Einstein Tensor

Once gravity is accepted as the curvature of spacetime, general relativity faces a single sharp question: which piece of the curvature should be set proportional to matter? The source is fixed and non-negotiable — the stress-energy tensor $T_{\mu\nu}$ , a symmetric rank-2 object that obeys local conservation $\nabla^{\mu}T_{\mu\nu}=0$ as an identity of physics. Whatever geometric quantity sits opposite it must therefore be symmetric, rank-2, and identically divergence-free, so that conservation is welded into the geometry rather than bolted on by fiat. The obvious first candidate, the Ricci tensor $R_{\mu\nu}$ , is symmetric and rank-2 but fails the last test: the contracted Bianchi identity forces $\nabla^{\mu}R_{\mu\nu}=\tfrac12\nabla_{\nu}R\neq0$ . Einstein spent two years chasing $R_{\mu\nu}\propto T_{\mu\nu}$ before the discrepancy defeated him. The cure is one term. Subtract half the metric times the scalar curvature and the leftover divergence cancels exactly: the combination $G_{\mu\nu}=R_{\mu\nu}-\tfrac12 g_{\mu\nu}R$ satisfies $\nabla^{\mu}G_{\mu\nu}=0$ for every metric, no exceptions, no conditions. This is the Einstein tensor, and it is the unique geometric partner the conserved source was waiting for. Lovelock's theorem sharpens uniqueness into rigidity: in four dimensions the only symmetric, divergence-free, rank-2 tensor built linearly from the metric and its first two derivatives is $a\,G_{\mu\nu}+b\,g_{\mu\nu}$ , and the constant $b$ is nothing but the cosmological constant. Geometrically $G_{\mu\nu}$ carries the trace part of curvature — the piece that changes the volume of a ball of freely falling particles — while the trace-free Weyl tensor, left untouched by the field equations, propagates freely as tides and gravitational waves.

Key takeaways

The Einstein tensor is the symmetric rank-2 combination $G_{\mu\nu}=R_{\mu\nu}-\tfrac12 g_{\mu\nu}R$ built from the Ricci tensor, the Ricci scalar, and the metric.
Its defining property is that it is automatically divergence-free, $\nabla^{\mu}G_{\mu\nu}=0$ , a pure identity following from the twice-contracted second Bianchi identity — not an equation of motion.
That divergence-free character is exactly what lets $G_{\mu\nu}$ couple to the conserved stress-energy tensor in the field equations $G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi G\,T_{\mu\nu}/c^{4}$ .
By Lovelock's theorem $G_{\mu\nu}$ is the unique such tensor in four dimensions up to the cosmological term, so the structure of general relativity is far more rigid than it looks.
Its trace in four dimensions is $G=g^{\mu\nu}G_{\mu\nu}=-R$ , which inverts the field equations to the trace-reversed form $R_{\mu\nu}=8\pi G\,(T_{\mu\nu}-\tfrac12 g_{\mu\nu}T)/c^{4}$ .
In vacuum $G_{\mu\nu}=0\Leftrightarrow R_{\mu\nu}=0$ , leaving only the Weyl tensor to carry curvature — the basis of Schwarzschild geometry and gravitational waves.
Key idea: $G_{\mu\nu}=R_{\mu\nu}-\tfrac12 g_{\mu\nu}R$ is the unique trace-adjusted Ricci tensor whose covariant divergence vanishes identically, making it the natural geometric partner of the conserved stress-energy tensor.

Where this lives today

The Einstein tensor is not a historical relic but the left-hand side of every calculation in relativistic gravitation. In numerical relativity, the codes that model merging black holes evolve the metric by casting the field equations $G_{\mu\nu}=8\pi G\,T_{\mu\nu}/c^{4}$ into a 3+1 form, in which four of the ten equations — precisely the ones protected by the identity $\nabla^{\mu}G_{\mu\nu}=0$ — become constraints that the initial data must satisfy and that the evolution must preserve; monitoring the drift of those constraints is how simulators know a run is trustworthy, and the waveforms they produce are what LIGO and Virgo match against their detections. In cosmology, the two Friedmann equations that fit the cosmic microwave background and the accelerating expansion are the time-time and spatial components of $G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi G\,T_{\mu\nu}/c^{4}$ , and the contracted Bianchi identity is what makes the fluid continuity equation $\dot\rho+3H(\rho+p/c^{2})=0$ a geometric consequence rather than a separate assumption. The measured value of the cosmological constant — the additive term Lovelock's theorem uniquely permits — is one of the deepest open problems in physics, because naive quantum estimates of the vacuum energy exceed the observed $\rho_{\Lambda}\approx6\times10^{-27}$ kg m⁻³ by many tens of orders of magnitude. And every test of general relativity in the strong field, from the shadow of M87* imaged by the Event Horizon Telescope to the precise timing of binary pulsars, is at bottom a test that the geometric tensor $G_{\mu\nu}$ tracks the matter tensor $T_{\mu\nu}$ exactly as equation (4) demands.

