Maximal Extension and the Penrose Diagram of Schwarzschild

Schwarzschild's 1916 vacuum solution is the first exact black hole, yet in its native coordinates it seems to break twice: the metric blows up at $r=2M$ , where $g_{tt}\to0$ and $g_{rr}\to\infty$ , and again at the centre $r=0$ . For half a century these loci were conflated and the horizon was mistaken for the edge of spacetime. The discipline of maximal extension separates a genuine tear in the manifold from a mere failure of the map, and the tool is simple: follow the light. Radial null geodesics are the surveyor's chain of general relativity — wherever they run off a chart at finite affine parameter, spacetime continues beyond it. The coordinate-invariant Kretschmann scalar $R_{abcd}R^{abcd}=48M^2/r^6$ is the arbiter: finite at $r=2M$ — the horizon is a coordinate artefact of the distant clock — and divergent only at $r=0$ , the one true curvature singularity. Removing the artefact and continuing the geodesics unfolds a far larger spacetime: the Kruskal-Szekeres extension reveals four regions — our exterior universe, a black-hole interior, a time-reversed white hole, and a second, causally disconnected exterior — joined at a bifurcation point and bridged by a non-traversable Einstein-Rosen throat. Compress this infinite manifold onto a finite page with a conformal rescaling that keeps light cones at $45°$ , and you have the Carter-Penrose diagram: the most economical picture in gravitational physics. The inevitability of the singularity, the one-way horizon, and the impossibility of signalling between the two universes are read off at a glance from straight lines. This article builds it from the metric up and re-derives every number.

Key takeaways

In Schwarzschild coordinates the metric degenerates at $r=2M$ , but the invariant $R_{abcd}R^{abcd}=48M^2/r^6$ stays finite there — the horizon is a coordinate singularity, and only $r=0$ is real.
Following radial null geodesics through the tortoise coordinate $r_*$ , then exponentiating the null coordinates, produces Kruskal-Szekeres coordinates $(T,X)$ : one chart, analytic everywhere except $r=0$ .
Analytic continuation reveals four regions — two exterior universes, a black hole, and a white hole — joined at a bifurcation two-sphere by a non-traversable Einstein-Rosen bridge.
The horizon $r=2M$ becomes the null pair $T=\pm X$ , the singularity $r=0$ the spacelike hyperbola $T^2-X^2=1$ , and constant- $t$ surfaces straight rays through the origin.
Conformal compactification ( $\tilde U=\arctan U$ ) pulls infinity onto the page as the boundaries $\mathscr{I}^\pm$ , $i^\pm$ , $i^0$ while keeping every light ray at $45°$ : the Penrose diagram.
The construction exposes the surface gravity $\kappa=1/4M$ as the rate that regularizes the horizon and fixes the Hawking temperature $T_{\mathrm H}=\kappa/2\pi=1/8\pi M$ .
Key idea: radial null geodesics are the diagnostic tool — coordinates whose level surfaces are those geodesics (Kruskal-Szekeres) dissolve the $r=2M$ pathology, analytic continuation unfolds a four-region manifold, and conformal rescaling compresses it into the finite Penrose diagram where every causal relation is a $45°$ line.

The core insight

Radial null geodesics are the diagnostic tool: by adopting coordinates whose constant surfaces are these geodesics (Kruskal-Szekeres), the $r=2M$ pathology vanishes and analytic continuation reveals a four-region manifold; conformal rescaling then compresses it into the finite Penrose diagram where every causal relationship is read off from $45°$ lines.

Where this lives today

The Penrose diagram is the working language, not a museum piece. When the Event Horizon Telescope resolved the shadows of M87 in 2019 and Sgr A in 2022, the photon ring around a dark disc was the signature of the region I geometry — the photon sphere at $r=3M$ and horizon at $r=2M$ . LIGO and Virgo probe it dynamically: a merger remnant's ringdown is a sum of quasinormal modes set by the surface gravity $\kappa$ and the light-ring, so each chirp measures the near-horizon geometry of derivation three. Horizon thermodynamics is now tested in the lab through analogue gravity — Steinhauer's sonic horizons in condensates reproduce the Hawking spectrum via the same map $V\propto e^{\kappa v}$ seen here. And two exteriors joined by a non-traversable bridge animate the "ER = EPR" conjecture linking bridges to entanglement, the information paradox, and firewalls, while cosmic censorship remains a great open problem posed entirely in the language of these diagrams.

