{"id":"ea75e2f8-3d33-4edd-a357-a8287e560b4f","arxiv_id":"2608.02808","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The 7.2-year InPTA DR2 data yield no detection of a stochastic gravitational wave background, a prior-dominated common-process posterior, and a 95% upper limit of A_GWB < 3.4e-14 at gamma = 13/3.","lead":"The Indian Pulsar Timing Array's second data release, 27 millisecond pulsars tracked over up to 7.2 years, shows no statistically significant sign of the gravitational wave background that longer-running arrays have reported. The paper places an upper limit on the background amplitude and argues that the array needs at least a 10-year baseline to begin detecting it.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'averages out' claim for chromatic contamination is asserted, not demonstrated; it could shift the headline upper limit, though the HD non-detection itself is robust.","rationale":"The central claim, 'no statistically significant evidence for a common red process or for Hellings-Downs correlations,' is well supported: four independent diagnostics agree, the Savage-Dickey Bayes factor is 2.5, the optimal-statistic S/N distributions peak near zero, and the HD-reweighted posterior is indistinguishable from CURN. These facts do not depend on the 'averages out' assertion. What depends on it is the precise value of the headline 95% upper limit. The paper presents direct evidence of chromatic contamination in the dropout factors of three high-precision pulsars, and its own simulation exhibits a chromatic high-amplitude bias at the current baseline. The argument that this contamination cannot affect the array-level posterior is plausible but not proven. However, if the contamination does bias the limit, the bias direction is conservative: extra per-pulsar red power can only add to the estimated common amplitude, making the upper limit weaker, not falsely strong. Chromatic noise is uncorrelated between pulsars and cannot create a false HD correlation. Therefore the non-detection and the qualitative conclusion that any GWB amplitude is at most around the reported level survive. The quantitative value deserves a caveat or a targeted test, but the reader's ACCEPT verdict does not need to change.","tokens_in":24407,"tokens_out":5314,"duration_ms":55488,"concrete_test":"Rerun the fiducial full-DR2 CURN analysis with the three dropout-excess pulsars' noise models augmented by refit high-dimensional chromatic processes, e.g., DMX plus a free chromatic spectral index, while keeping every other setup identical, and recompute the LinearExp 95% upper limit at gamma = 13/3. If the limit moves by more than about 0.1 in log10 A_GWB from -13.47, the 'averages out' claim in Secs. VB4 and VIIA fails and the upper limit must be reported with an explicit chromatic-model systematic; if it stays within 0.03-0.05, the concern is retired.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The weakest step is the paper's repeated assertion that residual chromatic power diagnosed by the dual-band dropout factors 'averages out' at the array level and 'does not bias' the upper limit (Secs. VB2, VB4, VIIA, VIII). The only quantitative support is the 0.05 difference in log10 A between the full DR2 and Band 5-only limits, but that is a comparison between two configurations, not a demonstration that either is unbiased. The CURN likelihood is block-diagonal (Eq. 2), so a per-pulsar chromatic leak adds directly to that pulsar's common-process red component; there is no cross-pulsar cancellation mechanism unless one is explicitly shown. The dropout analysis identifies exactly which pulsars leak (PSRs J1744-1134, J1909-3744, J1600-3053), and the paper's own simulation reports a short-baseline high-amplitude bias in the full DR2 configuration, with the 7.2-year validation explicitly acknowledged as weak (Sec. VIE). Thus the statement that the limit is unaffected by chromatic mismodelling is not quantitatively established. The direction of any bias is likely conservative, and per-pulsar chromatic noise cannot manufacture Hellings-Downs correlations, so the non-detection claim is not threatened; however, the numerical headline limit is less secure than the text suggests.