The Energy Conditions

Einstein's field equations $G_{\mu\nu}=8\pi G\,T_{\mu\nu}$ are, taken alone, empty of physics. Hand me any smooth metric $g_{\mu\nu}$ and I will hand you back a stress-energy tensor that sources it — just compute the left-hand side and divide by $8\pi G$ . Wormholes, warp bubbles, time machines all solve Einstein's equations for some $T_{\mu\nu}$ . What separates a black hole from science fiction is therefore not the geometry but the matter — and the only way general relativity gets to call matter "unreasonable" without committing to a specific field theory is to impose inequalities on $T_{\mu\nu}$ that every observer must agree on. Those inequalities are the energy conditions. Each one contracts the stress-energy tensor with causal vectors — a timelike four-velocity $u^\mu$ (a possible observer) or a null vector $k^\mu$ (a light ray) — and demands the result be non-negative. The weak condition says no observer ever measures negative energy density; the null condition is its ultra-relativistic limit; the dominant condition adds that energy never flows faster than light; and the strong condition, secretly not about energy at all, controls the trace-reversed source that appears in the Ricci tensor. For the perfect fluid they collapse to schoolroom inequalities between density $\rho$ and pressure $p$ . The reason any of this matters is geometric. Through the field equations, an inequality on $T_{\mu\nu}$ becomes an inequality on $R_{\mu\nu}$ — exactly the term in the Raychaudhuri equation that focuses bundles of geodesics. That single translation is the engine of the singularity theorems, the black-hole area theorem, and topological censorship. Yet modern physics treats the conditions as provisional: a positive cosmological constant violates the strong one, and quantum fields violate even the null one locally, forcing us toward averaged and quantum successors.

Key takeaways

Energy conditions are coordinate-invariant inequalities on $T_{\mu\nu}$ that encode "physically reasonable matter" without naming any particular field theory — you check them by contracting $T_{\mu\nu}$ with observer four-velocities $u^\mu$ and light rays $k^\mu$ .
The four classical conditions form a hierarchy: DEC ⟹ WEC ⟹ NEC, and SEC ⟹ NEC, while SEC and WEC are logically independent — a positive $\Lambda$ obeys WEC but breaks SEC.
For a perfect fluid they are elementary: NEC ⟺ $\rho+p\ge0$ ; WEC adds $\rho\ge0$ ; DEC adds $\rho\ge|p|$ ; SEC needs $\rho+p\ge0$ and $\rho+3p\ge0$ .
Via the trace-reversed Einstein equation $R_{\mu\nu}=8\pi G(T_{\mu\nu}-\tfrac12 Tg_{\mu\nu})$ , NEC and SEC become $R_{\mu\nu}k^\mu k^\nu\ge0$ and $R_{\mu\nu}u^\mu u^\nu\ge0$ — the focusing terms that drive the singularity and area theorems.
They are hypotheses, not laws: dark energy ( $w\approx-1$ ) violates SEC, and the Casimir vacuum gives $u\approx-4.3\times10^{-4}$ J/m³ at 1 μm plate separation, violating NEC pointwise.
Key idea: an energy condition is an inequality " $T_{\mu\nu}$ contracted with causal vectors is non-negative," chosen so that, through Einstein's equations, it forces the Ricci curvature to focus geodesics — turning a statement about matter into a theorem about spacetime geometry.

The core insight

An energy condition is an inequality of the form " $T_{\mu\nu}$ contracted with causal vectors is non-negative," chosen so that, through Einstein's equations, it forces the Ricci curvature to focus geodesics — turning a statement about matter into a theorem about spacetime geometry.

Where this lives today

The energy conditions are quietly load-bearing across modern gravitational physics. Every image of a black hole — LIGO–Virgo's ringdown waveforms since 2015, the Event Horizon Telescope's shadow of M87\ in 2019 and Sgr A\ in 2022 — is interpreted through singularity theorems whose only matter input is the NEC, so each detection is indirect evidence that collapsing matter respects it. In cosmology the game has flipped: DESI and Planck now measure the dark-energy equation of state to a few percent ( $w=-1.03\pm0.03$ ), so quantifying cosmic acceleration is, at bottom, a measurement of how badly the strong energy condition is violated. Precision Casimir-force experiments confirm the negative vacuum energy that breaks the pointwise NEC. The live frontier is quantum: Bousso, Fisher, Leichenauer and Wall's 2016 quantum null energy condition bounds $\langle T_{\mu\nu}k^\mu k^\nu\rangle$ by the second derivative of entanglement entropy and has been proven in quantum field theory, while Gao, Jafferis and Wall's 2017 traversable wormhole works by engineering an ANEC-respecting sliver of negative energy across the throat — turning the old "wormholes need exotic matter" verdict into a concrete quantum protocol. The classical inequalities have become the boundary along which classical and quantum gravity negotiate.

The physical picture

Picture the space of all possible matter as the plane of equations of state: energy density $\rho$ along one axis, pressure $p$ along the other. Ordinary matter — dust, gas, radiation — lives comfortably in the right half, $\rho>0$ , with modest pressure. The energy conditions are straight lines through the origin that fence off "reasonable" regions of this plane. Cross a fence and something exotic happens: gravity stops attracting, or energy starts outrunning light, or the vacuum acquires negative density. The first figure draws those fences and drops the standard fluids onto them, so you can read off at a glance which condition each kind of matter respects.

Figure

The plane of equations of state at fixed density: the lines ρ+p=0, ρ+3p=0 and p=±ρ fence off where each energy condition holds. Dust and radiation satisfy all four; vacuum energy p=−ρ sits exactly on the NEC edge and violates only the SEC; phantom matter with ρ+p<0 falls below every fence and violates all four

Now translate matter into geometry. A tight bundle of light rays — the generators of a horizon, say, or the light-front of a collapsing star — is described by its expansion $\theta$ , the fractional rate at which its cross-sectional area grows. The Raychaudhuri equation says $\theta$ is driven downward by shear and by the curvature term $R_{\mu\nu}k^\mu k^\nu$ , which the field equations tie directly to $T_{\mu\nu}k^\mu k^\nu$ — the null energy condition. When that term is non-negative, a bundle that starts converging ( $\theta_0<0$ ) is doomed: it must reach a caustic, a point where neighbouring rays cross and $\theta\to-\infty$ , within a bounded stretch of affine parameter. The second figure is that focusing theorem drawn as a picture, with the real bound attached.

Figure

A converging null congruence with initial expansion θ₀ negative: because the null energy condition makes R_μν kᵘ kᵛ ≥ 0, the Raychaudhuri equation forces the rays to a caustic within affine parameter Δλ of at most 2 divided by the magnitude of θ₀, no matter how small the shear — the focusing engine behind the singularity and area theorems

The core equations

The four classical conditions all have the same skeleton — contract $T_{\mu\nu}$ with a causal vector and demand the answer be non-negative — but they differ in which vector and which combination of $T_{\mu\nu}$ they use. We list them from weakest to strongest, then trace-reverse the field equations to expose why two of them are really statements about curvature, and finally record the Raychaudhuri equation and its averaged quantum successor. Units are $c=1$ throughout; the metric signature is $(-,+,+,+)$ . The null energy condition (NEC) is the weakest and the one every other condition implies. It asks only that light rays never see negative energy.

T_{\mu\nu}\,k^{\mu}k^{\nu}\ \ge\ 0\quad\text{for all null }k^{\mu} \qquad (1)

where: $T_{\mu\nu}$ — stress-energy tensor (energy density, momentum density, and stress); $k^{\mu}$ — arbitrary future-directed null vector, $g_{\mu\nu}k^\mu k^\nu=0$ . For a perfect fluid this reads $\rho+p\ge0$ : the density and pressure may each be negative, but not their sum. The weak energy condition (WEC) upgrades the test vector from null to timelike — from a light ray to a real observer — and demands that observer measure non-negative energy density.

T_{\mu\nu}\,u^{\mu}u^{\nu}\ \ge\ 0\quad\text{for all timelike }u^{\mu} \qquad (2)

where: $u^{\mu}$ — arbitrary future-directed unit timelike vector, $g_{\mu\nu}u^\mu u^\nu=-1$ , the four-velocity of a possible observer. Reading: no matter how fast you fly through the matter, the energy density you clock is $\ge0$ . For a fluid, $\rho\ge0$ and $\rho+p\ge0$ , so WEC ⟹ NEC. The dominant energy condition (DEC) keeps WEC's non-negative density but adds a causality clause: the energy-momentum an observer measures flowing past must itself be a causal (timelike or null) vector, i.e. energy cannot stream faster than light.

-T^{\mu}{}_{\nu}\,u^{\nu}\ \text{is future-directed causal, and}\ \ T_{\mu\nu}u^\mu u^\nu\ge0 \qquad (3)

where: $-T^{\mu}{}_{\nu}u^{\nu}$ — the energy-momentum flux four-vector measured by observer $u^\nu$ ; "causal" means $g_{\mu\nu}V^\mu V^\nu\le0$ . For a fluid this is $\rho\ge|p|$ , the tightest of the four fences, and DEC ⟹ WEC ⟹ NEC. The strong energy condition (SEC) is the odd one out. It is not about energy density at all: it constrains the trace-reversed combination of $T_{\mu\nu}$ , precisely the object that will turn out to source the Ricci tensor.