The physical picture

Schwarzschild time $t$ is the clock of a static observer infinitely far from the hole, infinitely blueshifted relative to one near the horizon. So when an infalling observer crosses $r=2M$ in finite proper time, the distant chart stamps it $t\to+\infty$ : spacetime does not end there, the labelling does — the chart runs out of paper. Coordinates tied to light rays put it back, and the first figure shows the result: the Kruskal-Szekeres plane, where the horizon is an ordinary $45°$ line through the middle and the manifold continues on both sides.

Figure

The Kruskal-Szekeres extension: horizons r = 2M are the 45° null lines T = ±X through the origin, the singularity r = 0 is the spacelike hyperbola T² − X² = 1 top and bottom, and the four sign choices of the null coordinates give two exteriors, a black hole II, and a white hole IV

Now picture light cones. Far away they open at $45°$ ; approach $r=2M$ in Schwarzschild coordinates and they appear to tip over and pinch shut — an illusion of the chart. In Kruskal coordinates they stand upright everywhere, and the truth is visible: inside the horizon (region II) every future-directed cone tilts toward the top, so every path leads to the singularity — escape is not merely difficult but geometrically forbidden. The second figure performs the final compression — the conformal map folding the infinite Kruskal plane into a finite diamond without tilting a cone off $45°$ .

Figure

The Carter-Penrose diagram of eternal Schwarzschild: conformal compactification pulls infinity onto the page, light still travels on 45° lines, the singularity r = 0 is a horizontal spacelike boundary top and bottom, and the two exteriors I and III meet only at the bifurcation point so they can never exchange a signal — surface gravity κ = 1/4M sets the Hawking temperature shown

The Penrose diagram trades true distance for a conformal picture — angles and light cones faithful, lengths distorted so infinity reaches the page's edge — so causal questions are answered by a $45°$ line while metric questions cannot be answered at all. The punchline is the singularity: $r=0$ is drawn horizontal, i.e. spacelike — a future time every interior observer runs into, not a place to steer around. That single orientation is the whole content of "you cannot avoid the singularity of a black hole."

The core equations

The starting point is the metric everything is extracted from — the unique static, spherically symmetric vacuum solution, carrying the two pathologies.

\mathrm{d}s^2=-\Big(1-\frac{2M}{r}\Big)\mathrm{d}t^2+\Big(1-\frac{2M}{r}\Big)^{-1}\mathrm{d}r^2+r^2\big(\mathrm{d}\theta^2+\sin^2\theta\,\mathrm{d}\phi^2\big) \qquad (1)

where: $M$ — geometric mass $=GM_{\mathrm{phys}}/c^2$ (units $G=c=1$ , dimensions of length); $t,r$ — Schwarzschild time and areal radius; $\mathrm{d}\Omega^2=\mathrm{d}\theta^2+\sin^2\theta\,\mathrm{d}\phi^2$ — unit two-sphere metric. Reading: $g_{tt}$ vanishes and $g_{rr}$ diverges at $r=2M$ , both blow up at $r=0$ — but a divergent component is not yet a divergent geometry. The referee is a coordinate-independent curvature scalar: where it stays finite, the geometry is regular however the coordinates misbehave.

R_{abcd}R^{abcd}=\frac{48M^2}{r^6} \qquad (2)

where: $R_{abcd}$ — Riemann tensor; the contraction is the Kretschmann scalar, an invariant. Reading: at $r=2M$ it equals $48M^2/(2M)^6=3/4M^4$ , finite; only as $r\to0$ does it diverge — the horizon is a coordinate mirage, the centre a genuine tear. To follow radial light rays we straighten them with the tortoise coordinate, defined so the null directions become diagonals.

r_*=r+2M\ln\Big|\frac{r}{2M}-1\Big|,\qquad \frac{\mathrm{d}r_*}{\mathrm{d}r}=\Big(1-\frac{2M}{r}\Big)^{-1} \qquad (3)

where: $r_*$ — tortoise (Regge-Wheeler) radial coordinate. Reading: the logarithm pushes the horizon to $r_*\to-\infty$ — the stretching an exponential will undo — and radial null rays become straight lines $t=\pm r_*+\text{const}$ . Naming the two null directions gives the Eddington-Finkelstein coordinates $u=t-r_*$ (retarded) and $v=t+r_*$ (advanced); each single null chart regularizes only one horizon. Exponentiating both at the surface-gravity rate is the master stroke: it cancels the vanishing $(1-2M/r)$ factor exactly.

