{"id":"22e6766a-a991-43c6-8a09-e26669435835","arxiv_id":"2511.22559","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A two-cosine axion dark energy potential fits CMB+BAO+supernova data better than the standard one-cosine potential, while preferring a sub-Planckian decay constant and a feeble dark-matter coupling.","lead":"An axion-style dark energy model with a potential built from two cosine waves (as string/supergravity models predict) fits a combination of Planck, DESI, and supernova data better than the usual one-cosine version. The fit also constrains the axion decay constant to sub-Planckian values and limits any dark matter-dark energy coupling to be very weak.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Sub-Planckian F constraint may be prior-driven: the MCMC prior ranges are not specified, so the high-scale parameter claim could be an artifact of restricting F to sub-Planckian values.","rationale":"The reader correctly identifies a weak spot in the SUGRA/string connection, but the truncation concern is not as load-bearing as stated: for N=2, the double-sum cross terms in Eq. (2.10) only contribute to the cos(φ/F) harmonic, so the fitted two-cosine potential is functionally the same family as the full potential with a redefined c1. The more serious issue is that the paper's flagship high-scale constraint—F is sub-Planckian—depends on flat prior ranges that are never specified. If the prior was restricted to F<1 m_Pl, the result is guaranteed by construction, and the abstract's wording overstates what the data alone show. The reported 68% intervals do not exclude F>1 at high significance, and PPS/Union3 provide only lower limits. This is a concrete, checkable reproducibility/correctness risk: a single rerun with an extended F prior would settle whether the constraint survives. The N=2 better-fit claim also deserves scrutiny because it is demonstrated for the special c1=c2 case, while the more general free-coefficient N=2 run gives worse ΔAIC; but that is a secondary nuance. Since the issue is addressable by respecifying priors and rerunning, the existing CONDITIONAL verdict remains appropriate rather than moving to reject or unverified.","tokens_in":23844,"tokens_out":11692,"duration_ms":108390,"concrete_test":"Rerun the N=2 equal-coefficient and free-coefficient MCMC chains with F prior [0.05, 10] m_Pl and φ_ini prior [0, 10] m_Pl, keeping all other settings and data identical, and with the general N=2 potential V=μ^4[c1'(1+cos(φ/F))+c2(1+cos(2φ/F))] where c1' includes the c21 cross-term contribution. If the F posterior still falls below 1 m_Pl and the ΔAIC improvement over N=1 persists, the central claims stand; if the posterior extends to large F or the improvement disappears, the high-scale and better-fit claims are prior/special-choice artifacts.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central high-scale claim is that the data determine the axion decay constant F to be sub-Planckian. But Sec. 6.2 states only 'We impose flat priors on all the parameters' and gives no numerical ranges. If the F prior has an upper bound at 1 m_Pl, the posterior trivially cannot exceed 1, making the abstract's 'determined to be sub-Planckian' a prior effect rather than a data-driven result. The reported intervals already hint at this: in Tables 1–3, PPS and Union3 give only lower limits on F, and the DESY5 68% interval F=0.62±0.22 has a 95% upper bound near 1.05 m_Pl, so F>1 is not strongly excluded by the data as presented. Without prior ranges or released chains, this part of the central claim is not independently checkable. The reader's truncation concern is weaker than stated: the double-sum terms in Eq. (2.10) involve only harmonics (r-l)<N, so for N=2 they just renormalize the c1 coefficient; the fitted V2 family already covers the full N=2 potential. However, the equal-coefficient benchmark used for the main N=2 claim is not the full potential if c21 is nonzero, and the free-coefficient N=2 run in Table 3 fits worse, so the 'N=2 better fit' claim is tied to a special choice rather than to the general SUGRA/string potential.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a quintessence dark-energy model in which the DE scalar is an ultralight axion with a superposed-cosine potential V(φ)=μ^4 Σ_n c_n [1+cos(nφ/F)], motivated by SUGRA/string axion landscapes, and is coupled to a scalar dark-matter field through V_int=(λ/2)χ^2φ^2. The authors implement the coupled background and perturbation equations in CLASS+Cobaya and compare N=1 and N=2 potentials using CMB (Planck PR4 + ACT DR6), DESI DR2 BAO, and three SN compilations (PPS, DESY5, Union3). They report that the N=2 potential with equal coefficients fits the data better than N=1 by ΔAIC differences of roughly 20 units, that the axion decay constant F is constrained to be sub-Planckian, and that the DM–DE coupling is feeble (λ≲4×10^{-6}). They also exhibit, for N=3,4 in a benchmark phenomenological study, a thawing-to-freezing