{"id":"6474eb70-6106-4f27-ade2-8fc00c1cc0a0","arxiv_id":"2607.28800","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Nonlinear axion-U(1) inflation, simulated on a lattice and fed into N-body simulations, boosts small-scale structure and high-redshift halo counts by up to ~200% relative to standard single-field initial conditions.","lead":"This paper runs the first cosmological simulations that start from a lattice simulation of inflation and evolve to present-day galaxy structure, applied to axion-U(1) inflation. The result is a new way to predict how nonlinear physics during inflation shows up in galaxy surveys and 21-cm observations.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Sourced and vacuum curvature perturbations are assumed uncorrelated, but no measurement is shown; if cross-correlation is non-negligible the PNG-only comparison (AU1 vs G) breaks down.","rationale":"Having read the paper, the most load-bearing assumption is indeed the statistical independence of the sourced and vacuum components of the curvature perturbation. The quantitative headline (60% halo excess after power matching) is only interpretable as non-Gaussianity if the G simulation is a valid Gaussian counterpart with the same power spectrum but no PNG. If ζ_vac and ζ_src are correlated, then the AU1 field contains a phase-coherent component that cannot be reproduced by scaling the SF field, and the G simulation does not share the same statistical properties except for the power spectrum. The paper asserts negligibility but gives no measured cross-spectrum. This is a concrete, checkable gap rather than a fatal error; the ALEF fields exist and the cross-power spectrum can be computed directly. The reader's weakest_assumption matches this, and we agree. No other concern appears equally load-bearing: the LPT/transfer-function mapping is standard, the N-body code is public, and the missing error bars are a reproducibility issue that does not invalidate the argument if the cross-correlation is indeed small. Therefore the appropriate verdict remains CONDITIONAL, with the condition being a demonstrated measurement of the cross-correlation and an error-bar estimate for the halo mass function ratios.","tokens_in":13327,"tokens_out":9840,"duration_ms":108477,"concrete_test":"Compute P_cross(k) = ⟨ζ_SF(k) ζ_src(−k)⟩ from the ALEF simulations, with ζ_src ≡ ζ_AU1 − ζ_SF, and the normalized correlation coefficient r(k) = P_cross(k) / √(P_vac(k) P_src(k)). If max_k |r(k)| exceeds ~0.05 over the k-range used in the N-body initial conditions, the uncorrelated-decomposition assumption fails; if also the halo mass function ratio n_AU1/n_G shifts by more than ~10% when the cross term is included in the Loverde-Smith prediction, the central 60% PNG-only enhancement claim is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"App. B asserts 'The cross-correlation between ζ_vac and ζ_src is negligibly small' and 'This decomposition into uncorrelated ζ_vac and ζ_src works well because we are far from the strong backreaction regime,' but no numerical evidence is presented. The central quantitative claim — a 60% enhancement in halo counts when comparing axion-U(1) to Gaussian initial conditions with matched power — is built on the G simulation being a pure Gaussian field with the same power spectrum as AU1. If P_cross ≠ 0, the matched-power Gaussian constructed by rescaling ζ_vac has a different phase structure and a different peak count than a Gaussian field with the same total power but independent sourced component. The halo mass function is sensitive to peaks and phases, so a non-negligible cross-correlation would bias the inferred PNG contribution. The paper also does not report error bars on the halo mass function differences, compounding the issue.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a new ab initio pipeline that connects lattice simulations of axion-U(1) inflation to cosmological N-body simulations: ALEF generates 3D curvature perturbations from nonlinear inflationary dynamics; the resulting field is converted to linear matter density with CLASS and used to set 3LPT initial conditions at z=15 for gadget-4. Three initial-condition sets are compared: AU1 (full axion-U(1)), SF (single-field with g_CS=0), and G (Gaussian with power spectrum matched to AU1). The authors report enhanced small-scale matter power and, more importantly, halo mass function enhancements of >200% at z=12 relative to SF and 60% relative to G, interpreting the residual after power matching as the effect of primordial non-Gaussianity. The Loverde-Smith/Tinker log-Edgeworth prediction is compared with the N-body