U=-e^{-u/4M},\qquad V=e^{v/4M} \qquad (4)

where: $U,V$ — Kruskal null coordinates; the constant $4M=1/\kappa$ is the inverse surface gravity, the only rate that makes the metric analytic at the horizon. Reading: in region I, $U<0$ , $V>0$ ; the horizon ( $r_*\to-\infty$ ) maps to the finite locus $U=0$ or $V=0$ . Substituting (4) into (1) gives the Kruskal metric, whose conformal factor is finite across the horizon.

\mathrm{d}s^2=-\frac{32M^3}{r}\,e^{-r/2M}\,\mathrm{d}U\,\mathrm{d}V+r^2\mathrm{d}\Omega^2,\qquad UV=\Big(1-\frac{r}{2M}\Big)e^{r/2M} \qquad (5)

where: $r=r(UV)$ is fixed implicitly by the second relation. Reading: at $r=2M$ the prefactor is $16M^2e^{-1}$ , finite and nonzero, so the metric is regular; it degenerates only at $r=0$ ( $UV=1$ ). Rotating the null axes to timelike and spacelike coordinates gives the Kruskal-Szekeres chart.

T=\tfrac12(V+U),\qquad X=\tfrac12(V-U)\ \Longrightarrow\ T^2-X^2=UV=\Big(1-\frac{r}{2M}\Big)e^{r/2M} \qquad (6)

where: $T$ — Kruskal time; $X$ — Kruskal space. Reading: constant- $r$ surfaces are hyperbolae; the horizon is $T=\pm X$ , the singularity $T^2-X^2=1$ , and constant- $t$ surfaces the straight rays $T/X=\tanh(t/4M)$ , with $t=\pm\infty$ the horizons. A conformal $\arctan$ rescaling maps the infinite $U,V$ range into a finite square — the Penrose diagram, with its boundary at infinity.

\tilde U=\arctan U,\qquad \tilde V=\arctan V \qquad (7)

where: $\tilde U,\tilde V\in(-\pi/2,\pi/2)$ — compactified null coordinates. Reading: since $\arctan$ preserves the $45°$ null lines, causal structure survives; the edges $\tilde U,\tilde V=\pm\pi/2$ become null infinity $\mathscr{I}^\pm$ , and the corners timelike infinity $i^\pm$ and spatial infinity $i^0$ . Finally, the rate hidden in equation (4) is a horizon invariant that fixes the black hole's temperature.

\kappa=\frac{1}{4M},\qquad T_{\mathrm H}=\frac{\kappa}{2\pi}=\frac{1}{8\pi M} \qquad (8)

where: $\kappa$ — surface gravity, the force-per-unit-mass at infinity to hold a body static just above the horizon; $T_{\mathrm H}$ — Hawking temperature (units $G=c=\hbar=k_B=1$ ). Reading: the map that regularizes the geometry and the spectrum that radiates the hole share one constant.

Derivation AKruskal-Szekeres extension — dissolving the horizon and unfolding four regions

The goal is to build the coordinate system analytic across $r=2M$ and then show that the relation it produces forces three regions beyond our own. We start from radial null geodesics — the only curves trustworthy near the horizon — straighten them with the tortoise coordinate, and exponentiate at the surface-gravity rate that cancels the metric's vanishing factor, leaving an implicit relation that, being analytic, cannot stop at the horizon.

Step 1. Introduce $r_*=r+2M\ln|r/2M-1|$ , so $\mathrm{d}r_*=(1-2M/r)^{-1}\mathrm{d}r$ ; the radial-null block becomes $(1-2M/r)(-\mathrm{d}t^2+\mathrm{d}r_*^2)$ and radial null rays are the straight lines $t\mp r_*=\text{const}$ . Why: this trades $r$ for the variable in which light travels at unit slope.

Step 2. Define $u=t-r_*$ , $v=t+r_*$ , giving $\mathrm{d}s^2=-(1-2M/r)\,\mathrm{d}u\,\mathrm{d}v+r^2\mathrm{d}\Omega^2$ . Still singular at $r=2M$ , but the pathology is now isolated in one scalar factor. Why: null coordinates concentrate the horizon problem into a single multiplicative zero.

Step 3. Exponentiate: $U=-e^{-u/4M}$ , $V=e^{v/4M}$ . Then $UV=-e^{r_*/2M}=(1-r/2M)e^{r/2M}$ and $\mathrm{d}u\,\mathrm{d}v=(16M^2/UV)\,\mathrm{d}U\,\mathrm{d}V$ , so the vanishing $(1-2M/r)$ is exactly cancelled by the matching pole $1/UV$ . Why: $4M=1/\kappa$ is the unique rate whose Jacobian swallows the metric's zero; any other leaves a residual singularity.

Step 4. Collect $\mathrm{d}s^2=-(32M^3/r)\,e^{-r/2M}\,\mathrm{d}U\,\mathrm{d}V+r^2\mathrm{d}\Omega^2$ , prefactor $16M^2e^{-1}$ at $r=2M$ . The metric is smooth for every $r>0$ ; the only degeneracy is at $r=0$ ( $UV=1$ ). Why: a finite nonzero smooth conformal factor certifies the horizon as pure coordinate illness.

Step 5. The relation $UV=(1-r/2M)e^{r/2M}$ , derived for region I ( $U<0<V$ ), is analytic and single-valued; its right side sweeps $(-\infty,1)$ as $r$ runs over $(0,\infty)$ , giving a good $r$ for all four sign choices — I, II (black hole, $U,V>0$ ), III, IV (white hole, $U,V<0$ ). Why: a manifold that stops where an analytic chart continues is incomplete.

Step 6. Pass to $T=\tfrac12(V+U)$ , $X=\tfrac12(V-U)$ , so $T^2-X^2=UV$ . Region I is $X>|T|$ , II is $T>|X|$ , III is $X<-|T|$ , IV is $T<-|X|$ ; the horizon is $T=\pm X$ , the singularity $r=0$ the spacelike hyperbola $T^2-X^2=1$ . Why: this exhibits upright $45°$ cones everywhere — the diagram we draw.

Go deeper · Derivation B · 1 minThe singularity is spacelike — the signature flip inside the horizon

Derivation BThe singularity is spacelike — the signature flip inside the horizon

The goal is to prove that a Schwarzschild singularity is a moment of time, not a location in space, by tracking the coordinate basis vectors as a worldline crosses $r=2M$ . The claim sounds paradoxical since $r$ is a radial distance outside, but at the horizon the signs of $g_{tt}$ and $g_{rr}$ both reverse, so the coordinate that measured space now measures time — and "reach $r=0$ " becomes like "reach next Tuesday", a theorem about the future rather than a navigation problem.

Step 1. For $r>2M$ , $(1-2M/r)>0$ , so $g_{tt}<0$ and $g_{rr}>0$ : $\partial_t$ is timelike (the time-translation Killing vector) and $\partial_r$ spacelike — the ordinary causal furniture of the exterior. Why: this fixes the baseline against which the interior's reversal is unmistakable.

Step 2. For $r<2M$ , $(1-2M/r)<0$ , so $g_{tt}>0$ and $g_{rr}<0$ : $\partial_t$ is now spacelike and $\partial_r$ timelike. The areal radius $r$ becomes a time coordinate and the interior is no longer static. Why: this one sign flip is the entire mechanism — no full extension is needed, only the metric on the two sides.

Step 3. Since $r$ is timelike inside, every future-directed worldline moves monotonically toward smaller $r$ : radial motion obeys $(\mathrm{d}r/\mathrm{d}\tau)^2=E^2-(1-2M/r)>0$ throughout region II, so $r$ cannot turn around, any more than an observer can decline to age. Why: the forced decrease converts "the singularity exists" into "it is in your future", whatever your rockets do.

Step 4. So $r=0$ is reached at finite proper time by every interior worldline and is drawn horizontal — spacelike. For radial free fall from rest at infinity ( $E=1$ ), integrating $\mathrm{d}\tau=-\sqrt{r/2M}\,\mathrm{d}r$ from $2M$ to $0$ gives $\tau=4M/3$ . Why: a finite time to a spacelike boundary is the quantitative face of inevitability — a countdown, not a cage. *The signature flip follows directly from equation (1), and the free-fall time $\tau=4M/3$ from $r=2M$ to $r=0$ gives $6.6\,\mu\mathrm{s}$ for a solar-mass hole.*

Go deeper · Derivation C · 1 minSurface gravity, the Rindler horizon, and the seed of the Hawking temperature

Derivation CSurface gravity, the Rindler horizon, and the seed of the Hawking temperature

The goal is to show that the constant $4M$ in the Kruskal exponentials is a geometric invariant — the surface gravity $\kappa$ — and that the near-horizon geometry it controls is a Rindler (uniformly accelerated) wedge, whose periodicity in imaginary time hands us the Hawking temperature almost for free. We expand the metric just outside $r=2M$ and recognize the Rindler form, reading off $\kappa$ , the rate that thermalizes the vacuum in the full quantum calculation.

Step 1. Set $r=2M+\delta$ ( $\delta\ll2M$ ) and use the proper distance $\rho=\int(1-2M/r)^{-1/2}\mathrm{d}r\approx2\sqrt{2M\delta}$ , so $1-2M/r\approx\rho^2/16M^2$ . Why: proper distance, not $r$ , is the physical ruler at the horizon, and it linearizes the metric.

Step 2. The metric becomes $\mathrm{d}s^2\approx-\kappa^2\rho^2\,\mathrm{d}t^2+\mathrm{d}\rho^2+(2M)^2\mathrm{d}\Omega^2$ with $\kappa=1/4M$ — the Rindler metric of a uniformly accelerated observer, $t$ a boost parameter, $\rho=0$ the acceleration horizon. Why: recognizing Rindler space transplants accelerated-frame results (Unruh effect) to the black hole.

Step 3. Read off $\kappa=1/4M$ : the redshifted acceleration of a static horizon observer, and the exponential rate in $U=-e^{-\kappa u}$ , $V=e^{\kappa v}$ that regularized the metric above. Why: one number appears as regularizing rate, Rindler acceleration, and inverse temperature — the unity the extension exposes.

Step 4. Continue to imaginary time $t\to-it_E$ : the metric becomes $\mathrm{d}s^2\approx\mathrm{d}\rho^2+\rho^2\,\mathrm{d}(\kappa t_E)^2$ , smooth at $\rho=0$ only if $\kappa t_E$ is an angle of period $2\pi$ , i.e. $t_E$ has period $\beta=2\pi/\kappa=8\pi M$ — the signature of temperature $T_{\mathrm H}=1/\beta=\kappa/2\pi=1/8\pi M$ . Why: merely demanding a smooth Euclidean geometry fixes the Hawking temperature — the extension is the geometric backbone of black-hole thermodynamics. *The near-horizon Rindler form yields $\kappa=1/4M$ , and the Euclidean periodicity $\beta=8\pi M$ gives $T_{\mathrm H}=1/8\pi M$ ; for a solar-mass hole $T_{\mathrm H}\approx6.2\times10^{-8}\,\mathrm{K}$ .*

Limiting cases

The extended geometry collapses to familiar pieces in the corners.

Case	Result	Meaning
Far exterior $r\gg2M$	$\mathrm{d}s^2\to-(1-2M/r)\,\mathrm{d}t^2+(1+2M/r)\,\mathrm{d}r^2+r^2\mathrm{d}\Omega^2$	Weak-field, asymptotically flat; the boundaries $\mathscr{I}^\pm$ , $i^0$ , $i^\pm$ live here.
Near horizon $r\to2M$	$\mathrm{d}s^2\approx-\kappa^2\rho^2\,\mathrm{d}t^2+\mathrm{d}\rho^2$ , $\kappa=1/4M$	Rindler wedge; $t$ a boost parameter, $\rho=0$ the acceleration horizon.
Inside horizon $r<2M$	$g_{rr}=(1-2M/r)^{-1}<0\Rightarrow r$ timelike	$r$ decreases to $r=0$ ; the singularity is a final moment, hence spacelike.
Massless limit $M\to0$	$UV\to$ Minkowski null coords	Horizons and singularity vanish; the four regions collapse to flat spacetime.
Realistic collapse	Regions III and IV → stellar interior	Oppenheimer-Snyder keeps only I and II — no white hole, no second universe.

Common trapThe wormhole is a picture, not a passage — and the white hole is an eternal-solution artefact

A constant- $T$ slice of the Kruskal manifold does look like two asymptotically flat sheets joined by a throat of minimum radius $2M$ , the Einstein-Rosen bridge, tempting to read as a shortcut. It is not traversable: the throat opens from zero at $T=-1$ , reaches $2M$ at $T=0$ , and pinches back to zero by $T=+1$ , faster than light can cross — any passage from I to III must be spacelike, hence unphysical, and on the diagram the two exteriors meet only at the bifurcation point. And the whole lower half — white hole IV and second exterior III — is an artefact of the eternal solution's time symmetry: a black hole formed by collapse replaces III and IV with the worldtube of the collapsing star, keeping only the future horizon and singularity. Real black holes have no white-hole twin.

Worked examples

1 · A solar-mass black hole by the numbers (fully worked)

Take $M=M_\odot=1.989\times10^{30}\,\mathrm{kg}$ , geometric mass $M_{\mathrm{geo}}=GM_\odot/c^2=1.477\,\mathrm{km}$ . Find the horizon radius, verify it is regular, and compute the surface gravity, Hawking temperature, and proper time from horizon to singularity.

Step 1. Horizon radius: $r=2M_{\mathrm{geo}}=2\times1.477\,\mathrm{km}=\mathbf{2.95\ km}$ — the null cross $T=\pm X$ , an event, not an obstacle.

Step 2. The horizon is regular: $R_{abcd}R^{abcd}=3/4M_{\mathrm{geo}}^4=3/(4\times1477^4)\approx\mathbf{1.6\times10^{-13}\ m^{-4}}$ , finite and small. Tidal forces there are lethal, but the geometry is smooth — nothing breaks at $r=2M$ .

Step 3. Surface gravity: $\kappa=c^4/4GM=c^2/2r_s\approx\mathbf{1.5\times10^{13}\ m/s^2}$ , the redshifted acceleration that regularizes the horizon.

Step 4. Hawking temperature: $T_{\mathrm H}=\hbar c^3/8\pi GMk_B\approx\mathbf{6.2\times10^{-8}\ K}$ — colder than the CMB, so a stellar hole absorbs far more than it radiates; scaling as $1/M$ , Sgr A* ( $4.3\times10^6\,M_\odot$ ) is $\approx1.4\times10^{-14}\,\mathrm{K}$ .

Step 5. Life expectancy inside: $\tau=4M_{\mathrm{geo}}/3=4\times1477/3=1969\,\mathrm{m}$ , i.e. $\tau=1969/c\approx\mathbf{6.6\ \mu s}$ . A solar-mass black hole is 2.95 km across, smooth at its horizon, and offers about 6.6 microseconds of proper time between crossing the horizon and reaching the spacelike singularity.

2 · Why the distant clock says "never" while Kruskal says "of course"

A probe approaches $r=2M(1+\epsilon)$ with $\epsilon=10^{-6}$ . Show that its horizon crossing is at finite Kruskal coordinate even though it is at infinite Schwarzschild $t$ .

Step 1. Tortoise value: $r_*=2M(1+\epsilon)+2M\ln\epsilon$ . With $\epsilon=10^{-6}$ , $\ln\epsilon=-13.8$ , so $r_*\approx\mathbf{-25.6\,M}$ — already far left in the tortoise chart, marching to $-\infty$ as $\epsilon\to0$ .

