Ⅵ: Asymptotic Safety

Loop quantum gravity quantises spacetime geometry directly, producing a discrete structure at the Planck scale. String theory replaces the point-like particles of the Standard Model with one-dimensional vibrating strings and requires ten dimensions to achieve internal consistency. Both introduce new fundamental ingredients that general relativity does not contain.

Asymptotic safety proposes something fundamentally different. It asks whether quantum gravity can be made consistent using only the degrees of freedom that general relativity already contains: the spacetime metric and the dynamics it obeys, described within the standard framework of quantum field theory, provided the theory behaves in a specific way at high energies. That specific behaviour is scale invariance.

A familiar example of scale invariance is a fractal. Zoom into a fractal at any magnification and the structure looks the same. There is no preferred scale at which the pattern changes character. Asymptotic safety proposes that gravity, at energies at and above the Planck scale, acquires this property. Not in a geometric sense, but in the sense that the equations governing gravity stop changing with energy. The theory freezes into a self-similar form, and the coupling constants that describe its strength approach fixed values rather than growing without bound.

Scale invariance means that the theory looks the same at all energy scales. There is no preferred scale, no coupling constant blowing up, no breakdown of the mathematical framework. At very high energies, at and above the Planck scale, gravity on the asymptotic safety proposal enters a regime in which it is exactly scale-invariant. The coupling constants* stop varying. They approach fixed, finite values and stay there, regardless of how high the energy goes. This is what it means for a theory to be asymptotically safe.

If this is true of gravity, then the specific failure identified in Article III, Newton’s constant growing without bound at high energies and producing uncontrollable infinities at each order of the perturbation series, is not a sign that quantum field theory cannot describe gravity. It is a sign that perturbation theory* is the wrong tool for gravity at the Planck scale. The theory may be finite at all energies and fully predictive without any modification of its fundamental content. No strings. No discrete geometry. No extra dimensions. Just general relativity, taken seriously at the highest energies.

Whether this is actually the case is the central question that has been investigating since 1976.

The name reflects the specific mathematical behavior the proposal describes. In physics, a value is asymptotic when it approaches a specific limit as a parameter—in this case, energy—grows without bound. Safety refers to the theory’s protection against ‘uncontrollable divergences.’ A theory is asymptotically safe if its coupling constants level off at a fixed, finite value as energy increases, rather than ‘blowing up’ into the infinities that usually render quantum gravity unpredictive. When Steven Weinberg coined the term in 1979, he chose a precise label: it describes exactly what the fixed point achieves by keeping the physics finite and ‘safe’ at the highest possible scales.

Why Asymptotic Safety Is Less Well Known

Before examining the technical content of the theory, it is worth stating plainly why asymptotic safety occupies a different place in the public and scientific discourse from loop quantum gravity or string theory.

String theory had a dramatic history and a claim to describe everything including particle physics. Loop quantum gravity has a clear philosophical appeal. It takes general relativity’s core lesson about background independence seriously and follows it to its logical conclusion. Both theories offer something vivid. String theory offers unification and extra dimensions. Loop quantum gravity offers a picture of space as a quantum structure woven from discrete pieces.

Asymptotic safety offers neither. It does not claim to unify gravity with the other forces. It does not predict extra dimensions or new particles at accessible energies. It does not replace the spacetime continuum with a discrete structure. Its central claim is that quantum field theory, applied to general relativity, already works, provided one uses the right non-perturbative tools rather than perturbation theory.

This conservatism is both its most distinctive feature and the reason it is less well known. It produces no new picture of reality. It asks only whether the framework that succeeded for every other fundamental force was abandoned too quickly when applied to gravity. The evidence for and against this possibility is technical, difficult to communicate, and still genuinely contested.

The Problem Statement

To understand the proposal precisely, we have to return to the specific failure of perturbative quantum gravity described in Article III.

Newton’s gravitational constant G has a negative mass dimension.

To see why G has a negative mass dimension, we look at Newton’s Law of Universal Gravitation,

$F = G \frac{m_1 m_2}{r^2}$

If we rearrange this to solve for G, we get,

$G = \frac{F \cdot r^2}{m_1 \cdot m_2}$

Force (F): $[M][L][T]^{-2}$ (from F = ma)

Distance squared ( $r^2$ ): $[L]^2$

Mass squared ( $m_1 m_2$ ): $[M]^2$

Plugging these in:

$[G] = \frac{([M][L][T]^{-2}) \cdot [L]^2}{[M]^2} = [M]^{-1} [L]^3 [T]^{-2}$

The exponent of -1 on the M is what we mean by a negative mass dimension.

