Metric Perturbation and Trace-Reversed Field

Nearly everything we can compute in general relativity by hand lives in the weak-field corner, where spacetime is almost flat and the full nonlinear machinery of curved geometry collapses into something a physicist trained on electromagnetism can read at a glance. The move that opens that corner is deceptively small: write the metric as a fixed Minkowski background plus a tiny correction, $g_{\mu\nu}=\eta_{\mu\nu}+h_{\mu\nu}$ with $|h_{\mu\nu}|\ll 1$ , and treat the symmetric tensor $h_{\mu\nu}$ as a field propagating on flat spacetime. Every geometric object — Christoffel symbols, Riemann, Ricci, and Einstein tensors — is then expanded in powers of $h$ and truncated at first order, converting Einstein's ten coupled nonlinear equations into a linear problem. But linear does not yet mean simple. The linearized Einstein tensor, written straight in terms of $h_{\mu\nu}$ , is a thicket of second derivatives entangled with the trace $h\equiv\eta^{\mu\nu}h_{\mu\nu}$ . This article is about the single algebraic manoeuvre that clears the thicket: the trace-reversed perturbation $\bar h_{\mu\nu}\equiv h_{\mu\nu}-\tfrac12\eta_{\mu\nu}h$ . Because the Einstein tensor is itself built from the Ricci tensor by a trace-reversal, pre-reversing the field undoes that operation, the trace terms cancel in pairs, and — after one gauge choice — the field equation becomes a flat-space wave equation, $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ . From this one line flow the gravitational-wave equation, the transverse-traceless gauge, the two polarizations, the quadrupole formula, and the whole modern industry of gravitational-wave astronomy. The trace-reverse is bookkeeping — but it is the bookkeeping that makes the rest possible.

Key takeaways

Linearized gravity writes $g_{\mu\nu}=\eta_{\mu\nu}+h_{\mu\nu}$ with $|h_{\mu\nu}|\ll1$ and treats the symmetric perturbation $h_{\mu\nu}$ as a spin-2 field on flat Minkowski spacetime, with indices raised and lowered by $\eta$ .
The trace-reversed field $\bar h_{\mu\nu}=h_{\mu\nu}-\tfrac12\eta_{\mu\nu}h$ is engineered to absorb the trace terms that clutter the linearized Einstein tensor; the map is an involution, so $h_{\mu\nu}=\bar h_{\mu\nu}-\tfrac12\eta_{\mu\nu}\bar h$ .
In four dimensions the trace is exactly reversed, $\bar h=-h$ (from $\eta^{\mu\nu}\eta_{\mu\nu}=4$ ), which is the literal origin of the name.
Imposing the Lorenz gauge $\partial^\mu\bar h_{\mu\nu}=0$ collapses the field equation to the flat-space wave equation $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ , whose retarded solution is $\bar h_{\mu\nu}=4G\!\int T_{\mu\nu}/|\mathbf x-\mathbf x'|\,\mathrm d^3x'$ .
The Newtonian limit fixes the coupling: a static dust source gives $\nabla^2\Phi=4\pi G\rho$ with $h_{00}=-2\Phi$ , and the metric $\mathrm ds^2=-(1+2\Phi)\mathrm dt^2+(1-2\Phi)\delta_{ij}\mathrm dx^i\mathrm dx^j$ .
The strain is genuinely tiny: $|h|=2GM/rc^2$ is $4\times10^{-6}$ at the Sun's surface and $10^{-21}$ for a wave at Earth, so the linear theory is excellent except at compact-object surfaces.
Key idea: Define $\bar h_{\mu\nu}=h_{\mu\nu}-\tfrac12\eta_{\mu\nu}h$ . Because the linearized Einstein tensor is the trace-reverse of the linearized Ricci tensor, this one substitution plus Lorenz gauge turns Einstein's equations into $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ .

The core insight

Define $\bar h_{\mu\nu}=h_{\mu\nu}-\tfrac12\eta_{\mu\nu}h$ . Because the linearized Einstein tensor is the trace-reverse of the linearized Ricci tensor, this single substitution turns the messy linearized field equation into $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ once the Lorenz gauge $\partial^\mu\bar h_{\mu\nu}=0$ is imposed.

