Special Relativity describes physics in flat spacetime — where there is no gravity. General Relativity extends this to curved spacetime, where gravity is not a force but the curvature of spacetime itself. This document traces the logical chain from Einstein's key insight to the field equations, explaining each mathematical object along the way.
Special Relativity requires algebra. General Relativity requires significantly more: multivariable calculus (partial derivatives, gradients), linear algebra (matrices, tensors), differential geometry (manifolds, curvature, geodesics), and partial differential equations. This document will explain each concept as it arises, but full mastery requires working through these subjects. This is the hardest part of your learning path — and also the most rewarding.
Einstein called this "the happiest thought of my life." It is the single insight on which all of General Relativity is built.
Why this matters: If gravity and acceleration are the same thing locally, then gravity is not a force acting at a distance (as Newton described it). Instead, gravity is a property of spacetime itself. A freely falling object is not "being pulled" — it is following the natural geometry of curved spacetime.
In an accelerating elevator, a horizontal beam of light would appear to curve downward (the elevator accelerates upward while the light crosses). By the equivalence principle, the same must happen in a gravitational field. Gravity bends light. This was confirmed during the 1919 solar eclipse.
In an accelerating elevator, a clock at the "floor" (toward the acceleration) ticks slower than a clock at the "ceiling." By the equivalence principle, clocks closer to a gravitational source tick slower. Time runs slower near massive objects.
GPS satellites must correct for this: their clocks run $\sim 45$ microseconds faster per day than ground clocks.
A photon climbing out of a gravitational well loses energy and its frequency decreases (shifts toward red). The fractional change:
% GPS time dilation correction
G = 6.674e-11; M = 5.972e24; c = 2.998e8;
r_surface = 6.371e6;
r_gps = 6.371e6 + 20200e3; % ~20,200 km altitude
% Gravitational time dilation (GR effect): GPS runs faster
dt_ratio_grav = sqrt(1-2*G*M/(r_gps*c^2)) / sqrt(1-2*G*M/(r_surface*c^2));
% Speed-related time dilation (SR effect): GPS runs slower
v_gps = sqrt(G*M/r_gps); % orbital velocity
dt_ratio_sr = sqrt(1-(v_gps/c)^2);
% Combined: GPS gains ~38 µs/day (GR) and loses ~7 µs/day (SR) = +31 µs/day
daily_gain_us = (dt_ratio_grav/dt_ratio_sr - 1)*86400*1e6;
fprintf('GPS clock drift: %.1f µs/day\n', daily_gain_us)In flat spacetime (Special Relativity), the distance between nearby events is:
This is the Minkowski metric. We can write it compactly using a matrix called the metric tensor $\eta_{\mu\nu}$:
Einstein's key mathematical move was to generalize this: in curved spacetime, replace the constant $\eta_{\mu\nu}$ with a position-dependent $g_{\mu\nu}(x)$:
The metric tensor $g_{\mu\nu}$ is a $4 \times 4$ symmetric matrix (10 independent components) that varies from point to point. It encodes all the information about the geometry of spacetime — distances, angles, the rate of clocks, and the curvature caused by gravity.
LaTeX: ds^2 = g_{\mu\nu}\,dx^\mu\,dx^\nu
When an index appears once up and once down (like $\mu$ in $g_{\mu\nu}\,dx^\mu$), it means sum over all values of that index (0, 1, 2, 3). So $g_{\mu\nu}\,dx^\mu\,dx^\nu$ is actually a double sum: $\sum_{\mu=0}^{3}\sum_{\nu=0}^{3}g_{\mu\nu}\,dx^\mu\,dx^\nu$ — that's 16 terms (but symmetry reduces unique components to 10). Einstein said introducing this convention was his "greatest contribution to mathematics."
% Minkowski metric (flat spacetime, c=1 units)
eta = diag([-1, 1, 1, 1]);
% Schwarzschild metric components (outside a spherical mass)
syms r theta G_const M c_val
g_tt = -(1 - 2*G_const*M/(r*c_val^2))*c_val^2;
g_rr = 1/(1 - 2*G_const*M/(r*c_val^2));
g_thth = r^2;
g_phph = r^2*sin(theta)^2;
g = diag([g_tt, g_rr, g_thth, g_phph]);
disp(g)On a flat surface, the shortest path between two points is a straight line. On a curved surface (like the surface of a sphere), the shortest path is a geodesic — a great circle. In curved spacetime, free-falling objects follow geodesics.
This replaces Newton's First Law. Instead of "objects move in straight lines unless acted upon by a force," we say "objects follow geodesics unless acted upon by a non-gravitational force." Gravity is not a force — it is the curvature of the path itself.
This is the equation of motion in General Relativity. The first term is the acceleration. The second term (with $\Gamma$) encodes how the curvature of spacetime deflects the object's path. In flat spacetime, $\Gamma = 0$ and the equation reduces to $\frac{d^2x}{d\tau^2} = 0$ — constant velocity, Newton's First Law.
LaTeX: \frac{d^2 x^\mu}{d\tau^2} + \Gamma^\mu_{\alpha\beta}\frac{dx^\alpha}{d\tau}\frac{dx^\beta}{d\tau} = 0
The $\Gamma^\mu_{\alpha\beta}$ in the geodesic equation are the Christoffel symbols (also called the connection coefficients). They tell you how the coordinate axes themselves change from point to point in curved spacetime.
They are constructed entirely from the metric tensor and its first derivatives. In 4D spacetime, there are $4 \times 4 \times 4 = 64$ components (reduced to 40 by symmetry in the lower indices).
On a flat map of the Earth, lines of longitude converge toward the poles. If you walk "straight north" along two different longitudes, the distance between your paths changes even though you're both walking straight. The Christoffel symbols encode exactly this kind of effect — how "straight" lines in curved space diverge or converge.
% Symbolic computation of Christoffel symbols
% (MATLAB Symbolic Math Toolbox)
syms r theta M G c
% For Schwarzschild metric (simplified, c=1, G=1):
% g_tt = -(1-2M/r), g_rr = 1/(1-2M/r), g_thth = r^2, g_phph = r^2*sin(theta)^2
% Example: Gamma^r_tt = M/r^2 * (1-2M/r) (gravitational "acceleration")
Gamma_r_tt = M/r^2 * (1 - 2*M/r);
fprintf('Christoffel symbol Gamma^r_tt = M/r^2 (1-2M/r)\n')
fprintf('At large r, this reduces to M/r^2 = Newtonian g\n')The Christoffel symbols tell you about coordinates. The Riemann curvature tensor $R^\rho{}_{\sigma\mu\nu}$ tells you about actual, physical curvature — the kind that cannot be removed by changing coordinates.
The key idea: If you parallel-transport a vector around a small closed loop in curved spacetime, it comes back rotated. The Riemann tensor measures this rotation.
This has $4^4 = 256$ components, reduced by symmetries to 20 independent components in 4D. It is built from the Christoffel symbols and their derivatives — which means it's built from the metric and its first and second derivatives.
If $R^\rho{}_{\sigma\mu\nu} = 0$ everywhere, spacetime is flat. If it's nonzero, spacetime is genuinely curved.
$g_{\mu\nu}$ (metric) → $\Gamma^\mu_{\alpha\beta}$ (Christoffel / connection) → $R^\rho{}_{\sigma\mu\nu}$ (Riemann curvature). Each level is built from the previous one via differentiation. The metric is the fundamental object; everything else is derived from it.
The Riemann tensor has 20 independent components — too many for a field equation. Einstein needed a simpler curvature object. The solution: contract (sum over) the Riemann tensor's indices.
The Ricci tensor is a $4 \times 4$ symmetric tensor (10 independent components). It measures how volumes change as they move through curved spacetime. Where mass is present, nearby geodesics converge — volumes shrink — and $R_{\mu\nu} \neq 0$.
Contract again to get a single number at each point — the Ricci scalar $R$, which measures the overall "average" curvature.
Einstein needed a curvature tensor that satisfies a conservation law — specifically, its covariant divergence must be zero (because energy-momentum is conserved). The Ricci tensor alone doesn't satisfy this. The correct combination is:
This is a $4 \times 4$ symmetric tensor (10 components) built from the Ricci tensor, the Ricci scalar, and the metric. It satisfies the contracted Bianchi identity:
This identity is purely geometric — it follows from the structure of the Riemann tensor — and it guarantees that whatever we set $G_{\mu\nu}$ equal to must also have zero divergence. Energy-momentum has exactly this property.
The right-hand side of Einstein's equation describes the matter and energy that cause spacetime to curve. This information is packaged in the stress–energy tensor $T_{\mu\nu}$.
For a perfect fluid (no viscosity, no shear stress):
where $\rho$ is the energy density, $p$ is the pressure, and $u_\mu$ is the four-velocity of the fluid. Conservation of energy-momentum requires:
This is why the Einstein tensor works — it also has zero divergence, so setting $G_{\mu\nu} \propto T_{\mu\nu}$ is mathematically consistent.
