $i^0$ is the point of $\mathscr{E}$ defined by $\tau=0$ and $\chi=\pi$
$\iff$ $U=-\pi/2$ and $V=\pi/2$;
\item $i^+$ is the point of $\mathscr{E}$ defined by $\tau=\pi$ and $\chi=0$
$\iff$ $U=\pi/2$ and $V=\pi/2$;
\item $i^-$ is the point of $\mathscr{E}$ defined by $\tau=-\pi$ and $\chi=0$
$\iff$ $U=-\pi/2$ and $V=-\pi/2$.
\end{itemize}
It is customary to pronounce $\scri$ as ``scri''\index{scri}, for \emph{script i}.
Note that in the above definitions, we have extended the coordinates $(U,V)$
to $\E$ by the transformations (\ref{e:glo:tau_chi_U_V}); $(U,V,\th,\ph)$
can be then considered as a chart on $\E$ with $(U,V)$ spanning the
infinite strip $0<V-U<\pi$ of $\R^2$ and the metric $\w{\tilde{g}}$ on $\E$
being given by expression (\ref{e:glo:tg_UV}).
\begin{figure}
\centerline{\includegraphics[width=0.45\textwidth]{glo_Omega_Mink.jpg}}
\caption[]{\label{f:glo:Omega_Mink}\footnotesize
Conformal factor $\Omega$ as a function of $(\tau,\chi)$ [cf. Eq.~(\ref{e:glo:Omega_tau_chi})].
Only the part above the yellow horizontal plane ($\Omega=0$) is physical.
\textsl{[Figure generated by the notebook \ref{s:sam:conformal_Mink}]}
}
\end{figure}
\begin{remark}
It is precisely because $\Omega$ vanishes at the
topological boundary of $\M$ in $\E$,
\be
\overline{\M} \setminus \M = \scri^+ \cup \scri^- \cup \left\{ i^0 \right\} \cup
\left\{ i^+ \right\} \cup \left\{ i^- \right\} ,
\ee
that the conformal transformation (\ref{e:glo:tilde_g_Omega}) brings the infinity
of Minkowski spacetime to a finite distance (cf. Fig.~\ref{f:glo:Omega_Mink}).
\end{remark}
\begin{remark}
On $\mathbb{S}^3$, the hyperspherical coordinates $(\chi,\th,\ph)$
are singular at $\chi=0$ and $\chi=\pi$, so that setting $\chi=0$ (or $\chi=\pi$)
defines a unique point of $\mathbb{S}^3$, whatever the value of $(\th,\ph)$.
Note also that the vertical left boundary of the diamond drawn in
Fig.~\ref{f:glo:conf_diag_Mink}, i.e. the segment defined by
$\tau\in(-\pi,\pi)$ and $\chi=0$, is \emph{not} a part of the boundary
of $\M$ but merely reflect the coordinate singularity at $\chi=0$, in the same
way that the left vertical boundary of Fig.~\ref{f:glo:null_coord}
is not a boundary of Minkowski spacetime but is
due to the coordinate singularity at $r=0$. Note by the way that
$\chi=0$ implies $r=0$ via (\ref{e:glo:tau_chi_t_r}).
\end{remark}
Let
\be
\scri := \scri^+ \cup \scri^-
\ee
and
\be \label{e:glo:def_tM_Mink}
\tilde{\M} := \M \cup \scri .
\ee
$\tilde{\M}$ is naturally a smooth manifold with
boundary\footnote{Cf. Sec.~\ref{s:bas:manif_boundary} for the precise definition.}\index{manifold!with boundary}
and its boundary is $\scri$:
\be
\partial \tilde{\M} = \scri.
\ee
\begin{remark}
Because the closure $\overline{\M}$ is self-intersecting at the point $i^0$
(cf. Fig.~\ref{f:glo:Einstcyl_Mink}), it is not a manifold with boundary: no open neighborhood of
$i^0$ is homeomorphic to a neighborhood of
$\mathbb{H}^4 = \mathbb{R}^3\times {[0,+\infty)}$,
as the definition of a manifold with boundary would
require, cf. Sec.~\ref{s:bas:manif_boundary}.
At the points $i^+$ and $i^-$, $\overline{\M}$ can be considered as a
topological manifold with boundary, but not as a \emph{smooth} manifold with boundary.
Hence, the three points $i^0$, $i^+$ and $i^-$ are excluded from the definition
of the manifold with boundary $\tilde{\M}$.
\end{remark}
\begin{figure}
\centerline{\includegraphics[width=0.35\textwidth]{glo_conf_Mink_null.pdf}}
\caption[]{\label{f:glo:conf_Mink_null}\footnotesize
Null radial geodesics in the conformal diagram of Minkowski spacetime.
The solid green lines are null geodesics $u=\mathrm{const}$ for
17 values of $u$ uniformly spanning $[-8,8]$, while the dashed green lines are
null geodesics $v=\mathrm{const}$ for 17 values of $v$ uniformly spanning $[-8,8]$.
