The Asymptotic-Safety Scenario

The asymptotic-safety scenario proposes that gravity remains a continuum quantum field theory because its dimensionless couplings approach a non-Gaussian fixed point at arbitrarily high renormalization scale. Predictivity would follow only if the fixed point has a finite-dimensional ultraviolet critical surface. The repeated appearance of such fixed-point candidates is important evidence, but their existence, number of relevant directions, and physical interpretation remain unresolved in the untruncated theory.

Required background. Flat-Spacetime Quantum Field Theory supplies Wilsonian flows, beta functions, and renormalized trajectories; Gravitational Actions and Boundary Terms supplies the generally covariant operator basis and its variational assumptions.

Helpful background. Einstein Equations identifies the infrared target; Quantum Information supplies physical-state and observable criteria; and Quantum Gravity supplies the volume-wide distinction between programme evidence and established prediction.

Fixed points in gravitational theory space

Write an effective action as a sum of diffeomorphism-invariant operators,

$$\Gamma_k[g]=\sum_i \bar u_i(k)\,\mathcal O_i[g], \qquad u_i(k)=k^{-d_i}\bar u_i(k),$$

where $d_i$ is the mass dimension of the dimensionful coupling $\bar u_i$. The dimensionless beta functions are $\beta_i(u)=\partial_tu_i$ with $t=\ln(k/k_0)$ for an arbitrary reference scale $k_0$. A fixed point $u_i^*$ satisfies $\beta_i(u^*)=0$. It is Gaussian if all essential interactions vanish and non-Gaussian otherwise.

Near a fixed point, linearization gives

$$\partial_t\delta u_i=B_i{}^j\delta u_j, \qquad \delta u_i(k)=\sum_I C_I V_i^{(I)} \left(\frac{k}{k_0}\right)^{-\theta_I},$$

where $B_i{}^j=(\partial\beta_i/\partial u_j)_*$ and its eigenvalues are $-\theta_I$. With this convention, $\operatorname{Re}\theta_I\gt0$ defines a direction attracted to the fixed point as $k\to\infty$. A trajectory that reaches the fixed point must lie on the ultraviolet critical surface spanned by these directions. If only finitely many are relevant, a finite number of measurements selects a trajectory and the remaining couplings become predictions.

This linear classification assumes no eigenvalue has zero real part. Marginal directions require the nonlinear beta functions: an initially vanishing linear exponent does not decide whether that coupling approaches or leaves the fixed point. A complex-conjugate pair with positive real part contributes two real relevant parameters, not one.

The couplings must be essential: directions produced by field redefinitions or gauge choices do not count as independent physics. In gravity, separating essential from redundant couplings is delicate because rescaling the metric can trade normalization against Newton’s coupling, background-field calculations introduce two metric arguments, and equations-of-motion operators can be redundant.

The Einstein–Hilbert illustration

The smallest common truncation retains

$$\Gamma_k[g]\approx \frac{1}{16\pi G_k}\int d^4x\sqrt{-g}\,(R-2\Lambda_k) +S_{\mathrm{gf}}+S_{\mathrm{gh}},$$

and studies $g_k=k^2G_k$ and $\lambda_k=\Lambda_k/k^2$. Functional-flow projections often yield a nonzero pair $(g_*,\lambda_*)$ and two relevant, sometimes complex-conjugate, critical exponents. This is an illustration, not a definition of the full scenario. The coordinates $(g_*,\lambda_*)$ depend strongly on regulator, gauge, field parametrization, and normalization; universal quantities and qualitative stability under enlarged truncations carry more weight.

Reuter’s effective-average-action calculation established the modern functional route and exhibited a non-Gaussian fixed-point candidate Reuter 1998, §§ III–V. Subsequent studies have added curvature invariants, fluctuation vertices, matter fields, and symmetry identities; reviews such as Eichhorn 2019, §§ 3–5 explain why recurring fixed points and modest relevant sectors are suggestive while not equivalent to a continuum construction.