Step 2. Schwarzschild time diverges: $T/X=\tanh(t/4M)$ , and the crossing $T=X$ needs $\tanh(t/4M)=1$ , i.e. $t\to+\infty$ — the origin of the "frozen star" myth, a statement about the label, not the probe.

Step 3. Kruskal coordinate is finite: on the future horizon $U=-e^{-u/4M}\to0$ while $V=e^{v/4M}$ stays finite and positive, so the crossing sits at the ordinary point $U=0$ , $V>0$ — a finite $(T,X)$ with $T=X$ , reached in finite proper time. The crossing is a regular finite event; only the infinitely blueshifted coordinate $t$ stretches it to $t=\infty$ — the freeze is a coordinate illusion, not a physical halt.

Physics of the portrait

Telling a coordinate accident from a real edge of the world took fifty years and turned a picture into a proof.

Origins · 1 minRoger Penrose and how this was discovered

Roger Penrose

1931– · conformal diagrams, trapped surfaces, and the global structure of black-hole spacetimes

When Karl Schwarzschild wrote his solution within weeks of Einstein's field equations in 1916, the loci $r=2M$ and $r=0$ were tangled together, and for decades $r=2M$ was feared to be the boundary of physics. Lemaître showed in 1933 that a freely falling observer sails through it untroubled; Einstein and Rosen in 1935 stumbled on the bridge geometry without recognizing it as a black hole; Finkelstein in 1958 read $r=2M$ as a one-way membrane; and Martin Kruskal (with George Szekeres independently) delivered in 1960 the maximal analytic extension — the four-region manifold, regular everywhere but the true singularity. Yet the extended spacetime was still infinite, awkward to reason about globally. Roger Penrose's decisive contribution, from 1963, was the conformal technique: rescale the metric so infinity is dragged to a finite boundary while light cones are left untouched, so a spacetime's entire causal skeleton — horizons, infinities, singularities — can be drawn on one finite diagram and read like a map (Brandon Carter systematized these for rotating and charged holes, so it is properly the Carter-Penrose diagram). The payoff was not merely pictorial: in 1965 Penrose used this global way of thinking — the trapped surface, whose future is bounded by a spacelike singularity — to prove that singularities are a generic, unavoidable consequence of collapse, not an artefact of symmetry, a theorem that won him a share of the 2020 Nobel Prize. The picture that began as a coordinate puzzle ended as a theorem about the fate of the universe.

History

1916

— Schwarzschild solves Einstein's equations; the loci r = 2M and r = 0 both appear

1935

— Einstein and Rosen describe the bridge joining two asymptotically flat sheets

1958

— Finkelstein interprets r = 2M as a one-way causal membrane

1960

— Kruskal and Szekeres publish the maximal analytic extension

1963

— Penrose introduces conformal diagrams with infinity drawn as a finite boundary

1965

— Penrose proves the singularity theorem using trapped surfaces

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §31.4-31.6 · Wald, General Relativity (1984), §6.4 and §12.5 · Carroll, Spacetime and Geometry (2004), §5.6-5.7 · Hartle, Gravity (2003), Ch. 12 · Hawking & Ellis, The Large Scale Structure of Space-Time (1973), §5.5. The key relations in this article are: the Kretschmann scalar $48M^2/r^6$ with horizon value $3/4M^4$ ; $\mathrm{d}r_*/\mathrm{d}r=(1-2M/r)^{-1}$ ; the Kruskal relation $UV=(1-r/2M)e^{r/2M}$ with prefactor $16M^2e^{-1}$ at $r=2M$ ; $T^2-X^2=UV$ with $T^2-X^2=1\Leftrightarrow r=0$ and $T=\pm X\Leftrightarrow r=2M$ ; $\kappa=1/4M$ , $T_{\mathrm H}=1/8\pi M$ ; and free-fall time $4M/3$ from $r=2M$ to $r=0$ . Note: with $T=\tfrac12(V+U)$ , $X=\tfrac12(V-U)$ the correct sign is $T^2-X^2=+UV$ (so $r=0\Rightarrow T^2-X^2=+1$ ), consistent with region II being $T>|X|$ .