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports the first independent pulsar-timing-array search for an isotropic stochastic gravitational-wave background using InPTA DR2: 27 millisecond pulsars with up to 7.2 years of uGMRT dual-band (Band 3 and Band 5) timing data. It applies a Bayesian common-uncorrelated-red-noise (CURN) analysis with per-pulsar dropout factors, a Savage-Dickey Bayes factor, noise-marginalized optimal statistics, and importance reweighting from CURN to Hellings-Downs correlations. The paper finds no statistically significant common red process (Bayes factor 2.5; monopole, dipole, and HD signal-to-noise ratios consistent with zero) and sets a fiducial 95% upper limit of log10 A_GWB < -13.47 (A_GWB < 3.4e-14) at spectral index gamma = 13/3. It also presents injection-recovery simulations projecting how the common-process posterior evolves for 10- and 15-year baselines, concluding that about 10 years are needed to begin recovering an injected signal at log10 A = -14.","tokens_in":24597,"tokens_out":7825,"duration_ms":73019,"significance":"If the results hold, this is a valuable and appropriate contribution: it is the first standalone SGWB constraint from InPTA, exploits the array's simultaneous dual-band capability to expose residual chromatic noise, and ships reproducible code. The analysis follows established enterprise-based PTA methodology and includes multiple cross-checks: dropout factors, solar-wind elongation cuts, Band 3 removal, CURN-to-HD reweighting, and importance reweighting with Kish effective sample sizes. The non-detection is consistent with the array's relatively short baseline and does not conflict with the evidence reported by longer-baseline arrays. The main caveat, the unsupported claim that chromatic contamination 'averages out' at the array level, does not threaten the HD-correlation non-detection but does affect how securely the numerical upper limit can be interpreted.","major_comments":[{"comment":"The statement in Section VB4 that residual chromatic power 'averages out in the array-marginalized amplitude posterior' and therefore 'does not bias the array-level amplitude limit' is not quantitatively established. Because the CURN covariance in Eq. (2) is block-diagonal, per-pulsar chromatic leakage enters that pulsar's common-process term additively, and the paper provides no explicit cancellation mechanism at the array level. The only quantitative evidence offered, the 0.05 difference between the full DR2 and Band 5-only limits, compares two configurations and does not show that either is unbiased. More importantly, the paper's own Section VIF and Table II report a short-baseline high-amplitude bias in the full DR2 configuration (log10 A recovered as -13.80 for an injected -14 signal at 7.2 yr) and describe this as 'the same chromatic-inflation signature seen in the real-data full DR2 upper limit.' This appears to contradict the 'does not bias' claim. Please either demonstrate by injection-recovery that the reported 95% upper limit has the stated coverage under the real noise model, or revise the text to state that the limit may be conservatively biased by residual chromatic power.","section":"VB4 and VIF"},{"comment":"The forecasting claim that 'it will take at least a 10 year baseline to start recovering the common red noise signal' is drawn from what appears to be a single realization per configuration: the text refers to 'all realizations' but does not state how many independent realizations were generated or averaged. With a single realization, the recovered posteriors in Table II do not carry ensemble error bars, so the 10-year values (log10 A = -13.56 with injected -14 for full DR2; gamma = 3.44 with injected 13/3) do not by themselves establish the onset of recovery. Please either add multiple realizations and report the spread of recovered medians, or soften the language to describe the 10-year posteriors as showing convergence toward the injected values rather than a demonstrated recovery.","section":"VIE-VIF"}],"minor_comments":[{"comment":"Please define the tilde notation and the matrices P_I and S_IJ explicitly before first use, since Eq. (11) uses P^{-1}_I \\tilde S_IJ P^{-1}_J without defining \\tilde S_IJ or the index ranges beyond the surrounding text.","section":"Section IVG, Eq. (11)"},{"comment":"The introductory sentence stating that current datasets 'do not yet support a statistically significant detection [8-11]' is in tension with Section VII, which describes [8-11] as having reported evidence in 2023; please harmonize the wording to avoid an apparent contradiction.","section":"Section I"},{"comment":"The caption phrase 'sorted by the InPTA-DR2 full DR2 value' is redundant; it should be simplified to 'sorted by the full DR2 dropout factor,' and the figure should explicitly state that the blue points are the Band 5-only configuration.","section":"Figure 4 caption"},{"comment":"The log-uniform reference quantile varies by 0.04 in log10 A across the elongation cuts (-13.55 to -13.59), while the text's 'stable to within 