{"id":"05aae182-32c6-41c0-9528-57929daed789","arxiv_id":"2411.08565","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":9,"one_line_summary":"The dark matter clump in lens J0946+1006 is a subhalo at z=0.207 with a steep density profile, a >5 sigma outlier from LCDM expectations, possibly pointing to self-interacting dark matter.","lead":"Astronomers measured the distance to a dark matter clump that bends light in the lens galaxy J0946+1006, finding it sits alongside the galaxy rather than far behind it. If the clump is a satellite, its dense, steep profile is hard to square with cold dark matter and may hint that dark matter particles interact.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The SIDM claim rests on the S1&S2 steep-slope measurement; the multipole expansion truncated at n=4 could let main-deflector complexity be absorbed by the tNFW perturber, biasing gamma_2D.","rationale":"The reader's weakest assumption is the same as my primary concern, and I find it genuinely load-bearing. The paper's key novelty is not the redshift measurement (which shifts little between S1-only and S1&S2) but the claimed steep profile and 5-sigma tension. That steep profile is entirely produced by the two-source configuration under a specific, finite harmonic model for the main deflector. The authors themselves note that without S2 a larger m200 subhalo fits if multipoles adjust, and that the S1-only shallow slope is an artifact. This admission demonstrates how sensitive the subhalo parameters are to the assumed multipole content. Since the multipole radial shape is fixed to the EPL slope and only orders 1,3,4 are included, there is no in-paper test that the S2 data are sufficient to break degeneracies with higher-order angular structure. A steepening of 0.8 in gamma_2D between S1-only and S1&S2 is a large shift, and the final value is close to the SIDM expectation; an incomplete basis for the main-deflector mass is exactly the kind of systematic that can produce a spurious detection. The proposed multipole-extension and injection tests are feasible with the existing Herculens code. I therefore do not object to the cautious CONDITIONAL verdict; if anything, these tests should be among the conditions. I credit the paper for clear reporting of the excluded chain, the ELBO approximation, and the skewed posterior, but those caveats do not remove the need to test the completeness of the angular basis.","tokens_in":23126,"tokens_out":8172,"duration_ms":79904,"concrete_test":"Fit the S1&S2 data with the same pipeline but add multipole orders n=5 and n=6, and also run a variant in which the multipole radial slope is freed from the EPL slope (or replaced by a pixelated potential correction near the subhalo). If gamma_2D stays below -1.6 and A_M5/A_M6 are consistent with zero, the concern is mitigated. Complement with an injection-recovery test: generate mock data from an EPL+shear+n=1..6 model with no steep subhalo, then analyze with the paper's n={1,3,4}+tNFW model; if the recovered gamma_2D is spuriously steep, the central claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing result is the S1&S2 measurement gamma_2D = -1.81 +/- 0.11 and the corresponding >5-sigma vmax-rmax outlier. This result appears only after adding the second source; with S1 alone the same model gives gamma_2D = -1.00, and the difference is attributed to a degeneracy between the M1 multipole and subhalo mass. That attribution assumes the main deflector's angular complexity is fully described by the EPL + shear + n = {1,3,4} multipole set of Eq. (7), with all multipoles sharing the EPL radial scaling. The S1&S2 model does reduce A_M1 from 0.18 to 0.036, but it does not test whether n>=5 multipoles, or multipoles with a different radial slope, can mimic the localized perturbation of a small-r_s tNFW subhalo at the well-measured offset (0.68\", 1.0\"). The resulting scale radius, log10 r_s = -1.42 (~0.09 kpc), is below the HST PSF FWHM, so the steep projected slope between 0.75 and 1.25 kpc is an extrapolation of the assumed tNFW shape rather than a direct measurement of the inner cusp. If omitted harmonic degrees of freedom or a different halo radial profile can absorb this signal, the gamma_2D and vmax-rmax outliers, and hence the SIDM gravothermal-collapse interpretation, would not survive. The paper's own caveats (ELBO rather than formal evidence, skewed posterior, no correlated drizzled noise, one excluded chain) are secondary to this systematic because they affect significance rather than the parameter values themselves.