\left(T_{\mu\nu}-\tfrac{1}{2}T\,g_{\mu\nu}\right)u^{\mu}u^{\nu}\ \ge\ 0\quad\text{for all timelike }u^{\mu} \qquad (4)

where: $T=g^{\mu\nu}T_{\mu\nu}=T^\mu{}_\mu$ — the trace of the stress-energy tensor; $T_{\mu\nu}-\tfrac12 Tg_{\mu\nu}$ — the trace-reversed source. For a fluid this becomes $\rho+p\ge0$ and $\rho+3p\ge0$ . Note the tension: large negative pressure ( $\rho+3p<0$ ) breaks SEC while leaving WEC intact. Trace-reversing the field equations is what converts the matter inequalities into geometry. Take the trace of $G_{\mu\nu}=8\pi G\,T_{\mu\nu}$ to get $R=-8\pi G\,T$ , substitute back, and the Ricci tensor stands alone.

R_{\mu\nu}=8\pi G\left(T_{\mu\nu}-\tfrac12 T g_{\mu\nu}\right)\ \Longrightarrow\ R_{\mu\nu}u^\mu u^\nu\ge0\ (\text{SEC}),\quad R_{\mu\nu}k^\mu k^\nu\ge0\ (\text{NEC}) \qquad (5)

where: $R_{\mu\nu}$ — Ricci tensor, the contraction of Riemann that measures how a geodesic ball's volume accelerates; the null trace term drops because $g_{\mu\nu}k^\mu k^\nu=0$ . Reading: SEC is exactly "timelike geodesics focus," NEC is exactly "null geodesics focus." Gravity, sourced by reasonable matter, always attracts. The Raychaudhuri equation is where that focusing lives. For a null congruence with affine parameter $\lambda$ in four dimensions,

\frac{\mathrm{d}\theta}{\mathrm{d}\lambda}=-\tfrac12\theta^{2}-\sigma_{\mu\nu}\sigma^{\mu\nu}+\omega_{\mu\nu}\omega^{\mu\nu}-R_{\mu\nu}k^{\mu}k^{\nu} \qquad (6)

where: $\theta$ — expansion (fractional rate of change of cross-sectional area); $\sigma_{\mu\nu}$ — shear (trace-free, $\sigma_{\mu\nu}\sigma^{\mu\nu}\ge0$ ); $\omega_{\mu\nu}$ — twist, which vanishes for a hypersurface-orthogonal congruence; the coefficient $\tfrac12$ is the $D=4$ null value $1/(n-2)$ . Reading: every term on the right except the twist pushes $\theta$ down once NEC makes the last term non-negative. Finally, the averaged null energy condition (ANEC) is the survivor when the pointwise NEC fails, as it does for quantum fields. It asks only that the integral of the energy density along a complete light ray be non-negative.

\int_{-\infty}^{\infty} T_{\mu\nu}\,k^{\mu}k^{\nu}\,\mathrm{d}\lambda\ \ge\ 0 \qquad (7)

where: the integral runs along an entire affinely-parametrized null geodesic. Reading: negative energy is allowed locally — quantum energy inequalities bound how much and for how long — but it must be paid back along the ray, and that suffices for averaged versions of the global theorems.

Derivation AReducing the four conditions to inequalities on density and pressure

The goal is to turn four abstract tensor inequalities into the four schoolroom relations between $\rho$ and $p$ that you actually use — NEC $\rho+p\ge0$ , WEC additionally $\rho\ge0$ , DEC $\rho\ge|p|$ , SEC $\rho+p\ge0$ with $\rho+3p\ge0$ — by the single trick of contracting the perfect-fluid stress tensor with a boosted observer and then taking the ultra-relativistic limit to recover the null case. We work in the fluid rest frame with $T^\mu{}_\nu=\mathrm{diag}(-\rho,p,p,p)$ and $\eta_{\mu\nu}=\mathrm{diag}(-1,1,1,1)$ , so that lowering an index gives $T_{\mu\nu}=\mathrm{diag}(\rho,p,p,p)$ .

Step 1. Take a generic boosted observer $u^\mu=\gamma(1,v,0,0)$ with $\gamma=(1-v^2)^{-1/2}$ , moving with speed $v<1$ through the fluid, and contract: $T_{\mu\nu}u^\mu u^\nu=\rho(u^0)^2+p(u^1)^2=\gamma^2(\rho+v^2p)$ . Why: WEC must hold for every observer, so we cannot pick a convenient frame — we must carry the boost parameter $v$ and demand non-negativity for all of it.

Step 2. Evaluate the two extremes. At $v=0$ the bracket is $\rho\ge0$ ; as $v\to1$ the physically meaningful combination $\gamma^{-2}T_{\mu\nu}u^\mu u^\nu\to\rho+p$ , which must also be $\ge0$ . Together, WEC $\Leftrightarrow\rho\ge0$ and $\rho+p\ge0$ . Why: a quadratic in $v^2$ that is non-negative on $[0,1)$ is pinned by its endpoints, so the whole family of observers reduces to just two inequalities.