Where this lives today

The trace-reversed field equation is the computational backbone of gravitational-wave astronomy. Every waveform template that LIGO, Virgo, and KAGRA match against their data begins as a solution of $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ : the post-Newtonian expansion modelling a black-hole inspiral iterates exactly this linear equation with higher-order source corrections, and the transverse-traceless projection of $\bar h_{\mu\nu}$ becomes the measured strain $h(t)\sim10^{-21}$ that stretches a 4-km interferometer arm by less than a thousandth of a proton width. The same formalism underlies pulsar-timing arrays like NANOGrav, whose 2023 evidence for a nanohertz background is read out as timing residuals with an angular pattern (the Hellings–Downs curve) fingerprinting the spin-2 character the trace-reverse exposes. Beyond radiation, the linearized metric $-(1+2\Phi)\mathrm dt^2+(1-2\Phi)\mathrm d\mathbf x^2$ is the working metric of geodesy: its factor-of-two spatial curvature makes light bending and Shapiro delay twice their naive Newtonian values, checked to parts in $10^5$ . And the open frontier — the LISA interferometer of the 2030s — will chase millihertz waves from massive black-hole mergers, every prediction still traceable to the one-line wave equation the trace-reverse makes possible.

The physical picture

Picture spacetime as a flat sheet — the Minkowski background $\eta_{\mu\nu}$ — with a faint, moving wrinkle laid on top. The wrinkle is $h_{\mu\nu}$ , a symmetric rank-2 tensor with ten independent components. We ask only that in some nearly-inertial coordinate system its components stay far below unity, $|h_{\mu\nu}|\ll1$ , so that products $h^2$ and higher are negligible. Under that single assumption gravity stops being geometry and becomes field theory: $h_{\mu\nu}$ is a field living on flat spacetime, its indices raised and lowered with $\eta$ rather than $g$ , behaving like a spin-2 cousin of the electromagnetic potential $A_\mu$ . The first figure shows this decomposition — the full curved metric resolved into an unchanging flat floor plus a small tensor perturbation.

Figure

The metric splits as g = η + h — a fixed flat Minkowski background plus a small symmetric perturbation field treated as living on flat spacetime, valid whenever the perturbation stays far below 1

Why does the trace-reversed field, and not $h_{\mu\nu}$ itself, become the natural variable? The clue is a shape match. The Einstein tensor is manufactured from the Ricci tensor by subtracting half its trace, $G_{\mu\nu}=R_{\mu\nu}-\tfrac12\eta_{\mu\nu}R$ — and that "subtract half the trace" is exactly the trace-reverse operation. So if we trace-reverse $h_{\mu\nu}$ first, the two reversals cancel and the trace clutter evaporates. The second figure makes the payoff quantitative: it shows how small $|h_{\mu\nu}|$ actually is across the situations where linearized gravity is used, and where the approximation breaks.

Figure

Strain magnitude h = 2GM/rc² across gravitational settings — a neutron-star surface reaches h ≈ 0.4 where linear theory fails, the Sun's surface gives 4.2×10⁻⁶, Earth's 1.4×10⁻⁹, and a passing wave from GW150914 only 1×10⁻²¹, so the weak-field condition holds everywhere but compact-object surfaces

The core equations

The starting point is the split of the metric into background plus perturbation — less an approximation than a choice of variables: we commit to a coordinate system in which the metric is manifestly close to Minkowski, and name the difference $h_{\mu\nu}$ .

g_{\mu\nu}=\eta_{\mu\nu}+h_{\mu\nu},\qquad |h_{\mu\nu}|\ll 1 \qquad (1)

where: $g_{\mu\nu}$ — full physical spacetime metric; $\eta_{\mu\nu}$ — flat Minkowski metric, signature $(-,+,+,+)$ ; $h_{\mu\nu}$ — small symmetric perturbation, ten independent components. Reading: the geometry is a fixed flat floor plus a small tensor field, keeping only terms linear in that field. The inverse metric follows by demanding $g_{\mu\nu}g^{\nu\rho}=\delta_\mu{}^\rho$ to first order; the perturbation flips sign, and — crucially — its indices are raised by the flat metric, so that $h_{\mu\nu}$ is genuinely a field on Minkowski space.

g^{\mu\nu}=\eta^{\mu\nu}-h^{\mu\nu}+O(h^2),\qquad h^{\mu\nu}=\eta^{\mu\alpha}\eta^{\nu\beta}h_{\alpha\beta} \qquad (2)

where: $g^{\mu\nu}$ — inverse physical metric; $h^{\mu\nu}$ — perturbation with indices raised by $\eta$ . Reading: to lowest order the inverse of "identity plus small" is "identity minus small". Two contractions organize the algebra that follows: the scalar trace $h$ , and the trace-reversed field $\bar h_{\mu\nu}$ built from it. The definition of $\bar h_{\mu\nu}$ copies the Ricci-to-Einstein recipe — subtract half the trace — so that later it cancels against it.