In quantum field theory, this means that the effective gravitational coupling* grows with energy. At low energies, this growth is negligible. Gravity is the weakest of the four forces by many orders of magnitude at the scale of elementary particles. But as energy approaches the Planck scale, the gravitational coupling becomes of order one. In the perturbative calculation, each new loop in the Feynman diagram series introduces a new type of infinity that cannot be absorbed into the existing parameters of the theory. New parameters must be introduced at each order. The theory requires infinitely many measured inputs and loses all predictive power. This was proven definitively by Goroff and Sagnotti in 1985.

The conclusion most physicists drew was that quantum field theory simply cannot describe gravity. Some genuinely new framework must be introduced at the Planck scale.

Asymptotic safety challenges that conclusion at its root. The problem is not that quantum field theory fails for gravity. The problem is that the gravitational coupling grows with energy when tracked using perturbation theory. But coupling constants do not always grow with energy. In quantum chromodynamics, the strong coupling constant decreases toward zero at high energies. This is asymptotic freedom. What if, instead of growing without bound, the gravitational coupling approaches a finite, non-zero fixed value? What if Newton’s constant does not blow up but simply stops running?

At such a fixed value, the theory is exactly scale-invariant. The coupling constants are finite. The infinities of perturbative quantum gravity are not cancelled or removed. They are rendered irrelevant, because the coupling constants that generate them never reach the values at which the perturbation series breaks down. The theory becomes safe from divergences at high energies.

Fixed Points and Theory Space

To understand what an asymptotic fixed point means mathematically, it helps to understand the space in which it exist.

The renormalisation group is a framework for tracking how the coupling constants of a theory change with energy scale. This change is called running. The electric charge runs with energy in quantum electrodynamics. Newton’s constant runs in gravity. The renormalisation group describes this running precisely.

As one follows the running of coupling constants to arbitrarily high energies, one of three things can happen. The couplings can grow without bound, which signals a breakdown of the theory. They can approach zero, which is asymptotic freedom. Or they can approach a finite, non-zero fixed value – a fixed point of the renormalisation group flow. At a fixed point, the rate of change of all coupling constants with energy is exactly zero. The theory is scale-invariant.

Now consider the space of all possible gravitational theories and every possible combination of curvature invariants with their own coupling constants. This is called theory space. It is infinite-dimensional. It contains the Einstein-Hilbert action* with its two parameters, plus all possible higher-curvature corrections, each weighted by its own coupling. Every point in theory space corresponds to a different theory of gravity. The renormalisation group flow defines a vector field on this space, a set of directions along which each theory evolves as the energy scale changes.

A fixed point is a specific point in theory space where this vector field vanishes, where all coupling constants simultaneously stop running. The structure of the flow near the fixed point determines whether the theory is predictive. The fixed point is approached from a specific surface in theory space called the ultraviolet critical surface. This surface is spanned by all the coupling constants that are attracted to the fixed point as energy increases. Its dimension equals the number of such relevant couplings.

If the critical surface has finite and small dimension, the theory is highly predictive. The vast majority of coupling constants in theory space are determined by the requirement that the renormalisation group trajectory lies on the critical surface. Only the few relevant couplings need to be measured. If the critical surface is infinite-dimensional, the theory requires infinitely many measurements and is not predictive. The existence of a fixed point with a finite-dimensional critical surface is both the mathematical requirement and the physical meaning of asymptotic safety.

In 1976 and 1979, Steven Weinberg proposed that gravity might possess exactly such a fixed point: a non-trivial, non-zero fixed point in the renormalisation group flow of the gravitational coupling constants. He called the property asymptotic safety.

Weinberg’s Proposal and Its Initial Difficulty

Weinberg’s original argument for this fixed point was an analogy. In exactly two spacetime dimensions, the renormalisation group flow of gravity can be computed perturbatively. A non-trivial fixed point exists in two dimensions and in a small perturbative expansion around two dimensions, the two plus epsilon expansion. Weinberg proposed that this fixed point persists in the continuation to four dimensions.

