{"id":"fd9dbf84-c155-4acc-acbb-f3bfc3096b4e","arxiv_id":"2411.10838","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A maximum-likelihood analysis of Pantheon+ supernovae finds a statistically significant dipole in the local Hubble expansion rate and in the deceleration parameter, which the authors interpret as evidence against a cosmological constant as the cause of cosmic acceleration.","lead":"This paper searches the Pantheon+ supernova catalogue for directional differences in the cosmic expansion rate, and reports a dipole pattern larger than the claimed precision of the Hubble constant measurement. A smart generalist should read it because it challenges the assumption that the local universe expands uniformly, which is central to how dark energy and the Hubble tension are interpreted.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The >5σ significance claims rest on a likelihood-ratio calibration that is invalid at the null point, where dipole direction and decay scale become unidentifiable; an empirical null calibration is needed before the anisotropy claim can be accepted.","rationale":"The reader's weakest assumption concerned the reliability of the Pantheon+ covariance matrix and peculiar-velocity corrections. That is a legitimate concern, but it is not the single most load-bearing issue for the paper's headline. The headline claim is explicitly a statistical significance claim: '>5σ', 'statistically significant', 'exceeding 1.5 km/s/Mpc'. Every one of those numbers is obtained from likelihood ratios and then converted to significances with Wilks' theorem. At the null point the extra parameters (dipole direction, and the decay scale S in the q0 model) are unidentifiable, so the standard regularity conditions for Wilks' theorem fail. This is not a minor technicality: in problems with an amplitude parameter and a direction constrained to a sphere, the null distribution of the profile likelihood ratio is not the naive χ² with the full parameter dimension, and the discrepancy can be many sigma in either direction. The paper's statement that the pull distribution is narrower than Gaussian does not resolve this, because pulls measure residual scatter rather than the sampling distribution of the profile likelihood ratio. The proposed null simulation is the standard, decisive check. It preserves all data treatment choices—including the removal of bias corrections and the peculiar-velocity model—so it tests exactly the statistical calibration, not ancillary cosmology. I am not claiming the detection is spurious; the consistency with the earlier JLA result and the decay with redshift are suggestive. But the confidence levels on which the abstract's conclusions rest are currently uncalibrated. This does not force a different verdict from the reader's CONDITIONAL, but it sharpens the condition: the significance should be verified by simulation before the >5σ statements are treated as established.","tokens_in":26840,"tokens_out":5534,"duration_ms":69162,"concrete_test":"Generate at least 1000 mock Pantheon+ catalogs from the best-fit isotropic model (Hm, qm, j0 with no dipole), preserving the real sky positions, redshifts, duplicate structure, covariance matrix, selection-bias removal, and peculiar-velocity corrections. For each mock, compute −2ΔlnL for (i) freeing Hd with direction and (ii) freeing qd, S and direction. Compare the observed values, e.g. ΔLLH = 65.3 in the LG frame C1, to the empirical 99.99994th percentile. If the empirical p-value exceeds about 1e-5, the >5σ statements in the abstract fail; if it is far below, the quoted significances are conservative.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claims are the likelihood-ratio significances in Tables 1–5 and Figures 5–9, calibrated by Wilks' theorem with '3 d.o.f.' or '4 d.o.f.' for free dipole directions. Wilks' theorem requires the null hypothesis to lie in the interior of a regular parameter space. Here the null is Hd = 0 in Eq. (14), or qd = 0 in q0 = qm + qd·n e^(−z/S) in Eq. (15). At that point the dipole direction n is undefined, and for q0 the decay scale S is also unidentifiable. The asymptotic null distribution of −2ΔlnL is therefore not the nominal χ² with 2, 3, or 4 d.o.f. The authors' check that the pull distribution is narrower than Gaussian (Sec. 3.1) does not address this: pull distributions characterize standardized residuals, not the profile likelihood-ratio statistic under the composite null. If the true null distribution has heavier or lighter tails than the assumed χ², the quoted α = 3.5σ–7.9σ and the abstract's >5σ statements can be substantially off. Since the abstract's central claim is precisely that the dipolar variation is 'statistically significant' and '>5σ', this calibration is load-bearing.