h\equiv\eta^{\mu\nu}h_{\mu\nu}=h^\mu{}_\mu,\qquad \bar h_{\mu\nu}\equiv h_{\mu\nu}-\tfrac12\,\eta_{\mu\nu}\,h \qquad (3)

where: $h$ — scalar trace contracted with the flat metric; $\bar h_{\mu\nu}$ — trace-reversed metric perturbation. Reading: the bar denotes "trace-reversed"; it is a linear, index-preserving operation on the tensor. The name is literal. Contracting the definition with $\eta^{\mu\nu}$ and using $\eta^{\mu\nu}\eta_{\mu\nu}=\delta^\mu_\mu=4$ in four dimensions, the trace comes out with the opposite sign — and the operation squares to the identity, so it can be undone.

\bar h=\eta^{\mu\nu}\bar h_{\mu\nu}=h-\tfrac12\cdot4\cdot h=-h,\qquad h_{\mu\nu}=\bar h_{\mu\nu}-\tfrac12\,\eta_{\mu\nu}\,\bar h \qquad (4)

where: $\bar h$ — trace of the trace-reversed field, equal to $-h$ . Reading: reversing twice returns the original tensor, so $h_{\mu\nu}$ and $\bar h_{\mu\nu}$ are equivalent descriptions. The engine of linearized gravity is the first-order Ricci tensor, obtained by expanding the Christoffel symbols to $O(h)$ and keeping the linear part of $R_{\mu\nu}=\partial_\alpha\Gamma^\alpha_{\mu\nu}-\partial_\nu\Gamma^\alpha_{\mu\alpha}+O(\Gamma^2)$ — symmetric, and second order in derivatives.

R^{(1)}_{\mu\nu}=\tfrac12\left(\partial^\alpha\partial_\mu h_{\nu\alpha}+\partial^\alpha\partial_\nu h_{\mu\alpha}-\Box h_{\mu\nu}-\partial_\mu\partial_\nu h\right) \qquad (5)

where: $R^{(1)}_{\mu\nu}$ — Ricci tensor to first order in $h$ ; $\Box=\eta^{\alpha\beta}\partial_\alpha\partial_\beta$ — flat d'Alembertian. Reading: two divergence terms, a wave term, and a trace term — the last two are what the trace-reverse tames. Building $G^{(1)}_{\mu\nu}=R^{(1)}_{\mu\nu}-\tfrac12\eta_{\mu\nu}R^{(1)}$ and rewriting through $\bar h_{\mu\nu}$ , every explicit trace term cancels, leaving a strikingly clean expression: a wave operator on $\bar h_{\mu\nu}$ plus three divergence terms.

G^{(1)}_{\mu\nu}=-\tfrac12\left(\Box\bar h_{\mu\nu}+\eta_{\mu\nu}\,\partial^\alpha\partial^\beta\bar h_{\alpha\beta}-\partial^\alpha\partial_\nu\bar h_{\mu\alpha}-\partial^\alpha\partial_\mu\bar h_{\nu\alpha}\right) \qquad (6)

where: $G^{(1)}_{\mu\nu}$ — Einstein tensor to first order. Reading: no bare traces survive; the only obstruction to a pure wave equation is the three terms containing a divergence $\partial^\alpha\bar h_{\alpha\nu}$ . Those three terms are pure gauge: they all carry the divergence $\partial^\alpha\bar h_{\alpha\nu}$ , which the residual coordinate freedom of linearized gravity can be used to set to zero. That is the Lorenz (harmonic) gauge, and it delivers the master equation of the whole subject.

\partial^\mu\bar h_{\mu\nu}=0\quad\Longrightarrow\quad \Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu} \qquad (7)

where: $T_{\mu\nu}$ — stress-energy tensor of matter; $G$ — Newton's gravitational constant; $c=1$ . Reading: in Lorenz gauge each of the ten components of $\bar h_{\mu\nu}$ obeys an independent, sourced, flat-space wave equation — the exact structure of electromagnetism's $\Box A_\mu=-\mu_0 J_\mu$ .

Derivation AFrom the linearized Einstein tensor to a flat-space wave equation

The goal is to carry the linearized Einstein equations all the way to the wave equation $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ , showing at each step how the trace-reversed variable and the Lorenz gauge conspire to remove every term except the d'Alembertian. We start from the first-order Ricci tensor, contract to the Ricci scalar, assemble the Einstein tensor, re-express it through $\bar h_{\mu\nu}$ so the trace terms cancel, and impose the gauge that kills the surviving divergences.