\textsl{[Figure generated by the notebook \ref{s:sam:conformal_Mink}]}
}
\end{figure}
The hypersurface $\scri^+$ is the location of $\tilde{\M}$ where all radial null geodesics
terminate, while $\scri^-$ is the location of $\tilde{\M}$ where all these geodesics originate (cf. Fig.~\ref{f:glo:conf_Mink_null}). For this
reason $\scri^+$ is called the
\defin{future null infinity}\index{future!null infinity}\index{null!infinity}\index{infinity!future null --}
of $(\M,\w{g})$
and $\scri^-$ the \defin{past null infinity}\index{past!null infinity}\index{infinity!past null --}
of $(\M,\w{g})$.
On the other side, any timelike geodesic of $(\M,\w{g})$ originates at $i^-$ and ends at
$i^+$ (cf. Fig.~\ref{f:glo:conf_diag_Mink}), while any spacelike geodesic
of $(\M,\w{g})$ originates at $i^0$ and terminates there
(after having completed a closed path on $\mathbb{S}^3$, cf. Fig.~\ref{f:glo:Einstcyl_Mink}).
The point $i^+$ is then called the
\defin{future timelike infinity}\index{future!timelike infinity}\index{timelike!infinity}\index{infinity!future timelike --}
of $(\M,\w{g})$,
$i^-$ the \defin{past timelike infinity}\index{past!timelike infinity}\index{infinity!past timelike --}
of $(\M,\w{g})$
and $i^0$ the \defin{spacelike infinity}\index{spacelike!infinity}\index{infinity!spacelike --} of $(\M,\w{g})$.
\begin{prop}
\label{p:glo:Mink_scri_null}
$\scri^+$ and $\scri^-$ are null hypersurfaces of $(\tilde{\M},\w{\tilde{g}})$.
\end{prop}
\begin{proof}
Since $\scri^+$ is defined by $V=\pi/2$, a valid coordinate
system on $\scri^+$ is $(U,\th,\ph)$ with $U$ spanning $(-\pi/2, \pi/2)$.
The metric induced by $\w{\tilde{g}}$ on $\scri^+$ is easily obtained by
setting $V=\pi/2$ in Eq.~(\ref{e:glo:tg_UV}):
\be
\left. \w{\tilde{g}} \right| _{\scri^+} =
\cos^2 U \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right) .
\ee
It appears clearly that the signature of this metric is $(0,+,+)$, i.e. it
is degenerate; hence $\scri^+$ is a null hypersurface of
$(\tilde{\M},\w{\tilde{g}})$. Similarly, $\scri^-$ being defined by
$U=-\pi/2$, a valid coordinate
system on $\scri^-$ is $(V,\th,\ph)$ with $V$ spanning $(-\pi/2, \pi/2)$
and the metric induced by $\w{\tilde{g}}$ on $\scri^-$ is obtained by
setting $U=-\pi/2$ in Eq.~(\ref{e:glo:tg_UV}):
\be
\left. \w{\tilde{g}} \right| _{\scri^-} =
\cos^2 V \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right) .
\ee
Again, it is clearly degenerate, so that $\scri^-$ is null hypersurface of
$(\tilde{\M},\w{\tilde{g}})$.
\end{proof}
The null character of $\scri^+$ and $\scri^-$ appears also clearly in
the conformal diagrams of Figs.~\ref{f:glo:conf_diag_Mink}
and \ref{f:glo:conf_Mink_null}, since
$\scri^+$ and $\scri^-$ are straight lines of slope $\pm 1$ in these diagrams.
\begin{hist}
The idea of using a conformal transformation to treat infinity as a boundary
``at a finite distance'' has been put forward by Roger Penrose\index[pers]{Penrose, R.}
in 1963 \cite{Penro63} and expanded in 1964 in the seminal paper \cite{Penro64},
where Penrose constructed the
conformal completion of Minkowski spacetime as a part of the Einstein cylinder.
In particular, Fig.~3 of Ref.~\cite{Penro64} is equivalent to Fig.~\ref{f:glo:Einstcyl_Mink}.
\end{hist}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Conformal completions and asymptotic flatness} \label{s:glo:conf_compl}
Having investigated the asymptotic structure of Minkowski spacetime
via a conformal completion, let us use the latter to define spacetimes
that ``look like'' Minkowski spacetime asymptotically.
A first step is the concept of conformal completion.