0.03' statement applies only to the LinearExp column; please specify which quantity the stability claim refers to.","section":"Table I and Section VB5"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is clearly within scope for a PTA data-release paper and the main headline finding, the non-detection of Hellings-Downs correlations, is robust. The required revision is tractable: either supply an injection-recovery coverage check for the upper limit under residual chromatic mismodelling or soften the 'does not bias' language. I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a competently executed null result from the first independent SGWB search of InPTA DR2. The headline 95% upper limit A<3.4e-14 at gamma=13/3 sits about an order of magnitude above the amplitudes reported by longer-baseline PTAs and is consistent with them. The genuinely novel piece is the dual-band dropout diagnostic: PSRs J1744-1134, J1909-3744, and J1600-3053 show dropout factors well above unity in the combined full DR2 analysis that collapse to ~1 when Band 3 is removed. That is a real, useful demonstration that simultaneous low-frequency coverage exposes chromatic mismodeling that single-band arrays would miss.\n\nThe paper does the standard things well: enterprise-based CURN/HD runs, importance reweighting with Kish effective sizes, solar-elongation cuts, NMOS, and a simulation forecast. The four diagnostics agree. The upper limit is stable to 0.03-0.05 in log10A across cuts and band configurations. The citation pattern is clean; the methodology follows established practice and the companion noise analysis is credited appropriately. The authors are also honest that the 7.2-year simulation validation is weak — the posterior is too broad to distinguish an injected signal from none. The circularity burden is low: the injected amplitude is chosen, not fitted, and the gamma=13/3 index is independently motivated.\n\nThe soft spot is loaded into the word 'averages out.' The paper repeats several times that chromatic contamination in the dropout factors does not bias the array-level amplitude limit, but the only quantitative support is that the full DR2 and Band 5-only limits differ by 0.05 in log10A. That is consistency between two configurations, not a demonstration that either is unbiased. The CURN covariance is block-diagonal, so a per-pulsar chromatic leak contributes directly to that pulsar's common red component; there is no cross-pulsar cancellation unless you show one. The direction of any bias is likely conservative, and per-pulsar chromatic noise cannot manufacture Hellings-Downs correlations, so the non-detection claim stands. But the numerical headline limit is less secure than the text implies.\n\nWho gets value: the PTA community, especially anyone working on chromatic noise mitigation and independent array constraints. It deserves a serious referee. My recommendation is accept after minor revision, with a request to either quantify the averaging claim (e.g., inject a chromatic mismodeling into the simulations) or soften the wording to 'consistent with no significant bias.'","headline":"Solid, honest null result from InPTA DR2 with a genuinely useful dual-band chromatic diagnostic; the 'averages out' claim is the only soft spot and is addressable.","tokens_in":25369,"tokens_out":3927,"would_cite":true,"duration_ms":28660,"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 reports no statistically significant common red-noise process in the Indian Pulsar Timing Array's second data release and places a 95% upper limit on any gravitational-wave background amplitude.","keywords":["stochastic gravitational wave background","pulsar timing array","common uncorrelated red noise","Hellings-Downs correlation","gravitational wave upper limit","millisecond pulsars","InPTA DR2","chromatic noise"],"falsifier":"Recompute the fiducial 95% upper limit after dropping the three pulsars whose Band 3 dropout factors are inflated (PSRs J1744-1134, J1909-3744, J1600-3053), or after replacing their chromatic noise models with more flexible ones; if the limit moves by more than about 0.05 in $\\log_{10} A_{\\rm GWB}$, the claim that per-pulsar contamination averages out is not supported.","tokens_in":24175,"feed_emoji":"📡","tokens_out":5654,"duration_ms":46787,"temperature":0.7,"pith_summary":"This paper reports the first independent search for an isotropic stochastic gravitational-wave background using the second data release of the Indian Pulsar Timing Array, 27 millisecond pulsars observed with the upgraded GMRT over up to 7.2 years. It concludes that the data show no statistically significant common