The analogy was suggestive but not calculable. The continuation from two to four dimensions requires crossing a regime where the perturbative expansion breaks down. Whether the two-dimensional fixed point has any relation to a fixed point in four dimensions could not be determined by available methods. Asymptotic safety remained an intriguing possibility without a systematic way to investigate it.

Reuter’s Breakthrough: The Effective Average Action

The situation changed in 1998, when Martin Reuter adapted a powerful non-perturbative technique to the gravitational field.

The technique is the effective average action, developed by Christof Wetterich and Tim Morris in the early 1990s. It is a scale-dependent version of the quantum effective action that interpolates between the classical action at high energy scales and the full quantum effective action at low energies. Its dependence on the energy scale is governed by an exact flow equation, the Wetterich-Morris equation, that describes how the action changes as degrees of freedom are progressively integrated out in momentum space.

The equation is exact in the sense that it makes no approximation about the strength of the coupling constants. It does not require gravity to be weakly coupled and does not rely on perturbative expansion around a fixed background metric. This is precisely what the quantum gravity problem requires. Perturbation theory fails because gravity becomes strongly coupled at the Planck scale. The effective average action method does not care. It tracks the renormalisation group flow exactly, regardless of coupling strength.

To understand why this requires a truncation*, recall the structure of theory space. The effective average action lives in this infinite-dimensional space. The exact flow equation describes its trajectory precisely but cannot be solved in the full infinite-dimensional space. It would require tracking infinitely many coupling constants simultaneously. One must restrict attention to a finite-dimensional subspace. The simplest restriction, the Einstein-Hilbert truncation, keeps only the two operators in Einstein’s original action: the Ricci scalar and the cosmological constant. Within this subspace, Reuter computed the renormalisation group flow and found a non-trivial fixed point in four spacetime dimensions.

This was the first direct, non-perturbative evidence for the asymptotic safety scenario in four dimensions, obtained without any extrapolation from two dimensions. The fixed point, now called the Reuter fixed point, is UV-attractive. As the energy scale increases toward the Planck scale, trajectories in theory space are pulled toward it. It corresponds to a low-dimensional ultraviolet critical surface. These are precisely the properties that Weinberg’s proposal required.

Physically, what Reuter found is when Newton’s constant and the cosmological constant are running is tracked non-perturbatively, they do not grow without bound at the Planck scale. They approach finite fixed values. The scale-invariant regime predicted by asymptotic safety appears in the renormalisation group flow exactly where it is needed, at and above the Planck scale.

The Evidence: What Enlarging the Truncation Reveals

The central question raised by Reuter’s result was whether the fixed point was real or an artifact of restricting to only two coupling constants. The full theory contains infinitely many. If the fixed point disappeared or changed character when more operators were added, it would not be a genuine feature of quantum gravity. If it persisted and its properties converged, that would be strong evidence for its reality.

Since 1998 the fixed point has been investigated in progressively more complex truncations, including higher powers of the Ricci scalar, Ricci tensor squared terms, Weyl curvature squared terms, and the full momentum-dependent structure of the graviton propagator. The finding has been consistent: the non-Gaussian fixed point persists across every truncation studied, and its critical exponents and the dimensionality of its critical surface remain qualitatively stable as the truncation is enlarged.

The evidence comes from several independent computational methods: the two plus epsilon expansion, perturbation theory of higher-derivative theories, a large-N expansion in the number of matter fields, and truncated functional flow equations. These approaches make different assumptions and encounter different technical limitations. Their agreement on the existence and qualitative properties of the fixed point is the programme’s strongest argument.

The evidence is not conclusive. All existing computations apply a truncation to the infinite-dimensional theory space before looking for a fixed point. Whether the fixed point found in the truncation reflects a genuine fixed point of the full theory, or is an artifact of the approximation, cannot currently be determined. A rigorous proof of the fixed point’s existence in the full, untruncated theory space does not yet exist. It is the programme’s central open problem.

Does the Fixed Point Survive Standard Model Fields?

One of the most important developments since Reuter’s original work has been the systematic investigation of whether the Reuter fixed point persists when the quarks, leptons, gauge bosons, and Higgs field of the Standard Model are coupled to gravity.