We now have all the ingredients. The left side describes geometry (curvature). The right side describes matter (energy-momentum). Einstein's great equation connects them:
Expanded:
The proportionality constant is fixed by requiring that the equations reduce to Newtonian gravity in the weak-field, slow-motion limit (see Section 10). The $G$ is Newton's gravitational constant. The $c^4$ comes from the conversion between mass and energy ($E = mc^2$) applied twice (once for each side of the equation). The $8\pi$ comes from matching Poisson's equation $\nabla^2\Phi = 4\pi G\rho$ in the Newtonian limit.
The left side ($G_{\mu\nu}$) describes how spacetime curves. The right side ($T_{\mu\nu}$) describes the energy, momentum, and pressure present. The equation says they are proportional.
Matter tells spacetime how to curve. Spacetime tells matter how to move.
Note: these look like one equation but are actually 10 coupled nonlinear partial differential equations (because $\mu$ and $\nu$ each range from 0 to 3, and $g_{\mu\nu}$ is symmetric). They are extraordinarily difficult to solve — exact solutions exist only for highly symmetric situations.
Equivalence Principle → gravity is curvature → need a metric $g_{\mu\nu}$ → compute Christoffel symbols $\Gamma$ → compute Riemann tensor $R^\rho{}_{\sigma\mu\nu}$ → contract to Ricci tensor $R_{\mu\nu}$ and scalar $R$ → form Einstein tensor $G_{\mu\nu} = R_{\mu\nu} - \frac{1}{2}Rg_{\mu\nu}$ → set equal to matter content $\frac{8\pi G}{c^4}T_{\mu\nu}$ → solve for $g_{\mu\nu}$
% The full chain symbolically (MATLAB Symbolic Math Toolbox)
% In practice, computing the Einstein tensor by hand is brutal.
% Computer algebra is essential.
% The Tensor Toolbox or symbolic packages handle this:
% Einstein's constant
G_const = 6.674e-11; c = 2.998e8;
kappa = 8*pi*G_const/c^4; % 2.076e-43 s²/(kg·m)
% This tiny number is why everyday matter barely curves spacetime
fprintf('Einstein constant kappa = %.3e\n', kappa)For General Relativity to be correct, it must reduce to Newtonian gravity when gravity is weak and speeds are slow. Here's how.
(1) Weak field: $g_{\mu\nu} \approx \eta_{\mu\nu} + h_{\mu\nu}$ where $h_{\mu\nu} \ll 1$. (2) Slow motion: $v \ll c$. (3) Static field: $\partial g_{\mu\nu}/\partial t = 0$. (4) Non-relativistic matter: $T_{00} \approx \rho c^2$, all other components negligible.
Under these approximations, the time-time component of the geodesic equation becomes:
Compare with Newton: $\frac{d^2 x^i}{dt^2} = -\frac{\partial \Phi}{\partial x^i}$, where $\Phi$ is the gravitational potential. So:
The 00-component of Einstein's field equations, under the same approximations, becomes:
This is exactly Poisson's equation for Newtonian gravity. For a point mass $M$, the solution is $\Phi = -GM/r$, which gives:
Newton's law of gravitation, recovered as a limiting case of Einstein's field equations.
Newton's gravity is not wrong — it's an approximation of General Relativity that works brilliantly when gravity is weak ($GM/rc^2 \ll 1$) and speeds are low ($v \ll c$). For your SLAM engine, for bridges, for rockets to the Moon — Newton is more than sufficient. GR only matters near black holes, neutron stars, or when extreme precision is needed (GPS).
The first exact solution to Einstein's field equations was found by Karl Schwarzschild in 1916 — just months after Einstein published GR, and while Schwarzschild was serving on the Russian front in WWI.
where $r_s = \frac{2GM}{c^2}$ is the Schwarzschild radius.
This describes the spacetime outside any spherically symmetric, non-rotating, uncharged mass. At $r = r_s$, the metric has a coordinate singularity — this is the event horizon of a black hole. At $r = 0$, there is a true physical singularity where curvature becomes infinite.
