{"id":"1a070f36-2ebb-40cd-9ef3-459cd6d93dcc","arxiv_id":"1908.03522","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Axion string networks show standard scaling with ζ0 = 1.19 ± 0.20, and the claimed logarithmic growth in string density is explained as an artifact of initial conditions.","lead":"New large simulations of axion cosmic strings find the string density reaches a steady 'scaling' value, with the recently reported growth over time being an artifact of how the simulations start. This tightens the predicted mass range for axion dark matter, which experiments are actively searching for.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Free t0 in Eq. (5) can absorb a genuine logarithmic ζ correction; the residual-log test in Eq. (10) is therefore not independent evidence for standard scaling.","rationale":"The reader identified the t0 assumption; this concern sharpens it into a concrete degeneracy rather than a general worry. The paper does real work: multiple box sizes, multiple initial correlation lengths, two width prescriptions, and a constant-comoving-width control (ζ0=1.34±0.22). Those features argue against gross simulation artifacts, and I would not reject the empirical result. However, the headline conclusion specifically requires that t0 encode only initial-condition physics. The finite dynamic range (Δlog ηt≈1.5–2) makes straight-line-plus-offset and log-plus-offset fits nearly interchangeable, and the paper's own Eq. (8) quantifies one direction without inverting it. A synthetic-injection test is cheap and would settle whether the residual bound αr=(−0.04±0.30)×10^-2 is meaningful. Until that check is done, CONDITIONAL is the right verdict; my read does not move it.","tokens_in":12655,"tokens_out":5725,"duration_ms":64859,"concrete_test":"Generate mock ξ(t) data from the logarithmic model ζ(t)=1.05+0.2 log(ηt) over the paper's 4k t-ranges (t≈150–500), add per-bin uncertainties matching Tables I/III, and run the paper's exact fitting pipeline: fit ξ=2β(t−t0) over the four windows, compute t0/tf, then fit Eq. (10) for αr. If the pipeline returns t0/tf≈−0.05 to −0.1 and αr consistent with zero within ~0.003, the diagnostic cannot separate the models and the scaling conclusion is not established by this analysis. A second synthetic set with standard scaling plus an injected initial offset should be included as a control.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the claim that Eq. (5), ξ=2β(t−t0), fixes a pure initial-condition offset, so the residual fit (10) can bound a genuine logarithmic correction. That step is insecure: over the paper's finite window, the two parameters (β,t0) of a straight line can absorb the leading trend of ζ=ζ*+α log(ηt). For small t0/t, the estimator used in the residual test is ζhat(t)=(t−t0)^2/ξ²=(1−t0/t)^2[ζ*+α log(ηt)], which expands as ζ*+α log(ηt)−2ζ* t0/t+...; choosing t0≈−α t_m/(2ζ*) at the window midpoint cancels the log trend, leaving only a second-order residual. Thus a fitted t0/tf≈−0.1 with αr≈0 is exactly what a genuine logarithm with α≈0.2 would produce under this pipeline, rather than evidence against it. Equation (8) demonstrates the t0→apparent-α direction only; the reverse direction is not excluded. Since the conclusion of standard scaling and the constant-ζ extrapolation to QCD both rest on this non-degeneracy, the central claim is conditional on an explicit degeneracy test.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents new lattice simulations of global axion strings and compares three phenomenological models for the string-density parameter ζ: standard scaling with a constant ζ0, a logarithmic correction ζ=ζ0*+α*log(ηt), and an inverse-logarithmic correction. Fitting the mean string separation ξ to ξ=2β(t−t0), the authors find ζ0=1.19±0.20 and report a residual logarithmic coefficient αr=(−0.04±0.30)×10^-2, concluding that the apparent logarithmic growth seen in earlier work is an artifact of initial conditions and that a constant-ζ extrapolation from the Peccei-Quinn scale to the QCD scale is justified. The paper also re-analyzes previous axion-number-density estimates and suggests that the axion dark-matter mass should be revised upward by about 50%.","tokens_in":12993,"tokens_out":9066,"duration_ms":98041,"significance":"If the central conclusion holds, the paper resolves an active controversy: recent simulations reporting logarithmic growth of ζ would imply substantially larger axion string densities and a different axion dark-matter mass, whereas a constant ζ0≈1.2 supports the traditional scaling extrapolation. The paper's strengths include a careful decomposition of statistical and systematic uncertainties, multiple box sizes and initial correlation lengths, an infinite-volume extrapolation, a physical-width versus constant-comoving-width cross-check, and explicit