Energy Conditions and Their Quantum Failure

Classical pointwise energy conditions are not properties of quantum matter. Even a free field in a smooth Hadamard state can have negative renormalized energy density in a bounded region. This does not make the theory unstable or acausal: spectral stability constrains the total Hamiltonian, microcausality constrains commutators, and quantum energy inequalities constrain suitably sampled negative energy. These are different statements from a pointwise lower bound.

Required background. Stress–Energy and Conservation provides the classical energy conditions and conserved tensor. Curvature supplies the geometric terms in curved-spacetime bounds.

Helpful background. Flat-Spacetime Quantum Field Theory provides Fock-state interference and Hamiltonian positivity. Partial Differential Equations and Green Functions supplies distributional smearing. Causality and Global Structure develops focusing and causal propagation. Quantum Fields in Curved Spacetime provides the Hadamard state class and renormalized stress tensor.

What fails pointwise

For a timelike unit vector $u^a$, the classical weak energy condition requires

$$T_{ab}u^au^b\ge 0.$$

The null energy condition replaces $u^a$ by every null vector $k^a$. These conditions support classical focusing theorems, but quantization replaces $T_{ab}$ by an operator-valued distribution. Smearing is necessary to obtain an operator; after renormalization its expectation in a Hadamard state can be a smooth tensor field evaluated pointwise. That smooth expectation need not be nonnegative.

A simple mechanism is interference. In Minkowski spacetime, take a normalized superposition of the vacuum and a two-particle wave packet,

$$|\psi\rangle=N\left(|0\rangle+\epsilon|2_f\rangle\right).$$

The normal-ordered energy density contains a term linear in $\epsilon$ from vacuum–two-particle interference and a positive term quadratic in $|\epsilon|$. At spacetime points where the interference has the negative phase, sufficiently small nonzero $\epsilon$ makes the linear term dominate. The local energy density is negative even though the expectation of the global Hamiltonian remains nonnegative.

Squeezed states implement the same physics over controlled regions. For the finite-energy Minkowski wave-packet example, positivity of the total Hamiltonian ensures that a negative spatial contribution cannot make the total energy negative. This is a statement about that Hamiltonian and vacuum normalization, not a universal rule that every negative density must later be followed by a positive pulse. The static cylinder Casimir state in Practical Stress-Tensor Calculations has a persistent negative vacuum energy relative to the infinite-line subtraction. Boundary conditions, topology, and the reference energy matter.

Sampling replaces point evaluation

A quantum energy inequality averages the measured density along an observer’s worldline $\gamma(\tau)$ with a smooth compactly supported sampling function $g$:

$$\int d\tau\,g(\tau)^2 \left\langle T_{ab}u^au^b\right\rangle_\omega \ge -\mathcal Q[g,\gamma,m,\xi,g_{ab}].$$

This form is a theorem only for appropriate field and coupling classes. In the minimally coupled scalar case, the functional $\mathcal Q$ is finite for the declared geometry, prescription, and Hadamard state class; a general nonminimal coupling need not admit a state-independent bound of this form. It must not be inferred from the notation $\xi$ in $\mathcal Q$.

For one massless minimally coupled scalar on four-dimensional Minkowski spacetime, along an inertial worldline and with vacuum normal ordering, a concrete bound is

$$\int d\tau\,g(\tau)^2\langle:T_{00}:\rangle_\omega \geq-\frac{1}{16\pi^2}\int d\tau\,|g''(\tau)|^2.$$

This is Fewster and Eveson 1998, §V.B, Eq. (5.5), PDF, with $f=g^2$. Choose $g\in C_0^\infty(\mathbb R)$ real and normalize $\int g^2d\tau=1$. A profile of duration $\tau_0$ is $g_{\tau_0}(\tau)=\tau_0^{-1/2}g_1(\tau/\tau_0)$, and substitution gives

$$\int |g_{\tau_0}''|^2d\tau =\tau_0^{-4}\int |g_1''|^2ds.$$

The normalization is why a bound on the averaged density scales as $-\tau_0^{-4}$. Holding the amplitude rather than the integral of the sampler fixed would change the scaling. In this flat inertial setting the bound limits the duration of a fixed negative density; it does not forbid a negative Casimir reference energy in a different boundary problem.