Predictivity and the infrared trajectory

A fixed point alone does not recover our world. A viable renormalized trajectory must leave its ultraviolet neighborhood, pass through a regime where general relativity is accurate, and reproduce the observed signs and magnitudes of $G$ and $\Lambda$ together with compatible matter couplings. Singularities in beta functions, phase boundaries, or failure of split-symmetry restoration can obstruct that path.

Nor is “two relevant directions” automatically two predictions. Relevant parameters must be mapped to physical renormalization conditions. Complex exponents imply spiraling approach in the chosen coordinates, not oscillating observables by themselves. Irrelevant directions can become predictable functions of the relevant ones only after the ultraviolet critical surface is shown to exist in the full theory space.

A useful comparison axis is therefore layered: fixed-point existence; stability under operator, regulator, and gauge changes; control of Ward or Nielsen identities; Lorentzian state-space definition; infrared crossover; and gauge-invariant observables. Evidence at an early layer cannot substitute for a later one.

Results and open questions through 2026-08-20

This discussion reflects results available through 2026-08-20. Competing interpretations persist. Proponents view the recurrence of fixed points across increasingly elaborate truncations and matter systems as universality evidence. Skeptics emphasize that the calculations remain finite projections of an exact flow, often on Euclidean backgrounds, with incomplete symmetry restoration and no general convergence theorem. A third, pragmatic position treats the scenario as a generator of testable scaling relations without claiming that the whole ultraviolet Hilbert space has been constructed.

Claim or formulation	Evidence or result	Scope and limitation
Einstein–Hilbert and polynomial-curvature flows	Non-Gaussian fixed-point candidates recur, often with a small relevant sector	Coordinates and exponents shift with gauge, regulator, parametrization, and truncation
Vertex and fluctuation formulations	Momentum-dependent vertices probe physics beyond background couplings	Operator completeness and split-symmetry control are still limited
Gravity coupled to matter	Broad regions of matter content retain candidate fixed points	Thresholds and admissible matter bounds depend on approximation and scheme
Predictive observables	Some scaling and phenomenological relations have been proposed	Few are both gauge invariant and quantitatively stable along a controlled infrared trajectory

The conclusions could change if systematic truncation sequences show stable essential critical exponents and restored symmetry identities, or if those sequences lose the fixed point; the interpretation could also change with a Lorentzian construction with a positive physical state space, or by a decisive obstruction to continuation toward the observed infrared regime.

What would count as stronger evidence?

Stronger evidence would combine independent formulations. Agreement between background and fluctuation flows after enforcing the relevant Ward identities would test split dependence. Operator-by-operator convergence with regulator bands would turn qualitative recurrence into an uncertainty estimate. A physical observable—such as an amplitude or relational correlation—should be evaluated along the same renormalized trajectory and reproduce known low-energy logarithms. Finally, analytic continuation must specify the contour and state space; Euclidean fixed-point data alone do not prove Lorentzian unitarity.

Exercise

Suppose a three-coupling truncation has critical exponents $\theta=\{3.0,\,0.7,\,-2.4\}$. At a reference scale $k_0$, which perturbations may be freely specified on a trajectory that reaches the fixed point as $k\to\infty$? What changes if the third direction is redundant?

Solution

Because $\delta u_I=C_I(k/k_0)^{-\theta_I}$, the first two perturbations decay toward the ultraviolet and span two relevant directions. Their coefficients may label the ultraviolet-complete trajectories. The third perturbation grows as $k^{2.4}$, so reaching the fixed point requires its coefficient to be fixed—zero in linearized coordinates, or a definite function of the relevant parameters on the nonlinear critical surface. If that third direction is redundant, it represents a field redefinition rather than an independent physical coupling; it should be removed before counting either predictions or tunable parameters. The truncation then still reports two essential relevant directions, subject to stability under enlargement.

References

• Eichhorn, Astrid (2019), “An Asymptotically Safe Guide to Quantum Gravity and Matter,” Frontiers in Astronomy and Space Sciences 5, 47, arXiv:1810.07615.
• Reuter, Martin (1998), “Nonperturbative Evolution Equation for Quantum Gravity,” Physical Review D 57, 971–985, arXiv:hep-th/9605030.