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"Enzi et al. model the double source plane lens J0946+1006 with a compound-lensing forward model, fitting jointly the main deflector (EPL plus external shear plus n = 1, 3, 4 multipoles), a truncated NFW dark perturber with free redshift, a Gaussian-component lens light model, and pixelated GP-regularized sources. With both sources they infer z_halo = 0.207 +/- 0.019, consistent with the main deflector at z = 0.222; an approximate ELBO-based comparison prefers a subhalo over a line-of-sight halo; and they derive vmax = 87.85 km/s, rmax = 0.28 kpc, and a projected slope gamma_2D = -1.81 at 0.75-1.25 kpc, which they interpret as a more than 5-sigma outlier from LambdaCDM expectations and as possible evidence for gravothermal collapse in self-interacting dark matter. The S1-only model yields a shallower gamma_2D = -1.00, which the authors attribute to a degeneracy between the M1 multipole and subhalo mass that the second source breaks.","tokens_in":23610,"tokens_out":12640,"duration_ms":111257,"significance":"If the results hold, the paper provides one of the first constraints on the redshift of a strong-lens dark perturber, and a rare measurement of an ultra-concentrated sub-kpc dark halo whose properties are in tension with LambdaCDM and consistent with SIDM gravothermal collapse; this would be an important step for the field. The analysis has notable strengths: it is fully forward-modeling, the priors for all parameters are tabulated, the pipeline is built on the public Herculens code, and the authors report their main caveats explicitly (ELBO-based model comparison, skewed posteriors, ignored drizzled noise, and one excluded chain). I do not see internal circularity: vmax, rmax, gamma_2D, and M2D are post-processing of the fitted tNFW parameters, and comparing them with external simulation relations is standard practice. The credibility of the SIDM interpretation, however, is limited by the untested completeness of the multipole expansion and by the assumed profile family.","major_comments":[{"comment":"The subhalo-versus-field-halo model comparison in Section 5.2 is based on an ELBO ratio, and the abstract states that 'lower bounds on the evidence strongly prefer a subhalo over a line-of-sight structure.' Because the ELBO is a separate lower bound on each model's log evidence, the difference between two ELBOs is not itself a lower bound on the log Bayes factor, so the quoted preference (log alpha_Bayes about 7; a 1-in-1000 chance of a field halo) is an uncalibrated approximation rather than a rigorous bound. Please compute a proper evidence (e.g., nested sampling or thermodynamic integration), or explicitly re-label the result as an indicative estimate and soften the abstract and conclusion language accordingly.","section":"§5.2 and §4.2"},{"comment":"The headline results (gamma_2D = -1.81 and the more than 5-sigma vmax-rmax outlier) depend on the assumptions that the main deflector's complexity is fully described by the n = {1,3,4} multipoles of Eq. (7) with the EPL radial scaling, and that the perturber is a spherical truncated NFW profile of Eq. (10). The paper convincingly shows that the second source breaks the specific M1 degeneracy (A_M1 drops from 0.18 to 0.036 and gamma_2D changes from -1.00 to -1.81), but it does not test whether higher-order multipoles (n >= 5) or multipoles with a different radial scaling can mimic the localized small-r_s perturbation, nor whether an alternative profile shape would yield a different slope. Since log10 r_s = -1.42 (about 0.09 kpc) is below the HST PSF FWHM, the measured steep slope at 0.75-1.25 kpc is largely an extrapolation of the assumed tNFW shape. I recommend adding an explicit robustness test (for example, including n = 5 and 6 multipoles, freeing the multipole radial scaling, or fitting a profile with a free inner slope) before presenting the SIDM gravothermal-collapse interpretation as supported.","section":"§5.2, §3.3.2, §3.3.3"},{"comment":"The abstract claims that the subhalo is a 'more than 5-sigma outlier' from the LambdaCDM vmax-rmax relation, whereas Section 5.2 states that the posterior is 'highly skewed' and that 'a concrete statement on the level of this tension (is) difficult due to a lack of samples close to the relation,' and also notes that the comparison relation does not account for redshift dependence. The significance should be defined precisely (for example, as a posterior probability or a quantile-based equivalent sigma value), and the abstract should carry the same caveats that the body of the paper states; as written, the abstract is stronger than the analysis supports.","section":"Abstract and §5.2"}],"minor_comments":[{"comment":"The exclusion of one S1-only chain that converged to a solution with an unconstrained r_s needs to be documented more transparently: please report the number of chains, the nature of that solution, and whether any S1 and S2 chains exhibited a similar mode, since 'its inclusion would not significantly change our results' is not a sufficient justification for removing it.","section":"§5.2"},{"comment":"The noise model ignores correlations introduced by