comparison of the three competing models. The result is also falsifiable: the quoted 3σ bound |α|<0.94×10^-2 gives a concrete target for independent simulations.","major_comments":[{"comment":"The conclusion that the residual logarithmic coefficient is tightly bounded rests on the assumption that the free intercept t0 in Eq. (5) absorbs only initial-condition effects. This assumption is not tested, and the concern lands: if the true density parameter were ζ(t)=ζ0*+α log(ηt), then the estimator in Eq. (9) gives ζhat(t)=(1−t0/t)^2[ζ0*+α log(ηt)] ≈ ζ0*+α log(ηt)−2ζ0*t0/t for small |t0/t|. Over a finite fitting window, a fitted value t0≈−α t_m/(2ζ0*), where t_m is the midpoint of the window, cancels the logarithmic trend to second order. Thus the small αr in Table III is exactly what a genuine log correction with α≈0.2 would produce under this pipeline. Equation (8) demonstrates only the direction from t0 to an apparent logarithmic coefficient; the reverse direction is not excluded. Since the central claims of standard scaling (Eq. (11)) and of a constant-ζ extrapolation to the QCD scale both depend on this non-degeneracy, I ask for an explicit test: fit a nested model that allows both t0 and a logarithmic correction, or perform an injection-recovery test in which a known α log(ηt) term is added to the scaling model and the residual procedure is required to recover it.","section":"§Measurements and results, Eqs. (5), (9)–(10), Table III"}],"minor_comments":[{"comment":"The statement that the direct log-model fits are 'apparently inconsistent' with α*≃0.2 is stronger than the numbers in Table III support: for the 4k box α*=0.11±0.11, so 0.2 lies within 1σ. The tight exclusion of α*≈0.2 comes mainly from the residual bound in Eq. (12), which is the degenerate test flagged in the major comment. Please present the direct-fit and residual evidence separately when characterizing the constraint.","section":"Table III and accompanying text"},{"comment":"The notation is confusing because Eq. (5) is written in terms of physical time t and ξ, while footnote 1 defines β as the slope of comoving separation ξ/a against conformal time τ=2t/a. Please state explicitly which time variable enters the plotted and fitted quantities, and define all symbols consistently.","section":"Eq. (5) and footnote 1"},{"comment":"The suggested upward revision of the axion dark-matter mass by about 50% is based on an estimate read from a figure in Ref. [30], with no propagated uncertainty. Please label this as indicative or provide an error estimate, since it appears in the abstract as a quantitative conclusion.","section":"Conclusions, paragraph beginning 'An implication...'"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a serious paper on a live controversy, and the new 4k simulations with varied initial correlation lengths are a step up in statistics and control. But the key inference — that apparent logarithmic growth is an artifact of initial conditions — rests on a degeneracy that the paper does not fully close.\n\nThe genuinely new pieces are the t0/tf diagnostic and the attempt to separate statistical from systematic error. Showing that the fitted log coefficient α* correlates with t0/tf, and that the correlation is consistent with Eq (8), is a useful way to organize the data. The uncertainty budget (stat vs sys, 1/L extrapolation) is more careful than most of what exists in this area. The constant-comoving-width checks are also a plus.\n\nThe soft spot is load-bearing. In the standard scaling fit, ξ = 2β(t−t0), t0 is free. Over the paper's finite window (only a factor of ~3 in time), a genuine logarithmic correction ζ = ζ* + α log(ηt) can be absorbed by shifting t0 negative. The residual test in Eq. (10) uses the same fitted t0, so it is not an independent test for a logarithm. The math is straightforward: for small t0/t, ζ̂ ≈ ζ* + α log(ηt) − 2ζ* t0/t, so a t0/tf around −0.1 with α around 0.2 produces exactly what the pipeline reports as zero residual. The paper's Eq (8) demonstrates one direction of this mapping; it does not rule out the other. The direct log-model fit in Table III has α* = 0.11±0.11 for the 4k runs, which is not a tight exclusion of α* ~ 0.2. So the headline claim 'no evidence for logarithmic growth' is only as strong as the assumption that t0 is a pure initial-condition offset, which is precisely the point under dispute.\n\nThe 50% axion mass revision is asserted without derivation; that's fine for a letter but should be flagged. No code or data release, which is a reproducibility gap for a simulation paper.