halo mass function and claimed to be in good agreement.","tokens_in":13599,"tokens_out":8645,"duration_ms":104193,"significance":"If the result holds, this is a genuinely novel methodological advance: it replaces template-based primordial non-Gaussianity with field-level predictions from nonlinear inflationary dynamics, and it yields falsifiable predictions for high-redshift surveys, 21-cm, and line-intensity mapping. The paper's strengths include paired seed-matched control simulations (SF and G), Planck-calibrated background and potential parameters, use of public simulation codes, and direct measurement of primordial bispectrum/trispectrum. The main caveats are that the central halo mass function claim lacks a halo finder/mass definition and error bars, and that the uncorrelated vacuum/sourced decomposition is asserted but not numerically demonstrated. These are fixable with additional analysis and reporting.","major_comments":[{"comment":"The central quantitative claim is not reproducible as stated. No halo finder (FoF linking length, SO overdensity) or mass definition (M200b, M200c, Mvir) is given, and the lower panels of Fig. 2 show nAU1/nSF−1 and nAU1/nG−1 without any error bars. Since ten realizations exist, realization scatter can be estimated directly and should be shown. In addition, the statement that the Loverde-Smith/Tinker dashed curves are in 'good agreement' is unsupported without a residual or goodness-of-fit statistic. Please state the finder and mass definition, add error bars, and quantify the agreement.","section":"§V.B, Fig. 2"},{"comment":"For the z=12 small-box results, the plotted mass range (Fig. 2, right panels, M ~1e10–1e11 M_sun) corresponds to halos of roughly 100–1,400 particles for a 4×512^3 run in a 100 Mpc box (particle mass ~7×10^7 M_sun). No minimum halo particle number or resolution/convergence test is reported. The 200% enhancement may therefore be affected by shot noise and mass-assignment systematics, especially at the lowest masses. Please report the particle mass, the minimum virial particle threshold, and a convergence test (e.g., comparing with a higher-resolution run or varying the minimum particle threshold).","section":"§V.B, Table I"},{"comment":"The decomposition ζ=ζ_vac+ζ_src relies on the assertion that the cross-correlation between ζ_vac and ζ_src is 'negligibly small,' but no numerical evidence is shown. The paired seed-matched ALEF simulations make the cross spectrum directly measurable: report ⟨ζ_SF (ζ_AU1−ζ_SF)⟩ or the cross power spectrum, normalized to P_vac and P_src, over the k-range used to construct the N-body initial conditions. If this cross term is non-negligible, the source-term isolation and the interpretation of the AU1−G comparison as isolating primordial non-Gaussianity would need to be revisited.","section":"§IV and App. B"}],"minor_comments":[{"comment":"'Three sets of ten simulations with the same random seed but different initial conditions' is ambiguous: does each of the ten realizations share the same seed across AU1/SF/G, or are the ten realizations independent? Please clarify.","section":"§IV, Table I"},{"comment":"There is a missing space in 'Thisab initioapproach' in the abstract; also, 'the entire history of the universe' should be qualified, since the N-body evolution starts at z=15 rather than at the end of inflation.","section":"Abstract"},{"comment":"The quoted 2% and 10% PNG effects in the matter power spectrum are given without uncertainties. With ten realizations, error bars should be reported in the text or figure.","section":"§V.A"},{"comment":"The parity-odd trispectrum is stated to be 'negligibly small' but no numerical bound or plot is shown; please state the suppression factor relative to the parity-even part.","section":"App. B, Fig. 4"}],"recommendation":"major_revision","confidential_remarks":"The paper is clearly within the scope of the journal and represents a significant step. The main revision should focus on making the halo mass function analysis reproducible (finder, mass definition, error bars, resolution test) and on directly measuring the vacuum-sourced cross-correlation. These are addressable without changing the core methodology."