{"id":"2d962bba-2ebb-483b-8cf1-3a294eb9882c","arxiv_id":"1909.01349","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Local convergence ratios and reduced shears for B0128+437, derived model-independently at image and subcomponent scales, indicate that elliptical lens models are too simple for this system when milliarcsecond substructure is included.","lead":"This paper applies a model-independent method to measure local gravitational lensing properties of the lensed quasar B0128+437, including at the scale of its milliarcsecond subcomponents. It finds that simple elliptical lens models cannot explain the subcomponent positions, while free-form and model-independent approaches agree, suggesting an asymmetric mass distribution in the lensing galaxy.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The subcomponent matching across images is the load-bearing link: the parity-based selection of the Biggs et al. labelling is not exhaustively tested, and the Lensmodel, PixeLens, and FSQ comparisons all inherit that labelling.","rationale":"The reader identified the subcomponent matching as the weakest assumption, and I agree that this is the most load-bearing element of the paper. The model-independent local lens properties, the Lensmodel fits, the PixeLens reconstructions, and the FSQ deviations all depend on which observed subcomponent is identified with which source component. The paper tests only a restricted set of permutations (swapping 1 and 3 in C and D, later also in B) and rejects them using a parity criterion that is itself assumed rather than derived from morphology. The central claim that elliptically symmetric models cannot reproduce the resolved structure would collapse if an alternative, physically motivated matching allowed an elliptical potential to fit all subcomponents within the 0.1 mas astrometric uncertainty. I also note the internal inconsistency between the abstract's 40% overlap and the body's 50% overlap, and the unremoved author note in Section 8.2, but these are correctable and do not affect the core reasoning as much as the matching degeneracy. Because the concern supports the reader's CONDITIONAL verdict rather than overturning it, I keep the verdict unchanged.","tokens_in":23216,"tokens_out":8636,"duration_ms":90944,"concrete_test":"Independently determine the relative parities of the subcomponents in images A, C, and D from the VLBI image morphology (e.g., the handedness or order of the three components along the jet) rather than from the smooth-lens topology assumption of Section 3. Then repeat Section 4.2.1's matching selection using those parities, and for every parity-consistent permutation of the subcomponent labels (at least the 8 swaps of 1 and 3 in C and D, ideally all 36 permutations for ACD) run the Lensmodel fits for configACD and configABCD and the FSQ analysis. If any permutation has the correct morphological parities and yields sub-mas rms with an elliptical potential, the conclusion that elliptical models are too simplistic fails; if all such permutations still give rms above the 0.1 mas astrometric uncertainty, the conclusion is robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that elliptically symmetric models are too simplistic is conditional on the subcomponent matching adopted from Biggs et al. (2004). Section 4.2.1 rejects alternative matchings (Table 4, Table B.1) solely because they do not produce the assumed relative parities of the magnification ratios J_C and J_D; the assumed parity pattern itself (A and C same parity, D opposite) is stated in Section 3 as an assumption, not derived from the VLBI morphology. The search over matchings is also narrow: only swaps of the outer labels 1 and 3 in images C and D are tested, keeping subcomponent 2 fixed as the core. If dust extinction or scatter-broadening changes the brightness ordering of the subcomponents, other permutations (including moving subcomponent 2) could be physical; the paper itself notes in Section 3 that the intensity-based labelling of A and D may be wrong. Because the Lensmodel fits, the PixeLens reconstructions, and the FSQ analysis all use the same Biggs et al. labelling, the whole comparison is not model-independent with respect to the matching. A parity-consistent alternative matching that also fits with an elliptical potential would overturn the central claim; the paper does not run such a test.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper applies the model-independent local lens-property formalism of Wagner & