{"id":"e3fe30c8-7369-4204-b777-f853780ccd62","arxiv_id":"2502.00584","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Weak lensing shows post-merger and non-merging galaxies have indistinguishable mass profiles, ruling out merger starbursts that form more than 60% of the final stellar mass.","lead":"This paper uses the bending of light by gravity to compare the dark matter halos of recently merged galaxies with matched galaxies that never merged. It finds no measurable difference in the current data, but sets a first lensing-based limit on how many stars mergers can create.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 95% upper limit of 60% on merger-induced starburst stellar mass is asserted without derivation and appears inconsistent with the paper's own mapping (Sec. 1: a 20% burst gives SHMR ratio R≈1.6); the limit needs an explicit posterior calculation.","rationale":"The reader's weakest_assumption focuses on Mummi selection/purity. That is a legitimate external-validity concern, and the authors disclose it (Sec. 2.1: no correction for the selection effect). However, the Mummi classifier has been published with >95% simulated purity, and the paper's null-lensing result is carefully phrased as 'no statistically significant difference' for the selected sample, so the sample issue is a generalization caveat rather than a threat to the internal validity of the measurement. The more load-bearing issue is the undereived 60% limit, which is the abstract's headline quantitative claim. The text contains no derivation connecting the measured SHMR ratio R to a posterior on the burst fraction f, and a back-of-the-envelope consistency check using the paper's own Section 1 numbers suggests the claimed 60% limit may be far too weak (or, if a heavily skewed posterior justifies it, that the 1σ uncertainties quoted in Table 1 are severely non-Gaussian and need to be shown). This is a verifiable internal-correctness risk: one can test it with the MCMC chains or a Gaussian reconstruction. The null lensing signal (p=0.41) is robust and not affected by this concern, so the verdict remains CONDITIONAL pending a clear derivation. I credit the paper for its careful matching, jackknife errors, and disclosure of limitations; the issue is presentation and verification of the key numerical claim, not the analysis as a whole.","tokens_in":20648,"tokens_out":15518,"duration_ms":144814,"concrete_test":"Take the MCMC posterior samples for M_halo and c from the post-merger fit (or recreate them from the reported covariance and chain summaries), compute the posterior for SHMR_post = M*/M_halo, combine with the Hudson et al. (2015) SHMR model as described in Section 1 to map each sample to the burst fraction f, and report the 95% upper limit on f. If this limit is not ≈0.6, the paper's headline claim is unsupported. Concretely, with the reported R=0.58+0.53−0.28 and the Section 1 mapping (f=0.2 → R≈1.6), a Gaussian approximation gives f_95% ≈ 0.2, a strong discrepancy.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline quantitative claim—ruling out starbursts with >60% of post-merger stellar mass at 95% confidence—is never derived. Section 4 simply states: 'Our observed ratio of the post-merger SHMR to control SHMR is R = 0.58+0.53−0.28... Using the models from M. J. Hudson et al. (2015)... our results rule out extreme (≳60%) bursts of SF at the 95% confidence level.' No formula, posterior, or confidence interval for the burst fraction f is given. This is not a cosmetic omission: the paper's own Section 1 example implies that a 20% starburst raises the SHMR ratio to R≈1.6, while f=0.6 would require R≈4 under the same equal-mass merger model. The reported R=0.58 with 1σ errors +0.53/−0.28 yields a 95% upper bound on R of roughly 1.6 if the posterior is approximately Gaussian, which would place the 95% upper limit on f near 20%, not 60%. Only a heavily skewed posterior extending to R≳4 would support the reported 60% limit. Without showing the chain for f (or the R posterior and the Hudson et al. mapping), the abstract's central numerical conclusion is unverifiable and potentially misstates the strength of the constraint. This is an internal correctness risk, independent of the (separately disclosed) Mummi selection effects.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper measures galaxy-galaxy lensing around 1623 Mummi-selected post-mergers and roughly 30,000 matched non-merging controls drawn from UNIONS/SDSS. The samples are weighted to share stellar mass, redshift, and geometric-mean neighbour distance, and the excess surface density is estimated with boost correction, random subtraction, and jackknife uncertainties. Navarro-Frenk-White plus point-mass halo fits are performed to radii of 1 Mpc. The paper reports no significant difference between the two lensing profiles (chi-square p = 0.41), obtains similar halo masses of about 4 x 10^12 M_sun for both samples, and claims a 95% confidence upper limit of about 60% on the fraction of post-merger stellar mass formed in merger-induced starbursts.","tokens_in":20973,"tokens_out":10499,"duration_ms":100253,"significance":"If the starburst limit is correct, this is the first statistical weak-lensing constraint on merger-induced