drizzling; please add a brief statement of the expected direction and size of the effect on the inferred parameter uncertainties, particularly in light of the supersampling sensitivity found by Minor (2024).","section":"§3.5, Eq. (14)"},{"comment":"Several lens-light amplitude posteriors (for example, A = 1.9^{+3931}_{-1.9}) are effectively unconstrained; please indicate which Gaussian components are identified by the data and report the actual standard-deviation agreement with previous work instead of rounding the agreement column to the next higher integer.","section":"Table 1"},{"comment":"The green 'S1 & S2 (*)' model, in which the halo does not affect the lens light, is important for the redshift argument but is never defined in the text; please state explicitly how this model was constructed and which prior it assumes for the halo position.","section":"§5.3, Fig. 7"},{"comment":"The closing statement that 'it seems likely that CDM will soon be definitively ruled out' overreaches the evidence from a single system and should be softened to match the caveats in Section 5.2.","section":"§6"},{"comment":"Please correct minor typographical issues, including 'ellitpical' in Eq. (6), the missing spaces in the Table 2 caption, and general proofreading of the compiled text.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"This is a well-executed and potentially important analysis; the free-redshift constraint on the J0946+1006 perturber is a genuine advance over prior work that assumed z = z_main. My concern is that the abstract and conclusions present the steep-profile/SIDM interpretation more definitively than the body, which is appropriately hedged, and that the headline significance and the subhalo/field model comparison rest on approximations (skewed posteriors and an ELBO ratio) whose calibration is not demonstrated. The paper fits the journal's scope well and is likely publishable after the robustness tests or claim-tempering described in the major comments."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should read this paper if you care about subhalo lensing or SIDM. The genuinely new thing is that they fit the redshift of the dark perturber rather than assuming it sits at the main deflector, and they get z_halo = 0.207 ± 0.019, consistent with a subhalo and robust to whether the second source is included. The evidence preference for subhalo over line-of-sight halo is based on an approximate ELBO ratio rather than a formal evidence calculation, but even a generous prior against subhaloes leaves it a ~1-in-1000 proposition. That part of the paper is solid.\n\nThe modeling is also more careful than earlier work: they fit the lens light simultaneously, include multipole perturbations (M1, M3, M4), use a pixelated source with a Matérn GP, and show honestly that the S1-only steep slope disappears when degeneracies with the M1 multipole are broken by the second source. The S1&S2 measurement of gamma_2D = -1.81 ± 0.11 is the load-bearing result, and the paper flags its own caveats: skewed posterior, no redshift dependence in the vmax–rmax comparison relation, an excluded chain, and uncorrelated drizzled noise.\n\nThat said, the stress-test concern lands. The multipole expansion is truncated at n = 4 and all multipoles share the EPL radial scaling, so there is no test of whether higher-order multipoles or multipoles with a different radial slope can mimic a compact tNFW perturber. The inferred scale radius is log10 r_s ~ -1.42, about 0.09 kpc, which is below the HST PSF FWHM; the steep inner slope is therefore an extrapolation of the assumed tNFW profile, not a direct measurement. The second source does break the M1 degeneracy, but it does not by itself validate the parametric form. So I read the paper's own caveats as secondary; the parameterization flexibility is the primary systematic. It does not sink the redshift constraint, which is the cleanest new result, but it should make you hesitate before treating a gravothermal-collapse interpretation as established.\n\nThe citation pattern is fine and the comparison to previous work (M24, B24, D24) is fair. The code and sampler are public, which is a real plus. This is a paper for strong-lensing specialists and dark-matter phenomenologists; it deserves a serious referee rather than a desk rejection. I would accept it for review, ask for a robustness test against higher-order or radially-varied multipoles and a formal evidence estimate, and see whether the >5-sigma claim survives.","headline":"A careful re-analysis that credibly constrains the perturber redshift in J0946+1006, but the SIDM steep-slope claim leans on parametric choices the current data may not be able to test.","tokens_in":24173,"tokens_out":1557,"would_cite":true,"duration_ms":16586,"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 dark perturber in J0946+1006 is a subhalo at the