Step 3. The null case is literally the $v\to1$ limit stripped of the diverging $\gamma^2$ : $T_{\mu\nu}k^\mu k^\nu=\rho+p$ for $k^\mu=(1,1,0,0)$ . Hence NEC $\Leftrightarrow\rho+p\ge0$ , and since it is the WEC endpoint, WEC ⟹ NEC automatically. Why: a null vector is a timelike vector boosted to the speed of light, so NEC is not an independent axiom but the boundary of WEC.

Step 4. For SEC, contract the trace-reversed source. Using $g_{\mu\nu}u^\mu u^\nu=-1$ , the $-\tfrac12 Tg_{\mu\nu}$ piece contributes $+\tfrac12 T$ , so $(T_{\mu\nu}-\tfrac12 Tg_{\mu\nu})u^\mu u^\nu=\gamma^2(\rho+v^2p)+\tfrac12(-\rho+3p)$ with $T=-\rho+3p$ . At $v=0$ this is $\tfrac12(\rho+3p)$ , and as $v\to1$ it is dominated by $\gamma^2(\rho+p)$ . Why: the trace term is the entire difference between SEC and WEC — it is what lets a positive $\Lambda$ ( $p=-\rho$ , so $\rho+3p=-2\rho<0$ ) satisfy WEC yet fail SEC.

Step 5. For DEC, form the measured flux $S^\mu=-T^\mu{}_\nu u^\nu$ . In the rest frame $S^\mu=(\rho,0,0,0)$ with $S_\mu S^\mu=-\rho^2\le0$ , future-directed iff $\rho\ge0$ ; carrying a general boost and requiring $S^\mu$ to stay causal for every observer adds $|p|\le\rho$ . Why: demanding the energy current never leave its own light cone is a stronger statement than mere non-negative density, which is why DEC sits at the top of the hierarchy.

Go deeper · Derivation B · 1 minTrace-reversing Einstein's equations to expose gravitational focusing

Derivation BTrace-reversing Einstein's equations to expose gravitational focusing

The goal is to show that the strong and null energy conditions are not really conditions on energy but disguised statements that gravity focuses geodesics, by algebraically inverting $G_{\mu\nu}=8\pi G\,T_{\mu\nu}$ to solve for the Ricci tensor $R_{\mu\nu}$ — the object that appears in the Raychaudhuri equation — and then contracting the result with a timelike and a null vector. This is the pivot on which the entire subject turns: it is the exact point where a hypothesis about matter becomes a theorem about spacetime.

Step 1. Start from $G_{\mu\nu}=R_{\mu\nu}-\tfrac12 Rg_{\mu\nu}=8\pi G\,T_{\mu\nu}$ and take the trace with $g^{\mu\nu}$ . Since $g^{\mu\nu}g_{\mu\nu}=4$ , we get $R-2R=8\pi G\,T$ , that is $R=-8\pi G\,T$ . Why: the Ricci scalar $R$ and the matter trace $T$ are locked together, so we can eliminate $R$ entirely and isolate the tensor $R_{\mu\nu}$ .

Step 2. Substitute $R=-8\pi G\,T$ back into the field equations and solve: $R_{\mu\nu}=8\pi G\big(T_{\mu\nu}-\tfrac12 Tg_{\mu\nu}\big)$ . The right-hand side is precisely the SEC combination from equation (4). Why: this "trace-reversed" form is the one that governs geodesic focusing, because it is $R_{\mu\nu}$ , not $G_{\mu\nu}$ , that appears in Raychaudhuri's equation.

Step 3. Contract with a timelike unit vector $u^\mu$ : $R_{\mu\nu}u^\mu u^\nu=8\pi G\,(T_{\mu\nu}-\tfrac12 Tg_{\mu\nu})u^\mu u^\nu$ . The SEC ( $\ge0$ on the matter side) is therefore identical to $R_{\mu\nu}u^\mu u^\nu\ge0$ . Why: this is the timelike focusing condition — the term that makes a ball of freely-falling dust accelerate inward, i.e. the statement that gravity is attractive.

Step 4. Contract instead with a null vector $k^\mu$ . The trace term drops because $g_{\mu\nu}k^\mu k^\nu=0$ , leaving $R_{\mu\nu}k^\mu k^\nu=8\pi G\,T_{\mu\nu}k^\mu k^\nu$ , so the NEC is identical to $R_{\mu\nu}k^\mu k^\nu\ge0$ . Why: the null case is cleaner than the timelike one — no trace subtlety survives — which is exactly why the area theorem, built on light-like horizon generators, needs only the weakest condition. *Why: the vanishing of the null trace term is the reason NEC (not WEC or SEC) governs horizon generators.*

Go deeper · Derivation C · 1 minFocusing a null congruence to a caustic

Derivation CFocusing a null congruence to a caustic

The goal is to prove the sharp focusing theorem drawn in the second figure: any hypersurface-orthogonal null congruence that starts converging, $\theta_0<0$ , is driven to a caustic ( $\theta\to-\infty$ ) within affine parameter $\Delta\lambda\le2/|\theta_0|$ , using nothing but the null energy condition and the Raychaudhuri equation. This is the geometric heart of Penrose's 1965 singularity theorem and Hawking's area theorem, and the derivation is a small masterpiece of turning an inequality into an inevitability.

