PENROSE DIAGRAMS AND CAUSAL STRUCTURE

The trouble with infinity is that you cannot fit it on a blackboard. A spacetime diagram of flat space has coordinates that run from minus infinity to plus infinity in both time and space, and the most interesting questions — can a light ray that leaves here ever reach there? does this worldline have an endpoint, or does it run forever? — are questions about what happens out at the edges, exactly where the diagram runs off the paper.

In the late 1950s and early 1960s a small group of relativists set out to tame the edges. The decisive move came from , then a young mathematician moving into physics, and independently from Brandon Carter and others: rescale the metric so that the infinite ranges of the coordinates are squeezed into a finite picture, while keeping the one feature that actually matters for causality — the 45° tilt of light rays. The rescaling is conformal: it changes distances but preserves angles, and in spacetime the angle that counts is the one a light ray makes. Penrose presented the technique in a 1963 paper, "Asymptotic Properties of Fields and Space-Times," and the resulting figures have been called Penrose diagrams (or Penrose–Carter diagrams) ever since.

The payoff is enormous. A Penrose diagram of Minkowski space fits the entire infinite expanse of special relativity into a single triangle. The diagram of a Schwarzschild black hole reveals, on one page, that the geometry Karl Schwarzschild wrote down in December 1915 secretly contains four regions — two universes, a black hole, and a white hole — and that the singularity is not a point in space you could steer around but a moment in time you cannot avoid. None of that is visible in the original coordinates. The compactification makes the causal skeleton of the spacetime literally drawable.

This essay builds the technique from the one gentle equation that does the squeezing, then uses it on flat space and on the Schwarzschild solution. The three figures let you fold infinity inward by hand, drag an event around the Minkowski diamond to read its causal relationships, and click through the four regions of the maximally-extended black hole.

Start with flat spacetime in spherical coordinates, and look only at the radial–time plane. Light moves along the two null directions

$u = t - r, \qquad v = t + r$

These are the null coordinates: an outgoing light ray keeps $u$ constant, an ingoing one keeps $v$ constant. In ordinary coordinates both $u$ and $v$ run over the entire real line, $(-\infty, \infty)$. To compactify, apply a function that squashes the whole real line into a finite interval. The arctangent is the canonical choice:

$\tilde{u} = \arctan u, \qquad \tilde{v} = \arctan v, \qquad \tilde{u}, \tilde{v} \in \left(-\tfrac{\pi}{2}, \tfrac{\pi}{2}\right)$

In plain terms: every event in the infinite spacetime gets a new pair of coordinates that lives inside a finite box. An event at $t = 10^{6}$ and an event at $t = 10^{12}$ both land near $\tilde u = \pi/2$ — they get crowded together near the boundary, but they never reach it. Infinity becomes an edge of the picture rather than a place that runs off the page.

The final step is cosmetic: define diagram axes $T = \tilde v + \tilde u$ (vertical, time-like) and $X = \tilde v - \tilde u$ (horizontal, space-like, with $X \ge 0$ since $r \ge 0$). The whole of Minkowski space now sits inside the triangle $X \ge 0$, $|T| + X \le \pi$.

The crucial property is that this rescaling leaves light cones alone. Because $u$ and $v$ were transformed by the same function, a line of constant $u$ is still a line of constant $\tilde u$, and constant-$u$ is exactly a 45° line in the $(X, T)$ plane. The conformal factor stretches the grid wildly near the boundary, but every light ray, everywhere, still runs at 45°.

The compactified picture of flat space has a structure worth memorizing, because the same boundary pieces reappear in every asymptotically-flat spacetime. The triangle (a half-diamond, since we folded the $r < 0$ side onto $r \ge 0$) has five distinct kinds of boundary, and each one is the destination of a different kind of trajectory.

A massive particle moving slower than light traces a time-like worldline. No matter how it accelerates, in the infinite future it ends at the single point $i^{+}$, future timelike infinity, at the top of the diamond; run it backward and it emerges from $i^{-}$, past timelike infinity, at the bottom. A light ray instead ends on $\mathscr{I}^{+}$, future null infinity (pronounced "scri-plus"), the slanted upper-right edge, and begins on $\mathscr{I}^{-}$, the lower-right edge. And a single instant of cosmic time — a constant-$t$ slice — stretches out to $i^{0}$, spatial infinity, the right-hand corner.

These are not five arbitrary labels; they are forced by the causal type of the boundary. Time-like things end at points ($i^{\pm}$), null things end on null surfaces ($\mathscr{I}^{\pm}$), and space-like slices end at a point ($i^{0}$). The diagram makes a precise statement that is awkward to phrase in coordinates: all outgoing radiation from any isolated system, gravitational or electromagnetic, eventually arrives at $\mathscr{I}^{+}$. This is why $\mathscr{I}^{+}$ is the natural home of scattering theory in general relativity and the place where gravitational-wave energy is defined.