Bounds come in two forms. A difference inequality compares $\omega$ with a reference Hadamard state and inherits that reference’s stress tensor. An absolute inequality has a state-independent right side built from local geometry and the sampling function. The rigorous absolute result used here is specifically for the four-dimensional globally hyperbolic, minimally coupled massive scalar field: Fewster and Smith 2008, §3.1, Theorem 3.1, and §§3.2.2–3.3. It should not be advertised as a theorem for every coupling, field, boundary condition, or averaged null observable.

Stability, causality, and focusing are distinct

Hamiltonian stability is a global spectral statement in a stationary representation. A positive Hamiltonian can coexist with negative density in a subregion because localization introduces interference and vacuum polarization. Conversely, a local quantum inequality is not a proof that a global Hamiltonian exists on a generic curved spacetime.

Causal propagation is controlled by the field equation and the vanishing of commutators at spacelike separation. A negative expectation value changes neither the principal part of the Klein–Gordon operator nor Einstein causality of the observable algebra. It may source geometric focusing or defocusing only when one solves an appropriate semiclassical metric equation, which is beyond the prescribed-background calculation here.

Classical singularity theorems use pointwise or averaged geometric hypotheses. Quantum violations show that one cannot insert $\langle T_{ab}\rangle$ and retain those hypotheses without modification. A quantum energy inequality supplies a constrained replacement in some regimes, but it is not automatically the averaged null energy condition used in every gravitational theorem.

Prescription and state checks

Every negative-energy claim should name the state, observer, sampling function, and stress-tensor prescription. A change of finite curvature counterterms can shift an absolute local density on a curved background. State differences computed in one scheme cancel that common ambiguity; flat-spacetime normal ordering removes it in the example above.

The Hadamard condition matters. A formally chosen state with an inadmissible ultraviolet singularity can produce divergent or regulator-dependent “negative energy” that lies outside the theorem. Smooth sampling also matters: a sharp characteristic function injects high frequencies and need not obey the same bound.

Finally, pointwise energy-condition failure does not mean unlimited negative energy within a theory admitting an applicable quantum inequality. Check the field’s coupling, the sampling scale, conservation, and the theorem’s geometric hypotheses. Compensation by positive energy is an additional conclusion in specific settings, not a universal substitute for those checks.

Exercise

Suppose a four-dimensional sampling profile has width $\tau_0$ and a massless flat-spacetime lower bound proportional to $-\tau_0^{-4}$. If the width is doubled while the profile shape is held fixed, by what factor does the magnitude of the lower bound change, and what does that mean physically?

Solution

Replacing $\tau_0$ by $2\tau_0$ multiplies $\tau_0^{-4}$ by $2^{-4}=1/16$. The allowed lower bound is therefore sixteen times closer to zero. Longer sampling suppresses a brief negative fluctuation and makes sustained negative energy much more tightly constrained. This scaling statement does not assert a universal numerical coefficient; that coefficient and curvature corrections depend on the precise field, trajectory, and normalized sampling function.

References

• Fewster, Christopher J., and Calvin J. Smith. “Absolute Quantum Energy Inequalities in Curved Spacetime.” Annales Henri Poincaré 9 (2008): 425–455. DOI. Open PDF.
• Fewster, Christopher J., and Simon P. Eveson. “Bounds on Negative Energy Densities in Flat Spacetime.” Physical Review D 58 (1998): 084010. DOI. Open PDF.
• Wald, Robert M. “The Back Reaction Effect in Particle Creation in Curved Spacetime.” Communications in Mathematical Physics 54 (1977): 1–19. DOI.