Quantum Gravity

Quantum gravity asks for a quantum description of geometry itself, but that phrase hides several different problems. At low energy, quantized metric fluctuations already form a predictive effective field theory. At shorter distances, candidate microscopic programmes must define physical states and observables, recover general relativity and local quantum field theory, and explain which calculations could distinguish them experimentally. This volume develops those tasks in that order. Its recurring discipline is to separate a dimensionful scale from a controlled expansion parameter, kinematics from dynamics, a continuum limit from a suggestive phase, and a bound on one operator from evidence for quantum gravity as a whole.

Helpful background. Classical Field Modes and Canonical Variables supplies the field-theory language used for gravitons; Einstein Equations supplies the classical dynamics and curvature conventions; and Quantum Information for Black-Hole Physics supplies states, channels, entanglement, and operational tests. Each chapter adds a bounded bridge where a more specialized prerequisite first enters.

What must a quantum theory of gravity accomplish?

A successful theory must answer more than “what happens near the Planck length?” It must identify a state space or algebra of observables, provide a dynamical rule, control redundancies generated by diffeomorphisms, and recover the phenomena already described by general relativity and quantum field theory. In a semiclassical regime it must explain why matter sees an approximately local Lorentzian geometry. In a microscopic regime it must say which structures are fundamental and how dimension, causal order, and metric relations emerge. For empirical use it must connect those structures to a source model, propagation or interaction law, detector observable, and statistical test.

These requirements operate at different logical levels:

Level	Question	Minimum evidence
Kinematics	What are the states, operators, constraints, and gauge identifications?	A defined representation with its hypotheses and physical interpretation.
Dynamics	Which amplitudes, constraints, or evolution law relate admissible states?	A well-defined rule with gauge, boundary, and unitarity checks.
Continuum recovery	Does a regime reproduce a smooth four-dimensional Lorentzian spacetime?	Scaling or coarse-graining control, not merely a large label or many degrees of freedom.
Correspondence	Are Einstein dynamics and local quantum fields recovered with controlled corrections?	A dimensionless expansion and discriminating limits.
Observables	Which gauge-invariant quantities can a physical observer access?	Relational or asymptotically anchored quantities with a stated algebra.
Empirical discrimination	Which data favor or exclude a specified mechanism?	A theory-to-observable map, calibrated likelihood, systematics, and alternatives.

No single success can silently stand in for the other five. A discrete area spectrum is not yet a continuum dynamics; a fixed point in a truncation is not yet a gauge-invariant observable; and a phenomenological dispersion bound is not a verdict on every microscopic completion.

Scales are not regimes

With $c=\hbar=k_{\mathrm B}=1$ and $G$ explicit, Newton’s constant has mass dimension $-2$. The Planck scales are therefore

$$M_{\mathrm{Pl}}=G^{-1/2}, \qquad \ell_{\mathrm{Pl}}=G^{1/2}, \qquad t_{\mathrm{Pl}}=G^{1/2}.$$

They are useful conversion scales, not by themselves a theorem about a minimum length or a sharp phase transition. A calculation is controlled by dimensionless combinations appropriate to its state and observable. Examples include $GE^2$ for a scattering process, $G\mathcal R$ for curvature scale $\mathcal R$, $G\sqrt{s}/b$ for the strength of high-energy deflection at impact parameter $b$ and center-of-mass energy $\sqrt{s}$, and ratios of higher-derivative terms to the Einstein term. Large local energy does not force large quantum-gravity corrections when the relevant invariant is small, while small individual quanta can collectively produce strong curvature or a large gravitational radius.

For example, high-energy scattering has several independent ratios, with $s=-(p_1+p_2)^2\gt0$ the squared center-of-mass energy. The eikonal phase is governed by $Gs$, the deflection by $G\sqrt{s}/b$, and quantum length corrections by $G/b^2$. Increasing the impact parameter can keep deflection and quantum corrections small even when $Gs$ is large. In that regime repeated exchanges must be resummed, but a large phase alone does not mean a large deflection. Planck units and invariant expansion parameters develops the dimensional analysis before the amplitude calculations.