Prerequisites. The Schwarzschild Metric · Schwarzschild Radius and the Geometric Mass · Geodesic Equations in Schwarzschild · Killing Vectors and Conserved Quantities in Schwarzschild. Leads to. Gravitational Redshift in Schwarzschild · Photon Orbits and the Photon Sphere.

Ask the tutor: ask me to walk you through any single arrow in the construction — from "why the tortoise coordinate?" to "how the Euclidean period becomes the Hawking temperature" — or to redraw the collapse Penrose diagram and show exactly which regions disappear.

Why does it take infinite Schwarzschild time to reach the horizon if the proper time is finite?

Because Schwarzschild $t$ is the clock of a static observer at infinity, infinitely blueshifted relative to a clock near the horizon. The relation $T/X=\tanh(t/4M)$ shows the horizon $T=X$ corresponds to $t\to+\infty$ : the distant chart labels the crossing "at the end of time" though it is an ordinary finite event in Kruskal coordinates. The infalling observer's own proper time converges — $4M/3$ from $r=2M$ to $r=0$ , about $6.6\,\mu\mathrm{s}$ for a solar mass. The "freezing" of infalling matter is an artefact of the distant clock, not a physical halt.

How do we know the maximal extension has exactly four regions and not more?

The relation $UV=(1-r/2M)e^{r/2M}$ is analytic and single-valued, its right side sweeping monotonically over $(-\infty,1)$ as $r$ runs over $(0,\infty)$ . Every $UV<1$ gives exactly one $r$ , and the four sign combinations of $(U,V)$ exhaust the plane below the singularity hyperbola $UV=1$ . The maximal-extension theorem guarantees completeness: every geodesic either runs to infinite affine parameter (reaching an $\mathscr{I}$ or $i$ boundary) or ends at $r=0$ . No analytic continuation is left, so no fifth region can exist.

Is the Einstein-Rosen bridge a usable wormhole?

No. A constant-Kruskal-time slice does show two asymptotically flat universes joined by a throat of minimum radius $2M$ , but the throat is dynamical: it opens from zero at $T=-1$ , peaks at $2M$ at $T=0$ , and pinches back to zero by $T=+1$ , faster than light can cross. Any worldline from I to III would have to be spacelike, which no physical object can be, and on the Penrose diagram the exteriors touch only at the bifurcation point. Traversable wormholes require exotic energy-condition-violating matter, absent here.

Why do real black holes have no white hole or second universe?

The white hole (IV) and second exterior (III) follow from the eternal solution's exact time-reversal symmetry — the vacuum geometry looks the same running backward, so it carries a past horizon and past singularity mirroring the future ones. A black hole formed by collapse breaks that symmetry: it has a stellar past, not a white hole. Its diagram replaces the bottom half (III and IV) with the worldtube of collapsing matter, keeping only region I, the future horizon, and the future singularity. This is why astrophysical black holes swallow but never spew, and why no white hole has been seen.

What is surface gravity physically, and why does the same number set the temperature?

Surface gravity $\kappa$ is the acceleration needed to hold a test mass static just above the horizon, redshifted to infinity; for Schwarzschild $\kappa=1/4M$ . It enters as the exponential rate $U=-e^{-\kappa u}$ , $V=e^{\kappa v}$ that cancels the metric's vanishing factor. The same rate governs the near-horizon Rindler geometry, and requiring its Euclidean (imaginary-time) version to be smooth forces a periodicity $\beta=2\pi/\kappa$ — the signature of a thermal state, so $T_{\mathrm H}=\kappa/2\pi=1/8\pi M$ . Regularization and radiation are two faces of one constant.

What does "conformal" mean in the Penrose diagram, and what is lost?

A conformal transformation rescales the metric by a positive function, $\tilde g_{ab}=\Omega^2 g_{ab}$ , preserving angles — hence light cones and all causal relations — while distorting lengths and areas. The Penrose diagram chooses $\Omega$ to shrink to zero at infinity, bringing points infinitely far away onto a finite boundary. Causal structure survives: two events can be checked for signalling with a $45°$ line, since null geodesics are conformally invariant. Metric information is lost — you cannot read distances, areas, or clock rates off it. It is a causal map, not a scale drawing: right for global questions, wrong for local ones.