Step 1. Contract the linearized Ricci tensor (5) with $\eta^{\mu\nu}$ to get the Ricci scalar $R^{(1)}=\partial^\mu\partial^\nu h_{\mu\nu}-\Box h$ , using $\eta^{\mu\nu}\partial^\alpha\partial_\mu h_{\nu\alpha}=\partial^\mu\partial^\nu h_{\mu\nu}$ and $\eta^{\mu\nu}\partial_\mu\partial_\nu h=\Box h$ . Why: the scalar is the ingredient the Einstein tensor subtracts, so we need it before we can build $G^{(1)}_{\mu\nu}$ .

Step 2. Form the linearized Einstein tensor $G^{(1)}_{\mu\nu}=R^{(1)}_{\mu\nu}-\tfrac12\eta_{\mu\nu}R^{(1)}$ by substituting the two expressions above. Written purely in $h_{\mu\nu}$ it still carries explicit trace terms $\partial_\mu\partial_\nu h$ and $\eta_{\mu\nu}\Box h$ , which is precisely the clutter we want gone. Why: this is the honest but ugly form — seeing it is what motivates the change of variables.

Step 3. Substitute $h_{\mu\nu}=\bar h_{\mu\nu}-\tfrac12\eta_{\mu\nu}\bar h$ (with $\bar h=-h$ ) everywhere and collect terms. The trace pieces cancel pairwise, leaving $G^{(1)}_{\mu\nu}=-\tfrac12\big(\Box\bar h_{\mu\nu}+\eta_{\mu\nu}\partial^\alpha\partial^\beta\bar h_{\alpha\beta}-\partial^\alpha\partial_\mu\bar h_{\nu\alpha}-\partial^\alpha\partial_\nu\bar h_{\mu\alpha}\big)$ , which is equation (6). Why: this is the whole point of the trace-reverse — the Ricci-to-Einstein trace subtraction is undone, so only $\bar h_{\mu\nu}$ and its divergences remain.

Step 4. Impose the Lorenz gauge $\partial^\mu\bar h_{\mu\nu}=0$ . It is always reachable: a coordinate shift $x^\mu\to x^\mu+\xi^\mu$ changes the divergence by $-\Box\xi_\nu$ , so solving $\Box\xi_\nu=\partial^\mu\bar h_{\mu\nu}$ enforces it. The three divergence terms in $G^{(1)}_{\mu\nu}$ then vanish. Why: gauge freedom is not decoration — it is exactly the tool that removes the last obstruction to a wave equation.

Step 5. What survives is $G^{(1)}_{\mu\nu}=-\tfrac12\Box\bar h_{\mu\nu}$ . Setting this equal to the source through Einstein's equation $G^{(1)}_{\mu\nu}=8\pi G\,T_{\mu\nu}$ and clearing the factor gives $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ . Why: the geometry has become a linear field sourced by matter, with the coupling constant $16\pi G$ carried along for the Newtonian limit to fix. *Verified: reproduces Carroll §7.1 eqs. (7.6)–(7.9) and Wald §4.4, in signature $(-,+,+,+)$ with $G_{\mu\nu}=8\pi G\,T_{\mu\nu}$ , $c=1$ ; the pairwise trace cancellation in Step 3 and the gauge shift in Step 4 were recomputed by hand.*

Go deeper · Derivation B · 1 minThe Newtonian limit fixes the coupling and recovers Poisson's equation

Derivation BThe Newtonian limit fixes the coupling and recovers Poisson's equation

The goal is to show that the factor $16\pi G$ is not free — it is forced by the requirement that a static, slow, weak source reproduce Newtonian gravity, $\nabla^2\Phi=4\pi G\rho$ . We take a nonrelativistic dust source, solve the static $00$ equation for $\bar h_{00}$ , invert the trace-reverse to get $h_{00}$ , read off the potential, and confirm the full weak-field metric drops out.

Step 1. For slow, cold matter the rest-energy density dominates: $T_{00}\approx\rho$ while $T_{0i}$ and $T_{ij}$ are smaller by powers of $v/c$ and $P/\rho c^2$ . So the only appreciable field-equation component is the $00$ one. Why: this is the defining physical content of "Newtonian" — mass, not momentum or pressure, sources the field.

Step 2. For a time-independent field $\Box\to\nabla^2$ , so the $00$ equation becomes $\nabla^2\bar h_{00}=-16\pi G\rho$ . Why: staticity turns the relativistic wave equation into an elliptic Poisson-type equation, the correct shape for a static potential.