\n\nBottom line: the paper deserves a serious referee. The referee should ask for a joint fit of ξ(t) with both t0 and a log term, and a statement of the degeneracy. If the log term survives that test, the conclusion changes. Until then, the verdict is conditional, not settled.","headline":"New simulations and a promising diagnostic, but the claim that logarithmic growth is an artifact of initial conditions rests on a degeneracy the paper does not close.","tokens_in":13488,"tokens_out":4002,"would_cite":true,"duration_ms":40293,"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":"The paper argues that the axion string density approaches a constant, $\\zeta_0 = 1.19 \\pm 0.20$, so recent reports of logarithmic growth are artifacts of initial conditions, and the implied axion dark matter mass rises by about 50%.","keywords":["axion strings","QCD axion dark matter","cosmic string scaling","Peccei-Quinn symmetry breaking","string length density","post-inflationary axion","lattice field theory simulations","axion dark matter mass"],"falsifier":"A longer simulation with $t_f/t_0 \\gtrsim 20$ using the same shifted estimator would settle the claim: if the residual $\\hat\\zeta - \\zeta_0$ stays flat within $|\\alpha| < 0.94 \\times 10^{-2}$, standard scaling stands; if it drifts upward at the rate $\\alpha_* \\simeq 0.2$ reported by the logarithmic-growth groups, the artifact explanation fails.","tokens_in":12465,"feed_emoji":"🧵","tokens_out":10444,"duration_ms":99704,"temperature":0.7,"pith_summary":"The paper sets out to settle whether the density of cosmic strings left over from the Peccei-Quinn symmetry breaking behind the QCD axion settles down to a constant, as predicted by the standard scaling picture, or grows logarithmically with time, as several recent simulation groups have reported. The answer matters because in the post-inflationary axion dark matter scenario the axion number density, and therefore the dark matter density, is proportional to the string length per unit volume at the QCD transition; an underestimated string density changes the axion mass needed for axions to make up all of the dark matter. Using new lattice simulations at three box sizes and three initial correlation lengths, the paper fits the mean string separation to linear growth in time and extracts $\\zeta_0 = 1.19 \\pm 0.20$. It concludes that apparent logarithmic and inverse-logarithmic corrections are artifacts of the initial conditions, absorbed by a single time offset, and that re-analysis of earlier axion number density estimates raises the required axion dark matter mass by about 50%.","feed_headline":"Axion strings keep constant density; axion dark matter mass rises ~50%","feed_subtitle":"New simulations rule out logarithmic growth, keep $\\zeta_0=1.19$, and point to a heavier QCD axion in the post-inflationary scenario.","key_machinery":"The central object is the string length density parameter $\\zeta$ (equivalently the mean string separation $\\xi = \\sqrt{V/\\ell_s} = t/\\sqrt{\\zeta}$), and the load-bearing identity is the standard scaling fit $\\xi = 2\\beta(t - t_0)$, with $\\zeta_0 = 1/4\\beta^2$. The time offset $t_0$ is the piece that carries the argument: it is the t-axis intercept of the linear fit and is interpreted as the entire memory of the initial conditions. The paper shows that for nonzero $t_0$ the raw estimator $\\zeta = t^2/\\xi^2$ approaches $\\zeta_0$ slowly, behaving like $\\zeta_0(1 + 2t_0/t)$, which a logarithmic fit misreads as growth with coefficient $\\alpha_*(t_f) \\simeq -2\\zeta_0 (t_0/t_f)(1 - t_0/t_f)^{-1}$. In this way, the apparent logarithmic and inverse-logarithmic corrections are tied to the ratio $t_0/t_f$, making them properties of the initial setup rather than of the scaling network.","core_discovery":"The paper's central claim is that the dimensionless string length density parameter $\\zeta$, defined by writing the string length per unit volume as $\\zeta/t^2$, reaches a constant $\\zeta_0 = 1.19 \\pm 0.20$ in the scaling regime, rather than growing logarithmically as claimed in Refs. [24–27]. The decisive evidence is that the mean string separation $\\xi = t/\\sqrt{\\zeta}$ follows a straight line in time, $\\xi = 2\\beta(t - t_0)$, where the intercept $t_0$ encodes the initial string configuration. Fits of the logarithmic and inverse-logarithmic models give coefficients consistent with zero after the ratio $t_0/t_f$ is extrapolated to zero, and a direct fit to the residuals of the linear model gives $\\alpha_r = (-0.04 \\pm 0.30) \\times 10^{-2}$. The paper concludes that no logarithmic correction larger than about 0.5 can occur between the Peccei-Quinn scale and the QCD scale, that constant-$\\zeta$ extrapolation from simulations to cosmology is justified, and that the axion dark matter mass estimate should be revised upward by roughly 50%.","pith_inferences":["Editorial extension: the $t_0/t_f$ diagnostic provides a cheap retest of previously published data; if those runs are re-fitted with a shifted estimator and their reported $\\alpha_* \\simeq 0.2$ survives the extrapolation to $t_0/t_f \\to 0$, the artifact explanation would be wrong.","Editorial extension: the same residual analysis could be applied to local (gauge) cosmic string networks, whose tension does not grow logarithmically, to test whether initial-condition offsets are the generic source of apparent scaling violations.","Editorial extension: replacing the paper's figure-based reading of $n_{\\rm ax}/n_{\\rm mis} \\simeq 0.5$ at $\\kappa = 1$ with a direct simulation measurement would turn the 50% axion mass revision into a sharper prediction."],"forward_implications":["The string network can be extrapolated at constant $\\zeta$ from the Peccei-Quinn transition to the QCD transition, so no large enhancement of string density near the QCD transition is expected.","The $3\\sigma$ bound $|\\alpha| < 0.94 \\times 10^{-2}$ on the residual logarithmic coefficient limits any drift in $\\zeta$ to less than about 0.5 over the full cosmological range.","The axion dark matter mass in the post-inflationary singlet scenario should be increased by about 50% relative to estimates based on the larger string densities.","For network models, the energy loss rate per unit length of string must rise at the same rate as the effective string tension, otherwise constant $\\zeta$ would be unsustainable.","Frustrated-string simulations with $\\kappa \\simeq 50$ produce $\\zeta \\simeq 20$, far above the $O(1)$ value allowed here, so their large-separation extrapolations of axion number density should not be used."],"supporting_citations":[{"why":"Introduces the Peccei-Quinn U(1) symmetry whose spontaneous breaking forms the axion strings studied here.","marker":"[1]"},{"why":"Provides the standard cosmic-string scaling framework in which the string length density parameter is expected to approach a constant.","marker":"[7, 8]"},{"why":"Reports the authors' earlier global-string simulations consistent with scaling, the methodological baseline this work extends.","marker":"[21]"},{"why":"Recent claims of a logarithmic increase in $\\zeta$, the competing hypothesis the new simulations are designed to test.","marker":"[24–27]"},{"why":"The argument that a logarithmically growing effective string tension should make $\\zeta$ grow, motivating the logarithmic and frustrated-string models.","marker":"[28–30]"},{"why":"Supplies the measurement $n_{\\rm ax}/n_{\\rm mis} \\simeq 0.5$ at $\\kappa=1$ used to re-estimate the axion dark matter mass.","marker":"[30]"},{"why":"Proposes the inverse-logarithmic correction model fitted as an alternative to standard scaling.","marker":"[32]"},{"why":"Provides the plaquette-winding algorithm that identifies string crossings and defines the measured string length.","marker":"[35]"},{"why":"Introduced the t-axis intercept $t_0$ as an initial string evolution parameter, the device that absorbs initial-condition artifacts in the scaling fit.","marker":"[36]"}],"fun_headline_variants":["Constant axion string density; dark matter mass estimate up 50%","No string density growth; axion mass revised 50% higher","Scaling density fixed; axion dark matter mass rises 50%","Axion strings scale steadily; mass estimate climbs 50%"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion rests on the assumption that one fitted time offset in the linear string separation fully captures the effect of the initial conditions, so that any residual logarithmic behaviour is just the disguised approach to scaling; if a genuine logarithmic drift could hide inside that offset over the fitted time window, the claim that $\\zeta$ is constant would collapse.","fun_headline_variants_meta":{"raw":{"variants":["Constant axion string density; dark matter mass estimate up 50%","No string density growth; axion mass revised 50% higher","Scaling density fixed; axion dark matter mass rises 50%","Axion strings scale steadily; mass estimate climbs 50%"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000281,"raw_usage":{"total_tokens":1752,"prompt_tokens":1123,"completion_tokens":629,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":739,"completion_tokens_details":{"reasoning_tokens":553}},"tokens_in":739,"tokens_out":629,"duration_ms":6914,"temperature":1.0,"reasoning_tokens":553,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:10:51.944602+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A longer simulation with $t_f/t_0 \\gtrsim 20$ using the same shifted estimator would settle the claim: if the residual $\\hat\\zeta - \\zeta_0$ stays flat within $|\\alpha| < 0.94 \\times 10^{-2}$, standard scaling stands; if it drifts upward at the rate $\\alpha_* \\simeq 0.2$ reported by the logarithmic-growth groups, the artifact explanation fails.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the Peccei-Quinn U(1) symmetry whose spontaneous breaking forms the axion strings studied here."}],"review_version":1}