lens redshift with a density slope of -1.81, a more than 5 sigma outlier from cold-dark-matter predictions that points toward self-interacting dark matter.","keywords":["strong gravitational lensing","dark matter substructure","self-interacting dark matter","gravothermal collapse","subhalo","SDSS J0946+1006","multipole perturbations","Bayesian inference"],"falsifier":"Re-fit the same HST image allowing additional multipole orders (for example $n=2,5,6$) or a non-parametric perturber profile; if the posterior on the projected slope $\\gamma_{\\rm 2D}$ moves to within about $2\\sigma$ of $-1$, or the $v_{\\rm max}$--$r_{\\rm max}$ tension drops below $5\\sigma$, the claimed evidence for a collapsing SIDM subhalo is falsified. Higher-resolution imaging that resolves scales below the current PSF and recovers a slope near $-1$ would directly contradict the steep-profile claim.","tokens_in":22844,"feed_emoji":"🔭","tokens_out":15024,"duration_ms":126753,"temperature":0.7,"pith_summary":"This paper asks where the known dark structure in the strong lens J0946+1006 really sits and what its mass profile implies for dark matter. By letting the perturber's redshift be a free parameter and modelling the first two source planes together with multipole perturbations of the main galaxy, the authors find that the perturber is a subhalo at $z_{\\rm halo}=0.207^{+0.019}_{-0.019}$, rather than a line-of-sight halo. With the second source included, the inferred subhalo has an average projected density slope $\\gamma_{\\rm 2D}=-1.81^{+0.15}_{-0.11}$ between 0.75 and 1.25\\,kpc and is a more than $5\\sigma$ outlier from the $\\Lambda$CDM $v_{\\rm max}$--$r_{\\rm max}$ relation. Because gravothermally collapsed self-interacting dark matter haloes are expected to have $\\gamma_{\\rm 2D}\\approx -2$, the paper presents this steep subhalo as evidence that dark matter may self-interact.","feed_headline":"Dark subhalo in J0946+1006 is a 5-sigma outlier for cold dark matter","feed_subtitle":"Free-redshift modelling places the perturber inside the lens galaxy with slope -1.8, matching self-interacting dark matter.","key_machinery":"The load-bearing machinery is compound lensing with two source planes and a free-redshift truncated NFW perturber, coupled to multipole perturbations of the main deflector. The recursive multi-plane lens equation propagates image positions through the redshift planes, with the family-ratio factors setting how strongly the perturber deflects each source; this is what lets the redshift of a dark, lightless halo be constrained from its lensing effect alone. The second source plane does the decisive work: because its arcs probe the mass distribution at larger radii, it breaks the degeneracy between the mass, scale radius, and slope of the subhalo and the amplitudes of the multipole perturbations, particularly the first-order lopsided term. The comparison benchmarks are the $\\Lambda$CDM $v_{\\rm max}$--$r_{\\rm max}$ relation from cosmological simulations and the $\\gamma_{\\rm 2D}\\approx -2$ prediction for SIDM gravothermal collapse.","core_discovery":"The central claim is that the compact dark mass perturbing J0946+1006 is a genuine subhalo of the main lens galaxy, not a line-of-sight structure, and that once modelling degeneracies are broken it is far too concentrated and steep for a cold dark matter halo. The posterior redshift is $z_{\\rm halo}=0.207^{+0.019}_{-0.019}$, consistent with the main deflector at $z_{\\rm main}=0.222$, and the evidence ratio corresponds to roughly a 1-in-1000 chance that the perturber is a field halo. Including the second source does not tighten the redshift, but it breaks the degeneracy between the halo parameters and the lopsided first-order multipole of the main galaxy: the single-source model allows a slope $\\gamma_{\\rm 2D}\\approx -1.0$ that is compatible with CDM, while the two-source model gives $\\gamma_{\\rm 2D}=-1.81^{+0.15}_{-0.11}$. The subhalo's maximum circular velocity and radius, $v_{\\rm max}=87.85^{+16.86}_{-9.81}\\,\\mathrm{km\\,s^{-1}}$ and $r_{\\rm max}=0.28^{+0.27}_{-0.16}\\,\\mathrm{kpc}$, place it more than $5\\sigma$ from the $\\Lambda$CDM relation, and even as a field halo it would be an approximately $4\\sigma$ outlier in mass--concentration. The steep slope is close to the $\\gamma_{\\rm 2D}\\approx -2$ signature of gravothermally collapsed SIDM haloes, so the paper concludes that self-interacting dark matter may explain the anomaly, while noting the current HST data cannot make this conclusive.","pith_inferences":["If this subhalo is really collapsed, a single strong-lens system may serve as the first direct astrophysical calibration point for the SIDM cross section and for gravothermal collapse timescales.","A natural test is to look for similarly steep, compact subhaloes in other compound lens systems and compare their abundance with SIDM-collapse