red-noise process in any of four diagnostics, and that the Hellings-Downs quadrupolar correlation expected from a gravitational-wave background is absent. The headline result is a 95% upper limit on the background amplitude of $A_{\\rm GWB} < 3.4\\times10^{-14}$ at the supermassive-black-hole-binary spectral index $\\gamma=13/3$, stable across solar-wind exclusion cuts. The work matters because it adds an independent, dual-frequency dataset to the global pulsar timing array effort and directly demonstrates that residual chromatic noise can masquerade as common-process support in the most precisely timed pulsars.","feed_headline":"Indian pulsar array finds no gravitational-wave background yet","feed_subtitle":"First independent InPTA search sets a stable 95% upper limit and anticipates a 10-year horizon.","key_machinery":"The central object is the common uncorrelated red-noise (CURN) process, a common power-law spectrum with independent phase realizations per pulsar, described by amplitude $A_{\\rm CURN}$ and spectral index $\\gamma_{\\rm CURN}$. The argument runs through four diagnostics built on it: the Bayesian CURN posterior with a Savage-Dickey Bayes factor; per-pulsar dropout factors that let each pulsar's participation in the common process switch on or off; importance reweighting of the CURN chain to the Hellings-Downs-correlated model; and the noise-marginalized optimal statistic, which averages the optimal statistic over the posterior of the noise parameters. The Hellings-Downs overlap reduction function $\\Gamma_{\\rm HD}(\\zeta_{ab})$ supplies the spatial-correlation template that must be matched for a gravitational-wave interpretation. The machinery also includes the fixed spectral index $\\gamma=13/3$ from the idealized supermassive-black-hole-binary population model, used to convert the non-detection into a prior-dependent 95% upper limit via importance reweighting from a log-uniform to a linear-in-amplitude prior.","core_discovery":"Using a Bayesian common-spectrum search and a noise-marginalized optimal statistic, the paper finds no evidence for a common uncorrelated red-noise process or for Hellings-Downs correlations in InPTA DR2. The free-spectral-index posterior is broad, $\\log_{10} A_{\\rm CURN} = -13.71^{+1.06}_{-3.28}$ with $\\gamma_{\\rm CURN}=2.98^{+3.62}_{-2.70}$, and the Savage-Dickey Bayes factor for a common process over noise alone is 2.5, which the paper describes as at most a bare mention on the Jeffreys scale. The optimal-statistic signal-to-noise ratios for monopole, dipole, and Hellings-Downs correlations all peak near zero. Fixing $\\gamma=13/3$, the paper places a 95% upper limit $\\log_{10} A_{\\rm GWB} < -13.47$ ($A_{\\rm GWB} < 3.4\\times10^{-14}$), robust across solar-elongation cuts of $10^\\circ$, $20^\\circ$, and $30^\\circ$ and consistent between the full dual-band data and Band 5-only data. The paper also shows that several of the most precisely timed pulsars have dropout factors well above unity in the full data that collapse to unity when Band 3 is removed, which it interprets as residual dispersion-measure and scattering power leaking into the common process; it argues this contamination averages out in the array-level amplitude posterior. Simulations with injected signals at $\\log_{10} A_{\\rm inj}=-14$ indicate that a baseline of at least 10 years is needed before the common red process starts to be recovered.","pith_inferences":["A natural extension is to apply the same Band 3 versus Band 5 dropout comparison to other multi-band pulsar datasets to flag pulsars whose apparent common-process support is chromatic in origin.","The claim that per-pulsar chromatic contamination averages out in the array-level posterior is asserted without a quantitative proof; a direct calculation of the array posterior with and without the three affected pulsars would test whether the upper limit remains stable to the stated 0.03-0.05 in $\\log_{10} A$.","A testable prediction of the paper's interpretation is that the elevated dropout factors for PSRs J1744-1134, J1909-3744, and J1600-3053 will disappear entirely once their chromatic noise models are improved, not merely when Band 3 data are removed.","The forecast implies that a 10-year InPTA baseline should show the common-process posterior beginning to concentrate near the injected amplitude; if it instead remains prior-dominated, the chromatic inflation seen at 7.2 years may persist longer than modeled."],"forward_implications":["The non-detection is consistent with the amplitudes reported by longer-baseline pulsar timing arrays; the paper attributes the difference to the shorter observing