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper analyses the Pantheon+ Type Ia supernova catalogue for large-scale dipolar anisotropies in the Hubble expansion rate and the deceleration parameter, using maximum-likelihood estimators in the heliocentric, CMB, and Local Group frames, with two treatments of light-curve standardisation (C1 and C2). In the redshift range 0.023 < z < 0.15, the authors report a dipolar variation in H0 exceeding 1.5 km/s/Mpc in all three frames, with significances ranging from about 3 to 8 sigma depending on analysis choices. For q0, they find a redshift-dependent dipole that decays with redshift, claiming >5 sigma significance in several frames, and interpret this as evidence of a 'tilted universe' scenario in which the apparent cosmic acceleration is an artefact of the local bulk flow, concluding that the acceleration cannot be due to a cosmological constant.","tokens_in":27120,"tokens_out":11681,"duration_ms":128135,"significance":"If the statistical claims hold, the results would challenge the standard FLRW interpretation of the supernova Hubble diagram and would tie together several existing anomalies: the CMB dipole, the bulk flow, and the anisotropy of the deceleration parameter. The paper is careful in several respects: it provides two independent analysis pipelines (C1 and C2), examines three different observer frames, releases the code, and presents falsifiable predictions, notably the redshift decay of the q0 dipole in Fig. 9 and the shell-resolved dipole of H0 in Fig. 3. These are genuine strengths. The main risk is statistical: the >5 sigma claim for the q0 dipole rests on an application of Wilks' theorem to a model with parameters that are unidentifiable at the null hypothesis, and the significance of the H0 dipole shows a notable sensitivity to the intrinsic-scatter parameter sigma_M0. These issues are load-bearing and require additional calibration before the central claims can be accepted.","major_comments":[{"comment":"The significance of the q0 dipole is computed from the profile likelihood ratio using Wilks' theorem with 2 or 4 degrees of freedom. At the null hypothesis qd = 0, however, the dipole direction n and the decay scale S are unidentifiable: the model q0 = qm + qd·n e^{−z/S} collapses to q0 = qm for all n and S. The likelihood-ratio statistic is therefore the supremum of a χ² process over the two-sphere and over S, and its null distribution is not χ² with 2 or 4 d.o.f. The check in §3.1 that the pull distribution is 'narrower than Gaussian' refers to standardized residuals, not to the null distribution of the profile likelihood ratio, so it does not validate the reported significances. Since the abstract's central claim is that the deceleration parameter has a '> 5 sigma' dipolar modulation, the authors should calibrate the null distribution by Monte Carlo simulation of isotropic synthetic data using the same two pipelines and frames, and report empirical p-values (or a properly computed trials factor). Without this, the statistical significance of the q0 dipole is not established.","section":"§3.2, Eq. (15), Table 5"},{"comment":"The significance of the Hubble-rate dipole in analysis C2 depends sensitively on the assumed intrinsic-scatter parameter sigma_M0. The footnote reports that if sigma_M0 is set to 0.2 (the best-fit value over the full 0 < z < 0.8 range), the significance in the heliocentric frame drops to 1.8 sigma, from values around 3.5–3.8 sigma in Tables 2 and 4. Since the abstract claims a 'statistically significant' dipolar variation 'in all frames', the authors should either demonstrate that profiling over sigma_M0 (rather than fixing it at a value chosen for a particular redshift range) yields a stable significance, or report the range of significances across reasonable treatments of sigma_M0 and temper the abstract accordingly. As written, the claim is not robust to this nuisance-parameter choice.","section":"§3.1, footnote to Table 2"}],"minor_comments":[{"comment":"The abstract says the analyses are carried out in the heliocentric, CMB, and Local Group frames and then states 'In all frames' the Hubble expansion rate is anisotropic; this should be clarified to mean those three frames only, since Fig. 3 shows that in the zHD frame (with peculiar-velocity corrections) the dipole is consistent with zero in the SH0ES redshift range.","section":"Abstract and §3.1"},{"comment":"The notation 'α' for statistical significance (e.g., '3.5σ') conflicts