\subsection{Conformal completion} \label{s:glo:def_conf_compl}
\begin{greybox}
A spacetime $(\M,\w{g})$ admits a
\defin{conformal completion at infinity}\index{conformal!completion!at infinity}\index{completion!conformal --}
iff there exists a Lorentzian manifold with boundary
$(\tilde{\M},\w{\tilde{g}})$ (cf. Sec.~\ref{s:bas:manif_boundary} for the definition)
equipped with a smooth non-negative scalar field
$\Omega: \tilde{\M} \rightarrow \mathbb{R}^+$
such that
\begin{enumerate}
\item $\tilde{\M} = \M \cup \scri$, with $\scri := \partial \tilde{\M}$
(the manifold boundary\footnote{As stressed in Remark~\ref{r:bas:manifold_boundary}
in Sec.~\ref{s:bas:manif_boundary}, the set $\partial \tilde{\M}$ is
the boundary of $\tilde{\M}$ as a manifold with boundary; it is not the boundary of
$\tilde{\M}$ as a topological space, the latter being $\varnothing$.} of $\tilde{\M})$;
\item on $\M$, $\w{\tilde{g}} = \Omega^2 \w{g}$;
\item on $\scri$, $\Omega=0$;
\item on $\scri$, $\dd \Omega \not= 0$.
\end{enumerate}
$\scri$ is called the \defin{conformal boundary}\index{conformal!boundary}\index{boundary!conformal --}
of $(\M,\w{g})$ within
the conformal completion $(\tilde{\M},\w{\tilde{g}})$.
\end{greybox}
Condition~1 expresses that $\M$ has been endowed with some boundary.
A rigorous formulation of it would be via an embedding $\Phi:\M \rightarrow \tilde{\M}$,
as in Eq.~(\ref{e:glo:embed_Mink_Einst}), so that
$\tilde{\M} = \Phi(\M) \cup \scri$. However, as above, we identify $\Phi(\M)$
with $\M$ and therefore simply write $\tilde{\M} = \M \cup \scri$.
Conditions~2 and 3 express that the boundary of $\M$, which ``lies at an infinite
distance'' with respect to $\w{g}$, has been brought to a
finite distance with respect to $\w{\tilde{g}}$. Indeed, in terms of
length elements [cf. Eq.~(\ref{e:fra:line_element})], condition~2 implies
\[
\D s^2 = \frac{1}{\Omega^2} \, \D {\tilde s}^2
\]
with $1/\Omega^2 \rightarrow +\infty$ as one approaches $\scri$
(condition~3).
Finally, condition~4 ensures
that $\scri$ is a regular hypersurface of $\tilde{\M}$.
It is of course fulfilled by Minkowski spacetime, as we can check graphically
on Fig.~\ref{f:glo:Omega_Mink}: the graph of $\Omega$ has no horizontal slope
at $\scri$.
\begin{remark}
The statement that $(\tilde{\M},\w{\tilde{g}})$ is a Lorentzian manifold with
boundary implies that $\w{\tilde{g}}$ is smooth everywhere on $\tilde{\M}$,
including at the boundary $\scri$.
\end{remark}
\begin{remark}
Since $\w{\tilde{g}}$ is a metric, it is by definition non-degenerate and
condition~2 implies that $\Omega$ cannot vanish on $\M$. Being non-negative,
we have necessarily $\Omega > 0$ on $\M$.
\end{remark}
\begin{remark}
The conformal boundary $\scri$ is not part of the physical spacetime
$\M$, but only of the conformal completion $\tilde{\M}$.
\end{remark}
\begin{remark}
One often speaks about
\emph{conformal compactification}\index{conformal!compactification}\index{compactification}
instead of \emph{conformal completion}, but in general $\tilde{\M}$ is not a
compact manifold. For instance,
the completion $\tilde{\M}$ of Minkowski spacetime defined by Eq.~(\ref{e:glo:def_tM_Mink})
is not compact, for the points $i^+$, $i^-$ and $i^0$ have been omitted in the construction of
$\tilde{\M}$.
\end{remark}
\begin{figure}
\centerline{\includegraphics[height=0.4\textheight]{glo_AdS_completion.pdf}}
\caption[]{\label{f:glo:AdS_completion} \footnotesize
Conformal completion of AdS$_{4}$ spacetime, depicted on the Einstein cylinder.
The conformal boundary $\scri$ is shown in yellow, red lines are lines
$\chi=\mathrm{const}$ (uniformly sampled in terms of $\tan\chi = \sinh\rho$),
green curves are radial null geodesics and the purple curve
is a radial timelike geodesic, bouncing back and forth around $\chi=0$.
\textsl{[Figure generated by the notebook \ref{s:sam:AdS}]}
}
\end{figure}
\begin{example}[conformal completion of AdS$_{4}$ spacetime] \label{x:glo:AdS}
The 4-dimensional anti-de Sitter spacetime\index{anti-de Sitter spacetime}
$(\M,\w{g})$
has been introduced in Example~\ref{x:neh:AdS} of Chap.~\ref{s:neh}.
The metric tensor expressed in the conformal coordinates
$(\tau,\chi,\th,\ph)$ is given by Eq.~(\ref{e:neh:metrix_AdS_conformal}):
\be
\w{g} = \frac{\ell^2}{\cos^2\chi} \left[ - \dd \tau^2
+ \dd \chi^2 + \sin^2\chi \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right) \right] ,
\ee
with $\tau\in\R$, $\chi \in (0,\pi/2)$, $\th\in(0,\pi)$ and $\ph\in(0,2\pi)$.