The physical picture

Curvature does two distinct jobs, and the Einstein tensor is the accountant that separates them. Release a small cloud of test particles at rest relative to one another and watch how it deforms as it falls freely. Two things can happen. The cloud can change volume — shrink or swell as a whole — and it can change shape at fixed volume, squashing a sphere into an ellipsoid. The volume change is the Ricci part of curvature, and it is precisely what local matter causes: pack energy and pressure inside the cloud and the volume accelerates inward. The shape change is the Weyl part, the tidal distortion that a distant mass produces and that survives in empty space. The Einstein tensor packages the volume-changing curvature, which is exactly why it, and not the full Riemann tensor, is the piece coupled to matter.

Figure

Curvature splits into two jobs. A ball of freely-falling test particles has its volume changed by the Ricci or trace part of curvature — the piece matter sources through G_μν — while its shape is sheared at fixed volume by the trace-free Weyl part, the tidal distortion that survives in vacuum. The volume focusing obeys V̈ over V equals minus R_μν u^μ u^ν

The extra term $-\tfrac12 g_{\mu\nu}R$ is not cosmetic bookkeeping; it is a conservation enforcer. The Ricci tensor "leaks" divergence at the rate $\tfrac12\nabla_\nu R$ , and subtracting half the metric times $R$ manufactures an equal and opposite leak that cancels it exactly, for any metric whatsoever. Geometry hands physics a conserved current for free, and it does so through an identity — the contracted Bianchi identity — that has nothing to do with the field equations and everything to do with the internal consistency of Riemannian geometry. Because $\nabla^\mu G_{\mu\nu}=0$ holds off-shell, the four relations it encodes act as constraints among the ten field equations, reflecting the freedom to relabel spacetime points (diffeomorphism invariance) and reducing the count of genuinely dynamical equations. The same term also fixes the coupling constant. When $G_{\mu\nu}=\kappa T_{\mu\nu}$ is dragged down to the Newtonian limit, $\kappa$ is forced to a specific, tiny number — and its smallness is the whole reason planets, not pebbles, are needed to bend spacetime enough to notice.

Figure

The Einstein equation as a balance sheet. Geometry G_μν equals matter times the coupling κ = 8πG/c⁴ = 2.08×10⁻⁴³ s²kg⁻¹m⁻¹, absurdly small so only astrophysical masses curve spacetime. Applied to the expanding universe the time-time component gives 3H² = 8πG ρ, so today's critical density is ρ_c = 8.5×10⁻²⁷ kg/m³, about five protons per cubic metre

The take-away from the two pictures is one sentence: $G_{\mu\nu}$ is the volume-changing, matter-sourced, automatically conserved part of curvature, and setting it proportional to the stress-energy tensor is the most economical way geometry can be told what matter does.

The core equations

The definition is the starting point. It combines the Ricci tensor, the Ricci scalar, and the metric into a single symmetric object, and every property below is a consequence of this one line together with the identities of Riemannian geometry.

G_{\mu\nu}=R_{\mu\nu}-\tfrac12\,g_{\mu\nu}R \qquad (1)

where: $G_{\mu\nu}$ — Einstein tensor (symmetric, rank 2); $R_{\mu\nu}$ — Ricci tensor, $R_{\mu\nu}=R^{\alpha}{}_{\mu\alpha\nu}$ ; $R$ — Ricci scalar, $R=g^{\mu\nu}R_{\mu\nu}$ ; $g_{\mu\nu}$ — metric tensor. Physical reading: the first term is the volume-focusing curvature; the second is the trace correction that makes the whole thing conserved. The single most important property is that the covariant divergence vanishes identically. This is not an equation of motion imposed on special metrics — it holds for the Einstein tensor of every metric, on-shell or off, and it is the reason the field equations are consistent with $\nabla^\mu T_{\mu\nu}=0$ .

\nabla^{\mu}G_{\mu\nu}=0 \qquad (2)

where: $\nabla^{\mu}$ — covariant divergence, $g^{\mu\alpha}\nabla_{\alpha}$ . Physical reading: whatever conservation law the source obeys, the geometry obeys the identical one automatically, so no inconsistency can arise between the two sides. Contracting the definition with the inverse metric gives the trace. Because $g^{\mu\nu}g_{\mu\nu}=n$ , the trace flips sign in four dimensions, and this sign flip is the mechanism behind "trace reversal."

G=g^{\mu\nu}G_{\mu\nu}=\left(1-\tfrac{n}{2}\right)R\;\xrightarrow{\ n=4\ }\;-R \qquad (3)

where: $n$ — spacetime dimension; $G$ — trace of the Einstein tensor. Physical reading: in four dimensions the trace of the geometric tensor is minus the scalar curvature, which lets the field equations be algebraically inverted for $R_{\mu\nu}$ . Setting $G_{\mu\nu}$ proportional to the source, plus the one term Lovelock's theorem permits, yields the Einstein field equations. The constant of proportionality is fixed by the Newtonian limit, and $\Lambda$ is the divergence-free additive freedom.

G_{\mu\nu}+\Lambda\,g_{\mu\nu}=\dfrac{8\pi G}{c^{4}}\,T_{\mu\nu} \qquad (4)

where: $\Lambda$ — cosmological constant; $T_{\mu\nu}$ — stress-energy tensor; $G$ — Newton's gravitational constant; $c$ — speed of light. Physical reading: ten coupled second-order equations for the metric, four of them constraints because of identity (2). Using the four-dimensional trace, the field equations invert to the trace-reversed form, in which the Ricci tensor is sourced directly by the stress-energy minus half its trace. This is the version one actually solves in the weak-field and Newtonian limits.

R_{\mu\nu}=\dfrac{8\pi G}{c^{4}}\left(T_{\mu\nu}-\tfrac12 g_{\mu\nu}T\right) \qquad (5)

where: $T$ — trace of stress-energy, $T=g^{\mu\nu}T_{\mu\nu}$ . Physical reading: in vacuum ( $T_{\mu\nu}=0$ ) this collapses to $R_{\mu\nu}=0$ , the equation that Schwarzschild solved. Finally, the Einstein tensor is not merely postulated — it is what the variation of the simplest curvature action produces. Extremizing the Einstein–Hilbert action over the metric yields $G_{\mu\nu}$ exactly, boundary terms aside.

S=\dfrac{c^{4}}{16\pi G}\int\sqrt{-g}\,R\,\mathrm{d}^{4}x,\qquad \frac{1}{\sqrt{-g}}\frac{\delta S}{\delta g^{\mu\nu}}\propto R_{\mu\nu}-\tfrac12 g_{\mu\nu}R=G_{\mu\nu} \qquad (6)

where: $g$ — determinant of $g_{\mu\nu}$ ; $\delta$ — functional variation. Physical reading: the divergence-free property is then guaranteed a second way, by the diffeomorphism invariance of the action through Noether's second theorem.

Derivation AThe divergence-free identity from the second Bianchi identity

The goal is to prove that $\nabla^{\mu}G_{\mu\nu}=0$ is a genuine identity of Riemannian geometry, true for the Einstein tensor of every metric with no field equations assumed, by contracting the differential (second) Bianchi identity of the Riemann tensor twice and watching the trace term supply exactly the divergence needed to cancel the Ricci leak. We start from the Bianchi identity, contract once to reach the Ricci level, contract again to reach the scalar level, and then assemble $G_{\mu\nu}$ .

Step 1. Write the second Bianchi identity for the Riemann tensor of any metric connection, $\nabla_{\lambda}R_{\rho\sigma\mu\nu}+\nabla_{\rho}R_{\sigma\lambda\mu\nu}+\nabla_{\sigma}R_{\lambda\rho\mu\nu}=0$ , the cyclic sum over the three derivative-and-first-pair indices. Why: this identity holds automatically for any metric-compatible, torsion-free connection, so anything derived from it is an identity, never a dynamical equation.

Step 2. Contract with $g^{\lambda\mu}$ . Using the antisymmetries of the Riemann tensor and the definition $R_{\sigma\nu}=g^{\lambda\mu}R_{\lambda\sigma\mu\nu}$ , the three terms collapse to $\nabla^{\mu}R_{\rho\sigma\mu\nu}+\nabla_{\rho}R_{\sigma\nu}-\nabla_{\sigma}R_{\rho\nu}=0$ . Why: the first contraction converts the four-index identity into a relation among the Ricci tensor and the once-contracted Riemann tensor, the level at which matter lives.

Step 3. Contract again with $g^{\rho\nu}$ . The mixed Riemann term becomes a second Ricci divergence, and the two Ricci-divergence terms add while the scalar term appears once: $\nabla^{\mu}R_{\sigma\mu}+\nabla^{\nu}R_{\sigma\nu}-\nabla_{\sigma}R=0$ , that is $2\nabla^{\mu}R_{\mu\sigma}=\nabla_{\sigma}R$ . Why: the second contraction produces the contracted Bianchi identity $\nabla^{\mu}R_{\mu\sigma}=\tfrac12\nabla_{\sigma}R$ , the precise statement of how much the Ricci tensor fails to be conserved.

Step 4. Take the divergence of the definition (1) and substitute. Metric compatibility $\nabla_\mu g_{\alpha\beta}=0$ lets $\nabla$ pass through the metric, so $\nabla^{\mu}G_{\mu\sigma}=\nabla^{\mu}R_{\mu\sigma}-\tfrac12\nabla_{\sigma}R=\tfrac12\nabla_{\sigma}R-\tfrac12\nabla_{\sigma}R=0$ . Why: the $-\tfrac12 g_{\mu\sigma}R$ term contributes exactly $-\tfrac12\nabla_\sigma R$ , cancelling the Ricci leak term for term, which is the entire reason the definition carries that coefficient.

Go deeper · Derivation B · 1 minWhy the trace coefficient must be exactly one-half

Derivation BWhy the trace coefficient must be exactly one-half

The goal is to show that the number $\tfrac12$ in the definition is not a convention or an aesthetic choice but is forced uniquely by the single requirement that the tensor be divergence-free for every metric. We introduce an unknown coefficient, demand $\nabla^\mu G_{\mu\nu}=0$ identically, and let the contracted Bianchi identity pin the coefficient down, exposing precisely why Einstein's earlier guess $R_{\mu\nu}\propto T_{\mu\nu}$ could not survive.

Step 1. Write the trial tensor with an undetermined coefficient, $G_{\mu\nu}^{(\alpha)}=R_{\mu\nu}-\alpha\,g_{\mu\nu}R$ , and take its divergence. Metric compatibility moves $\nabla$ past $g_{\mu\nu}$ , leaving $\nabla^{\mu}G_{\mu\nu}^{(\alpha)}=\nabla^{\mu}R_{\mu\nu}-\alpha\,\nabla_{\nu}R$ . Why: only the scalar $R$ is differentiated in the second term, so the whole divergence reduces to two gradients of $R$ .

Step 2. Insert the contracted Bianchi identity $\nabla^{\mu}R_{\mu\nu}=\tfrac12\nabla_{\nu}R$ from the previous derivation, giving $\nabla^{\mu}G_{\mu\nu}^{(\alpha)}=\bigl(\tfrac12-\alpha\bigr)\nabla_{\nu}R$ . Why: the geometry itself supplies the value $\tfrac12$ for the Ricci divergence, so the residual is a single multiple of $\nabla_\nu R$ .

Step 3. Demand the divergence vanish for a generic metric, where $\nabla_{\nu}R\neq0$ . The only way $(\tfrac12-\alpha)\nabla_{\nu}R=0$ can hold identically is $\alpha=\tfrac12$ . Why: because $\nabla_\nu R$ is nonzero for almost every spacetime, the coefficient cannot be tuned away case by case — it is forced once and for all, which is why $R_{\mu\nu}\propto T_{\mu\nu}$ (the case $\alpha=0$ ) is inconsistent with $\nabla^\mu T_{\mu\nu}=0$ . *Note: the coefficient $\tfrac12$ is independent of spacetime dimension.*

Go deeper · Derivation C · 1 minTrace reversal and the inverted field equations

Derivation CTrace reversal and the inverted field equations

The goal is to derive the trace-reversed field equation $R_{\mu\nu}=\tfrac{8\pi G}{c^{4}}(T_{\mu\nu}-\tfrac12 g_{\mu\nu}T)$ from the standard form, making explicit that "trace reversal" is nothing more mysterious than solving for the scalar curvature and substituting it back, and confirming that the vacuum case correctly reduces to Ricci-flatness. We set $\Lambda=0$ and abbreviate $\kappa=8\pi G/c^{4}$ , take the trace in four dimensions, solve for $R$ , and reinsert.

Step 1. Write the field equations in standard form, $R_{\mu\nu}-\tfrac12 g_{\mu\nu}R=\kappa T_{\mu\nu}$ . Why: this is the object to be inverted; the Ricci tensor is buried inside $G_{\mu\nu}$ and we want it alone on the left.

Step 2. Take the four-dimensional trace by contracting with $g^{\mu\nu}$ . Using $g^{\mu\nu}g_{\mu\nu}=4$ and $T=g^{\mu\nu}T_{\mu\nu}$ , the left side is $R-\tfrac12\cdot4\cdot R=-R$ , so $-R=\kappa T$ . Why: identity (3) told us the trace of $G_{\mu\nu}$ is $-R$ in four dimensions, and here it directly ties the scalar curvature to the trace of the matter.

Step 3. Solve for the scalar, $R=-\kappa T$ . Why: the scalar curvature is fixed pointwise by the trace of the stress-energy, so it can be eliminated everywhere it appears.

Step 4. Substitute back into the standard form, $R_{\mu\nu}=\kappa T_{\mu\nu}+\tfrac12 g_{\mu\nu}R=\kappa\bigl(T_{\mu\nu}-\tfrac12 g_{\mu\nu}T\bigr)$ . Why: this "trace-reversed" source is what actually drives the Ricci tensor; for vacuum $T_{\mu\nu}=0$ it gives $R_{\mu\nu}=0$ , and for static pressureless dust in signature $(-,+,+,+)$ the $00$ -component recovers Poisson's equation $\nabla^{2}\Phi=4\pi G\rho$ in the weak field. *The Newtonian-limit check $\nabla^{2}\Phi=4\pi G\rho$ is precisely what fixes $\kappa=8\pi G/c^{4}$ .*

Limiting cases

The Einstein tensor simplifies dramatically in the standard special regimes, and each row below is a place where the general definition collapses to something a working relativist uses daily.

Case	Result	Meaning
Vacuum	$G_{\mu\nu}=0\iff R_{\mu\nu}=0$	Tracing $G_{\mu\nu}=0$ gives $-R=0$ , so $R=0$ and the equation reduces to Ricci-flatness. Curvature survives entirely in the Weyl tensor — the case of Schwarzschild and of gravitational waves.
Maximally symmetric (de Sitter)	$R_{\mu\nu}=\Lambda g_{\mu\nu}\Rightarrow G_{\mu\nu}=-\Lambda g_{\mu\nu}$	With $R=4\Lambda$ in four dimensions, $G_{\mu\nu}=\Lambda g_{\mu\nu}-2\Lambda g_{\mu\nu}=-\Lambda g_{\mu\nu}$ , so $G_{\mu\nu}+\Lambda g_{\mu\nu}=0$ is satisfied in empty de Sitter space.
Two dimensions	$G_{\mu\nu}\equiv0$	In $n=2$ , $R_{\mu\nu}=\tfrac12 g_{\mu\nu}R$ holds identically, so $G_{\mu\nu}$ vanishes everywhere — general relativity has no local dynamics in 2D.
Weak field (linearized)	$G_{\mu\nu}\approx-\tfrac12\,\Box\,\bar h_{\mu\nu}$	In harmonic gauge with the trace-reversed perturbation $\bar h_{\mu\nu}=h_{\mu\nu}-\tfrac12\eta_{\mu\nu}h$ , the Einstein tensor becomes a wave operator, and the field equations turn into $\Box\bar h_{\mu\nu}=-2\kappa T_{\mu\nu}$ .

Common trapThe Einstein tensor is just the Ricci tensor with a cosmetic term, and its trace is always minus R

Both halves of that sentence are traps. The $-\tfrac12 g_{\mu\nu}R$ term is the opposite of cosmetic: without it $\nabla^{\mu}R_{\mu\nu}=\tfrac12\nabla_{\nu}R\neq0$ , so a field equation $R_{\mu\nu}=\kappa T_{\mu\nu}$ would force the unphysical constraint $\nabla_\nu R=0$ on all matter — the very inconsistency that defeated Einstein's first attempts. It is precisely the added term that supplies the cancelling divergence and makes $\nabla^{\mu}G_{\mu\nu}=0$ . As for the trace: $G=-R$ holds only in four dimensions. In general $G=(1-\tfrac{n}{2})R$ , which vanishes in $n=2$ and is positive for $n<2$ ; do not carry $G=-R$ into dimensionally regularized or higher-dimensional calculations. Note also that the coefficient $\tfrac12$ in the definition is genuinely dimension-independent even though the trace factor $(1-\tfrac{n}{2})$ is not — the two must not be conflated.

Worked examples

1 · The critical density of the universe from G_00 (fully worked)

The flat Friedmann–Lemaître–Robertson–Walker metric is $\mathrm{d}s^{2}=-c^{2}\mathrm{d}t^{2}+a(t)^{2}\delta_{ij}\mathrm{d}x^{i}\mathrm{d}x^{j}$ . Its Einstein tensor has time-time component $G_{tt}=3(\dot a/a)^{2}=3H^{2}$ (with $c=1$ ). Use the field equation $G_{\mu\nu}=8\pi G\,T_{\mu\nu}$ with $\Lambda=0$ to find the mass density that makes space spatially flat today, given the Hubble constant $H_{0}=67.4$ km s⁻¹ Mpc⁻¹.