The causal reading is immediate once the cones are drawn. Two events can exchange a signal only if one lies inside the other's light cone — and because the cones are exactly 45°, you can read the relationship off the slope of the line connecting them. Steeper than 45°: time-like, a massive signal connects them. Exactly 45°: null, only light connects them. Shallower than 45°: space-like, nothing connects them at all.

Apply the same machinery to the Schwarzschild geometry and something startling appears. The line element found in 1915 looks like it has a catastrophe at the Schwarzschild radius $r_s = 2GM/c^2$: a term blows up there. For decades this was read as the edge of the world. It is not. In 1960 Martin Kruskal and George Szekeres found coordinates that pass smoothly through $r = r_s$, and the conformally-compactified version of those coordinates is the Schwarzschild Penrose diagram.

It has four regions. Begin with the metric in Kruskal–Szekeres null coordinates $U$ and $V$; the line element takes the form

$ds^2 = -\frac{32 G^3 M^3}{r}\, e^{-r/2GM}\, dU\, dV + r^2\, d\Omega^2$

where $r$ is now an implicit function of the product $UV$, and the metric coefficient is finite and nonzero everywhere except at $r = 0$. The horizon, where the original coordinates failed, sits harmlessly at $U = 0$ or $V = 0$ — the two diagonal lines that cross at the centre. The genuine catastrophe, the curvature singularity, sits at $r = 0$, which in these coordinates is the hyperbola

$UV = 1$

— two branches, one in the future and one in the past. The future branch caps off the black-hole interior; the past branch floors the white-hole interior.

Reading the diagram: Region I is our exterior universe, $r > r_s$, with its own $\mathscr{I}^{\pm}$ and $i^0$. Region II, above the crossing horizons, is the black-hole interior. Region III is a second asymptotically-flat exterior, a mirror universe causally disconnected from ours. Region IV, below, is the white hole — the time-reverse of a black hole, a region things can only leave. The two exteriors are joined at the central crossing point by a non-traversable "throat" (the Einstein–Rosen bridge); you cannot send a signal from I to III, because any attempt to cross the throat lands you in region II and on the singularity instead.

The single most counterintuitive lesson of the Schwarzschild diagram is the orientation of the singularity. In the original coordinates, $r = 0$ reads like the centre of a sphere — a place, a location you might in principle hover above and observe from a safe distance. The Penrose diagram shows this is wrong. Inside the horizon, the singularity $r = 0$ is a horizontal line across the top of region II. Horizontal in a Penrose diagram means space-like, and a space-like surface is a surface of constant time, not constant position.

Once you have crossed the horizon into region II, the singularity is in your future the way next Tuesday is in your future. Every future-directed path — every possible motion, however you fire your rockets — runs upward in the diagram, and every upward path in region II terminates on that horizontal line. You can no more avoid it than you can avoid tomorrow. This is the precise statement behind the popular claim that "inside a black hole, the singularity is everywhere you can go": it is in the future of every interior event, and the future is unavoidable.

This is also where the diagram's honesty about causality and dishonesty about distance pay off together. Firing your engines to "escape" the singularity makes things worse, not better: in region II, thrust that would carry you outward in flat space instead carries your worldline closer to the 45° horizon lines, shortening the proper time remaining before $r = 0$. The maximal proper time to the singularity is achieved by free-fall — by doing nothing. The diagram displays this directly: the longest time-like path from the horizon to the singularity is the one that stays most vertical, and burning fuel only tilts you toward a cone.

Penrose diagrams are not merely pedagogical cartoons; they are the working language in which the global structure of spacetime is stated and argued. The black-hole information question, the Hawking-radiation geometry, the structure of cosmological horizons, and the careful definitions of mass and radiation at infinity are all phrased on these compactified pictures. When a relativist asks whether a spacetime is "asymptotically flat," the operational test is whether it admits a well-behaved $\mathscr{I}^{+}$ of the Minkowski type drawn in FIG.46a.

Two cautions travel with the technique. First, the four-region maximally-extended Schwarzschild diagram is a vacuum idealization. A black hole that actually forms from collapsing matter has no white hole and no mirror universe — regions III and IV are replaced by the infalling star, and only regions I and II survive. The eternal diagram is the exact solution; the astrophysical diagram is the relevant one. Second, rotation changes everything: the Kerr geometry of a spinning black hole has a far richer Penrose diagram, with an inner horizon, a time-like (avoidable) singularity, and a chain of regions that, taken at face value, suggests passage to other universes — a face value no one fully trusts, because the inner horizon is unstable.

From here the trail runs two ways. The structure of the horizon itself — what an infaller feels versus what a distant observer sees — is the subject of the event horizon. And the question the diagram raises but cannot answer — what happens to information that crosses into region II — drives the modern story of black-hole thermodynamics and the information paradox. The diagram tells you the causal skeleton with perfect fidelity. It is silent on what the bones are made of.