Defining $\Omega := \ell^{-1}\cos\chi$, we notice that
a conformal completion of $(\M,\w{g})$ is $(\tilde{\M},\w{\tilde{g}})$
where (i) $\tilde{\M}$ is the part $\chi \leq \pi/2$ of the Einstein cylinder\footnote{Recall that on the
Einstein cylinder the range of $\chi$ is $(0,\pi)$, cf. Eq.~(\ref{e:glo:range_tau_chi}).}
introduced in Sec.~\ref{s:glo:conf_complet_Mink}
and (ii) $\w{\tilde{g}}$ is the metric (\ref{e:glo:tg_Einstein}).
The boundary $\scri = \partial\tilde{\M}$ is then the hypersurface $\chi=\pi/2$
of the Einstein cylinder (cf. Fig.~\ref{f:glo:AdS_completion});
$\scri$ is spanned by the coordinates $(\tau,\th,\ph)$
and its topology is that of a 3-dimensional cylinder: $\scri \simeq \mathbb{R}\times\mathbb{S}^2$.
We notice that conditions 3 and 4 of the definition of a conformal completion
are satisfied: $\Omega = \ell^{-1} \cos\chi = 0$ at $\scri$ and
$\dd\Omega = - \ell^{-1} \sin\chi\, \dd\chi = -\ell^{-1} \dd\chi \not = 0 $
at $\scri$.
The metric induced by $\w{\tilde{g}}$ on $\scri$ is obtained by
setting $\chi=\pi/2$ in (\ref{e:glo:tg_Einstein}):
$- \dd \tau^2 + \dd\th^2 + \sin^2\th \, \dd\ph^2$.
This 3-metric is clearly Lorentzian, which shows that $\scri$ is a timelike
hypersurface of $(\tilde{\M},\w{\tilde{g}})$.
\end{example}
The above example shows that $\scri$ is not necessarily a null hypersurface,
as it is for Minkowski spacetime (cf. Sec.~\ref{s:glo:conf_complet_Mink}).
Actually the causal type of $\scri$ is determined by the cosmological
constant, as follows:
\begin{prop}[causal type of $\scri$ and sign of the cosmological constant]
\label{p:glo:type_scri_sign_Lambda}
If the spacetime dimension obeys\footnote{Cf. Remark~\ref{r:fra:Einstein_eq_n_2} in
Sec.~\ref{s:fra:Einstein_eq}.}
$n\geq 3$ and $\w{g}$ is a solution of Einstein equation with
a cosmological constant $\Lambda$ [Eq.~(\ref{e:fra:Einstein_eq})]
and the trace $T$ of the energy-momentum tensor tends to zero in the
vicinity of $\scri$ (i.e. when $\Omega\rightarrow 0$), then
\begin{itemize}
\item $\scri$ is a null hypersurface of $(\tilde{\M},\w{\tilde{g}})$ iff $\Lambda=0$;
\item $\scri$ is a spacelike hypersurface of $(\tilde{\M},\w{\tilde{g}})$ iff $\Lambda>0$;
\item $\scri$ is a timelike hypersurface of $(\tilde{\M},\w{\tilde{g}})$ iff $\Lambda<0$.
\end{itemize}
\end{prop}
\begin{proof}
It follows from $\w{\tilde{g}} = \Omega^2 \w{g}$ that the Ricci scalars $\tilde{R}$
and $R$ of respectively $\w{\tilde{g}}$ and $\w{g}$ are related by\footnote{This relation is
easily established by starting from Eq.~(2.30) of Hawking \& Ellis' textbook~\cite{HawkiE73} or Eq.~(2.19) on p.~645 of Choquet-Bruhat's one~\cite{Choqu09} and inverting the roles of $\w{\tilde{g}}$ and $\w{g}$, thereby substituting
$\Omega^{-1}$ for $\Omega$.}
\be \label{e:glo:tildeR_R}
\Omega^2 \tilde{R} = R - (n-1) \left( 2 \Omega \, \tilde{g}^{\mu\nu} \tilde{\nabla}_\mu
\tilde{\nabla}_\nu \Omega - n \, \tilde{g}^{\mu\nu} \partial_\mu \Omega \partial_\nu \Omega
\right) ,
\ee
where $n = \mathrm{dim}\, \M$ and $\tilde{\nabla}$ stands for the Levi-Civita connection of
$\w{\tilde{g}}$. Using the trace of the Einstein equation (\ref{e:fra:Einstein_eq_n}) to
express $R$, we get
\[
\Omega^2 \tilde{R} = \frac{2}{n-2}\left( n \Lambda - 8\pi T \right)
- (n-1) \left( 2 \Omega \, \tilde{g}^{\mu\nu} \tilde{\nabla}_\mu
\tilde{\nabla}_\nu \Omega - n \, \tilde{g}^{\mu\nu} \partial_\mu \Omega \partial_\nu \Omega
\right)
\]
This equation is a priori valid in $\M = \tilde{\M}\setminus\scri$ only.
