Cosmological Backreaction and Apparent Acceleration
Scrollable formula. Use Left and Right arrow keys to reveal hidden content.
Scrollable table. Use Left and Right arrow keys to reveal additional columns.
Scrollable figure. Use Left and Right arrow keys to inspect hidden content.
General relativity is nonlinear, so averaging an inhomogeneous Universe need not reproduce the evolution of an exactly homogeneous metric. That backreaction is a real mathematical and numerical question. Explaining the observed late-time acceleration is a much stronger claim: one needs constraint-satisfying realistic structure, gauge-robust observables on the past light cone, and a joint fit to distance, CMB, BAO, and growth data without a cosmological constant or dark-energy component. Through 2026-08-20, full-relativistic studies in tested late-Universe regimes find small corrections rather than acceleration-sized effects. Exhaustive exclusion is absent, but evidence for backreaction as the acceleration mechanism is insufficient.
Helpful background. Cosmological Backreaction derives the averaging problem and Buchert equations. This record assesses the stronger observational claim while keeping foliation, gauge, domain, and light-cone effects explicit.
From averaged evolution to apparent acceleration
Section titled “From averaged evolution to apparent acceleration”The finite proposition is whether statistically realistic, constraint-satisfying inhomogeneous initial data in general relativity, with specified matter, can produce a gauge-robust forward calculation that matches the apparent acceleration and the joint distance, growth, and light-cone observations without dark energy. Positive acceleration of one domain scale factor in one slicing is not sufficient.
The result depends on spatial versus null or light-cone averaging; observer congruence and foliation; domain scale and window; kinematical versus averaged-curvature terms; Newtonian, post-Friedmann, or full-relativistic dynamics; dust versus velocity dispersion, radiation, baryons, and compact objects; topology and boundaries; initial constraints; and the distinction between an expansion statistic and an inferred observable.
| Assessment at 2026-08-20 | Meaning |
|---|---|
| Question status | Partially resolved: tested realistic regimes disfavor acceleration-sized backreaction, while exhaustive mathematical exclusion and precision light-cone effects remain open |
| Evidence status | Insufficient evidence for explaining acceleration; supported for smaller relativistic and averaging corrections |
| Confidence | Moderate; the numerical trend is strong, but dynamic range, gauge, matter content, and observation matching prevent a universal theorem |
| Domain | Late-time inhomogeneous cosmologies studied with scalar averaging, perturbative gauge analyses, and current full-GR numerical-relativity methods |
| What could change the conclusion | A gauge-robust observation-matched relativistic calculation producing acceleration, a general no-go theorem, or multi-code evidence changing the expected observable magnitude |
A starting project: check averaging before simulating acceleration
Section titled “A starting project: check averaging before simulating acceleration”This topic rewards comfort with the Einstein constraints, fluid kinematics, and numerical error analysis. Read the linked cosmological-backreaction background, then Buchert 2000, the averaging theorem, Eqs. (10a)–(10d), and Corollary 3, Eqs. (13a)–(13b), open PDF. Write down the dust, foliation, and comoving-domain assumptions. Derive the integrability relation from the averaged evolution and conservation equations, keeping the domain dependence throughout.
For a first numerical task, implement those averages on a known homogeneous dust solution and on one explicitly supplied inhomogeneous dust dataset from a collaborating group. Verify volume weighting, the expansion variance, shear normalization, and the integrability residual under refinement. If no appropriate dataset is available, complete the analytic identity and homogeneous implementation before proposing a simulation. Arbitrarily prescribing a backreaction function produces a closure model, not an independently evolved spacetime.
The deliverable is a small averaging routine with analytic controls and a residual-versus-resolution comparison, accompanied by the slicing and domain definitions. This baseline can run on a laptop when data are modest. Generating reliable inhomogeneous data requires numerical-relativity supervision, initial-data and evolution software, and potentially cluster access; predicting apparent acceleration also requires null-geodesic observables and observer sampling.
A candidate extension is to compare one averaging diagnostic with one light-cone distance statistic in the same converged spacetime. Check existing calculations and code capabilities before claiming novelty. Pivot if the effect scales with the grid, the constraints deteriorate, or the observer prescription changes alongside the foliation. A nonzero averaged term is an interesting diagnostic but is not yet an explanation of supernova distances. This staged project makes the missing observable calculation explicit before investing in a large simulation.