Step 3. Invert the trace-reverse. With $\bar h_{00}$ the only source-driven component, the trace is $\bar h=\eta^{00}\bar h_{00}=-\bar h_{00}$ , so $h_{00}=\bar h_{00}-\tfrac12\eta_{00}\bar h=\bar h_{00}-\tfrac12(-1)(-\bar h_{00})=\tfrac12\bar h_{00}$ . Hence $\nabla^2h_{00}=-8\pi G\rho$ . Why: the physical metric perturbation, not the trace-reversed one, is what couples to a test particle's geodesic, so we must translate back.

Step 4. Identify the potential. A slow test particle obeys $\ddot x^i=-\tfrac12\partial_i h_{00}$ , matching Newton's $\ddot x^i=-\partial_i\Phi$ only if $h_{00}=-2\Phi$ . Then $\nabla^2h_{00}=-8\pi G\rho$ becomes $\nabla^2\Phi=4\pi G\rho$ — Poisson's equation, exactly. Why: this fixes $16\pi G$ ; any other coefficient gives the wrong Newtonian constant.

Step 5. Assemble the full metric. The same source also drives $h_{ij}=\tfrac12\bar h_{00}\delta_{ij}=-2\Phi\,\delta_{ij}$ , so $\mathrm ds^2=-(1+2\Phi)\mathrm dt^2+(1-2\Phi)\delta_{ij}\mathrm dx^i\mathrm dx^j$ . Why: the spatial part is what bends light by twice the Newtonian value — essential for the correct deflection and Shapiro delay. *Verified: matches MTW §18.4, Carroll §7.1 (with $\Phi=-\tfrac12h_{00}$ ), and Schutz §8.4; the numerical factor $16\pi G$ is precisely what makes $\nabla^2\Phi=4\pi G\rho$ come out, and $|h_{00}|=2GM/rc^2=4.2\times10^{-6}$ at the Sun's surface was checked numerically.*

Go deeper · Derivation C · 1 minThe retarded solution and the gateway to gravitational radiation

Derivation CThe retarded solution and the gateway to gravitational radiation

The goal is to solve $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ explicitly and show that its causal solution is a retarded integral over the source — the single formula from which the quadrupole formula and all of gravitational-wave generation descend. We use the Green's function of the flat d'Alembertian, impose outgoing boundary conditions, and read off the far-field structure.

Step 1. The retarded Green's function of $\Box$ satisfies $\Box G(x-x')=\delta^4(x-x')$ with $G=-\delta(t_{\text{ret}}-t')/4\pi|\mathbf x-\mathbf x'|$ , where $t_{\text{ret}}=t-|\mathbf x-\mathbf x'|$ . Why: this is the identical Green's function used for the electromagnetic potentials, which is why gravity in Lorenz gauge looks so much like electromagnetism.

Step 2. Convolving the source with the Green's function gives $\bar h_{\mu\nu}(t,\mathbf x)=4G\displaystyle\int\frac{T_{\mu\nu}(t_{\text{ret}},\mathbf x')}{|\mathbf x-\mathbf x'|}\,\mathrm d^3x'$ , with prefactor $16\pi G/4\pi=4G$ . Why: the field is the sum of source contributions emitted early enough for their influence, travelling at $c$ , to arrive — causality made explicit.

Step 3. Far from a slow compact source, expand $|\mathbf x-\mathbf x'|\approx r$ and use conservation $\partial^\mu T_{\mu\nu}=0$ to trade stress components for second time-derivatives of the mass distribution: $\bar h_{ij}\approx\dfrac{2G}{r}\ddot I_{ij}(t-r)$ with $I_{ij}=\int\rho\,x^ix^j\,\mathrm d^3x'$ . Why: conservation forbids monopole and dipole radiation, so the leading wave is quadrupolar — the origin of the quadrupole formula.

Step 4. Project into the transverse-traceless gauge to isolate the two physical polarizations $h_+$ and $h_\times$ ; the radiated power is the Isaacson flux $\propto\langle\dddot{Q}_{ij}\dddot{Q}^{ij}\rangle$ . Why: only the transverse-traceless part carries energy to infinity, so the projected trace-reversed field is the natural variable for luminosity. *Verified: the prefactor $4G=16\pi G/4\pi$ ; the retarded integral matches Maggiore Vol. 1 §3.1 and MTW Ch. 36; the scaling $\bar h_{ij}\sim(2G/r)\ddot I_{ij}$ is standard.*

Limiting cases

The trace-reverse and its field equation simplify sharply in the regimes that matter most.