transmutation without DM coupling. The model comparison is the central quantitative result; the high-scale parameter claims are the central interpretive claims.","tokens_in":24347,"tokens_out":8201,"duration_ms":70488,"significance":"If correct, the paper demonstrates that multi-cosine axion potentials, of the type generated in SUGRA/string constructions, are viable and can be preferred over the single-cosine pNGB potential by current cosmological data, while yielding one-sided constraints on the effective decay constant and the DM–DE coupling. The analysis has genuine strengths: the underlying Lagrangian treatment of the DM–DE interaction is internally consistent; the background and perturbation equations are implemented in the standard CLASS+Cobaya pipeline; the comparison is made with standard information criteria across three independent SN data sets; and the paper explicitly quantifies the relative improvement (Tables 1–3). The significance is, however, moderated by three issues: the prior ranges for the sampled parameters are not reported, which is crucial for the sub-Planckian F claim; the headline N=2 result is obtained for a special equal-coefficient choice rather than the generic SUGRA/string potential; and the abstract's λ bound is not what the tables' 68% intervals imply. These issues are fixable and are detailed below.","major_comments":[{"comment":"Section 6.2 states 'We impose flat priors on all the parameters' but gives no numerical ranges. The abstract's headline that F is 'determined to be sub-Planckian' is therefore not independently checkable: if the prior on F is truncated at 1 m_Pl, the posterior cannot exceed it. Moreover, Tables 1–3 mostly report one-sided lower limits on F (e.g., F>0.620, F>0.691, F>0.599); the only two-sided interval, F=0.62±0.22 for DESY5 in Table 2, has a 95% upper endpoint near 1.05 m_Pl, so F>1 is not strongly excluded. Please quote the prior ranges for all sampled parameters and, ideally, release the chains; the sub-Planckian claim should be rephrased or supported by a prior-independent statement.","section":"§6.2, Tables 1–3"},{"comment":"The fitted potential (3.5) uses only the single-sum part of Eq. (2.10). For N=2 the neglected double-sum term c21[1+cos(a/F)] has the same harmonic as the c1 term, so the free-coefficient potential (6.3) in fact spans the full N=2 potential. However, the main N=2 result (Table 2) imposes c1=c2=1, which is not the general SUGRA/string potential when c21≠0; the free-coefficient fit (Table 3) gives ΔAIC = 20.70, 9.48, 17.58 versus 8.83, −0.52, 6.45 for Table 2, with poorly constrained c1 and c2. The relative improvement over N=1 survives in both variants, but its magnitude and the claimed connection to the SUGRA/string potential depend on the special equal-coefficient choice. Justify this choice or present the free-coefficient run as the fiducial SUGRA-motivated case.","section":"Eq. (2.10), Eq. (3.5), Tables 2–3"},{"comment":"The bound λ≲4×10^{-6} m_Pl^{-2} Mpc^{-2} quoted in the abstract does not follow from the reported 68% intervals. Table 1 (PPS) gives log λ = −5.49^{+0.97}_{−2.2}, corresponding to a 68% upper bound near 3×10^{-5}; Table 2 (DESY5) gives log λ = −5.5^{+1.4}_{−1.7}, i.e. an upper bound near 10^{-4}. If a different data set or confidence level is intended, it should be stated explicitly; as written the abstract understates the uncertainty by roughly an order of magnitude, and the same overstatement appears in §7.","section":"Abstract and Tables 1–3"}],"minor_comments":[{"comment":"The summand is written as cos(N a/F) under a sum over k; this should likely be cos(k a/F). Please correct the index.","section":"Eq. (5.1)"},{"comment":"The caption lists 'log μ4 = 7.0', but the text in §6.2 says the c1≠c2 run fixes log μ4 = −7.0. One of these is a typo.","section":"Table 3 caption"},{"comment":"The tables use asymmetric intervals and one-sided limits without specifying whether the one-sided entries are 68% or 95% bounds, and without stating units for log μ4 and log λ. Please add a footnote with the confidence level and units (m_Pl^2 Mpc^{-2} for μ^4 and m_Pl^{-2} Mpc^{-2} for λ).","section":"Tables 1–3"},{"comment":"The sentence 'c_i are coefficients that are multiples of μ^4' is ambiguous. State explicitly that μ^4 is fixed to 10^{-7} m_Pl^2 Mpc^{-2} in the c1≠c2 run and that c1 and c2 are dimensionless coefficients.","section":"Eq. (6.3) and §6.2"},{"comment":"The transmutation phenomenon for N=3,4 is demonstrated only for a benchmark with equal coefficients and log μ4 = −7.0. The abstract presents it as a general result; please qualify it as a benchmark demonstration unless a wider scan is performed.","section":"§6.1"}],"recommendation":"major_revision","confidential_remarks":"The internal model comparison is interesting and the pipeline is sound, but the headline high-scale constraints are oversold and are not reproducible without the prior ranges. The editor may wish to require prior ranges and chain release before publication, and to ask the authors to bring the abstract and conclusion in line with the intervals actually reported in the tables."