This is not a technical detail. A correct theory of quantum gravity must describe gravity in the presence of the matter that actually exists in the universe. A fixed point that exists for pure gravity but disappears when matter is added would not constitute a physically viable ultraviolet completion.

Extensive calculations coupling the gravitational flow equations to the renormalisation group flows of Standard Model matter fields have found that, for the particle content of the Standard Model, the Reuter fixed point persists. Gravitational and matter fluctuations collectively approach a fixed point in the ultraviolet. This constrains the allowed matter content of asymptotically safe theories and connects the programme directly to particle physics phenomenology.

More than that, the requirement that the theory be asymptotically safe places constraints on the values of coupling constants at low energies. The fixed point determines their Planck-scale values, and the renormalisation group running from the Planck scale to observable energies predicts what should be measured in experiments.

The Higgs Mass and Beyond

The most striking result of the matter-coupled asymptotic safety programme is a prediction for the Higgs boson mass made before the Higgs was observed.

In 2010, Mikhail Shaposhnikov and Christof Wetterich showed that if quantum gravity is asymptotically safe and the Standard Model is the correct description of matter up to the Planck scale, then the quartic self-coupling of the Higgs field is an irrelevant coupling at the ultraviolet fixed point. An irrelevant coupling is one whose value at the fixed point is entirely determined by the fixed point itself, not by any independent experimental input. Its Planck-scale value is therefore predicted, not free. Running this predicted value down from the Planck scale to the electroweak scale using the Standard Model’s renormalisation group equations produces a prediction for the Higgs mass of approximately 126 GeV. In 2012, the ATLAS and CMS experiments at the Large Hadron Collider announced the discovery of the Higgs boson with a measured mass of approximately 125.1 GeV.

A 3D event display from the CMS detector showing a Higgs boson candidate decaying into two photons. This experimental data from 2012 confirmed the Higgs mass at approximately 125 GeV—a value strikingly close to the 126 GeV prediction made by the Asymptotic Safety framework two years prior.

Breaking down the visuals –

The Yellow Lines (Tracks): These represent the paths of charged particles (like pions or protons) produced in the collision. They are measured by the inner “tracker” of the CMS detector.

The Green Towers/Lines: They represent large deposits of energy in the Electromagnetic Calorimeter* (ECAL). Because there are no yellow tracks leading directly to these green blocks, physicists can infer they were created by photons—neutral particles that don’t leave tracks but carry a lot of energy.

The Blue Translucent Cylinder: This is a 3D wireframe representing the volume of the CMS detector itself, which is roughly 15 meters in diameter and 21 meters long.

The Dashed Yellow Lines: These are “extrapolated” paths showing exactly where the two high-energy photons traveled from the central vertex to the calorimeter.

The prediction was made two years before the measurement it anticipates, and falls within the measurement’s uncertainty. Whether this constitutes a genuine confirmed prediction of asymptotic safety or a numerical coincidence is contested. The calculation is sensitive to the top quark mass and to assumptions about the particle content above the Planck scale. The agreement is real, but its significance requires caution.

Beyond the Higgs mass, asymptotic safety has produced constraints on the top quark mass, predictions for gauge coupling values, and investigations of whether the fine-structure constant can be derived from first principles within the framework. These are genuine quantitative predictions that distinguish the programme from approaches making no contact with low-energy particle physics.

Dimensional Reduction

A separate and unexpected result of the asymptotic safety framework concerns the effective dimension of spacetime at the Planck scale.

In four-dimensional flat spacetime, a diffusing random walk spreads in a way that scales with the dimension of the space. The spectral dimension is defined by how this spreading changes with diffusion time, and in flat four-dimensional space it is exactly four. Within the asymptotic safety framework, the spectral dimension can be computed as a function of energy scale by analysing the structure of the graviton propagator near the Reuter fixed point.

The result: at macroscopic distances the spectral dimension is four. At the Planck scale, in the vicinity of the fixed point, it approaches two. This dimensional reduction is not assumed. It emerges from the renormalisation group flow as the theory approaches the fixed point, where the effective degrees of freedom of gravity reorganise into an essentially two-dimensional structure. The smooth four-dimensional spacetime of general relativity is not fundamental in this picture. It is what the scale-invariant regime looks like when probed at low energies.