with the standard use of α as the significance level; consider using 'Nsig' or 'significance' in the table headers and captions.","section":"Tables 1–5"},{"comment":"The sentence 'Assuming that this follows a χ² distribution (with 3 d.o.f.)' should read 'Assuming that 2ΔLL follows a χ² distribution'; additionally, the likelihood expression in Eq. (8) would benefit from an explicit definition of the vectors and matrices involved.","section":"§3.1"},{"comment":"There are typographical errors in the captions ('wich' instead of 'which') and in Table A3 the entry '.03/ln 2' should be '0.03/ln 2'; these should be corrected.","section":"Appendix A, Tables A1–A4"},{"comment":"The statement that the pull distribution is 'narrower than Gaussian' is too terse to be evaluated; the authors should specify how the pull distribution is defined, report its variance, and explain why it bears on the likelihood-ratio calibration.","section":"§3.1, pull-distribution check"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the journal's scope and the topic is timely. The main request is methodological: the >5σ significance of the q0 dipole must be re-calibrated empirically, and the dependence of the H0 dipole significance on sigma_M0 must be addressed. If the Monte Carlo calibration confirms the quoted significances, the paper could become a strong contribution; if it does not, the abstract's central claims will need to be substantially revised."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, what you should know: this is a serious, reproducible re-analysis, not a routine dipole search. The H0 dipole in the SH0ES redshift range is new, and the code is on GitHub. The two pipelines and three frames make the result easy to probe. The comparison with earlier Pantheon+ anisotropy papers is thorough and fair, including the papers that find no anisotropy.\n\nWhat is actually new: an H0 dipole of order 2 km/s/Mpc in 0.023<z<0.15, present in heliocentric, CMB, and Local Group frames; and a q0 dipole that extends the JLA result at higher significance, with the shell analysis showing redshift decay. The C1/C2 split and the explicit treatment of bias corrections are good practice. Credit where due: code availability, clear statements about model-dependent peculiar-velocity corrections, and the honest note that removing bias corrections changes the monopole but not the dipole.\n\nSoft spots. The q0 significance is the load-bearing claim, and the Wilks calibration is not valid at the null. In q0 = qm + qd*n*exp(-z/S), if qd=0 then both n and S are unidentifiable; the profile likelihood ratio will not follow a chi-square with 2 or 4 dof. The pull-distribution check in Sec. 3.1 does not fix this, because it describes standardized residuals, not the profile likelihood under the composite null. So the abstract's >5 sigma for q0 is, as written, not supported. That is exactly the stress-test concern, and I think it lands for q0. For the H0 dipole I am less worried: Hd can be absorbed into an unconstrained 3-vector, and the model is linear in that vector, so Wilks with 3 dof is more defensible, though an empirical null would still be better.\n\nSecond: the zHD frame is not a neutral check. The Pantheon+ covariance and zHD redshifts are built using peculiar-velocity corrections derived from a Lambda-CDM-tuned Newtonian model. The authors flag this, and their main results in zhel/zLG do not depend on it, but zHD results should not be used as evidence either for or against anisotropy.\n\nThird: the conclusion \"cannot therefore be due to Lambda\" overreaches. A dipole in q0 would be strong evidence against isotropic dark energy, but the bulk-flow interpretation is a theoretical model, not a direct consequence of the fit.\n\nWho should read it: anyone working on SNe cosmology, the Hubble tension, or tests of the cosmological principle. The citation pattern is self-referential in the interpretation section, but those are the direct antecedents. It deserves a serious referee, with an explicit request for an empirical null calibration of the q0 test and softer wording in the abstract until that is done.","headline":"A careful, reproducible re-analysis that makes a strong H0-dipole claim and a stronger q0-dipole claim; the q0 significance needs a null calibration before the >5 sigma wording can be trusted.","tokens_in":27645,"tokens_out":4685,"would_cite":true,"duration_ms":55573,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that the local Hubble expansion and the deceleration parameter carry statistically significant dipoles in every reference frame, so the acceleration inferred from