stellar mass growth, and the null lensing difference itself is a useful step for a sample this large. The matched-control weighting, random subtraction, boost correction, and jackknife covariance treatment are careful and largely transparent, and the paper makes its input catalogues publicly available. However, the headline quantitative claim, the 60% starburst upper limit, is not derived in the text and appears inconsistent with the paper's own numerical examples, so the significance of the central result currently rests on an unverified calculation.","major_comments":[{"comment":"The 95% upper limit on the starburst fraction is asserted without a posterior calculation. After Table 1 the text states, \"Using the models from M. J. Hudson et al. (2015)... our results rule out extreme (≳60%) bursts of SF at the 95% confidence level,\" but no formula maps the fitted R = 0.58+0.53-0.28 to a burst fraction f, and no posterior for f is shown. This is load-bearing because the Introduction says a 20% burst gives R ≈ 1.6 and Section 4 says a 10% burst gives R ≈ 1.4, while a Gaussian approximation to the quoted R uncertainty would place the 95% upper bound on R near 1.2-1.5, not near the R ≈ 4 that a 60% burst would require under the same mapping. The authors should present the R posterior, the Hudson et al. model mapping, and the resulting posterior or upper limit on f, and then reconcile the quoted 60% limit with their own examples.","section":"Section 4 / Abstract"},{"comment":"The paper explicitly states that Mummi post-mergers tend toward higher stellar masses and redshifts and that no correction for this selection effect is attempted. Because the Abstract and Conclusions interpret the null lensing signal and the starburst limit as constraining \"the merger process\" generally, this selection effect is load-bearing. If the Mummi-selected sample is not representative of post-mergers as a whole, the constraints apply only to the selected class. The authors should either quantify the selection using the IllustrisTNG mocks or explicitly qualify the Abstract and Conclusions to state that the results apply to Mummi-selected post-mergers.","section":"Section 2.1"}],"minor_comments":[{"comment":"The notations \"ND30048:76\" and \"ND2903\" appear to be LaTeX errors for N = 30048.76 and N = 2903; please fix these labels.","section":"Figure 4 caption"},{"comment":"The statement that stellar mass uncertainties from 10,000 bootstrap iterations are under 2% would benefit from a one-sentence description of the bootstrap procedure, such as the resampling unit and whether it was applied to each lens sample separately.","section":"Section 3.5"},{"comment":"The caption says the fit uses only data points not in the shaded region, but the shaded region is not labelled in the figure; please add a legend or explicitly state that the excluded points are those with R > 1.0 Mpc.","section":"Figure 7 caption"},{"comment":"For the comparison chi-square quoted as chi2_1,2 = 14.50 with p = 0.41, the number of degrees of freedom used should be stated; this is especially useful because the two profiles are compared using a covariance matrix that is not described in detail.","section":"Section 4"},{"comment":"There are several typographical and typesetting issues, including the caption \"R ΔΣ\" in Figure 6 and the repeated use of \"ND\" in figure captions; a careful proofread would improve readability.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The central qualitative result, the null lensing difference, is solid and appropriate for the journal. The main risk is the unverified 60% starburst claim: if the authors cannot provide the posterior calculation, the abstract should be revised to state only the robust null result and a qualitatively bounded burst fraction. The Mummi selection effect is disclosed in the text, but the Abstract and Conclusions should be checked for over-generalization beyond the selected sample."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I read the Cheng et al. paper on galaxy-galaxy lensing of post-mergers. The core measurement is a well-executed null result: 1623 Mummi-selected post-mergers and ~30,000 matched controls show statistically indistinguishable lensing signals (p=0.41), and the NFW fits give similar halo masses. The sample construction is careful—matching on stellar mass, redshift, and R_avg, restricting to low-density environments, jackknife errors—and the paper is honest about the selection effects in the Mummi catalogue. That part deserves credit.\n\nThe problem is the abstract's headline quantitative claim: the 95% upper limit of 60% on merger-induced starburst stellar mass. I cannot find a derivation anywhere in the text. The paper just states it, citing Hudson et al. (2015). Worse, it appears inconsistent with the paper's own Section 1 model: a 20% starburst gives R≈1.6, and the observed R=0.58+0.53−0.28 would give a 95% upper bound on R around 1.5 under a roughly Gaussian posterior. To rule out f>60%, you would need R≳4 under that model. So either the posterior is extremely skewed (which is not shown) or the limit is simply wrong. This is a load-bearing flaw because the abstract's main quantitative conclusion rests on it. The authors need to show the actual posterior of the burst fraction, or at least the R posterior and the mapping.