Case	Result	Meaning
Vacuum, Lorenz gauge	$\Box\bar h_{\mu\nu}=0$	Each component obeys the massless wave equation; gravitational waves propagate at exactly the speed of light.
Traceless perturbation	$h=0\ \Rightarrow\ \bar h_{\mu\nu}=h_{\mu\nu}$	The two fields coincide — this is the transverse-traceless gauge used for radiation.
Static Newtonian source	$\nabla^2\Phi=4\pi G\rho,\ h_{00}=-2\Phi$	The wave equation reduces to Poisson's equation, fixing the coupling $16\pi G$ .
General dimension $n$	$\bar h=\left(1-\tfrac n2\right)h$	Only in $n=4$ is the trace exactly reversed; in $n=2$ the map is the identity and the involution fails.
Strong field, $\|h\|\sim1$	linear theory invalid	At a neutron-star surface $\|h\|\approx0.4$ ; the neglected $O(h^2)$ terms are no longer small and full GR is required.

Common trapThinking the trace-reverse is new physics, or that it always flips the sign

The trace-reverse is pure algebra — a linear relabelling of the same ten components, carrying no new physics; it earns its keep only by undoing the Ricci-to-Einstein trace subtraction. Two traps follow. First, the sign flip $\bar h=-h$ holds only in four dimensions: the general result is $\bar h=(1-\tfrac n2)h$ , so in $n=2$ the trace is unchanged and the map is not even an involution — never quote " $\bar h=-h$ " without the $n=4$ caveat. Second, do not confuse the flat-metric trace $h=\eta^{\mu\nu}h_{\mu\nu}$ the definition uses with $g^{\mu\nu}h_{\mu\nu}$ ; they agree only to leading order, and the whole formalism contracts with $\eta$ , not $g$ .

Worked examples

1 · The weak-field strain at the Sun's surface (confidence builder)

Estimate the size of the metric perturbation $h_{00}$ at the surface of the Sun, and confirm that the linear theory is justified. Use $GM_\odot=1.327\times10^{20}\ \mathrm{m^3\,s^{-2}}$ and $R_\odot=6.96\times10^8\ \mathrm m$ .

Step 1. The Newtonian limit gives $h_{00}=-2\Phi$ with $\Phi=-GM/r$ , so $h_{00}=2GM/(rc^2)$ at the surface — the dimensionless "compactness" of the object. The strain is set entirely by $GM/rc^2$ .

Step 2. Insert the numbers: $rc^2=R_\odot c^2=(6.96\times10^8)(8.99\times10^{16})=6.26\times10^{25}\ \mathrm{m^3\,s^{-2}}$ , so $h_{00}=2(1.327\times10^{20})/(6.26\times10^{25})=\mathbf{4.2\times10^{-6}}$ .

Step 3. Interpret. Since $|h_{00}|\approx4\times10^{-6}\ll1$ , the neglected $O(h^2)\sim10^{-11}$ terms are negligible — linearized gravity is superb here, and it is what predicts the $1.75''$ light deflection at the solar limb. The same formula gives $|h|\approx0.4$ at a neutron-star surface, where $O(h^2)$ is a $16\%$ correction and the linear theory has broken down. At the Sun: $h_{00}\approx4.2\times10^{-6}$ , deep in the weak-field regime.

2 · Trace-reversing the weak-field metric (algebra you can check by hand)

Take the static weak-field perturbation $h_{00}=-2\Phi$ , $h_{ij}=-2\Phi\,\delta_{ij}$ , $h_{0i}=0$ . Compute its trace $h$ , its trace-reversed field $\bar h_{\mu\nu}$ , and verify the involution.

Step 1. Trace: $h=\eta^{\mu\nu}h_{\mu\nu}=\eta^{00}h_{00}+\delta^{ij}h_{ij}=(-1)(-2\Phi)+(-2\Phi)(3)=2\Phi-6\Phi=\mathbf{-4\Phi}$ . The spatial part, summed over three directions, dominates and flips the sign.

Step 2. Trace-reverse the time-time component: $\bar h_{00}=h_{00}-\tfrac12\eta_{00}h=-2\Phi-\tfrac12(-1)(-4\Phi)=-2\Phi-2\Phi=\mathbf{-4\Phi}$ . The whole Newtonian field concentrates into $\bar h_{00}$ .

Step 3. Trace-reverse the spatial part: $\bar h_{ij}=h_{ij}-\tfrac12\eta_{ij}h=-2\Phi\,\delta_{ij}-\tfrac12\delta_{ij}(-4\Phi)=-2\Phi\,\delta_{ij}+2\Phi\,\delta_{ij}=\mathbf 0$ . In the trace-reversed picture the static field is purely time-time — which is exactly why the $00$ equation alone reproduces Poisson.