Step 1. Begin with the null Raychaudhuri equation (6). For a hypersurface-orthogonal congruence the twist vanishes, $\omega_{\mu\nu}=0$ , leaving $\mathrm{d}\theta/\mathrm{d}\lambda=-\tfrac12\theta^2-\sigma_{\mu\nu}\sigma^{\mu\nu}-R_{\mu\nu}k^\mu k^\nu$ . Why: twist represents rotation, and a rotating bundle can resist collapse; ruling it out is what makes the focusing unconditional.

Step 2. Apply the NEC through equation (5): $R_{\mu\nu}k^\mu k^\nu=8\pi G\,T_{\mu\nu}k^\mu k^\nu\ge0$ . Since shear-squared is also non-negative, both of the last two terms are $\le0$ , so $\mathrm{d}\theta/\mathrm{d}\lambda\le-\tfrac12\theta^2$ . Why: every neglected term only makes $\theta$ fall faster, so the inequality we keep is a rigorous lower bound on the rate of collapse — matter cannot slow the focusing, only speed it.

Step 3. Change variable to $\theta^{-1}$ . Then $\mathrm{d}(\theta^{-1})/\mathrm{d}\lambda=-\theta^{-2}\,\mathrm{d}\theta/\mathrm{d}\lambda\ge\tfrac12$ , a clean linear inequality. Integrating from $\lambda=0$ gives $\theta^{-1}(\lambda)\ge\theta_0^{-1}+\tfrac12\lambda$ . Why: the reciprocal linearizes the Riccati inequality, converting an awkward nonlinear ODE into a straight line we can integrate by hand.

Step 4. If the bundle starts converging, $\theta_0<0$ , then $\theta_0^{-1}<0$ , and the right-hand side climbs through zero at $\lambda=-2/\theta_0=2/|\theta_0|$ . But $\theta^{-1}\to0$ means $\theta\to-\infty$ : a caustic, where neighbouring rays cross and the congruence description breaks down. Hence $\Delta\lambda\le2/|\theta_0|$ . Why: this finite bound is the whole point — focusing is not merely eventual but happens within a computable stretch, which is what lets the singularity theorems reach a contradiction with geodesic completeness. *Why: the coefficient $\tfrac12$ is the $D=4$ null value $1/(n-2)$ .*

Limiting cases

For an isotropic perfect fluid, $T^\mu{}_\nu=\mathrm{diag}(-\rho,p,p,p)$ , every condition collapses to inequalities between $\rho$ and $p$ . The table walks the equation-of-state parameter $w=p/\rho$ from dust up through phantom matter and records which fences survive.

Case	Result	Meaning
Dust / cold matter ( $w=0$ )	$\rho\ge0,\ p=0$	Satisfies all four; the textbook "reasonable matter."
Radiation ( $w=\tfrac13$ )	$p=\tfrac13\rho,\ \rho\ge0$	All four hold: $\rho+p=\tfrac43\rho$ , $\rho+3p=2\rho$ , $\rho\ge\|p\|$ .
Stiff fluid ( $w=1$ )	$p=\rho,\ \rho\ge0$	Saturates DEC ( $\rho=\|p\|$ ); sound speed $c_s=c$ , the causal maximum.
Cosmological constant ( $w=-1$ )	$p=-\rho,\ \rho>0$	Sits on the NEC and DEC edges ( $\rho+p=0$ ); violates SEC ( $\rho+3p=-2\rho<0$ ), so it accelerates the expansion.
Phantom energy ( $w<-1$ )	$\rho+p<0$	Falls below the NEC fence and violates all four; the exotic matter of wormholes and Big Rip models.
Quantum vacuum (Casimir)	$\rho<0$ locally	Violates WEC and NEC pointwise; rescued only by the averaged ANEC.

Common trapthe energy conditions are laws of nature that all real matter must obey

They are nothing of the sort — they are convenient hypotheses, and every one of them is violated by matter we know exists. The strong energy condition is the most fragile: any positive cosmological constant breaks it, which is why the observed cosmic acceleration is, read literally, a measured SEC violation; the inflaton that drove the early universe breaks it too. The weak and null conditions, though far more robust classically, fail pointwise for ordinary quantum fields — the Casimir vacuum carries negative energy density, and squeezed light carries negative energy flux. The correct discipline is never to assume a condition "because the matter is physical," but to check it against the actual $T_{\mu\nu}$ , and to reach for the averaged or quantum versions (ANEC, quantum energy inequalities, the quantum null energy condition) whenever you invoke a global theorem in a setting where quantum effects matter.

