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).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.