Step 1. Read off the field equation in the time-time slot. For a perfect fluid at rest, $T_{tt}=\rho$ (mass density in units $c=1$ ), so $G_{tt}=8\pi G\,T_{tt}$ becomes $3H^{2}=8\pi G\rho$ . Solving for the density that closes the universe gives the critical density $\rho_{c}=\dfrac{3H^{2}}{8\pi G}$ . This is the Friedmann equation, and it is nothing but the $00$ -component of the conserved Einstein tensor.

Step 2. Convert $H_{0}$ to SI. One megaparsec is $3.086\times10^{22}$ m, so $H_{0}=\dfrac{67\,400\ \text{m/s}}{3.086\times10^{22}\ \text{m}}=\mathbf{2.184\times10^{-18}\ s^{-1}}$ .

Step 3. Insert the numbers. The numerator is $3H_{0}^{2}=3\times(2.184\times10^{-18})^{2}=3\times4.770\times10^{-36}=1.431\times10^{-35}\ \text{s}^{-2}$ . The denominator is $8\pi G=8\pi\times6.674\times10^{-11}=1.677\times10^{-9}\ \text{m}^{3}\text{kg}^{-1}\text{s}^{-2}$ .

Step 4. Divide: $\rho_{c}=\dfrac{1.431\times10^{-35}}{1.677\times10^{-9}}=\mathbf{8.53\times10^{-27}\ kg\,m^{-3}}$ . Dividing by the proton mass $1.673\times10^{-27}$ kg gives about 5.1 protons per cubic metre — a fantastically dilute average. The full curvature budget of the observable universe is set by the time-time component of $G_{\mu\nu}$ , and it corresponds to roughly five hydrogen atoms per cubic metre.

2 · The cosmological constant as an effective vacuum (numeric)

The observed cosmological constant is $\Lambda\approx1.1\times10^{-52}\ \text{m}^{-2}$ . Treating the $\Lambda g_{\mu\nu}$ term as a source by moving it to the right of the field equations, find (a) the effective vacuum mass density it represents, and (b) the de Sitter curvature radius $R_{\mathrm{dS}}=\sqrt{3/\Lambda}$ . Compare the density to the critical density from Example 1.

Step 1. Move $\Lambda$ to the source side. Writing $G_{\mu\nu}=8\pi G\,T_{\mu\nu}/c^{4}-\Lambda g_{\mu\nu}$ identifies an effective vacuum stress-energy with mass density $\rho_{\Lambda}=\dfrac{\Lambda c^{2}}{8\pi G}$ . The same $g_{\mu\nu}$ that Lovelock's theorem allows as an additive term reappears here as a uniform, conserved energy density.

Step 2. Evaluate the density. The numerator is $\Lambda c^{2}=1.1\times10^{-52}\times(2.998\times10^{8})^{2}=1.1\times10^{-52}\times8.988\times10^{16}=9.89\times10^{-36}\ \text{s}^{-2}$ . Dividing by $8\pi G=1.677\times10^{-9}$ gives $\rho_{\Lambda}=\mathbf{5.9\times10^{-27}\ kg\,m^{-3}}$ .

Step 3. Compare to critical. The ratio is $\rho_{\Lambda}/\rho_{c}=5.9/8.53=\mathbf{0.69}$ , matching the observed dark-energy fraction $\Omega_{\Lambda}\approx0.69$ — the cosmological term supplies about seven-tenths of the total.

Step 4. Compute the de Sitter radius. $R_{\mathrm{dS}}=\sqrt{3/\Lambda}=\sqrt{3/(1.1\times10^{-52})}=\sqrt{2.73\times10^{52}}=\mathbf{1.65\times10^{26}\ m}$ . Dividing by $9.461\times10^{15}$ m per light-year gives about $1.7\times10^{10}$ ly, or 17 billion light-years — the length scale over which the cosmological piece of $G_{\mu\nu}$ curves spacetime. A curvature term that measures $10^{-52}$ per square metre still sets the density and the horizon of the entire cosmos.

Physics of the portrait

The Einstein tensor carries one name for good reason: its assembly, in the November of 1915, was the act that completed general relativity, and it capped an eight-year search that ran through Prague, Zurich, and Berlin.