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline is: the equal-coefficient N=2 superposed-cosine axion potential does fit the data substantially better than N=1 (ΔAIC improves by ~20 across all three datasets), and that is the genuinely new piece. I also credit the authors for building the DM-DE interaction from a Lagrangian rather than ad hoc sources—that's a real upgrade over much of the literature. The CLASS+Cobaya pipeline is standard and the ΔAIC tables make it easy to see that ΛCDM is still preferred in five of six comparisons.\n\nThe soft spots are in the interpretation. The abstract's 'F determined to be sub-Planckian' is stronger than the tables: PPS and Union3 give only lower limits on F, and the DESY5 interval (0.62±0.22 m_Pl) has an upper bound that is effectively set by the prior. The paper says 'flat priors on all parameters' but gives no ranges, so the sub-Planckian conclusion could be prior-driven. That's the most serious issue, and it is fixable by reporting priors and testing sensitivity.\n\nSecond, the better-fit result hinges on the special choice c1=c2. The free-coefficient N=2 run fits worse, but it also fixes log μ^4 = -7 rather than letting it vary, so the comparison is not clean. The double-sum truncation in the SUGRA potential is not a real problem for N=2—those terms only renormalize c1—but the generic SUGRA/string potential has a coefficient ratio that should be marginalized over, and that has not been shown to fit better than N=1.\n\nThird, no code or chains are released, so the prior issue is not independently checkable. That is a minor complaint in a paper like this, but it matters here because the central high-scale claim rests on it.\n\nOverall: a modest, honest contribution. The relative improvement is real for a specific benchmark, the high-scale parameter constraints are overstated, and the prior issue needs to be addressed. I would send it to peer review rather than desk-reject; the problems are addressable in revision.","headline":"N=2 axion-potential fit improves on N=1, but the sub-Planckian F and the preference over a general N=2 potential are weaker than the abstract suggests.","tokens_in":24767,"tokens_out":6494,"would_cite":true,"duration_ms":56499,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["83F05","85A40"],"pacs":["98.80.-k","95.36.+x"],"model":"deepseek-v4-flash","headline":"A dark-energy field with a two-term cosine potential fits cosmological data better than the standard one-cosine potential, the authors argue, and pins the axion decay constant below the Planck mass.","keywords":["axion dark energy","quintessence","superposed cosine potential","axion decay constant","dark matter-dark energy interaction","supergravity","string theory","cosmological parameter constraints"],"falsifier":"A cosmological re-fit of the full two-sum potential of Eq. (2.10), including the c_rl cross terms, would settle whether the N=2 preference and the sub-Planckian F bound survive; if the full-potential fit no longer prefers N=2, the central claim fails. Alternatively, an independent estimate of c_rl from explicit instanton data in a concrete string compactification could show the cross terms are large enough to change the N=2 fit.","tokens_in":23762,"feed_emoji":"🔭","tokens_out":4882,"duration_ms":43093,"temperature":0.7,"pith_summary":"The paper argues that the dark-energy field driving cosmic acceleration could be an ultralight axion whose potential is a superposition of N cosine terms, as arises generically in supergravity and string models, rather than the single cosine usually assumed. It shows that the N=2 superposition fits current cosmological data (CMB, baryon acoustic oscillations, and three supernova compilations) better than the N=1 case, while remaining competitive with the cosmological constant for some data sets. The fits also deliver constraints on high-energy parameters: the effective axion decay constant F is sub-Planckian but near the Planck scale, consistent with string-theoretic expectations, and the dark-matter–dark-energy interaction strength is feeble, with an upper bound around 4e-6 in the model's natural units. For N=3 and N=4, the potential alone causes thawing quintessence to transmute into freezing quintessence even with no coupling to dark matter.","feed_headline":"Double-cosine axion dark energy out-fits the single-cosine model","feed_subtitle":"Cosmological fits pin the axion decay constant below the Planck mass and the dark energy–dark matter coupling to near zero.","key_machinery":"The central object is the superposed-cosine axion potential V(φ) = μ^4 Σ c_n (1 + cos(nφ/F)), which the paper derives from a single anomalous U(1) shift symmetry broken by instanton effects in supergravity/string models, with F the effective axion decay constant. The analysis truncates the full