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core of this paper is new and worth engaging with: it actually runs lattice simulations of axion-U(1) inflation through to N-body structure formation, which no one has done before. The field-level transfer from curvature to matter density, the paired seed-matched SF control, and the G simulations matched to the AU1 power spectrum are all sensible designs. The claim that sourced non-Gaussianity boosts high-redshift halo counts by tens to hundreds of percent is clearly stated and physically plausible, and the agreement with the Loverde-Smith log-Edgeworth expansion is a nice consistency check. The authors have built a real pipeline, not just a toy. The soft spots are real but not fatal. The biggest one is the asserted negligibility of the cross-correlation between zeta_vac and zeta_src. The paper gives no numerical evidence; it is an analytic handwave. If that cross-correlation is not actually small in the simulated regime, then the G comparison no longer isolates PNG cleanly, because the phases in AU1 differ from a pure Gaussian with the same power spectrum. This is checkable from the ALEF outputs--they should just measure it and show it. I would want to see that before fully trusting the fractional differences. The halo mass function results also lack three specifics that a referee will need: the halo finder and mass definition are never stated, the main figure has no error bars (only the primordial statistics plot in the appendix does), and the resolution at z=12 in the 100 Mpc box is likely marginal for 10^9-10^10 solar mass halos. None of these are necessarily wrong, but they are exactly the details that determine whether the 60-200% numbers are robust or artifacts. The 'entire history of the universe' phrasing is oversold--this is dark-matter-only with a transfer-function bridge--but that is rhetorical, not a technical flaw. Overall, the logic is sound and the paper is honest about its setup. I would send it to a competent referee who knows both lattice inflation and N-body methods, and ask them to verify the cross-correlation and demand the missing numerical details. It deserves a proper review, not a desk rejection. Even if the numbers shift after revision, the pipeline itself is a genuine methodological contribution.","headline":"A genuinely new inflation-to-N-body pipeline with suggestive halo mass function predictions, but the headline numbers need a few missing details (halo finder, error bars, cross-correlation check) before they should be taken as conclusive.","tokens_in":732,"tokens_out":820,"would_cite":true,"duration_ms":29554,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["98.80.Cq"],"model":"deepseek-v4-flash","headline":"The first simulation pipeline that runs from deep inflation to today shows axion-U(1) inflation leaves a measurable boost in early massive halos.","keywords":["inflation","primordial non-Gaussianity","ab initio simulation","N-body simulation","axion-U(1) inflation","halo mass function","matter power spectrum","21-cm cosmology"],"falsifier":"Search for the predicted 40% enhancement of matter power at k≈10 h/Mpc or the 200% halo-count excess at z≈12 in 21-cm or high-redshift galaxy surveys; if the observed structure on these scales matches Gaussian initial conditions with the same power spectrum, the axion-U(1) prediction is ruled out.","tokens_in":13230,"feed_emoji":"🌌","tokens_out":5918,"duration_ms":61274,"temperature":0.7,"pith_summary":"This paper establishes a new way to predict the cosmic web: instead of assuming Gaussian or template-based initial conditions, it simulates the nonlinear dynamics of inflation on a lattice, converts the resulting curvature field into the linear matter density, and evolves it with an N-body simulation to z=0. Applied to axion-U(1) inflation, the pipeline predicts that gauge-field-sourced fluctuations produce a blue-tilted power spectrum and a full hierarchy of non-Gaussian correlations. These make small-scale structure at high redshift much more abundant: at z=12, halo counts are more than 200% higher than in single-field inflation, and still 60% higher than Gaussian simulations with the same power spectrum, isolating a genuine non-Gaussianity effect. The paper argues that this makes high-redshift 21-cm, line-intensity, and galaxy surveys direct probes of inflationary physics.","feed_headline":"First inflation-to-today simulation raises early halo counts by 200%","feed_subtitle":"Axion-U(1) inflation's non-Gaussian signature survives to z=12, giving 21-cm and galaxy surveys a direct test of inflationary physics.","key_machinery":"The core mechanism is the decomposition of the primordial curvature perturbation into an uncorrelated sum ζ = ζ_vac + ζ_src, obtained by running paired lattice simulations of inflation with identical seeds and the axion-gauge coupling switched off in one. ζ_src carries a blue-tilted, scale-dependent, non-separable hierarchy of higher-order correlations produced by the Chern-Simons interaction between the axion and the U(1) gauge field. These curvature