Step 4. Verify the involution: $\bar h=\eta^{\mu\nu}\bar h_{\mu\nu}=(-1)(-4\Phi)+0=4\Phi=-h$ , and $h_{00}=\bar h_{00}-\tfrac12\eta_{00}\bar h=-4\Phi-\tfrac12(-1)(4\Phi)=-4\Phi+2\Phi=-2\Phi$ ✓, recovering the original. $\bar h_{00}=-4\Phi$ , $\bar h_{ij}=0$ , and $\bar{\bar h}_{\mu\nu}=h_{\mu\nu}$ as required.

Physics of the portrait

The trace-reversed field is a piece of algebra, but the physics it unlocks — a linear wave theory of gravity — is the personal legacy of one man who worked it out within months of finishing general relativity itself.

Origins · 1 minAlbert Einstein and how this was discovered

Albert Einstein

1879–1955 · linearized gravity and the prediction of gravitational waves

Barely half a year after presenting the full nonlinear field equations in November 1915, Einstein asked what they predict when gravity is weak. In his 1916 paper Näherungsweise Integration der Feldgleichungen der Gravitation he wrote $g_{\mu\nu}=\eta_{\mu\nu}+h_{\mu\nu}$ , linearized, and introduced harmonic coordinates — the condition we now call the Lorenz gauge — precisely to reduce the field equations to a wave equation with the same form as electromagnetism. From it he concluded that gravity propagates as waves at the speed of light, the first prediction of gravitational radiation. His 1916 analysis had an error in the energy flux, which he fixed in a celebrated 1918 paper deriving the quadrupole formula. Einstein stayed ambivalent about whether the waves were real: in 1936, with Nathan Rosen, he submitted a paper arguing they did not exist, only for a referee to show the apparent singularity was a coordinate artifact — a plane wave in disguise. Their reality was settled by the Bondi–Pirani–Feynman "sticky bead" argument in the late 1950s, and finally, a century after 1916, by LIGO's detection of GW150914 in 2015. The trace-reversed field, harmonic gauge, and wave equation of this article are, line for line, the tools Einstein forged in that first weak-field paper.

History

1915

— Einstein completes the nonlinear field equations of general relativity

1916

— Einstein linearizes the equations, adopts harmonic coordinates, and predicts gravitational waves

1918

— Einstein derives the quadrupole formula for radiated power, correcting his earlier flux

1936

— Einstein and Rosen doubt the waves' reality; the referee shows it is a coordinate artifact

2015

— LIGO records GW150914, the first direct detection of a gravitational wave

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §18.1–18.2 and Ch. 35–36 · Wald, General Relativity (1984), §4.4 · Carroll, Spacetime and Geometry (2004), Ch. 7 · Schutz, A First Course in General Relativity (2nd ed., 2009), Ch. 8 · Hartle, Gravity (2003), Ch. 21 · Maggiore, Gravitational Waves Vol. 1 (2008), §1.1 and Ch. 3.

Provenance. Definitions and conventions follow Carroll (Ch. 7), Wald (§4.4), MTW (§18), Schutz (Ch. 8), and Maggiore (Vol. 1), all in signature $(-,+,+,+)$ with $G_{\mu\nu}=8\pi G\,T_{\mu\nu}$ , $c=1$ . A verifier should recompute: (i) $\eta^{\mu\nu}\eta_{\mu\nu}=4$ giving $\bar h=-h$ ; (ii) the involution $\bar{\bar h}_{\mu\nu}=h_{\mu\nu}$ ; (iii) the pairwise cancellation of trace terms turning $G^{(1)}_{\mu\nu}$ into equation (6); (iv) the Newtonian reduction $\nabla^2\Phi=4\pi G\rho$ with $h_{00}=-2\Phi$ and coupling $16\pi G$ ; (v) the retarded prefactor $4G=16\pi G/4\pi$ ; (vi) the anchor strains $2GM/rc^2$ equal to $4.2\times10^{-6}$ (Sun), $1.4\times10^{-9}$ (Earth), $0.41$ (neutron star). All numerical constants are standard.

Prerequisites. Minkowski Spacetime and the Flat Metric · Einstein Field Equations · Riemann and Ricci Tensors. Leads to. Linearized Einstein Equations · Lorenz (Harmonic) Gauge · Gravitational Wave Equation · Quadrupole Formula.