supernovae is a local bulk-flow effect rather than the…","keywords":["Type Ia supernovae","Hubble parameter anisotropy","deceleration parameter","peculiar velocities","bulk flow","tilted universe","cosmological principle","Pantheon+ catalogue"],"falsifier":"A decisive check is to repeat the dipole fit on a supernova sample that has not been pre-corrected with model-dependent peculiar velocities and that has uniform sky coverage in redshift bins: if the dipole in H0 or q0 disappears or shifts direction when only the raw redshifts and a complete-sky selection are used, the claim would be refuted. The tilted-universe interpretation specifically predicts that the q0 dipole should decay on a scale of order S ~ 0.01 and align with the local bulk flow, so a future dataset with thousands of low-redshift supernovae that shows an undiminished or misaligned dipole at z > 0.1 would also falsify it.","tokens_in":26632,"feed_emoji":"🌌","tokens_out":9400,"duration_ms":85321,"temperature":0.7,"pith_summary":"This paper re-analyzes the Pantheon+ Type Ia supernova catalogue with maximum-likelihood estimators to test the isotropy of the cosmic expansion, working in the heliocentric, CMB and Local Group frames. The authors find a statistically significant dipolar variation in the Hubble expansion rate exceeding 1.5 km/s/Mpc over 0.023 < z < 0.15 in all three frames, larger than the claimed 1% precision of the local H0 measurement. They also find a >5 sigma, redshift-dependent dipole in the deceleration parameter q0, consistent with earlier JLA results. They conclude that the inferred cosmic acceleration cannot be due to a cosmological constant and is instead a general-relativistic effect of the local bulk flow. A sympathetic reader cares because this directly challenges the standard ΛCDM interpretation of dark energy and reframes the Hubble tension as a breakdown of the assumption that the local universe is isotropic.","feed_headline":"Cosmic acceleration shows a >5 sigma dipole in supernova data","feed_subtitle":"If true, the signal attributed to dark energy is an artifact of our local bulk flow.","key_machinery":"The argument is carried by a maximum-likelihood estimator that analytically marginalises over an intrinsic Gaussian scatter in absolute magnitude (and, in analysis C2, over stretch and colour), combined with a third-order cosmographic expansion of the distance modulus in redshift. Anisotropy is parametrised as $H = H_m + H_d \\hat{n}$ for the Hubble parameter and $q_0 = q_m + q_d \\hat{n} e^{-z/S}$ for the deceleration parameter, with significance assessed by Wilks' theorem. The frame choice is the critical handle: the same data are fit in heliocentric ($z_{\\mathrm{hel}}$), CMB ($z_{\\mathrm{CMB}}$) and Local Group ($z_{\\mathrm{LG}}$) frames, and compared with the 'Hubble diagram' frame $z_{\\mathrm{HD}}$ in which redshifts are corrected for both observer and host-galaxy peculiar velocities using a Newtonian flow model that assumes ΛCDM beyond the survey volume. The physical interpretation invokes the relativistic tilted-universe equations, in which two observers in relative motion assign different deceleration tensors to the same spacetime, producing a Doppler-like dipole in $H$ and $q_0$ that decays with redshift.","core_discovery":"The paper's central claim is that the local expansion of the universe is anisotropic at a level that is too large to ignore. On the Pantheon+ sample, a dipole in the Hubble parameter of amplitude greater than 1.5 km/s/Mpc is reported in the redshift range 0.023 < z < 0.15 in the heliocentric, CMB and Local Group frames. The deceleration parameter q0 shows a dipolar modulation that decays with redshift, with a significance above 5 sigma when the sample includes SNe at z > 0.00937. The direction and decay length of the q0 dipole match the signature expected from an observer moving with a contracting bulk flow in the 'tilted universe' scenario, so the authors conclude that the cosmic acceleration inferred from supernovae cannot be due to a cosmological constant.","pith_inferences":["A natural extension is to map the dipole as a function of redshift and sky position using upcoming wide-field supernova samples; if the dipole direction is stable and scales as the predicted decay, this would strengthen the bulk-flow interpretation without relying on the Pantheon+ covariance.","If the tilted-universe interpretation is right, the same frame-dependent anisotropy should appear in other cosmological observables, such as the kinetic Sunyaev-Zeldovich effect or gravitational lensing, which could be tested with CMB and galaxy-survey data.","The paper