Origins · 2 minAlbert Einstein and how this was discovered

Albert Einstein

1879–1955 · the geometric side of the field equations G_μν = 8πG T_μν/c⁴

Albert Einstein reached the Einstein tensor not by elegant deduction but by a long, error-strewn climb. The seed was what he later called the happiest thought of his life, in 1907: a person falling freely feels no gravity, so acceleration and gravitation are locally the same. Turning that equivalence principle into a field theory of geometry took the mathematics of Riemann, which Einstein did not command; in 1912 he turned to his friend Marcel Grossmann, and together they produced the 1913 "Entwurf" theory, which correctly used the metric as the gravitational potential but proposed field equations that were not generally covariant. For two years Einstein defended the mistake, half-convinced by a flawed argument that full covariance was impossible. Through 1914 and 1915 he circled the correct form, at times setting the source proportional to the Ricci tensor alone and being unable to reconcile it with energy conservation. In November 1915, in a burst of four weekly communications to the Prussian Academy, he corrected the trajectory, at last writing the field equations with the trace term that guarantees conservation — the combination $R_{\mu\nu}-\tfrac12 g_{\mu\nu}R$ now bearing his name. The same month David Hilbert, to whom Einstein had explained the theory in Göttingen, obtained the equations from a variational principle; the priority was briefly tense but both men were generous, and the physical theory was unambiguously Einstein's. In 1917 he added the cosmological constant to force a static cosmos, a step he is said to have later regretted, though the term has returned as dark energy. What makes the tensor his is not the algebra but the physical insistence, held against his own repeated errors, that geometry and conserved matter must be married exactly — and the Einstein tensor is that marriage made precise.

History

1902

— Bianchi publishes the contracted differential identities that make ∇^μ G_μν = 0 automatic

1913

— Einstein and Grossmann cast gravitation in Riemannian geometry in the Entwurf theory

1915

— Einstein writes G_μν = 8πG T_μν/c⁴ in November; Hilbert derives it from an action the same month

1917

— Einstein introduces the cosmological constant Λ to model a static universe

1971

— Lovelock proves G_μν + Λg_μν is the unique divergence-free tensor of its class in four dimensions

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §17.1–17.3 · Wald, General Relativity (1984), §3.2 and §4.3 · Carroll, Spacetime and Geometry (2004), §3.8 and §4.1 · Hartle, Gravity (2003), Ch. 22 · Schutz, A First Course in General Relativity (2009), §8.1 · Lovelock, "The Einstein tensor and its generalizations," J. Math. Phys. 12, 498 (1971). All numerical values use CODATA $G=6.674\times10^{-11}$ m³ kg⁻¹ s⁻² and $c=2.998\times10^{8}$ m s⁻¹. Key results: the coupling $\kappa=8\pi G/c^{4}=2.08\times10^{-43}\ \text{s}^{2}\text{kg}^{-1}\text{m}^{-1}$ from the Newtonian limit; the critical density $\rho_{c}=3H_{0}^{2}/8\pi G=8.5\times10^{-27}$ kg m⁻³ for $H_{0}=67.4$ km s⁻¹ Mpc⁻¹; the vacuum density $\rho_{\Lambda}=\Lambda c^{2}/8\pi G=5.9\times10^{-27}$ kg m⁻³ and de Sitter radius $\sqrt{3/\Lambda}=1.65\times10^{26}$ m for $\Lambda=1.1\times10^{-52}$ m⁻².

Builds on: Ricci Tensor · Ricci Scalar · Second (Differential) Bianchi Identity · Contracted Bianchi Identities. Leads to: Einstein Field Equations · Weyl Conformal Tensor · Ricci Decomposition of the Riemann Tensor.

Ask the tutor: Ask me to compute $G_{\mu\nu}$ for any metric you like — Schwarzschild, FLRW, de Sitter, or a weak-field perturbation — and watch the trace term enforce conservation step by step.

Why is the specific term subtracted half the metric times R, and not something else?

Because that and only that term makes the tensor divergence-free. The Ricci tensor obeys $\nabla^{\mu}R_{\mu\nu}=\tfrac12\nabla_{\nu}R$ , so it "leaks" divergence at the rate $\tfrac12\nabla_{\nu}R$ . Subtracting $\tfrac12 g_{\mu\nu}R$ contributes exactly $-\tfrac12\nabla_{\nu}R$ to the divergence (metric compatibility lets $\nabla$ slide past $g_{\mu\nu}$ ), cancelling the leak identically. If you tried any other coefficient $\alpha$ , the divergence would be $(\tfrac12-\alpha)\nabla_{\nu}R$ , which is nonzero for generic spacetimes. The coefficient is forced by geometry, not chosen for elegance, and it is the same $\tfrac12$ in every spacetime dimension.

Is the Einstein tensor conserved because matter is conserved, or the other way around?