Exact averaging results and their scope
Section titled “Exact averaging results and their scope”For irrotational dust on flow-orthogonal hypersurfaces, scalar averaging yields exact domain evolution equations. Kinematical backreaction and averaged spatial curvature enter the effective acceleration and integrability conditions; Buchert derives them in §§2–3, especially Eqs. (9)–(13), with closure issues discussed in §4 Buchert 2000, §§2–4. The equations establish that nonlinear averaging need not commute with evolution. They do not select the foliation, close the system, or produce an observed distance–redshift relation by themselves.
Current numerical-relativity cosmology provides the strongest empirical-theoretical adverse evidence to an acceleration-sized effect. Aurrekoetxea, Clough, and Lim review formulations, gauge and boundary issues, diagnostics, backreaction, observables, and open problems in §2.3, §§5.1–5.2, and §7 2025, §2.3, §§5.1–5.2, and §7. Across tested realistic late-time regimes, relativistic corrections can matter at precision-survey levels but do not replace accelerated expansion.
Gauge dependence is not a cosmetic issue. Spatial averages can acquire large terms under slicing changes that do not correspond to a changed observation. Verweg, Jones, and van de Weygaert analyze foliation dependence and a route toward gauge-robust quantities in §§2–6 2023, §§2–6. Their proposal is a conceptual check, not yet a complete nonlinear survey likelihood.
Results and open questions at 2026-08-20
Section titled “Results and open questions at 2026-08-20”| Claim or evidence | What it supports | What it does not establish |
|---|---|---|
| Buchert scalar equations | Exact noncommutation of averaging and evolution for irrotational dust domains | Closure, gauge-independent observables, or acceleration data |
| Current full-GR simulation synthesis | Small corrections and no acceleration-sized effect in tested realistic regimes | Exhaustive coverage of every spacetime, observer, or matter model |
| Gauge and foliation analysis | Large apparent averaging terms can be coordinate or slicing dependent | A completed nonlinear gauge-invariant forward model |
| Restricted no-go calculations | Negative results in specified matter and perturbative settings | A universal theorem covering all averaging and light-cone constructions |
| Weak-limit critique literature | Limits of promoting one mathematical framework to every fitting problem | Positive evidence that backreaction explains acceleration |
| Independent Einstein-Toolkit averaging study | Negligible backreaction on selected Newtonian-gauge-like hypersurfaces, with resolved subvolume curvature and mass-flow effects | A comparison between slicings, full survey dynamic range, or a joint observational fit |
The simulation synthesis receives the greatest weight for magnitude in realistic regimes, while exact equations carry greatest weight for mechanism existence. Critiques of no-go results preserve logical possibilities; they do not supply the missing observation-matched positive calculation.
The strongest negative and permissive interpretations
Section titled “The strongest negative and permissive interpretations”The negative interpretation says that weak-field structure, full-GR simulations, and scale separation consistently keep late-time corrections too small to mimic acceleration. A representative restricted calculation shows why in a matter-dominated perturbative setup, while explicitly presenting itself as a step toward rather than a complete proof Kasai, Asada, and Futamase 2006, Eqs. (1)–(14) and final paragraph. Its scope must remain attached.
The permissive interpretation emphasizes that spatial averaging, observational fitting, weak limits, and null-cone statistics are different problems. Buchert et al. challenge broad readings of Green–Wald-type results in §§2–6 2015, §§2–6. This successfully blocks a universal inference from one framework, but does not demonstrate acceleration. The unresolved discriminator is a gauge-robust, converged light-cone calculation compared with the same observations used to infer acceleration.
Counterexamples, null results, and failed routes
Section titled “Counterexamples, null results, and failed routes”A large kinematical backreaction term that changes under slicing is a counterexample to treating its magnitude as an observable. A domain-averaged acceleration without ray tracing is a counterexample to equating an effective scale factor with supernova or BAO inference. Restricted no-go calculations are counterexamples to particular broad claims but fail when promoted beyond their matter, perturbative, and averaging assumptions.