Taking the limit $\Omega\rightarrow 0$ and
assuming that $T\rightarrow 0$ in that limit, we get, by continuity, an identity
on $\scri$:
\be \label{e:glo:normal_scri_square}
\tilde{g}^{\mu\nu} \partial_\mu \Omega \partial_\nu \Omega \stackrel{\scri}{=}
- \frac{2}{(n-1)(n-2)} \Lambda .
\ee
Since $\scri$ corresponds to a constant value of the scalar field $\Omega$ ($\Omega=0$),
the left-hand side of this equation is nothing but the scalar square
$\w{\tilde{g}}(\w{n},\w{n})$ of the vector $\w{n}$ normal to $\scri$
defined as the dual with respect to $\w{\tilde{g}}$ of the 1-form
$\dd\Omega$: $n^\alpha = \tilde{g}^{\alpha\mu} \partial_\mu \Omega$
(remember that by hypothesis 4 in the definition
of a conformal completion, $\dd\Omega$ is non-vanishing on $\scri$, so that
$\w{n}$ is a valid normal
vector to $\scri$). Equation~(\ref{e:glo:normal_scri_square}) implies
that the sign of $\w{\tilde{g}}(\w{n},\w{n})$ is the opposite of that of $\Lambda$.
Given the link between the causal type of a hypersurface and the causal type of its normal
(cf. Sec.~\ref{s:def:hor_as_null}), this completes the proof.
\end{proof}
One may distinguish two subparts of the conformal boundary:
\begin{greybox}
Let $(\M,\w{g})$ be a time-orientable\footnote{Cf. Sec.~\ref{s:fra:time_orientation}.} spacetime admitting a conformal completion at infinity $(\tilde{\M},\w{\tilde{g}})$, with conformal boundary $\scri$.
One defines the \defin{future infinity}\index{future!infinity} of $(\M,\w{g})$ as the subset $\scri^+$ of $\scri$ whose points can be reached from a point in $\M$ by a future-directed causal curve in $\tilde{\M}$.
Similarly the \defin{past infinity}\index{past!infinity} of $(\M,\w{g})$ is the subset $\scri^-$ of
$\scri$ whose points can be reached from a point in $\M$ by a past-directed causal curve in $\tilde{\M}$.
If $\scri^+$ (resp. $\scri^-$) is a null hypersurface, it is called the
\defin{future null infinity}\index{future!null infinity}\index{null!future -- infinity}
(resp. \defin{past null infinity}\index{past!null infinity}\index{null!past -- infinity})
of $(\M,\w{g})$.
Furthermore if $\scri = \scri^+ \cup \scri^-$, one says that $(\tilde{\M},\w{\tilde{g}})$ is a
\defin{conformal completion at null infinity}\index{conformal!completion!at null infinity}
of $(\M,\w{g})$.
\end{greybox}
\begin{remark}
The above definitions of $\scri^+$ and $\scri^-$
generalize those given for the Minkowski spacetime in Sec.~\ref{s:glo:conf_complet_Mink}.
Note that, contrary to the Minkowski case, $\scri^+$ and $\scri^-$ are not null
for spacetimes with a non-zero cosmological constant (cf. Property~\ref{p:glo:type_scri_sign_Lambda}).
In particular, the following
examples exhibit respectively timelike and spacelike $\scri^+$ and $\scri^-$.
\end{remark}
\begin{example}[Future and past infinities of AdS$_{4}$ spacetime]
Let us consider the conformal completion of AdS$_{4}$ discussed in Example~\ref{x:glo:AdS}.
It is evident from behavior of radial null geodesics (the green curves plotted in Fig.~\ref{f:glo:AdS_completion}) that any point of $\scri$ can be connected to $\M$ by a
future-directed null geodesic as well as by a past-directed one. It follows that
$\scri^+ = \scri^- = \scri$ and both $\scri^+$ and $\scri^-$ are timelike hypersurfaces.
\end{example}
\begin{example}[Conformal completion of dS$_{4}$ spacetime]
\label{x:glo:dS4}
The 4-dimensional \defin{de Sitter spacetime}\index{de Sitter spacetime} is
$(\M,\w{g})$ with $\M\simeq \mathbb{R}\times\mathbb{S}^3$ and $\w{g}$ is the metric
whose expression in the so-called \emph{global coordinates}
$(t,\chi,\th,\ph)$ is
\be \label{e:glo:dS4}
\w{g} = \ell^2 \left[ - \dd t^2
+ \cosh^2 t \left(
\dd \chi^2 + \sin^2\chi \left( \dd\th^2 + \sin^2\th \, \dd \ph^2 \right) \right) \right] ,
\ee
where $\ell$ is a positive constant. Note that $t$ spans $\mathbb{R}$
while $(\chi,\th,\ph)$ are standard polar coordinates on $\mathbb{S}^3$:
$\chi\in(0,\pi)$, $\th\in(0,\pi)$ and $\ph\in(0,2\pi)$.