\n\nThe selection effect is a lesser concern: the Mummi post-mergers skew toward higher mass and redshift, and the authors do not correct for it. They disclose this, so it is a caveat rather than a hidden flaw, but it does mean the null result constrains the selected post-merger population, not necessarily all mergers.\n\nThe low concentrations (c≈0.76 for post-mergers) are puzzling, and the authors spend a lot of text on possible explanations. That is fine, but it does suggest the lensing model or shear calibration may have unmodelled systematics. This is minor relative to the derivation issue.\n\nBottom line: the null lensing measurement is solid and the method scales to DESI/Euclid-era samples. The paper deserves a serious referee, but the referee should insist on a proper derivation of the starburst limit, or the claim should be removed from the abstract. I would likely cite the null result, but not the 60% limit as it stands.","headline":"A careful null lensing measurement of post-mergers, but the paper's 60% starburst upper limit is asserted without derivation and looks inconsistent with its own model mapping.","tokens_in":21700,"tokens_out":4856,"would_cite":true,"duration_ms":44862,"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":"Post-merger galaxies and non-merging controls show statistically indistinguishable weak-lensing profiles (p = 0.41), and the fitted halo models rule out starbursts forming more than 60% of the post-merger stellar mass at 95% confidence.","keywords":["galaxy-galaxy lensing","galaxy mergers","post-mergers","dark matter haloes","stellar-to-halo mass ratio","NFW profile","UNIONS survey","weak gravitational lensing"],"falsifier":"An independent post-merger sample of roughly 15,000 objects, selected without the Mummi classifier, should yield a statistically significant separation in ΔΣ(R) and an SHMR ratio R above 1 if massive merger-induced starbursts are common; if the current null is real, the same sample would tighten the 95% upper limit on the burst fraction to about 10% while keeping the two profiles consistent.","tokens_in":20375,"feed_emoji":"🌌","tokens_out":5864,"duration_ms":55470,"temperature":0.7,"pith_summary":"The paper asks whether galaxy mergers change the dark-matter haloes and stellar content of the merged product. To answer it, the authors stack the weak gravitational lensing signal around 1,623 post-merger galaxies and roughly 30,000 non-merging controls matched in stellar mass, redshift, and environment, all drawn from the UNIONS survey. They find no statistically significant difference between the two excess surface density profiles, and both are consistent with a model of a point-like stellar component plus a Navarro-Frenk-White dark-matter halo of mass about 4×$10^{12}$ solar masses. The ratio of stellar to halo mass implies that, at 95% confidence, merger-induced starbursts cannot have formed more than 60% of the final stellar mass. The result matters because it demonstrates that weak lensing can constrain merger-induced star formation, and it sets a target for next-generation surveys.","feed_headline":"Merged galaxies look just like non-mergers in lensing","feed_subtitle":"Post-mergers and matched controls show identical dark-matter profiles, ruling out starbursts above 60 percent.","key_machinery":"The central observable is the excess surface density profile ΔΣ(R), the projected mass-density contrast around a lens, measured by stacking the tangential shears of background galaxies from the UNIONS ShapePipe catalogue. The analysis weights the control sample to match the post-mergers in stellar mass, redshift, and geometric-mean distance to the three nearest neighbours, restricts both samples to low-density environments to suppress the two-halo term, applies boost-factor and random-subtraction corrections, and fits a two-component model: a fixed point-like stellar mass plus an NFW dark-matter halo with free M_halo and concentration c. The stellar-to-halo mass ratio derived from these fits carries the starburst constraint.","core_discovery":"Using galaxy-galaxy lensing excess surface density ΔΣ(R) measured around post-mergers and non-merger controls, the paper finds statistically indistinguishable lensing signals: the chi-square test over the full profile gives p = 0.41, and restricting to R ≤ 1 Mpc gives p = 0.34. Fitting a point-like stellar component plus an NFW dark-matter halo yields M_halo ≈ 4×$10^{12}$ M_sun for both samples, with a moderately negative correlation between M_halo and concentration, and the post-merger concentration is not significantly different from the control concentration. The derived stellar-to-halo mass ratio for post-mergers is 1.7% versus 2.96% for controls; interpreted with the Hudson et al. (2015) SHMR models, this rules out at 95% confidence a merger-induced starburst that forms more than 60% of the post-merger stellar mass. The paper concludes that weak lensing is a viable probe of merger properties and that a sample roughly ten times larger would be sensitive to starbursts at the ten percent level.","pith_inferences":["Beyond the paper, a direct test of the Mummi selection bias