Worked examples

1 · Dark energy and the strong energy condition (fully worked)

The universe's expansion is accelerating. Treating dark energy as a perfect fluid with the measured value $w=p/\rho\approx-1$ and energy density $\rho_\Lambda c^2\approx5.3\times10^{-10}\ \mathrm{J/m^3}$ (about $6\times10^{-27}\ \mathrm{kg/m^3}$ ), determine which energy conditions it obeys and confirm that it must accelerate the cosmos.

Step 1. Write the four fluid tests with $p=-\rho$ (taking $w=-1$ exactly for a pure cosmological constant). NEC: $\rho+p=\rho-\rho=0\ge0$ holds (saturated). WEC: also needs $\rho\ge0$ , and $\rho_\Lambda>0$ , so holds. DEC: needs $\rho\ge|p|=\rho$ , holds (saturated). SEC: needs $\rho+3p=\rho-3\rho=-2\rho<0$ , which fails.

Step 2. So dark energy satisfies NEC, WEC and DEC but violates the SEC — it lands exactly on the vacuum-energy point of the first figure. Dark energy is the cleanest laboratory case of strong-energy-condition violation by matter that is otherwise perfectly reasonable.

Step 3. Feed this into the second Friedmann (acceleration) equation, $\ddot a/a=-\tfrac{4\pi G}{3}(\rho+3p)$ . With $\rho+3p=-2\rho_\Lambda<0$ , the right-hand side is positive: $\ddot a/a=+\tfrac{8\pi G}{3}\rho_\Lambda>0$ . Because SEC is exactly the condition $\rho+3p\ge0$ that forces $\ddot a\le0$ , breaking it is not just consistent with acceleration — it is the whole reason for it.

Step 4. The observational payoff: current data give $w=-1.03\pm0.03$ , statistically indistinguishable from a cosmological constant and firmly on the SEC-violating side of $w=-\tfrac13$ . Any $w<-\tfrac13$ suffices to violate SEC and accelerate the universe; the measured $w\approx-1$ clears that bar with room to spare.

2 · The Casimir vacuum violates the NEC (fully worked)

Two parallel perfectly-conducting plates a distance $a$ apart reshape the electromagnetic vacuum, lowering its energy density between them to $u=-\pi^2\hbar c/(720\,a^4)$ . Evaluate $u$ for $a=1\ \mu\mathrm{m}$ and show that the pointwise NEC fails.

Step 1. Gather constants: $\hbar c=1.055\times10^{-34}\times2.998\times10^{8}=3.16\times10^{-26}\ \mathrm{J\cdot m}$ , and $\pi^2=9.87$ . With $a=10^{-6}\ \mathrm m$ , $a^4=10^{-24}\ \mathrm{m^4}$ .

Step 2. Plug in: $u=-\dfrac{9.87\times3.16\times10^{-26}}{720\times10^{-24}}=-\dfrac{3.12\times10^{-25}}{7.2\times10^{-22}}\ \mathrm{J/m^3}$ . This gives $u\approx-4.3\times10^{-4}\ \mathrm{J/m^3}$ — a genuinely negative energy density in the region between the plates.

Step 3. Now test the NEC. In the fluid language the Casimir vacuum has $\rho=u<0$ with a pressure of comparable magnitude; more directly, the local $T_{\mu\nu}k^\mu k^\nu$ evaluated for a ray crossing the gap picks up this negative $\rho$ and comes out below zero at points inside. So $T_{\mu\nu}k^\mu k^\nu<0$ somewhere along the ray: the pointwise NEC is violated by nothing more exotic than the quantum electromagnetic vacuum.

Step 4. The framework copes not by denying the negative energy but by integrating it. The ANEC integral $\int T_{\mu\nu}k^\mu k^\nu\,\mathrm d\lambda$ along a complete geodesic remains $\ge0$ because the ray must also traverse the positive-energy regions outside the plates, and quantum energy inequalities cap how negative and how prolonged the dip can be. Negative energy is real and local; it is the averaged condition, not the pointwise one, that survives to constrain global structure.

3 · How fast does a light bundle focus?

A converging light-front leaves a collapsing shell with initial expansion $\theta_0=-0.10\ \mathrm{m^{-1}}$ . Using only the NEC, bound the affine distance to its caustic — and explain why shear only tightens the bound.

Step 1. The focusing theorem of Derivation 3 gives $\Delta\lambda\le2/|\theta_0|$ directly. Numerically, $\Delta\lambda\le2/0.10=20\ \mathrm m$ (of affine parameter): the bundle cannot avoid a caustic within that span.

Step 2. This bound used only $R_{\mu\nu}k^\mu k^\nu\ge0$ and $\sigma^2\ge0$ dropped to zero. Restoring shear makes $\mathrm{d}\theta/\mathrm{d}\lambda$ more negative, so the true caustic arrives sooner: the 20 m is a worst case. Shear and matter both accelerate collapse; nothing in the NEC-obeying world can defer it past $2/|\theta_0|$ .