Neither causes the other; they are logically independent and happen to match. $\nabla^{\mu}G_{\mu\nu}=0$ is a pure identity of Riemannian geometry, true for any metric whatsoever, proved by contracting the second Bianchi identity with no reference to matter or field equations. Separately, $\nabla^{\mu}T_{\mu\nu}=0$ is a law of physics about the source. The field equations $G_{\mu\nu}=\kappa T_{\mu\nu}$ are consistent precisely because both sides are independently divergence-free. In fact the logic can be run forward: once you impose the field equations, the geometric identity forces $\nabla^{\mu}T_{\mu\nu}=0$ , so energy-momentum conservation becomes a consequence of the geometry rather than an extra postulate.

Does G_μν = 0 in vacuum mean there is no gravity there?

No — it means there is no matter there, which is very different. In vacuum $G_{\mu\nu}=0$ is equivalent to $R_{\mu\nu}=0$ (Ricci-flatness), but the full Riemann curvature can be large. The part that survives is the Weyl tensor, the trace-free piece the field equations leave unconstrained. Schwarzschild spacetime is the standard example: outside the star $R_{\mu\nu}=0$ , yet tidal forces stretch and squeeze, light bends, and orbits precess, all encoded in a nonzero Weyl tensor. Gravitational waves are the same story — ripples of Weyl curvature travelling through a region where $G_{\mu\nu}=0$ . So $G_{\mu\nu}$ tells you where the sources are, not where the gravity is.

How does the cosmological constant fit into the Einstein tensor?

Since the metric is covariantly constant, $\nabla^{\mu}g_{\mu\nu}=0$ , the combination $G_{\mu\nu}+\Lambda g_{\mu\nu}$ is divergence-free for any constant $\Lambda$ . Lovelock's theorem then says this is the most general symmetric, divergence-free, rank-2 tensor you can build in four dimensions from the metric and up to its second derivatives with linear dependence on the second derivatives. So $\Lambda$ is not an optional decoration — it is the unique additive freedom the mathematics permits. You can read it two ways: as a modification of the geometric left-hand side, or as a uniform vacuum energy $\rho_{\Lambda}=\Lambda c^{2}/8\pi G$ moved to the matter side. The two are algebraically identical, which is why "dark energy" and "cosmological constant" describe the same term.

Why does the trace G = −R hold only in four dimensions?

Because the trace picks up the dimension through $g^{\mu\nu}g_{\mu\nu}=n$ . Contracting the definition gives $G=g^{\mu\nu}R_{\mu\nu}-\tfrac12 g^{\mu\nu}g_{\mu\nu}R=R-\tfrac{n}{2}R=(1-\tfrac{n}{2})R$ . In four dimensions the factor is $1-2=-1$ , so $G=-R$ ; in two dimensions it is $0$ , which is why $G_{\mu\nu}$ vanishes identically and 2D gravity has no local dynamics; in three dimensions it is $-\tfrac12 R$ . Note the contrast with the coefficient $\tfrac12$ inside the definition, which stays $\tfrac12$ in every dimension — the divergence-free requirement that fixes it does not know about $n$ . Conflating the two factors is a common slip in higher-dimensional work.

Where does the coupling constant 8πG/c⁴ come from?

It is fixed entirely by demanding agreement with Newtonian gravity in the weak, slow, static limit. In that limit the $00$ -component of the trace-reversed field equation reduces to Poisson's equation $\nabla^{2}\Phi=4\pi G\rho$ , and matching the two determines the proportionality between $G_{\mu\nu}$ and $T_{\mu\nu}$ to be $\kappa=8\pi G/c^{4}$ . Numerically $\kappa\approx2.08\times10^{-43}\ \text{s}^{2}\text{kg}^{-1}\text{m}^{-1}$ , an extraordinarily small number. That tininess is why gravity seems weak: it takes a planetary or stellar mass to produce curvature large enough to detect, and why bending starlight past the Sun deflects it by only 1.75 arcseconds. The factor $8\pi$ rather than a bare $4\pi$ traces to the trace-reversal in the $00$ -equation.

Is the Einstein tensor derivable from an action, or is it postulated?

It is derivable. Varying the Einstein–Hilbert action $S=\tfrac{c^{4}}{16\pi G}\int\sqrt{-g}\,R\,\mathrm{d}^{4}x$ with respect to the inverse metric $g^{\mu\nu}$ yields exactly $G_{\mu\nu}=R_{\mu\nu}-\tfrac12 g_{\mu\nu}R$ , with the variation of the Ricci tensor contributing only a boundary term. This is the route Hilbert took in 1915. The variational origin gives the divergence-free property a second, independent explanation: because the action is invariant under diffeomorphisms, Noether's second theorem guarantees that the field equations obey $\nabla^{\mu}G_{\mu\nu}=0$ off-shell. So the same $\tfrac12$ that the Bianchi identity forces also drops out of the action automatically — two roads to the identical tensor.