span, since the signal-to-noise of a $\\gamma=13/3$ background grows steeply with baseline.","The reported upper limit is stable under solar-elongation cuts, indicating that unmodelled solar-wind power does not bias the array-level constraint.","Dual-band data expose chromatic contamination that single-band arrays cannot resolve: the same pulsars whose dropout factors are inflated by Band 3 would otherwise appear to support a common signal.","Simulated extensions of the InPTA data show that extending the observing baseline alone, with 27 pulsars fixed, recovers an injected signal at $\\log_{10} A_{\\rm inj}=-14$ by roughly 15 years, with a biased high amplitude at 10 years in the full configuration.","Because the simulated results hold the pulsar count fixed, adding pulsars remains a complementary route to sensitivity that this forecast does not quantify."],"supporting_citations":[{"why":"Supplies the single-pulsar noise models for all 27 pulsars that the array-level searches build on.","marker":"[36]"},{"why":"Provides the dropout method used to identify which pulsars drive the common process.","marker":"[42]"},{"why":"Provides the likelihood-reweighting technique that converts the CURN chain to a Hellings-Downs-correlated posterior.","marker":"[45]"},{"why":"Provides the noise-marginalized optimal statistic used for the frequentist signal-to-noise checks.","marker":"[49]"},{"why":"Provides the simulation pipeline adapted to forecast sensitivity at longer baselines.","marker":"[50]"},{"why":"Defines the Hellings-Downs overlap reduction function used as the gravitational-wave correlation template.","marker":"[25]"},{"why":"Gives the scaling laws used to argue that the non-detection reflects the short baseline and that timespan is the binding constraint.","marker":"[53]"}],"fun_headline_variants":["No gravitational-wave background in InPTA's second data release","Indian pulsar array finds no cosmic background, sets limit","InPTA sees no gravitational-wave background despite 7-year watch","Pulsar array rules out gravitational-wave background for now","InPTA: no gravitational-wave background in 7.2-year search"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that the single-pulsar noise models from the companion analysis fully capture each pulsar's chromatic noise, so that any remaining dispersion-measure or scattering power does not bias the common-process amplitude; the paper itself shows this assumption fails for PSRs J1744-1134, J1909-3744, and J1600-3053, and it argues without a quantitative proof that the contamination averages out at the array level.","fun_headline_variants_meta":{"raw":{"variants":["No gravitational-wave background in InPTA's second data release","Indian pulsar array finds no cosmic background, sets limit","InPTA sees no gravitational-wave background despite 7-year watch","Pulsar array rules out gravitational-wave background for now","InPTA: no gravitational-wave background in 7.2-year search"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000925,"raw_usage":{"total_tokens":4130,"prompt_tokens":1278,"completion_tokens":2852,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":894,"completion_tokens_details":{"reasoning_tokens":2764}},"tokens_in":894,"tokens_out":2852,"duration_ms":17444,"temperature":1.0,"reasoning_tokens":2764,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:58:06.207034+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the fiducial 95% upper limit after dropping the three pulsars whose Band 3 dropout factors are inflated (PSRs J1744-1134, J1909-3744, J1600-3053), or after replacing their chromatic noise models with more flexible ones; if the limit moves by more than about 0.05 in $\\log_{10} A_{\\rm GWB}$, the claim that per-pulsar contamination averages out is not supported.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the dropout method used to identify which pulsars drive the common process."},{"cited_title":"Swarup, S","cited_arxiv_id":null,"evidence_quote":"Provides the likelihood-reweighting technique that converts the CURN chain to a Hellings-Downs-correlated posterior."},{"cited_title":"Noise analysis of the Indian Pulsar Timing Array data release I","cited_arxiv_id":"2303.12105","evidence_quote":"Provides the simulation pipeline adapted to forecast sensitivity at longer baselines."},{"cited_title":"enterprise: Enhanced Numerical Toolbox Enabling Ro- bust Estimates of Pulsar Inherent Signals, enterprise Software,","cited_arxiv_id":null,"evidence_quote":"Gives the scaling laws used to argue that the non-detection reflects the short baseline and that timespan is the binding constraint."}],"review_version":1}