Violations and Limitations of Classical Energy Conditions

Pointwise energy conditions are assumptions used by particular geometric arguments, not consequences of covariance or stress-energy conservation. Classical nonminimal couplings can violate them, and renormalized quantum stress tensors generically admit negative local energy densities. When a condition fails, only the proof steps that use it are lost. Averaged conditions or quantum energy inequalities can support different results, but they do not restore the original pointwise premise.

Required background. We use congruence kinematics, Einstein’s equation, and the metric definition of stress-energy.

Helpful background. The causal inequalities use special relativity and smooth tensor fields. Any incompleteness conclusion additionally requires global causal structure.

What must be specified before declaring a violation

An energy-condition claim requires a theory and a stress tensor. One must state whether the configuration is on shell, whether the tensor is canonical, Hilbert, improved, effective, or renormalized, and whether curvature terms have been moved between the two sides of a field equation. In quantum field theory one must also name the state and renormalization prescription. For an averaged claim, the curve, tangent normalization, sampling function, and domain are part of the statement.

These choices are not semantic. A nonminimal term $\xi R\phi^2$ can be retained as part of the matter stress tensor or partially moved into an effective gravitational coupling. The full equation is unchanged, while a pointwise inequality imposed on the rearranged “effective matter” tensor can change form.

Kontou and Sanders survey classical violations and identify which matter models satisfy which conditions Kontou and Sanders 2020, § 2.3, pp. 14–16 and Table 2, PDF. Their conclusions are conditional on the displayed stress tensors and field equations, not universal verdicts about every scalar theory.

Classical scalar examples

For a canonical minimally coupled scalar,

$$T_{ab}=\nabla_a\phi\nabla_b\phi -\frac12g_{ab}\left[(\nabla\phi)^2+2V(\phi)\right].$$

Null contraction gives

$$T_{ab}k^ak^b=(k^a\nabla_a\phi)^2\ge0,$$

so the classical canonical field satisfies NEC pointwise. In $D\gt2$ dimensions its trace-reversed tensor is

$$T_{ab}-\frac{T}{D-2}g_{ab} =\nabla_a\phi\nabla_b\phi+\frac{2V}{D-2}g_{ab}.$$

For a unit timelike observer, the SEC contraction is consequently $(u\cdot\nabla\phi)^2-2V/(D-2)$. A positive potential can make it negative even while every null contraction remains nonnegative. For example, a constant scalar at a positive stationary point of $V$ obeys its field equation and has $T_{ab}=-Vg_{ab}$: the same algebra as positive vacuum energy. This already shows why “the scalar satisfies the energy conditions” is too coarse.

For a nonminimally coupled scalar, metric variation produces terms such as

$$\xi\left(g_{ab}\Box-\nabla_a\nabla_b+G_{ab}\right)\phi^2.$$

Null contraction includes $-\xi k^ak^b\nabla_a\nabla_b\phi^2$, whose sign is not fixed by a square. Smooth on-shell configurations can therefore violate NEC locally. The failure is classical and need not involve negative-norm states or quantum fluctuations.

The counterexample must be used carefully. It shows that a pointwise NEC-based focusing step cannot be asserted for every nonminimal scalar solution. It does not prove that every such solution defocuses, nor that no averaged or geometric lower bound exists.

Quantum local energy

Renormalized quantum stress-energy is an expectation value after subtraction of state-independent singular structure. Quantum interference permits $\langle T_{ab}u^au^b\rangle$ to be negative in bounded regions or intervals. Casimir-type states and squeezed states supply familiar mechanisms. A classical pointwise condition imposed on every quantum state is therefore too strong.

Negative local energy does not imply arbitrarily large negative energy for arbitrarily long duration. A quantum energy inequality (QEI) has a different form, schematically

$$\int d\tau\,f(\tau)^2 \langle T_{ab}u^au^b\rangle_\omega \ge-\mathcal Q[f,\text{geometry},\text{field}],$$

for a specified class of states and sampling functions. The lower bound depends on the field theory, worldline, scale, and reference prescription. Kontou and Sanders review the hypotheses and limitations of these bounds Kontou and Sanders 2020, § 3, pp. 17–27, PDF.

A QEI is not WEC with an integral sign. It allows pointwise violations, has a state/domain-dependent lower bound, and usually controls a weighted timelike average rather than every causal contraction.

Averaged replacements and focusing

An averaged energy condition integrates a contraction along a complete or specified causal geodesic, for example an averaged null condition of the form

$$\int_\gamma T_{ab}k^ak^b\,d\lambda\ge0.$$

Its meaning depends on affine normalization, convergence of the integral, completeness or endpoint data, and whether the inequality is achronal, weighted, or damped. The review compares these variants and their relation to classical theorems Kontou and Sanders 2020, § 4.1, pp. 28–31, PDF.

Raychaudhuri’s pointwise comparison used

$$\theta'\le-\frac{\theta^2}{D-2}$$

at every parameter value. If $R_{ab}k^ak^b$ becomes negative locally, that exact inequality fails. An averaged lower bound may still control the integrated expansion or force focusing under different initial and genericity assumptions, but the proof must be reconstructed. It is invalid simply to insert an averaged hypothesis into the pointwise line.

Kontou and Sanders show how weakened convergence premises enter theorem-specific focusing arguments Kontou and Sanders 2020, § 5.1, pp. 41–42, Eqs. (111)–(115), PDF.

Logical conclusions and nonconclusions

Violation of SEC can permit accelerated expansion and block the standard timelike focusing bound while NEC still holds. Violation of NEC blocks the standard null-convergence route in Einstein gravity, but a given congruence may still focus because of negative initial expansion and shear. Conversely, satisfying NEC does not ensure a trapped surface, global causality, or geodesic incompleteness.

No pointwise condition should be advertised as a universal definition of “reasonable matter.” Different conditions encode energy density, flux causality, or trace-reversed convergence, and known field theories treat them differently. Their defensible role is conditional: if a theorem assumes a named inequality and a model satisfies it in the relevant domain, that step is available.

Exercises

Why does the canonical scalar’s NEC calculation not establish SEC?

Solution

NEC uses a null contraction, which removes every metric-proportional term and leaves the square $(k\cdot\nabla\phi)^2$. SEC uses the trace-reversed tensor contracted with timelike vectors. The potential survives that operation with a sign that can make the contraction negative. The two conditions test different algebraic combinations, so proving one does not prove the other.

References

• Kontou, Eleni-Alexandra, and Ko Sanders. “Energy Conditions in General Relativity and Quantum Field Theory.” Classical and Quantum Gravity 37 (2020): 193001. DOI; arXiv.