two-sum potential to its single-sum terms, couples the axion to a dark-matter scalar through V_int = (λ/2)χ²φ², and rewrites the Klein-Gordon and perturbation equations in variables that tame the rapid dark-matter oscillations. These equations are integrated numerically and fit to cosmological data, yielding posterior constraints on μ^4, F, φ_ini, λ, and the coefficients c_n.","core_discovery":"The central claim is that a single axionic field with the multi-cosine potential V(φ) = μ^4 Σ_{n=1}^N c_n (1 + cos(nφ/F)), truncated to the single-sum terms of a supergravity/string axion landscape, is a viable and testable dark-energy model. Fitting the N=2 version to a combination of CMB, baryon-acoustic-oscillation, and supernova data within a Lagrangian-based interacting quintessence–dark-matter model, the authors find the N=2 potential improves the fit over the standard N=1 axion potential and constrains the axion decay constant to sub-Planckian values (F around 0.4–0.8 times the Planck mass, with lower limits near 0.6) and the interaction strength to λ ≲ 4e-6 in the model's units. The","pith_inferences":["The main caveat is the paper's own: the fits use only the single-sum part of the derived potential, dropping the double-sum cross terms c_rl. If those cross terms are not negligible, the N=2 preference, the F bound, and the λ bound may shift; a full two-sum fit would test this directly.","The success of the N=2 potential suggests that future CMB and large-scale-structure surveys with percent-level distance measurements could distinguish specific axion landscapes through the coefficients c_n, turning cosmology into a probe of instanton-generated superpotential terms.","The λ constraint could be tightened by cross-correlating with structure-growth observables such as S8, since interacting quintessence imprints on the matter power spectrum; the paper's predicted power spectra provide a concrete target.","The same single-field superposition framework could be applied to N>4 or to explicit string compactifications with computed c_k, testing whether specific axiverse models survive the same data."],"forward_implications":["If correct, supergravity/string-motivated multi-axion potentials become viable dark-energy models that current cosmological data can actually discriminate among (N=1 versus N=2).","The data-driven bound that F is sub-Planckian supports the string-theoretic prohibition of trans-Planckian axion decay constants.","The dark-matter–dark-energy coupling is constrained to be feeble (λ ≲ 4e-6 in the model's units), so significant interaction between the two dark sectors is ruled out at this level.","The N=2 potential's better information-criterion score than N=1 indicates that adding one extra cosine term is enough to improve the description of the expansion history, even though ΛCDM remains competitive for some data sets.","For N=3 and N=4, the thawing-to-freezing transmutation is a purely potential-driven phenomenon, independent of any dark-matter coupling."],"fun_headline_variants":["Multi-cosine axion dark energy beats single-cosine fit","Axion dark energy pins decay constant below Planck mass","Dark energy-dark matter coupling found ultra-feeble","Axion transmutation: thawing to freezing quintessence","Supergravity axion landscape tightens cosmological constraints"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The truncation of the axionic potential to the single-sum terms, dropping the double-sum cross terms c_rl in Eq. (2.10), is the load-bearing simplification; if those cross terms are not negligible, the fitted potential is not the full supergravity/string potential and the derived constraints would have to be redone.","fun_headline_variants_meta":{"raw":{"variants":["Multi-cosine axion dark energy beats single-cosine fit","Axion dark energy pins decay constant below Planck mass","Dark energy-dark matter coupling found ultra-feeble","Axion transmutation: thawing to freezing quintessence","Supergravity axion landscape tightens cosmological constraints"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000744,"raw_usage":{"total_tokens":3225,"prompt_tokens":888,"completion_tokens":2337,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":632,"completion_tokens_details":{"reasoning_tokens":2258}},"tokens_in":632,"tokens_out":2337,"duration_ms":15681,"temperature":1.0,"reasoning_tokens":2258,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T19:45:36.456060+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A cosmological re-fit of the full two-sum potential of Eq. (2.10), including the c_rl cross terms, would settle whether the N=2 preference and the sub-Planckian F bound survive; if the full-potential fit no longer prefers N=2, the central claim fails. Alternatively, an independent estimate of c_rl from explicit instanton data in a concrete string compactification could show the cross terms are large enough to change the N=2 fit.","supporting_citations":[],"review_version":1}