implicitly predicts that a correction based on a fully relativistic (rather than Newtonian, ΛCDM-extended) peculiar-velocity model would still leave a residual dipole; fitting this would quantify how much of the signal is model-dependent.","The observed anisotropy may also affect distance-ladder estimates of the Hubble constant, suggesting that the 'Hubble tension' could be partly a geometric effect of the local flow."],"forward_implications":["The local measurement of H0 to 1% precision is not valid if the local expansion is anisotropic at the level reported, since the anisotropy exceeds the claimed precision in the same redshift range.","The cosmic acceleration inferred from supernovae cannot be attributed to a cosmological constant; it would instead be a general-relativistic effect of the local bulk flow.","Redshift corrections that apply model-dependent peculiar velocities ($z_{\\mathrm{HD}}$) can mask a genuine dipole, so the standard 'isotropisation' of supernova data assumes the very cosmological model being tested.","The monopole deceleration parameter is consistent with zero unless light-curve stretch and colour are allowed to be sample- and redshift-dependent, which would undermine the standard-candle assumption.","The dipole in $q_0$ decays with redshift as predicted by the tilted-universe scenario, providing a consistency check against the data."],"supporting_citations":[{"why":"The Pantheon+ supernova dataset analysed throughout the paper.","marker":"[22]"},{"why":"The Pantheon+ analysis that supplies the covariance matrices, bias corrections and the 'constrained chi2' methodology the authors critique.","marker":"[2]"},{"why":"The maximum-likelihood estimator with Gaussian intrinsic scatter used for all fits.","marker":"[23]"},{"why":"The JLA catalogue, the previous dataset whose anomaly this paper extends.","marker":"[15]"},{"why":"Previous detection of a q0 dipole in JLA that motivates the present analysis.","marker":"[16]"},{"why":"The peculiar-velocity model used to build the Hubble-diagram frame that the uncorrected frames are compared against.","marker":"[19]"},{"why":"The tilted-universe theory that predicts a decaying dipole in the deceleration parameter.","marker":"[17]"},{"why":"The companion tilted-universe result connecting bulk flows to apparent acceleration.","marker":"[18]"},{"why":"Independent challenge to the FLRW assumption via the radio-source dipole, used to question the validity of peculiar-velocity corrected redshifts.","marker":"[7]"},{"why":"The local H0 measurement whose claimed 1% precision the reported dipole exceeds.","marker":"[14]"}],"fun_headline_variants":["Cosmic expansion shows a >1.5 km/s/Mpc dipole in supernovae","Local bulk flow may explain cosmic acceleration, study finds",">5 sigma dipole in supernova data challenges dark energy","Cosmic acceleration likely an artifact of bulk flow"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central premise is that the Pantheon+ covariance matrix and the peculiar-velocity corrections are accurate enough for the claimed precision; those corrections assume a ΛCDM background beyond the survey volume, so a wrong flow model could create or hide the dipole.","fun_headline_variants_meta":{"raw":{"variants":["Cosmic expansion shows a >1.5 km/s/Mpc dipole in supernovae","Local bulk flow may explain cosmic acceleration, study finds",">5 sigma dipole in supernova data challenges dark energy","Cosmic acceleration likely an artifact of bulk flow"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000976,"raw_usage":{"total_tokens":4110,"prompt_tokens":874,"completion_tokens":3236,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":490,"completion_tokens_details":{"reasoning_tokens":3165}},"tokens_in":490,"tokens_out":3236,"duration_ms":21217,"temperature":1.0,"reasoning_tokens":3165,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T19:13:29.723121+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive check is to repeat the dipole fit on a supernova sample that has not been pre-corrected with model-dependent peculiar velocities and that has uniform sky coverage in redshift bins: if the dipole in H0 or q0 disappears or shifts direction when only the raw redshifts and a complete-sky selection are used, the claim would be refuted. The tilted-universe interpretation specifically predicts that the q0 dipole should decay on a scale of order S ~ 0.01 and align with the local bulk flow, so a future dataset with thousands of low-redshift supernovae that shows an undiminished or misaligned dipole at z > 0.1 would also falsify it.","supporting_citations":[],"review_version":1}