Step 3. This is the mechanism behind Penrose's theorem in one line: a trapped surface guarantees $\theta_0<0$ on both outgoing and ingoing null congruences, and the NEC then forces a focal point on every generator within finite affine parameter. Penrose's theorem adds the global hypotheses that make this a contradiction — global hyperbolicity with a non-compact Cauchy surface — for then those focal points would close the boundary $\partial J^+(\mathcal{S})$ into a compact set, which such a spacetime cannot admit. The only escape is that the spacetime is null geodesically incomplete, i.e. singular. A finite focusing bound is precisely what upgrades "gravity attracts" into "a singularity must form."

Physics of the portrait

The energy conditions are inseparable from the man who first showed what they could prove. Before 1965 a "singularity" in general relativity was suspected to be an artifact of the perfect symmetry assumed in exact solutions; Roger Penrose changed that by importing global, topological reasoning and pairing it with the null energy condition to make singularities generic.

Origins · 1 minRoger Penrose and how this was discovered

Roger Penrose

b. 1931, Nobel Prize in Physics 2020 · global methods and singularity theorems in general relativity

Roger Penrose is the mathematician who taught relativists to think globally. Trained in algebraic geometry at Cambridge, he brought to gravitation a topologist's habit of reasoning about a spacetime as a whole rather than solving equations point by point. His 1965 paper "Gravitational Collapse and Space-Time Singularities" — barely three pages — introduced the notion of a trapped surface, a closed 2-surface from which even the outgoing light rays converge, and proved that once one forms, the null energy condition forces geodesic incompleteness: a singularity is inevitable, with no symmetry assumed. This was the first theorem to show that black holes are not mathematical accidents but a robust prediction of collapse. With Stephen Hawking he extended the argument to cosmology in 1970, applying the strong energy condition to prove the Big Bang itself must be singular. Penrose went on to seed a generation of ideas — the Penrose process for extracting black-hole rotational energy, the cosmic censorship conjecture, twistor theory, conformal diagrams — but it is the 1965 focusing argument, and its dependence on a humble inequality about $T_{\mu\nu}k^\mu k^\nu$ , that the 2020 Nobel Prize honoured as "the discovery that black hole formation is a robust prediction of the general theory of relativity."

History

1955

— Raychaudhuri publishes the focusing equation governing the expansion of geodesic congruences

1965

— Penrose proves the first singularity theorem from a trapped surface plus the null energy condition

1970

— Hawking and Penrose combine timelike and null focusing into the general singularity theorem

1971

— Hawking proves the black-hole area theorem, needing only the null energy condition

1973

— Hawking and Ellis codify the null, weak, strong and dominant conditions in their monograph

Going further

Texts. Wald, General Relativity (1984), §9.2 (energy conditions) and §9.5 (singularity theorems) · Misner, Thorne & Wheeler, Gravitation (1973), §5.6 and Box 33.1 · Carroll, Spacetime and Geometry (2004), §4.6 · Hartle, Gravity (2003), Ch. 22 (gravitational collapse) · Schutz, A First Course in General Relativity (2009), §4.5 · Hawking & Ellis, The Large Scale Structure of Space-Time (1973), §4.3 · Poisson, A Relativist's Toolkit (2004), §2.1 · Curiel, "A Primer on Energy Conditions," in Towards a Theory of Spacetime Theories (2017). The four fluid inequality sets — NEC $\rho+p\ge0$ ; WEC adds $\rho\ge0$ ; DEC $\rho\ge|p|$ ; SEC $\rho+p\ge0,\ \rho+3p\ge0$ — are re-derived above by contracting $T^\mu{}_\nu=\mathrm{diag}(-\rho,p,p,p)$ with a boosted timelike $u^\mu$ and a null $k^\mu$ . The trace-reversed field equation $R_{\mu\nu}=8\pi G(T_{\mu\nu}-\tfrac12 Tg_{\mu\nu})$ and the null Raychaudhuri equation (denominator $n-2=2$ in $D=4$ ) are textbook; the focusing bound $\mathrm{d}\theta/\mathrm{d}\lambda\le-\tfrac12\theta^2\Rightarrow\Delta\lambda\le2/|\theta_0|$ follows from those equations. The acceleration equation $\ddot a/a=-\tfrac{4\pi G}{3}(\rho+3p)$ is the second Friedmann equation. The Casimir energy density $u=-\pi^2\hbar c/(720\,a^4)$ gives $-4.3\times10^{-4}\ \mathrm{J/m^3}$ at $a=1\ \mu\mathrm m$ (independently, the plate pressure $\pi^2\hbar c/240a^4\approx1.3\ \mathrm{mPa}$ agrees). Dark-energy numbers ( $\rho_\Lambda c^2\approx5.3\times10^{-10}\ \mathrm{J/m^3}$ , $w=-1.03\pm0.03$ ) are current Planck/DESI-era values. Units $c=1$ except where SI numbers are quoted; signature $(-,+,+,+)$ .