Independent Einstein-Toolkit work measures averaged quantities on Newtonian-gauge-like hypersurfaces and finds negligible cosmic backreaction on those hypersurfaces, while resolving curvature fluctuations up to roughly ten percent and matter-flow density shifts up to roughly five percent in selected approximately 200 Mpc subvolumes Oestreicher and Koksbang 2024, abstract and §§2–5. This is a hypersurface-specific null and magnitude check, not a quantitative comparison between slicings. Its limited box, matter model, and dynamic range prevent it from being the final cosmological likelihood. The strongest current null is the absence of a verified calculation that produces acceleration and matches the joint data without dark energy.
What would change the assessment
Section titled “What would change the assessment”A positive result must start from constraint-satisfying statistically realistic data, demonstrate resolution, box, boundary, gauge, and observer convergence, ray-trace observables, and fit supernova, BAO, CMB, and growth likelihoods without a hidden acceleration component. Independent codes should reproduce the decisive observable. A negative result would become stronger if a theorem covered realistic matter, appropriate asymptotics, and the actual light-cone fitting problem, or if multi-code simulations bounded every plausible correction well below the needed level under controlled dynamic-range extrapolation.
Research checklist
Section titled “Research checklist”- Specify matter, topology, boundaries, initial constraints, gauge, lapse and shift, observer congruence, and averaging hypersurface.
- Define the domain window and distinguish kinematical backreaction from averaged curvature and lapse terms.
- Demonstrate resolution, box-size, boundary, and dynamic-range convergence against matched FLRW and Newtonian controls.
- Track gauge-invariant or operational observables rather than coordinate averages alone.
- Ray-trace redshift, distance, lensing, and number-count observables for a documented observer sample.
- Include velocity dispersion, radiation, baryons, and compact structures when their omission affects the claimed scale.
- Compare with a joint likelihood using the same calibration and nuisance assumptions as the standard model.
- Release initial data, code revision, analysis windows, null geodesics, and likelihood interface.
Source and data map
Section titled “Source and data map”Buchert provides the exact scalar-averaging formalism. Aurrekoetxea, Clough, and Lim provide the current numerical-relativity synthesis and magnitude assessment. Verweg et al. analyze the gauge problem; Oestreicher and Koksbang provide an independent numerical null and magnitude check on one specified Newtonian-gauge-like hypersurface, not a cross-slicing experiment. Kasai et al. supply a restricted adverse calculation; Buchert et al. supply the strongest scope critique.
No end-to-end simulation was rerun. No public artifact was verified that connects realistic constraint-satisfying initial data, full-GR evolution, null geodesics, and a joint cosmological likelihood. Reproduction therefore requires the initial-data generator, matter and gauge, boundaries, formulation and code commit, resolution and box hierarchy, domain and observer sampling, ray-tracing definitions, matched controls, and likelihood. Availability of an evolution code alone is not reproduction of the acceleration claim.
References
Section titled “References”- Aurrekoetxea, J. C., Clough, K., and Lim, E. A. 2025, “Cosmology using numerical relativity,” Living Reviews in Relativity 28, 5, DOI 10.1007/s41114-025-00058-z.
- Buchert, T. 2000, “On Average Properties of Inhomogeneous Fluids in General Relativity: Dust Cosmologies,” General Relativity and Gravitation 32, 105–125, DOI 10.1023/A:1001800617177.
- Buchert, T., et al. 2015, “Is There Proof That Backreaction of Inhomogeneities Is Irrelevant in Cosmology?” Classical and Quantum Gravity 32, 215021, DOI 10.1088/0264-9381/32/21/215021.
- Kasai, M., Asada, H., and Futamase, T. 2006, “Toward a No-Go Theorem for an Accelerating Universe through a Nonlinear Backreaction,” Progress of Theoretical Physics 115, 827–832, DOI 10.1143/PTP.115.827.
- Oestreicher, A., and Koksbang, S. M. 2024, “Backreaction in Numerical Relativity: Averaging on Newtonian gauge-like hypersurfaces in Einstein Toolkit cosmological simulations,” Open Journal of Astrophysics 7, arXiv:2408.03049v1, DOI 10.33232/001c.124744.
- Verweg, A., Jones, B. J. T., and van de Weygaert, R. 2023, “Averaging over cosmic structure: Cosmological backreaction and the gauge problem,” arXiv:2310.19451v2, arXiv.