% Schwarzschild radius for various objects
G = 6.674e-11; c = 2.998e8;
M_sun = 1.989e30; M_earth = 5.972e24; M_human = 70;
rs_sun = 2*G*M_sun/c^2; % 2.95 km
rs_earth = 2*G*M_earth/c^2; % 8.87 mm
rs_human = 2*G*M_human/c^2; % 1.04e-25 m (subatomic)
fprintf('Sun: %.2f km\n', rs_sun/1e3)
fprintf('Earth: %.2f mm\n', rs_earth*1e3)
fprintf('70 kg human: %.2e m\n', rs_human)| Prediction | Equation | Confirmed |
|---|---|---|
| Perihelion precession of Mercury | $\Delta\phi = \frac{6\pi GM}{c^2 a(1-e^2)}$ | 1915 — matched the anomalous 43"/century |
| Deflection of starlight by the Sun | $\delta = \frac{4GM}{c^2 b}$ | 1919 — Eddington's solar eclipse expedition |
| Gravitational redshift | $z = \frac{\Delta\lambda}{\lambda} = \frac{GM}{Rc^2}$ | 1959 — Pound-Rebka experiment |
| Gravitational time dilation | $\Delta t = \Delta t_0\sqrt{1-2GM/rc^2}$ | 1971 — Hafele-Keating (clocks on planes) |
| Gravitational waves | $h \sim \frac{GM}{rc^2}\frac{v^2}{c^2}$ | 2015 — LIGO detected merging black holes |
| Black holes | Schwarzschild solution | 2019 — Event Horizon Telescope imaged M87* |
| Gravitational lensing | Einstein ring: $\theta_E = \sqrt{\frac{4GM}{c^2}\frac{d_{LS}}{d_L d_S}}$ | Multiple observations since 1979 |
| Frame dragging (Lense-Thirring) | Rotating masses drag spacetime | 2011 — Gravity Probe B satellite |
Every prediction of General Relativity has been confirmed to extraordinary precision. No experiment has ever contradicted it in its domain of validity.
In 1917, Einstein added a term $\Lambda g_{\mu\nu}$ to his field equations to allow for a static universe:
$\Lambda$ is the cosmological constant. Einstein later called it his "biggest blunder" after Hubble discovered the universe is expanding. But in 1998, observations of distant supernovae showed the expansion is accelerating — and a positive $\Lambda$ provides exactly this effect.
The cosmological constant is now understood as the energy density of the vacuum itself — dark energy — and constitutes approximately 68% of the total energy content of the universe. Its observed value is:
Explaining why $\Lambda$ has this particular (tiny but nonzero) value is one of the greatest unsolved problems in physics — the cosmological constant problem.
General Relativity is spectacularly successful. So why do we need to go beyond it? Because GR and quantum mechanics are fundamentally incompatible.
GR treats spacetime as a smooth, continuous manifold. Quantum mechanics requires that physical quantities are quantized — they come in discrete packets. When you try to apply quantum field theory to gravity (treating the graviton as the force carrier, as we do for photons in electromagnetism), the calculations produce infinities that cannot be removed by the standard techniques (renormalization) that work for the other forces.
The incompatibility becomes physically relevant in two regimes: (1) inside black holes at $r = 0$ where GR predicts infinite curvature — a singularity — and (2) at the Big Bang, where the entire universe was compressed to a single point. Both situations require simultaneously strong gravity (GR) and small scales (QM). Neither theory alone can handle these.
| Approach | Core Idea |
|---|---|
| String Theory | Particles are 1D strings vibrating in 10+ dimensions. Gravity emerges from closed string modes. |
| Loop Quantum Gravity | Spacetime itself is quantized — made of discrete "atoms" of geometry at the Planck scale ($10^{-35}$ m). |
| Causal Set Theory | Spacetime is a discrete set of events with only causal ordering, no continuous manifold. |
| Asymptotic Safety | Gravity may be renormalizable after all at a non-trivial ultraviolet fixed point. |
The scale where quantum gravity effects become important is set by combining the three fundamental constants $G$, $\hbar$, and $c$:
These are unimaginably small (and the mass, unimaginably large for a particle). Below the Planck length, the concept of smooth spacetime likely breaks down entirely.
% Planck units
hbar = 1.055e-34; G = 6.674e-11; c = 2.998e8;
l_P = sqrt(hbar*G/c^3); % 1.616e-35 m
t_P = sqrt(hbar*G/c^5); % 5.391e-44 s
m_P = sqrt(hbar*c/G); % 2.176e-8 kg
E_P = m_P*c^2; % 1.956e9 J = 1.22e19 GeV
fprintf('Planck length: %.3e m\n', l_P)
fprintf('Planck time: %.3e s\n', t_P)
fprintf('Planck energy: %.3e GeV\n', E_P/1.602e-10)This is where your learning path diagram ends — at the frontier of human knowledge. The equation that unifies General Relativity with quantum mechanics has not been found. It is, quite literally, the last equation. Everything you're studying — from the algebra of exponents to the Lorentz transformations to the Riemann curvature tensor — is building the mathematical vocabulary and physical intuition needed to contribute to this search.