This distinction also identifies the one firm starting point. Around a background in a low-curvature regime, write

$$g_{\mu\nu}=\bar g_{\mu\nu}+\kappa h_{\mu\nu}, \qquad \kappa^2=32\pi G.$$

The massless spin-2 field $h_{\mu\nu}$ can be gauge fixed and quantized. The most general low-energy action compatible with diffeomorphism invariance is an operator expansion,

$$S_{\mathrm{EFT}} =\int d^4x\,\sqrt{-g}\left[ \frac{R-2\Lambda}{16\pi G} +c_1R^2+c_2R_{\mu\nu}R^{\mu\nu} +c_3R_{\mu\nu\rho\sigma}R^{\mu\nu\rho\sigma} +\cdots\right].$$

At momenta well below the matching scale, loop and derivative orders organize the error. Local analytic terms contain coefficients sensitive to ultraviolet physics; nonanalytic infrared terms can yield universal long-distance predictions. Perturbative nonrenormalizability means that infinitely many operators are needed at arbitrarily high energy, not that low-energy quantum general relativity is inconsistent. Donoghue gives this modern separation and its observable qualifications in Donoghue 2023, §§2–6.

Constraints change what counts as local

In an ordinary nongauge theory, a field value at a coordinate point can be an operator. In gravity, an infinitesimal diffeomorphism moves both the field and the coordinate label. Physical observables must instead be relational, asymptotically anchored, or defined after a controlled gauge choice. A matter operator $\phi(x)$ may be dressed perturbatively as

$$\Phi(x)=\phi\!\left(x^\mu+V^\mu[h;x]\right),$$

where the functional $V^\mu$ transforms so that $\Phi$ is invariant to the stated order. The dressing necessarily carries a gravitational field to an anchor and different choices differ by radiative data. Consequently, exact local tensor factorization and ordinary microcausality do not transfer to gravity without qualifications. The explicit perturbative construction and its obstruction to compactly supported gauge-invariant operators are set out in Donnelly and Giddings 2016, §§2–5.

This issue connects apparently separate chapters. Canonical constraints ask which states are physical; gravitational path integrals divide by the same gauge volume; loop representations quantize a particular holonomy–flux algebra; emergence programmes must recover an approximately local algebra; and experimental claims must specify which dressed or asymptotic observable a detector measures.

The thirteen-part route

The chapters follow a dependency chain rather than a list of competing schools.

Chapter	Central question	Result carried forward
Quantum-Gravity Scales, Regimes, and Evidence	Which invariant parameters control an approximation, and what would count as evidence?	A common language for regimes, claims, and comparison.
Gauge Symmetry, Relational Observables, and Locality	What survives diffeomorphism constraints as a physical observable or subsystem?	Relational and dressed observables with explicit boundary conditions.
Gravitons, Gauge Fixing, and Quantum Amplitudes	How are weak metric fluctuations quantized and used in scattering?	Propagators, ghosts, vertices, amplitudes, and loop organization.
General Relativity as a Quantum Effective Field Theory	Which quantum-gravity predictions are controlled without knowing the ultraviolet completion?	Power counting, universal infrared terms, and a matching boundary.
Canonical Quantum Gravity and the Problem of Time	How do constraints define physical states and relational evolution?	Wheeler–DeWitt, inner-product, clock, and semiclassical-time questions.
Gravitational Path Integrals and Boundary Conditions	What integration cycle, measure, boundary data, and saddles define a sum over geometries?	A contour-aware path-integral specification.
Loop Quantum Geometry and Spin Foams	What do connection variables, holonomies, geometric operators, and covariant amplitudes establish?	Precise kinematics and explicit tests for dynamics and continuum recovery.
Discrete, Causal, and Combinatorial Spacetime	Can order, simplices, or combinatorial quanta produce macroscopic spacetime?	Universality and finite-size criteria shared across discrete programmes.
Continuum Quantum Gravity and Renormalization-Group Flows	Can continuum couplings approach a predictive ultraviolet fixed point?	Perturbative limits and regulator-, gauge-, and truncation-aware fixed-point evidence.
Spacetime Emergence and Quantum Causal Structure	How could geometry, locality, dimension, and causal order emerge?	Recovery conditions that can compare otherwise unlike microscopic proposals.
Quantum Cosmology	What do canonical, contour, loop, and decoherence methods say in cosmological truncations?	Explicit models whose boundary, clock, and truncation dependence remains visible.
Quantum Black Holes and Microscopic Interfaces	Which microscopic data must a non-string programme supply at horizons and interiors?	Common entropy, regularity, evaporation, and correspondence targets.
Phenomenology and Operational Quantum-Gravity Tests	Which tabletop, propagation, cosmological, or black-hole observations discriminate a mechanism?	Operator-level falsifiers and uncertainty-aware inference.