fields are converted to linear matter density perturbations through the standard transfer function and growth factor, then used as third-order Lagrangian perturbation theory initial conditions for N-body evolution. The signature in the halo mas","core_discovery":"The central discovery is a field-level prediction that bypasses standard template-based parametrizations of primordial non-Gaussianity. By isolating the vacuum and sourced components of the curvature perturbation via paired lattice simulations of inflation, and mapping the curvature field to linear matter density through the standard transfer function, the authors find that the sourced axion-gauge fluctuations survive to late times. They enhance matter power by about 10% at k=1 h/Mpc (z≥2) and about 40% at k=10 h/Mpc (z≥8), and boost the halo mass function far more strongly, especially for the most massive halos at z=12. The excess over power-matched Gaussian simulations (60%) demonstrates t","pith_inferences":["The same pipeline could be applied to other nonlinear inflationary or reheating models, turning any lattice-simulable early-universe field into a halo-level prediction without template assumptions.","The non-separable, blue-tilted bispectrum shape implies template-based primordial non-Gaussianity searches (local/equilateral/orthogonal) may miss the signal; simulation-calibrated estimators would be needed.","If confirmed, the predicted high-redshift halo excess would offer an alternative explanation for the abundance of massive galaxies at early times, independent of astrophysical efficiency arguments.","A testable extension would be to split the sourced field into bispectrum-only and trispectrum-only parts (e.g., by setting higher cumulants to zero) to quantify which statistic drives the halo boost; the paper's observed hierarchy suggests the bispectrum dominates at low mass and the trispectrum contributes at high mass."],"forward_implications":["High-redshift observations (21-cm, line-intensity mapping, galaxy surveys at z~6-12) should see a scale-dependent excess of small-scale power and an overabundance of massive halos relative to ΛCDM with Gaussian initial conditions.","The 60% excess over power-matched Gaussian simulations shows that primordial non-Gaussianity alone can be constrained from the halo mass function nearly independently of the power spectrum.","The enhancement grows with halo mass, so the rarest, most massive high-redshift halos are the most sensitive test of this class of models.","Nonlinear gravitational clustering erases much of the signal by z=0, so low-redshift surveys will see little; the cleanest constraints come from quasilinear scales at high redshift.","The log-Edgeworth prediction from the measured cumulants reproduces the N-body halo counts, enabling fast analytic forecasts from lattice-generated primordial fields."],"fun_headline_variants":["First inflation-to-today simulation predicts 200% halo boost","Axion inflation leaves 40% small-scale power signature","From Big Bang to cosmic web: first ab initio simulation","Inflation's nonlinearity imprints halos at z=12","Simulation maps inflation's direct signature in galaxy surveys"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The sourced curvature perturbation is an adiabatic scalar mode uncorrelated with the vacuum mode, so it can be linearly mapped to late-time matter through the standard transfer function; if that mapping fails, the halo-mass-function prediction loses its foundation.","fun_headline_variants_meta":{"raw":{"variants":["First inflation-to-today simulation predicts 200% halo boost","Axion inflation leaves 40% small-scale power signature","From Big Bang to cosmic web: first ab initio simulation","Inflation's nonlinearity imprints halos at z=12","Simulation maps inflation's direct signature in galaxy surveys"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000233,"raw_usage":{"total_tokens":1307,"prompt_tokens":700,"completion_tokens":607,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":444,"completion_tokens_details":{"reasoning_tokens":523}},"tokens_in":444,"tokens_out":607,"duration_ms":5556,"temperature":1.0,"reasoning_tokens":523,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T00:21:10.808767+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search for the predicted 40% enhancement of matter power at k≈10 h/Mpc or the 200% halo-count excess at z≈12 in 21-cm or high-redshift galaxy surveys; if the observed structure on these scales matches Gaussian initial conditions with the same power spectrum, the axion-U(1) prediction is ruled out.","supporting_citations":[],"review_version":1}