Ask the tutor: Want to see how the two surviving polarizations of $\bar h_{\mu\nu}$ become the plus-and-cross stretching pattern a detector actually feels?

Why bother with the trace-reverse instead of working with h_μν directly?

Because the Einstein tensor is itself the trace-reverse of the Ricci tensor: $G_{\mu\nu}=R_{\mu\nu}-\tfrac12\eta_{\mu\nu}R$ . Written in $h_{\mu\nu}$ , the linearized field equation carries explicit trace terms — $\partial_\mu\partial_\nu h$ and $\eta_{\mu\nu}\Box h$ — that clutter every calculation. Trace-reversing $h_{\mu\nu}$ first undoes the trace subtraction built into $G_{\mu\nu}$ , so those terms cancel and the equation collapses to a clean wave operator on a single object, $\Box\bar h_{\mu\nu}$ . No new physics — just the variable in which the equations are simplest.

Is the sign flip h̄ = −h a general fact or special to four dimensions?

It is special to four dimensions. Contracting the definition gives $\bar h=(1-\tfrac n2)h$ in $n$ spacetime dimensions, because $\eta^{\mu\nu}\eta_{\mu\nu}=n$ . Only at $n=4$ does the coefficient become $1-2=-1$ , so the trace is exactly negated. At $n=2$ the coefficient is zero, so $\bar h=0$ regardless of $h$ and the map is degenerate — not even an involution. In higher dimensions it stays linear and invertible, but the tidy " $\bar h=-h$ " is a four-dimensional accident, so always attach the $n=4$ caveat.

How does the Lorenz gauge condition ∂^μ h̄_μν = 0 relate to electromagnetism?

It is the exact gravitational analogue of the electromagnetic Lorenz gauge $\partial^\mu A_\mu=0$ , named for the same physicist, Ludvig Lorenz (not the more famous Hendrik Lorentz). There that condition turns Maxwell's equations into $\Box A_\mu=-\mu_0 J_\mu$ ; here $\partial^\mu\bar h_{\mu\nu}=0$ turns Einstein's equations into $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ . In both cases the gauge is reachable because of an underlying symmetry — electromagnetic gauge invariance there, coordinate (diffeomorphism) invariance here — and in both a residual freedom survives, which in gravity is what lets you further reach the transverse-traceless gauge.

Why does the coupling come out as 16πG and not 8πG?

The Einstein equation itself carries $8\pi G$ : $G_{\mu\nu}=8\pi G\,T_{\mu\nu}$ . But the linearized Einstein tensor in Lorenz gauge is $G^{(1)}_{\mu\nu}=-\tfrac12\Box\bar h_{\mu\nu}$ — note the $-\tfrac12$ . Setting $-\tfrac12\Box\bar h_{\mu\nu}=8\pi G\,T_{\mu\nu}$ and multiplying by $-2$ gives $\Box\bar h_{\mu\nu}=-16\pi G\,T_{\mu\nu}$ . The extra factor of two is not arbitrary: only $16\pi G$ makes the static $00$ equation reduce to Poisson's $\nabla^2\Phi=4\pi G\rho$ after you invert the trace-reverse to recover $h_{00}=-2\Phi$ . Any other coefficient gives the wrong strength of gravity.

The linear field h_μν has ten components — why do only two physical ones survive?

Start with ten independent components of the symmetric $h_{\mu\nu}$ . The Lorenz gauge is four conditions ( $\partial^\mu\bar h_{\mu\nu}=0$ , one per $\nu$ ), leaving six. But it does not exhaust the freedom: any coordinate shift $\xi^\mu$ with $\Box\xi^\mu=0$ preserves the gauge, and those four functions eliminate four more components. Ten minus four minus four is two — the physical polarizations, plus and cross, that a gravitational wave carries. This counting is the linearized statement that the graviton is a massless spin-2 particle with helicity $\pm2$ , and why gravitational waves have exactly two independent strain patterns.

Does the perturbation h_μν really live on flat spacetime, or is that a fiction?

It is a well-defined approximation, not a fiction, but its domain is exactly $|h_{\mu\nu}|\ll1$ . There you genuinely treat $h_{\mu\nu}$ as a tensor field on a fixed Minkowski background: indices raised with $\eta$ , derivatives ordinary partials, the field obeying a flat-space wave equation. The picture is self-consistent because the dropped corrections are $O(h^2)\sim10^{-11}$ at the Sun. It fails where $|h|$ approaches unity — near black-hole horizons or neutron-star surfaces — where the "background plus field" split loses meaning and the full nonlinear equations on a curved manifold are required.