Three recurring calculations

The first recurring example is a weak perturbation about Minkowski space. It starts with the transverse–traceless graviton, continues through gauge-fixed propagators and amplitudes, and becomes the low-energy EFT expansion. When the same excitation is dressed to spatial infinity, it also exposes why a local matter operator and a gravitational observable are not identical.

The second is homogeneous FLRW minisuperspace with a scalar field used as a candidate clock. It turns the Hamiltonian constraint into a Wheeler–DeWitt equation, makes WKB time and branch selection concrete, and supplies a finite- dimensional setting for lapse contours, no-boundary and tunneling proposals, decoherence, and loop-cosmology bounces. It is intentionally a truncation made before quantization; success there does not establish the result for inhomogeneous modes or the full theory.

The third is a common programme matrix. For each approach, ask: What is the kinematical state space? What defines dynamics? Which parameter is taken to a continuum limit? Which observables recover GR and local QFT? What adverse result or missing limit matters most? Which measurement could discriminate the proposal? Reusing the same questions prevents vocabulary internal to one programme from setting the standard for all others.

Choose a reader route

If this is your first encounter with quantum gravity, begin with three concrete questions. Planck units explain when a correction could be small; graviton polarizations show what is being quantized; and power counting explains why predictions remain possible with infinitely many allowed operators. Work the solved exercises before moving to competing microscopic theories. You do not need to master every programme to understand the controlled low-energy results.

For a systematic graduate route, read in order through quantum EFT, then study canonical and path-integral formulations before comparing microscopic programmes. This sequence first establishes what is already calculable and then makes the ultraviolet questions precise.

For controlled predictions, begin with scales, gravitons, quantum EFT, and phenomenology. For foundations, connect relational observables to canonical constraints, path-integral contours, and emergence. For non-string microscopic programmes, compare the loop, discrete, and continuum-RG chapters using the shared six-level table above. For cosmology and black holes, read the path-integral chapter before the two application chapters so that boundary conditions and contour choices do not disappear from the conclusions.

Computational readers can follow three transparent method benchmarks: a finite-dimensional Picard–Lefschetz contour problem, a return-probability and spectral-dimension estimator, and a finite functional-RG beta-function system. They teach convergence and sensitivity checks; they are not substitutes for production calculations in four-dimensional quantum gravity.

Before accepting any result, ask four questions:

• What is the dimensionless control parameter and first omitted correction?
• Which quantity is gauge invariant, relational, or asymptotically anchored?
• Is the result kinematical, dynamical, a continuum statement, or an empirical inference?
• Which changed assumption, adverse limit, or competing mechanism could alter the conclusion?

Boundaries with neighboring subjects

Generic flat-spacetime renormalization, BRST theory, and amplitude methods are developed in Flat-Spacetime Quantum Field Theory; this volume supplies their gravitational specialization. Quantum Fields in Curved Spacetime develops quantum fields on a prescribed geometry, while Semiclassical Gravity develops renormalized stress tensors, backreaction, stable horizon thermodynamics, and semiclassical evaporation. Here the metric or its microscopic precursor is itself a quantum degree of freedom.

Black-Hole Information develops Page curves, islands, fine-grained recovery, reconstruction, and interior consistency. The quantum-black-hole chapter here asks what microscopic inputs a non-string programme supplies to those questions. String Theory and Holographic Black Holes develops string spectra, dualities, AdS/CFT dictionaries, matrix models as realization-specific constructions, D-brane counting, and explicit string microstates. String and holographic examples enter here only to test a general emergence or evidence criterion.

Classical modified-gravity models remain distinct from quantum counterterms or proposed ultraviolet dynamics even when their actions contain the same higher-curvature invariants. Likewise, a detector constraint belongs first to the named operator, source population, and likelihood; translating it into a microscopic claim requires a valid matching calculation.

Results and open questions

The comparison below is current through 2026-08-20. “Controlled” refers to the stated regime, not to a universal completion.

