JOVANA
Explore Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
All guides

The Geometry of General Relativity & the Singularity Theorems

Einstein turned gravity into curvature and spacetime into a Lorentzian manifold. We read his equation as geometry, follow how the focusing of nearby geodesics forces light cones to converge, and watch Penrose and Hawking parlay that single focusing fact into a theorem: a black hole, or the universe itself, must contain a singularity.

From Riemannian to Lorentzian: one sign changes everything

Everything you built on the Riemannian rung — the Levi-Civita connection, the curvature tensor, geodesics, Jacobi fields — carries over almost verbatim to general relativity. The single change is in the metric. A Riemannian metric g is positive-definite: every nonzero vector has positive length-squared. A Lorentzian metric has signature (-,+,+,+): one minus sign and three plusses, so g(v,v) can be negative, zero, or positive. That one sign is the whole physics. Vectors split into three kinds — timelike (g(v,v) < 0, the worldlines of massive particles), null or lightlike (g(v,v) = 0, the paths of photons), and spacelike (g(v,v) > 0) — and a Lorentzian manifold of dimension four is what relativity calls spacetime.

At each point the timelike directions form a double cone — the light cone — and choosing one half as 'future' consistently across the manifold is a time-orientation. A geodesic is still the curve with zero covariant acceleration, nabla of the tangent along itself equals zero, but now it splits by type: timelike geodesics are free-falling massive bodies, null geodesics are light rays. The arc length integral that geodesics extremize is, for a timelike curve, the proper time an ideal clock reads along it. And here intuition must flip: a free-falling timelike geodesic between two events LOCALLY MAXIMIZES proper time, the opposite of the Riemannian length-minimizing geodesic. The minus sign reverses the variational sense, which is the geometric heart of the twin 'paradox'.

Einstein's equation: matter tells spacetime how to curve

Einstein's idea, stripped to geometry, is that gravity is not a force but the curvature of spacetime, and that matter is its source. The unknown is the metric g itself; the field equation relates the curvature of g to the energy and momentum present. Writing Ric for the Ricci curvature and S for the scalar curvature, the Einstein field equation reads Ric - (1/2) S g = 8 pi T, where T is the stress-energy tensor encoding the density and flow of matter. The left side is pure geometry, the right side pure physics; the equation says they are equal. Wheeler's slogan is exact: matter tells spacetime how to curve, and curved spacetime tells matter how to move (along geodesics).

The combination Ric - (1/2) S g is the Einstein tensor, and it is not an arbitrary choice. The second Bianchi identity from the bundles rung forces this exact combination to be divergence-free, which is precisely what is needed for energy-momentum to be conserved (the divergence of T must vanish). So the geometry hands you conservation for free. In empty space T = 0, and tracing the equation shows S = 0, leaving the vacuum Einstein equation Ric = 0 — exactly the Einstein manifold condition you met on the Riemannian rung, now in Lorentzian signature. Vacuum spacetimes are Ricci-flat Lorentzian manifolds; the Schwarzschild black hole and gravitational waves both solve Ric = 0.

Geodesic focusing: the engine of the theorems

The deepest single mechanism in this guide is one you already own from the Riemannian rung, reread in Lorentzian signature: the focusing of nearby geodesics by curvature. Recall a Jacobi field J measures how an infinitesimally nearby geodesic drifts away from a reference geodesic, and it obeys the Jacobi equation, second covariant derivative of J plus R(J, tangent) tangent equals zero. Positive curvature in the relevant direction makes J oscillate back toward zero — neighbouring geodesics reconverge. The point where J hits zero again is a conjugate point: a place where infinitesimally separated geodesics from a common source refocus, like rays of light brought back together by a lens.

Now make it physical. Track not one nearby geodesic but a whole bundle of them — a congruence — and measure how fast the bundle's cross-sectional volume shrinks. That rate is the expansion theta, and it obeys the Raychaudhuri equation: the time-derivative of theta equals minus (theta-squared over the dimension) minus shear-squared minus Ric(tangent, tangent). Every term on the right that we can sign is negative, dragging theta downward. The crucial hypothesis is on the matter: the strong energy condition says Ric(v,v) is non-negative for every timelike v — physically, gravity attracts, energy density stays non-negative. Under that single curvature condition, an initially converging bundle of geodesics is focused to a point — theta plunges to minus infinity — in finite proper time. Gravity, geometrically, always pulls geodesics together.