would be to rerun the analysis on post-mergers identified by an independent method, such as visual tidal-feature classification, and check whether the null ΔΣ result persists; if it does not, the current limit would apply only to the neural-network-selected subset.","The low concentrations relative to dark-matter-only simulation predictions leave room for an alternative explanation that the paper only sketches; splitting the sample by satellite likelihood, using deeper spectroscopic data, would test whether satellite contamination inflates the low-concentration signal.","An untested extension is to split post-mergers by time since coalescence: simulations predict the halo response decays over roughly a gigayear, so a stack spanning 0 to 1.7 Gyr may wash out a signal that finer time binning would reveal."],"forward_implications":["If the null signal is real, mergers do not, on average, produce markedly heavier dark-matter haloes or higher stellar fractions than non-merging galaxies of the same stellar mass and environment.","The 95% confidence upper limit of about 60% on the starburst fraction directly constrains models of merger-induced star formation at stellar masses near 10^11 M_sun.","With a post-merger sample roughly ten times larger, the same methodology is expected to detect the weak-lensing signatures of mergers and constrain starbursts at the ten percent level.","The weighting and environment-control procedure provides a template for isolating merger effects from selection effects and environment in future lensing analyses."],"supporting_citations":[{"why":"Supplies the Mummi neural-network classification and the UNIONS/SDSS post-merger and non-merger catalogue used as the lens samples.","marker":"L. Ferreira et al. (2024b)"},{"why":"Provides the stellar-to-halo-mass relation used to translate the observed SHMR ratio into the 60% upper limit on the burst fraction.","marker":"M. J. Hudson et al. (2015)"},{"why":"The previous galaxy-galaxy lensing study of mergers, which this work extends with a roughly 56 times larger post-merger sample.","marker":"D. Harvey & F. Courbin (2015)"},{"why":"Simulations predicting how mergers change dark-matter density profiles, used as the theoretical expectation that the lensing measurement confronts.","marker":"K. Wang et al. (2020)"},{"why":"Provides the RealSim mock-image pipeline used to generate the realistic IllustrisTNG images on which Mummi is trained.","marker":"C. Bottrell et al. (2019)"},{"why":"Supplies the IllustrisTNG100-1 simulation used to build the mock galaxy sample for Mummi's training and validation.","marker":"D. Nelson et al. (2019)"},{"why":"Independent measurement that merger-triggered star formation contributes about 10% of post-merger stellar mass, the consistency target for the null result.","marker":"L. Ferreira et al. (2024a)"},{"why":"Dark-matter-only concentration-mass relation used as the baseline against which the fitted low concentrations are compared.","marker":"A. R. Duffy et al. (2008)"}],"fun_headline_variants":["Lensing reveals no dark matter halo change from galaxy mergers","Merged galaxies have identical dark matter profiles in lensing","Galaxy mergers don't alter dark matter halos, lensing shows","Lensing rules out massive starbursts from galaxy mergers","Post-merger and control galaxies show same lensing signal"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the Mummi neural-network classifier, trained on IllustrisTNG mock images, having identified true post-mergers in UNIONS data with high purity and on those identified post-mergers being representative of the post-merger population; the paper does not correct for the selection bias that Mummi post-mergers have higher stellar masses and redshifts.","fun_headline_variants_meta":{"raw":{"variants":["Lensing reveals no dark matter halo change from galaxy mergers","Merged galaxies have identical dark matter profiles in lensing","Galaxy mergers don't alter dark matter halos, lensing shows","Lensing rules out massive starbursts from galaxy mergers","Post-merger and control galaxies show same lensing signal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000582,"raw_usage":{"total_tokens":2822,"prompt_tokens":1113,"completion_tokens":1709,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":729,"completion_tokens_details":{"reasoning_tokens":1623}},"tokens_in":729,"tokens_out":1709,"duration_ms":13512,"temperature":1.0,"reasoning_tokens":1623,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T18:26:19.149305+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"An independent post-merger sample of roughly 15,000 objects, selected without the Mummi classifier, should yield a statistically significant separation in ΔΣ(R) and an SHMR ratio R above 1 if massive merger-induced starbursts are common; if the current null is real, the same sample would tighten the 95% upper limit on the burst fraction to about 10% while keeping the two profiles consistent.","supporting_citations":[{"cited_title":"2015, MNRAS, 451, L95, doi: 10.1093/mnrasl/slv073","cited_arxiv_id":null,"evidence_quote":"The previous galaxy-galaxy lensing study of mergers, which this work extends with a roughly 56 times larger post-merger sample."}],"review_version":1}