and CDM predictions.","Before settling on SIDM, the same data should be re-fit with higher-order multipoles and a more flexible perturber profile, since baryonic contraction or an insufficiently flexible host model could mimic a steep dark slope.","The free-redshift technique could be generalised to map dark matter along the line of sight rather than only in the lens plane, turning multi-plane lenses into tomographic probes of halo location."],"forward_implications":["The system becomes one of the very few dark-matter-only subhaloes detected in strong lensing, making it a single-object benchmark for subgalactic dark matter models.","A more than $5\\sigma$ outlier in the $v_{\\rm max}$--$r_{\\rm max}$ plane means this subhalo is very rare in $\\Lambda$CDM; if such outliers are common, Euclid's large lens sample will reveal it.","A slope near $-2$ is the expected signature of gravothermal collapse, so the result supports self-interacting dark matter over cold dark matter on subgalactic scales.","The single-source-only reconstruction, which looks CDM-compatible, is shown to be an artifact of multipole degeneracy, establishing that future substructure searches should include all available source planes."],"supporting_citations":[{"why":"Defines the data reduction and shows that modelling the second source breaks lens-model degeneracies; this paper extends it by fitting the lens light simultaneously and freeing the halo redshift.","marker":"B24"},{"why":"Shows that supersampling and the second source are important for the inferred subhalo, and finds the same steep slope that this paper reproduces.","marker":"M24"},{"why":"Previous analysis of the perturber's luminosity and concentration that provides the comparison values for the halo parameters.","marker":"D24"},{"why":"First identified the overconcentration of the J0946+1006 dark halo relative to $\\Lambda$CDM, the tension this paper revisits with a free redshift.","marker":"Minor et al. (2021)"},{"why":"Supplies the simulation-based $v_{\\rm max}$--$r_{\\rm max}$ relation used to quantify the more than $5\\sigma$ outlier.","marker":"O'Riordan et al. (2023)"},{"why":"Simulations of SIDM haloes showing that gravothermal collapse produces $\\gamma_{\\rm 2D}\\approx -2$, the benchmark for the steep-slope interpretation.","marker":"Turner et al. (2021)"},{"why":"Discovered the dark perturber in J0946+1006 and defines the object whose redshift and profile are constrained here.","marker":"Vegetti et al. (2010)"},{"why":"Establishes that compound lensing can produce images qualitatively different from single-plane lensing, motivating the free-redshift treatment.","marker":"Collett & Bacon (2016)"},{"why":"Shows that unmodelled multipole structure can bias subhalo recovery, motivating the inclusion of multipole perturbations in the mass model.","marker":"Nightingale et al. (2024)"}],"fun_headline_variants":["J0946+1006 subhalo is 5-sigma too steep for cold dark matter","Subhalo in J0946+1006: slope -1.8 challenges CDM, supports SIDM","Dark halo in J0946+1006 is a subhalo, too steep for LambdaCDM","Steep collapsed subhalo in J0946+1006 suggests self-interacting dark matter"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The main galaxy's mass distribution is assumed to be exactly the chosen smooth power law plus shear plus multipole orders 1, 3 and 4, with the second source breaking the degeneracy completely; if the galaxy has additional angular structure that this model cannot absorb, the inferred steep subhalo could be an artifact.","fun_headline_variants_meta":{"raw":{"variants":["J0946+1006 subhalo is 5-sigma too steep for cold dark matter","Subhalo in J0946+1006: slope -1.8 challenges CDM, supports SIDM","Dark halo in J0946+1006 is a subhalo, too steep for LambdaCDM","Steep collapsed subhalo in J0946+1006 suggests self-interacting dark matter"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00084,"raw_usage":{"total_tokens":3829,"prompt_tokens":1281,"completion_tokens":2548,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":897,"completion_tokens_details":{"reasoning_tokens":2441}},"tokens_in":897,"tokens_out":2548,"duration_ms":17352,"temperature":1.0,"reasoning_tokens":2441,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:31:42.403022+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-fit the same HST image allowing additional multipole orders (for example $n=2,5,6$) or a non-parametric perturber profile; if the posterior on the projected slope $\\gamma_{\\rm 2D}$ moves to within about $2\\sigma$ of $-1$, or the $v_{\\rm max}$--$r_{\\rm max}$ tension drops below $5\\sigma$, the claimed evidence for a collapsing SIDM subhalo is falsified. Higher-resolution imaging that resolves scales below the current PSF and recovers a slope near $-1$ would directly contradict the steep-profile claim.","supporting_citations":[],"review_version":1}