Result or programme	Strongest present evidence	Boundary of the claim
Low-energy quantum general relativity	A systematic EFT expansion with universal infrared terms and renormalized local operators, summarized in Donoghue 2023, §§2–6	It does not determine the ultraviolet completion or remain predictive once its expansion parameters cease to be small.
Canonical and covariant formulations	Exact classical constraints, formal quantum constraint equations, perturbative gauge fixing, and controlled finite-dimensional saddle analyses	A symbol such as $\widehat{\mathcal H}\Psi=0$ does not by itself define an operator domain, physical inner product, contour, or anomaly-free dynamics.
Loop and spin-foam approaches	A mathematically precise holonomy–flux kinematics, geometric operators, and large-spin vertex asymptotics; see Ashtekar and Lewandowski 2004, §§II–VI and Perez 2013, Parts II–IV	Physical dynamics, a controlled four-dimensional continuum limit, and complete recovery of low-energy observables remain separate tests.
Causal, triangulated, and combinatorial approaches	Covariant discrete structures, nontrivial phase behavior, dimension diagnostics, and analytic large-$N$ organizations in specified models	Manifoldlikeness, universality, Lorentzian continuum dynamics, and observables are not guaranteed by discreteness or large size alone.
Functional-RG asymptotic safety	Recurrent non-Gaussian fixed-point patterns across many finite truncations, reviewed with limitations in Eichhorn 2019, §§2–5	Fixed-point coordinates depend on gauge, regulator, background, parametrization, and projection; robust observables and convergence remain decisive.
Quantum-cosmology proposals	Solvable minisuperspace and symmetry-reduced models, semiclassical branches, contour analyses, and decoherence mechanisms; see Kiefer and Sandhöfer 2022, §§2–6	Truncation, factor ordering, clock, state, contour, and probability prescriptions limit extrapolation to the full universe.
Operational searches	Strong model-specific propagation, cosmological, and gravitational-wave constraints, plus advancing massive-quantum-system tests	No accepted observation uniquely identifies quantum spacetime; current null results constrain declared operators and source models, not every theory.

The ultraviolet completion and its empirical identification remain unresolved. This assessment should be reviewed after a microscopic programme demonstrates a controlled continuum limit with gauge-invariant observables and quantitative matching to GR/QFT; after a fixed-point calculation exhibits regulator- and truncation-stable physical observables; after a full-theory result upgrades a symmetry-reduced singularity resolution; or after an experiment discriminates a quantum-gravity mechanism from backgrounds and alternative models. A new kinematical spectrum, saddle, finite truncation, or isolated anomaly alone is not enough.

Check your understanding

A calculation reports a correction proportional to $GE^2$ in a scattering observable and an experiment places a tight upper bound on its coefficient. What follows, and what does not?

Solution

If the amplitude, state, gauge-invariant observable, matching convention, and experimental likelihood are specified, the calculation and bound constrain that operator or mechanism in the regime where $GE^2$ and all other expansion parameters are small. One may compare the inferred coefficient with a named microscopic model whose matching calculation is valid.

The result does not locate a universal onset at $E=M_{\mathrm{Pl}}$, prove a minimum length, exclude quantum gravity, or rank unrelated programmes. Those claims require additional control of the ultraviolet regime and a mapping from each programme to the measured operator. A dimensionful scale and a model-specific null bound cannot replace that mapping.

References

• Ashtekar, Abhay, and Jerzy Lewandowski. “Background Independent Quantum Gravity: A Status Report.” Classical and Quantum Gravity 21 (2004): R53–R152. Open manuscript.
• Donnelly, William, and Steven B. Giddings. “Observables, Gravitational Dressing, and Obstructions to Locality and Subsystems.” Physical Review D 94 (2016): 104038. DOI. Open manuscript.
• Donoghue, John F. “Quantum General Relativity and Effective Field Theory.” arXiv:2211.09902v2 (2023). Open manuscript.
• Eichhorn, Astrid. “An Asymptotically Safe Guide to Quantum Gravity and Matter.” Frontiers in Astronomy and Space Sciences 5 (2019): 47. DOI. Open manuscript.
• Kiefer, Claus, and Barbara Sandhöfer. “Quantum Cosmology.” Zeitschrift für Naturforschung A 77 (2022): 543–559. Open manuscript.
• Perez, Alejandro. “The Spin-Foam Approach to Quantum Gravity.” Living Reviews in Relativity 16 (2013): 3. DOI. Open manuscript.