Raychaudhuri equation (for a timelike geodesic congruence, expansion theta):

    d theta / d tau   =   - theta^2 / 3   -   |shear|^2   -   Ric(u, u)
                            \________________________________________/
                                   every signed term is <= 0

Strong energy condition:   Ric(u, u) >= 0   for all timelike u

   ==>   if theta < 0 anywhere (the bundle starts converging),
         theta -> -infinity  in finite proper time  tau
   ==>   a CONJUGATE POINT forms: nearby geodesics refocus.

Lorentzian twist:  past a conjugate point a timelike geodesic
                   no longer MAXIMIZES proper time (Morse index theory).
The Raychaudhuri equation: curvature plus the strong energy condition force any converging bundle of geodesics to focus to a conjugate point in finite time. This single inequality is the analytic engine behind both singularity theorems.

Penrose and Hawking: turning focusing into a theorem

Here is the conceptual leap, and it is genuinely a leap. In the Riemannian world, a conjugate point along a geodesic merely means the geodesic stops being length-minimizing past that point — the Morse index theorem you saw on the comparison rung counts exactly this. In the Lorentzian world the same fact says a timelike geodesic stops MAXIMIZING proper time past a conjugate point: a longer (in proper time) timelike curve between the events exists nearby. The singularity theorems of Penrose (1965) and Hawking (1967) run this backward. Suppose spacetime were geodesically complete — every geodesic extends to infinite parameter, nothing ever 'runs off the edge'. Then between suitable events a longest timelike geodesic must exist and be free of conjugate points. But focusing forces a conjugate point. Contradiction. Therefore spacetime is NOT geodesically complete.

Read that conclusion carefully, because it is subtle and routinely misstated. Geodesic incompleteness means some freely-falling observer's worldline cannot be extended — it simply ends after finite proper time, with nowhere to go. THAT is what 'singularity' rigorously means in these theorems: not necessarily a point where curvature blows to infinity, not a puncture you can locate on a map, but an incomplete geodesic, a path that falls off the edge of existence. The theorems prove a singularity must exist; they say almost nothing about its nature. Penrose's version applies to gravitational collapse and predicts a singularity inside a black hole; Hawking's, run on the time-reversed expanding universe, predicts the Big Bang as a past singularity. Same focusing engine, two cosmic conclusions.

Positivity of mass, and honest limits

One more landmark closes the geometric picture and ties this rung together. An isolated gravitating system — a star, a galaxy, viewed from far away — sits in a spacetime that looks flat at infinity, and one can read off a single number, the ADM mass, from how fast the metric approaches flatness. The positive mass theorem (Schoen-Yau 1979, then Witten 1981) says: if the matter obeys a non-negative energy condition (the dominant energy condition), then the ADM mass is non-negative, and it is zero only for flat Minkowski space. Geometrically it is a statement about a Riemannian 3-manifold — a 'time slice' of the spacetime — whose scalar curvature is non-negative; the theorem says such a slice, asymptotically flat, cannot have negative total mass. Gravity, properly set up, never weighs less than nothing.

The two proofs are a beautiful microcosm of this whole rung. Schoen and Yau argued through minimal surfaces and the geometry of scalar curvature — the same minimal-surface machinery this volume built. Witten's proof, astonishingly, used the Dirac operator and a Weitzenbock formula straight out of Guide 2 of this very rung: positivity of mass falls out of an integration-by-parts identity for harmonic spinors, exactly the technique that powered Seiberg-Witten theory in Guide 3. The arc of this rung — Ricci flow, index theory, gauge theory, relativity — is not four separate subjects but one circle of ideas, curvature controlling global geometry, seen from four windows.

Hold the honest limits. First, 'survey, not proof': we stated the singularity theorems and positive mass theorem and motivated their engine, but each genuine proof is a chapter of hard global Lorentzian causality theory or hard elliptic analysis — a course, not a paragraph. Second, a singularity theorem predicts incompleteness, NOT a curvature blow-up; whether physical singularities are always hidden behind horizons is the still-open cosmic censorship conjecture, one of relativity's deepest unsolved problems. Third, the energy conditions are physical assumptions, not theorems: quantum fields can violate them, which is exactly the loophole wormholes and the accelerating universe exploit. The geometry is rigorous; the physical inputs are honest hypotheses you must keep visible. Guide 5 closes the rung with mirror symmetry and Calabi-Yau manifolds, where geometry meets string theory and the open problems multiply.