The metric (\ref{e:glo:dS4}) is a solution
of the vacuum Einstein equation\index{Einstein!equation!vacuum --}\index{vacuum!Einstein equation} (\ref{e:fra:vac_Einstein_Lambda}) with
the positive cosmological constant $\Lambda = 3/\ell^2$.
Using coordinates $(\tau,\chi,\th,\ph)$ with
$\tau := 2\arctan(\tanh(t/2)) \in (-\pi/2,\pi/2)$, one gets
\be
\w{g} = \frac{\ell^2}{\cos^2\tau} \left[ - \dd \tau^2
+ \dd \chi^2 + \sin^2\chi \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right) \right] .
\ee
Defining $\Omega := \ell^{-1}\cos\tau = (\ell\cosh t)^{-1}$, we notice that
a conformal completion of $(\M,\w{g})$ is $(\tilde{\M},\w{\tilde{g}})$
where (i) $\tilde{\M}$ is the part $-\pi/2\leq \tau \leq \pi/2$ of the Einstein cylinder
introduced in Sec.~\ref{s:glo:conf_complet_Mink}
and (ii) $\w{\tilde{g}}$ is the metric (\ref{e:glo:tg_Einstein}).
The boundary $\scri = \partial\tilde{\M}$ has two connected components:
$\scri^+$, which is the hypersurface $\tau = \pi/2$ of $\tilde{\M}$, and
$\scri^-$, which is the hypersurface $\tau = -\pi/2$.
Both $\scri^+$ and $\scri^-$ are spanned by the coordinates $(\chi,\th,\ph)$
and their topology is that of $\mathbb{S}^3$.
We notice that conditions 3 and 4 of the definition of a conformal completion
are satisfied: $\Omega = \ell^{-1} \cos\tau = 0$ at $\scri$ and
$\dd\Omega = - \ell^{-1} \sin\tau\, \dd\tau = \pm \ell^{-1} \dd\tau \not = 0 $
at $\scri$.
The metric induced by $\w{\tilde{g}}$ on $\scri$ is obtained by
setting $\tau=\pm\pi/2$ in (\ref{e:glo:tg_Einstein}):
$\dd \chi^2 + \sin^2\chi \left( \dd\th^2 + \sin^2\th \, \dd\ph^2 \right) $.
This 3-metric is clearly Riemannian (this is actually the standard round metric
of $\mathbb{S}^3$), which shows that $\scri$ is a spacelike
hypersurface of $(\tilde{\M},\w{\tilde{g}})$. This of course agrees with
Property~\ref{p:glo:type_scri_sign_Lambda}, given that $\Lambda > 0$.
Finally, it is clear that $\scri^+$ (resp. $\scri^-$)
matches the definition of a future (resp. past) infinity given above.
We conclude that $(\tilde{\M},\w{\tilde{g}})$ is a conformal completion at null infinity
of de Sitter spacetime.
\end{example}
\subsection{Asymptotic flatness} \label{s:glo:asymp_flat}
Penrose\index[pers]{Penrose, R.} \cite{Penro64,Penro68} has defined
a spacetime $(\M,\w{g})$ to be \defin{asymptotically simple}\index{asymptotically!simple} iff there exists
a conformal completion at infinity $(\tilde{\M},\w{\tilde{g}})$
of $(\M,\w{g})$
such that every null geodesic in $\M$ has two endpoints in $\scri$.
The last condition, which is verified by Minkowski spacetime (cf. Fig.~\ref{f:glo:conf_Mink_null}),
de Sitter spacetime and anti-de Sitter spacetime (cf. the null geodesics in
Fig.~\ref{f:glo:AdS_completion}), is rather restrictive. In particular, it excludes
black hole spacetimes, since, almost by definition, the latter contain null
geodesics that have no endpoint on $\scri^+$, having only a past endpoint
on $\scri^-$, as far as $\scri$ is concerned. To cope with these spacetimes,
Penrose\index[pers]{Penrose, R.} has also introduced the following definition \cite{Penro68}:
a spacetime $(\M,\w{g})$ is
\defin{weakly asymptotically simple}\index{weakly!asymptotically simple} iff
there exists an open subset $\mathscr{U}$ of $\M$ and
an asymptotically simple spacetime $(\M_0, \w{g}_0)$
with an open neighborhood $\mathscr{U}_0$ of $\scri_0 = \partial \tilde{\M}_0$
in $\tilde{\M}_0$ such that $(\mathscr{U}_0\cap \M_0,\w{g}_0)$ is
isometric to $(\mathscr{U},\w{g})$.
\begin{remark}
For a given weakly asymptotically simple spacetime, there may be different
(non overlapping) regions $\mathscr{U}$ satisfying the above property.