Prerequisites. The Stress-Energy Tensor · Perfect Fluid Stress-Energy · The Einstein Field Equations · The Einstein Tensor. Leads to. The Cosmological Constant · Local Conservation of Energy-Momentum · The Newtonian Limit.

Ask the tutor: Ask me to check any equation of state — inflaton, quintessence, a modified-gravity source, or your own $T_{\mu\nu}$ — against all four conditions and trace what it means for focusing, singularities, and cosmic acceleration.

Are the energy conditions actual laws of physics?

No — and this is the single most important thing to internalize. They are hypotheses about what "reasonable" matter looks like, adopted because they let us prove powerful theorems, not because nature guarantees them. Every one is violated by something real: the strong condition by any positive cosmological constant and by inflation, the null and weak conditions by ordinary quantum fields such as the Casimir vacuum and squeezed light. A theorem that "assumes the WEC" is really saying "here is what follows if the matter behaves; check that it does before you trust the conclusion." Treat them as premises to be verified, never as facts to be assumed.

Why is the strong energy condition called "strong" if a positive Λ violates it?

The name is about logical restrictiveness in a specific direction, not about being hardest to violate. "Strong" refers to how tightly it constrains the trace-reversed source $T_{\mu\nu}-\tfrac12 Tg_{\mu\nu}$ — it forbids large negative pressures ( $\rho+3p<0$ ) that the weak condition happily allows. The confusing consequence is that SEC and WEC are logically independent: neither implies the other. A cosmological constant ( $p=-\rho$ ) satisfies WEC but breaks SEC, while one can imagine matter that does the reverse. So "strong" does not mean "implies all the others" — only DEC does that. It means "strongly restricts the gravitational, attracting part of the source."

What is the physical difference between the null and weak conditions?

The weak condition tests real observers — timelike four-velocities $u^\mu$ — and demands each measure non-negative energy density, $T_{\mu\nu}u^\mu u^\nu\ge0$ . The null condition tests light rays — null vectors $k^\mu$ — and is the limit of the weak one as the observer is boosted toward the speed of light. That limit isolates the combination $\rho+p$ and discards the bare $\rho$ , which is why WEC ( $\rho\ge0$ and $\rho+p\ge0$ ) is strictly stronger than NEC ( $\rho+p\ge0$ alone). Physically: NEC is what survives when you look along a light cone, and because horizon generators are light rays, it is NEC — not WEC — that governs black-hole horizons.

How exactly does an inequality on matter become a theorem about geometry?

Through one algebraic move: trace-reverse the Einstein equations. Solving $G_{\mu\nu}=8\pi G\,T_{\mu\nu}$ for the Ricci tensor gives $R_{\mu\nu}=8\pi G(T_{\mu\nu}-\tfrac12 Tg_{\mu\nu})$ , so an inequality on the right-hand matter side is identical to an inequality on $R_{\mu\nu}$ . And $R_{\mu\nu}$ contracted with a geodesic's tangent is precisely the term in the Raychaudhuri equation that focuses the geodesic bundle. So NEC becomes $R_{\mu\nu}k^\mu k^\nu\ge0$ (null focusing) and SEC becomes $R_{\mu\nu}u^\mu u^\nu\ge0$ (timelike focusing). Once bundles are guaranteed to focus, caustics and then incompleteness follow — that is the whole route from "matter is reasonable" to "spacetime is singular."

Why does cosmic acceleration count as evidence for SEC violation?

Because the second Friedmann equation reads $\ddot a/a=-\tfrac{4\pi G}{3}(\rho+3p)$ , and the strong energy condition is exactly $\rho+3p\ge0$ . If SEC held, the bracket would be non-negative and $\ddot a\le0$ — the expansion could only decelerate. The measured $\ddot a>0$ therefore forces $\rho+3p<0$ , i.e. $w=p/\rho<-\tfrac13$ , a direct violation of SEC. The observed dark-energy value $w\approx-1$ is comfortably past that threshold. So "the universe accelerates" and "the dominant matter violates the strong energy condition" are the same statement, read once as observation and once as inequality.

Do wormholes and warp drives really require breaking an energy condition?

Yes — specifically the null energy condition. A traversable wormhole must flare open at its throat, which geometrically demands defocusing of the light rays threading it, and defocusing requires $R_{\mu\nu}k^\mu k^\nu<0$ , i.e. $T_{\mu\nu}k^\mu k^\nu<0$ : exotic matter with $\rho+p<0$ . The Alcubierre warp metric needs the same. Classical fields obey NEC, so such geometries can only be sourced by quantum negative energy — and there the averaged null energy condition and quantum energy inequalities bite hard, generally rendering macroscopic wormholes marginal. The 2017 Gao–Jafferis–Wall construction threads this needle by using a quantum coupling to produce just enough ANEC-consistent negative energy, showing the door is not bolted shut, only very narrow.