The same dimensional reduction has been independently found in loop quantum gravity and in causal dynamical triangulations. The convergence of different theoretical frameworks on the same effective dimension at the Planck scale is either a significant clue about the structure of quantum spacetime, or a coincidence arising from some shared but unstated mathematical assumption. Article VII examines this convergence in the context of the broader question of whether spacetime is fundamentally discrete.

The Unitarity Problem and the Swampland

Asymptotic safety faces two serious criticisms beyond the truncation problem.

The first is unitarity. Enlarging the truncation beyond the Einstein-Hilbert approximation introduces higher-derivative operators, terms in the gravitational action involving more than two derivatives of the metric. Such terms generically introduce new propagating modes in the graviton spectrum with negative norm. These ghost states carry negative probability, violating the basic requirement that a physical quantum theory produce non-negative probabilities. Whether the Reuter fixed point is consistent with unitarity has not been established. Some researchers argue the ghost modes are artefacts of the truncation. Others argue they represent a genuine obstruction. The question remains open.

The second criticism comes from the swampland programme, which originated in string theory. The swampland is the set of quantum field theories that appear internally consistent but cannot be consistently coupled to quantum gravity. String theorists have developed a set of conjectures, including the weak gravity conjecture, the swampland distance conjecture, and the de Sitter conjecture, that characterise the boundary between consistent and inconsistent theories. Recent work has investigated whether asymptotically safe gravity satisfies these conjectures. The results are mixed and the relationship between asymptotic safety and the swampland remains an active frontier. One proposal suggests that the two approaches occupy different energy regimes: string theory in the deep ultraviolet, an asymptotically safe scaling regime at intermediate energies, and the Standard Model in the infrared. Whether this reconciliation is viable has not been established.

Background Independence

One structural feature of asymptotic safety deserves attention against the series’ earlier articles.

Loop quantum gravity is explicitly background-independent. String theory in its standard perturbative formulation is background-dependent. Asymptotic safety, as standardly formulated using the effective average action, requires a background metric to define the coarse-graining procedure. It is therefore, at least in its present formulation, background-dependent in a technical sense, a limitation acknowledged by practitioners of the programme.

The background metric in the effective average action plays a double role: it defines both the gauge fixing and the regulator used to implement the scale separation. Separating these roles cleanly is a technically difficult problem. Work on background-independent formulations of the functional renormalisation group for gravity is ongoing, but a fully background-independent formulation of asymptotic safety does not yet exist. Whether this is a fundamental obstacle or a technical challenge awaiting resolution is a question the programme has not yet answered definitively.

Where Asymptotic Safety Stands

Asymptotic safety is the most conservative approach to quantum gravity examined in Group 2. It requires no new fundamental objects, no extra dimensions, and no modification of general relativity’s degrees of freedom. Its central physical claim is that gravity, at energies at and above the Planck scale, enters a regime of exact quantum scale invariance: a fixed point at which coupling constants cease to run, divergences are rendered irrelevant, and the theory is fully predictive from a finite number of measured inputs.

The theory has produced evidence for this from multiple independent computational approaches, a quantitative prediction for the Higgs mass consistent with observation, constraints on Standard Model parameters, and a picture of Planck-scale spacetime as effectively two-dimensional. None of these results constitutes a proof, and the truncation problem remains the central obstacle to any definitive conclusion.

Its unresolved difficulties, the truncation problem, the unitarity question, the swampland tension, and the absence of full background independence, are genuine obstacles, not minor technical details. Any one of them could prove fatal. What asymptotic safety offers, if it works, is the most economical possible resolution of the quantum gravity problem: not a new picture of reality, but the discovery that the existing picture was already sufficient, provided one looked for its fixed point.

The three approaches of Group 2 have each made genuine progress, and each faces a different set of unresolved difficulties. Loop quantum gravity has not proven its classical limit. String theory has the landscape. Asymptotic safety has not proven its fixed point. What unites all three is the absence of experimental confirmation, reflecting the fundamental difficulty of probing physics at scales sixteen orders of magnitude beyond the reach of current accelerators.

Next, in this series, we turn from the proposed theories to the physical consequences they must explain and ask the question that all three approaches must eventually answer: is spacetime fundamentally discrete?

coSym. | Exploring the Universe: Within & Beyond

Special Series

Topics