For instance we shall see in Chap.~\ref{s:ker}
that there are an infinite series of them in the (maximally extended) Kerr spacetime.
\end{remark}
Finally one says that a spacetime $(\M,\w{g})$ is
\defin{asymptotically flat}\index{asymptotically!flat}\index{flat!asymptotically --}
(or more precisely \defin{weakly asymptotically simple and empty}\index{weakly!asymptotically simple and empty} \cite{HawkiE73})
iff $(\M,\w{g})$ is weakly asymptotically simple and the Ricci tensor of
$\w{g}$ vanishes in an open neighborhood of $\scri$: $\w{R} = 0$.
\begin{example}
The de Sitter and anti-de Sitter spacetimes are asymptotically simple but
are not asymptotically flat.
\end{example}
Penrose \cite{Penro65b} (see also \cite{Fraue04}) has shown that if $(\M,\w{g})$
is asymptotically simple and empty, the Weyl tensor of $\w{g}$ (cf. Sec.~\ref{s:bas:Weyl})
vanishes at $\scri$. Since the
Ricci tensor is zero, this implies that the full Riemann curvature tensor vanishes
at $\scri$ [cf. Eq.~(\ref{e:bas:Weyl})], hence the qualifier \emph{asymptotically flat}.
The following property holds:
\begin{prop}[null $\scri$ for asymptotically flat spacetimes]
\label{p:glo:null_scri}
The conformal boundary $\scri$ of an asymptotically flat spacetime $(\M,\w{g})$
is a null hypersurface of the conformal completion $(\tilde{\M},\w{\tilde{g}})$.
\end{prop}
\begin{proof}
Consider Eq.~(\ref{e:glo:tildeR_R}). Near $\scri$, we have $R=0$ by the
very definition of asymptotic flatness. The limit $\Omega\rightarrow 0$
results then in
$\tilde{g}^{\mu\nu} \partial_\mu \Omega \partial_\nu \Omega \stackrel{\scri}{=} 0$,
which, following the argument in the proof on p.~\pageref{e:glo:tildeR_R}, implies that
$\scri$ is a null hypersurface.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Black holes} \label{s:glo:BH}
\subsection{Preliminaries regarding causal structure} \label{s:glo:causal_struct}
Before we proceed to the precise definition of a black hole, let us introduce
some concepts regarding the causal structure of a given time-orientable spacetime $(\M,\w{g})$.
For any subset $S$ of $\M$, one defines
\begin{itemize}
\item the \defin{chronological future of $S$}\index{chronological!future}\index{future!chronological --} as the set $I^+(S)$ of all
points of $\M$ that can be reached from a point of $S$ by a future-directed
timelike curve of nonzero extent;
\item the \defin{causal future of $S$}\index{causal!future}\index{future!causal --} as the set $J^+(S)$ of
all points that either are in $S$ or can be reached from a point of $S$ by a future-directed
causal curve;
\item the \defin{chronological past of $S$}\index{chronological!past}\index{past!chronological --} as the set $I^-(S)$ of all
points of $\M$ that can be reached from a point of $S$ by a past-directed
timelike curve of nonzero extent;
\item the \defin{causal past of $S$}\index{causal!past}\index{past!causal --} as the set $J^-(S)$ of
all points that either are in $S$ or can be reached from a point of $S$ by a past-directed
causal curve.
\end{itemize}
From the above definitions, one has always $S \subset J^\pm(S)$ and
$I^\pm(S) \subset J^\pm(S)$.
\begin{remark}
One has not necessarily $S \subset I^\pm(S)$. For instance,
if $\M$ does not contain
any closed timelike curve, one has $S \cap I^\pm(S) = \varnothing$ for
$S = \{p\}$ with $p$ being any point of $\M$.
\end{remark}
Here are some basic topological properties of the future and past sets
defined above (see e.g. \S~6.2 of \cite{HawkiE73} or Chap.~14 of
\cite{ONeil83} for proofs):
\begin{itemize}
\item
$I^\pm(S)$ is always an open subset\footnote{This property is a direct
consequence of Lemma~\ref{p:glo:lem1} in Sec.~\ref{s:glo:properties_H} below.} of $\M$, while
$J^\pm(S)$ is not necessarily a closed subset.
\item The interior of $J^\pm(S)$ is $I^\pm(S)$:
\be \label{e:glo:int_JS_IS}
\mathrm{int}\, J^\pm(S) = I^\pm(S).
\ee
\item Both sets have the same closure:
\be \label{e:glo:clos_JS_IS}
\overline{J^\pm(S)} = \overline{I^\pm(S)} .
\ee
\item
It follows from (\ref{e:glo:int_JS_IS}) and (\ref{e:glo:clos_JS_IS})
that both sets share the same (topological) boundary:
\be \label{e:glo:boundary_JS_IS}
\partial J^\pm(S) = \partial I^\pm(S).
