Observables and Evidence for Continuum Completion

Fixed-point coordinates are not measurements. The decisive question for a proposed continuum completion is whether a gauge-invariant observable can be computed along a controlled renormalized trajectory, shown stable under truncation and scheme changes, and distinguished from ordinary effective-field-theory freedom. As of the evidence cutoff used here, continuum-flow studies provide substantial structural evidence but no single gravitational observable satisfying that entire chain.

Required background. Flat-Spacetime Quantum Field Theory supplies amplitudes, Wilson coefficients, and scheme-independent observables; Gravitational Actions and Boundary Terms supplies the action and boundary data from which observables are defined.

Helpful background. Einstein Equations supplies the infrared target; Quantum Information supplies relational and operational criteria; and Quantum Gravity supplies the distinction between controlled results and programme-specific extrapolations.

From a fixed point to a prediction

A defensible prediction requires a sequence:

1. identify essential couplings and a non-Gaussian fixed point;
2. determine the ultraviolet critical surface and select a trajectory by physical renormalization conditions;
3. follow that trajectory into a semiclassical regime without crossing singular flow boundaries;
4. construct a gauge-invariant or relational observable with a specified Lorentzian state and boundary condition;
5. propagate regulator, gauge, truncation, and matching uncertainties to the result.

Each arrow can fail. A universal critical exponent is stronger than a fixed-point coordinate, but it is still not automatically a detector observable. A running Newton coupling $G(k)$ becomes physical only after the arbitrary identification of $k$ with a momentum, curvature, or relational scale is replaced by a derivation in the observable itself.

At low energy, the continuum programme must reproduce gravitational effective field theory. Nonanalytic amplitude terms generated by massless propagation are invariant under local counterterm redefinitions, whereas analytic terms mix with unknown Wilson coefficients Donoghue 2023, §§ 3–5. This provides a stringent independent check: an ultraviolet trajectory may constrain analytic coefficients, but it must not erase the established infrared logarithms.

Candidate observable classes

Scattering is conceptually clean in asymptotically simple backgrounds because an on-shell amplitude avoids many off-shell gauge ambiguities. Gravity complicates the construction through infrared dressing, the absence of a conventional S-matrix in cosmology, and Euclidean-to-Lorentzian continuation. Momentum-dependent fluctuation vertices are progress toward amplitudes, but background effective couplings alone are insufficient.

Cosmological and black-hole applications often “RG improve” a classical solution by replacing $G$ or $\Lambda$ with running quantities and choosing $k=k(x)$. Such models can reveal possible scaling regimes, yet the scale setting and improved field equation need not follow from the effective action. A corrected metric that violates the relevant Ward identity or stress-energy conservation is not a controlled prediction.

The spectral dimension,

$$d_s(\sigma)=-2\,\frac{d\ln P(\sigma)}{d\ln\sigma},$$

measures the return probability $P(\sigma)$ of an auxiliary diffusion process. Many quantum-gravity approaches find dimensional reduction at short diffusion time. It is a useful comparative probe, but depends on the diffusion operator, ensemble, and resolution prescription; it is not by itself a measured spacetime dimension or unique evidence for asymptotic safety.

Matter-sector relations can be more directly testable. If gravitational fixed-point dynamics makes a matter coupling irrelevant, matching along the infrared trajectory can constrain masses or interaction strengths. Yet thresholds, truncation of gravitational vertices, and the assumed ultraviolet matter content affect the result. A numerical agreement obtained after choosing those inputs is not a parameter-free prediction.

Unitarity, causality, and the state space

Euclidean functional flows naturally compute Euclidean correlation functions. A physical completion also needs reflection positivity or another route to a positive Lorentzian state space, analytic structure compatible with causal propagation, and gauge-invariant observables. Higher derivatives generated by the flow do not automatically imply a physical ghost, because the full momentum-dependent propagator need not equal a finite polynomial truncation; conversely, a Euclidean fixed point does not automatically remove negative spectral weight.

Credible positions compete here. One position expects the complete nonlocal effective action to possess acceptable spectral properties even when finite derivative truncations do not. Another demands an explicit reconstruction before assigning physical meaning to the fixed point. A pragmatic position tests Lorentzian observables within controlled subsectors while withholding a claim of full reconstruction. The choice of contour, background, operator algebra, and asymptotic state is therefore part of the result, not technical decoration.

Results and open questions through 2026-08-20

This discussion reflects results available through 2026-08-20. Reviews document increasingly sophisticated fixed-point and matter calculations Eichhorn 2019, §§ 4–6, but the gap between fixed-point evidence and a complete observable remains open. The matrix separates mathematical outputs from their interpretation.

Candidate result or mechanism	Evidence at the cutoff	Scope and limitation
Essential critical exponents	Recurrent finite relevant sectors across several truncations	Values retain approximation dependence; the full critical surface is unresolved
Momentum-dependent fluctuation vertices	Flows resolve more kinematics than background couplings	On-shell Lorentzian amplitudes and infrared dressing are incomplete
Short-scale spectral dimension	Dimensional reduction appears in multiple continuum calculations	Diffusion definition is auxiliary and not programme-specific
Cosmological or black-hole improvement	Running couplings generate concrete phenomenological models	Scale identification and backreaction are often ansatz-dependent
Matter-coupling relations	Some trajectories restrict infrared matter parameters	Results depend on matter content, thresholds, and gravitational truncation

Evidence for continuum completion is therefore cumulative rather than decisive. Proponents stress cross-truncation recurrence and growing vertex control; critics stress the absence of a nonperturbative definition, physical Hilbert space, and scheme-independent quantitative observable. Neither position is settled by one stable zero of beta functions.

The conclusions could change if a gauge-invariant Lorentzian amplitude is computed with quantified truncation uncertainty and correct EFT matching, or if systematic enlargement destroys the proposed fixed point; the interpretation could also change with a positive reconstruction theorem, a robust violation of positivity, or an experimentally discriminating prediction fixed by the same ultraviolet trajectory.

A practical evidence test

For every proposed observable, ask four adverse questions. Does a local field redefinition change it? Does varying the regulator within an admissible class move it beyond the reported numerical error? Does an alternative scale-setting prescription alter the signal? Can the same effect be produced by an unconstrained low-energy Wilson coefficient? Failure of any test need not falsify the fixed point, but it downgrades the claim from prediction to model-dependent illustration.

Exercise

A calculation reports $G(k)=g_*/k^2$ near a fixed point and substitutes $k=\xi/r$ into the Schwarzschild lapse to predict a shifted horizon. Identify two independent ambiguities and state what would turn the result into a stronger observable claim.

Solution

First, $k=\xi/r$ is a scale-setting ansatz: changing $\xi$ or using a curvature invariant changes the corrected lapse. Second, inserting a running coupling into a classical solution need not solve the field equations obtained by varying the full scale-dependent effective action; derivative and nonlocal terms can contribute at the same order. A stronger claim would derive a gauge-invariant relational quantity from a specified effective action and Lorentzian state, solve the corrected equations consistently, show stability under regulator, gauge, and truncation changes, and match the same trajectory to known low-energy EFT amplitudes. A horizon-coordinate shift alone is not sufficient.

References

• Donoghue, John F. (2023), “Quantum General Relativity and Effective Field Theory,” in Handbook of Quantum Gravity, arXiv:2211.09902.
• Eichhorn, Astrid (2019), “An Asymptotically Safe Guide to Quantum Gravity and Matter,” Frontiers in Astronomy and Space Sciences 5, 47, arXiv:1810.07615.