\ee
\end{itemize}
The subset of the causal future (resp. past) of $S$ formed by points that cannot be connected
to $S$ by a timelike curve is called the \defin{future horismos}\index{future!horismos}\index{horismos}
(resp. \defin{past horismos}\index{past!horismos})
of $S$ and is denoted by $E^+(S)$ (resp. $E^-(S)$):
\be
E^+(S) := J^+(S) \setminus I^+(S) \qand
E^-(S) := J^-(S) \setminus I^-(S) .
\ee
The horismos $E^\pm(S)$ is formed by null geodesics emanating from points in $S$
(cf. Proposition~4.5.10 of Ref.~\cite{HawkiE73}).
One has $E^\pm(S) \subset \partial J^\pm(S)$.
The spacetime $(\M,\w{g})$
is said to be \defin{causally simple}\index{causally simple}\index{simple!causally --}
iff for every compact set $K\subset \M$, $E^\pm(K) = \partial J^\pm(K)$. This is
equivalent to saying that $J^+(K)$ and $J^-(K)$ are closed subsets of $\M$.
\subsection{General definition of a black hole} \label{s:glo:def_BH}
We are now in position to give the general definition of a black hole.
We shall do it for a spacetime $(\M,\w{g})$ that admits a conformal completion
at infinity as defined in Sec.~\ref{s:glo:def_conf_compl} and
such that the future infinity $\scri^+$ is
\defin{complete}\index{complete!future null infinity}: if $\scri^+$
is a null hypersurface, which occurs if $(\M,\w{g})$ is asymptotically flat
(cf. Propery~\ref{p:glo:null_scri}),
this means that $\scri^+$ is generated by complete\footnote{Let us recall
that a geodesic is \emph{complete} iff its affine parameters range through the whole of
$\R$, cf. Sec.~\ref{s:geo:existence_uniqueness}
in Appendix~\ref{s:geo}. In particular, such a geodesic is inextendible.}
null geodesics.
The completeness condition is imposed to avoid ``spurious'' black holes,
such as black holes in Minkowski space (cf. Remark~\ref{s:glo:spurious_bh} below).
The neighborhood of $\scri^+$
in $\tilde{\M}$ can then be considered as the infinitely far region
reached by outgoing null geodesics. If a null geodesic does not reach this
region, it can be considered as being trapped somewhere else in spacetime: this
``somewhere else'' constitutes the black hole region. More precisely:
\begin{greybox}
Let $(\M,\w{g})$ be a time-oriented spacetime with a conformal completion at infinity
such that the future infinity $\scri^+$ is complete;
the \defin{black hole region}\index{black!hole!region}
is the set of points of $\M$ that do not belong to the causal past of $\scri^+$ (cf. Fig.~\ref{f:glo:def_bh}):
\be \label{e:glo:def_BH}
\encadre{\mathscr{B} := \M \setminus (J^-(\scri^+)\cap\M) } .
\ee
If $\mathscr{B} \neq \varnothing$, one says
that the spacetime $(\M,\w{g})$ \defin{contains a black hole}\index{black!hole} or that
$(\M,\w{g})$ is a \defin{black hole spacetime}\index{black!hole!spacetime}.
\end{greybox}
The black hole region is thus the set of points of $\M$
from which no future-directed causal curve in $\tilde{\M}$ reaches $\scri^+$.
\begin{figure}
\centerline{\includegraphics[width=0.6\textwidth]{glo_def_bh.pdf}}
\caption[]{\label{f:glo:def_bh} \footnotesize
The black hole region $\mathscr{B}$ defined as the complement of
the causal past of the future infinity, $J^-(\scri^+)$.}
\end{figure}
\begin{example}
The Minkowski spacetime contains no black hole, for all future-directed null geodesics
terminate at $\scri^+$ (cf. Fig.~\ref{f:glo:conf_Mink_null}).
More generally, any asymptotically simple spacetime contains no black hole
(cf. Sec.~\ref{s:glo:asymp_flat}).
\end{example}
\begin{example} \label{x:glo:Schwarztschild_BH}
The prototype of a black hole is the \emph{Schwarzschild black hole}\index{Schwarzschild!black hole};
it will be shown in Sec.~\ref{s:sch:BH} that the Schwarzschild spacetime contains
a region $\mathscr{B}$ that fulfills the above definition of a black hole region.
\end{example}
\begin{remark} \label{s:glo:spurious_bh}
If we release the assumption of $\scri^+$-completeness in the above definition,
we may end up with unphysical or ``spurious'' black holes.
For instance, let us consider the conformal completion of Minkowski spacetime
$(\M,\w{g})$ resulting from its embedding in the Einstein cylinder
$(\E,\w{\tilde{g}})$, as in
Sec.~\ref{s:glo:conf_complet_Mink},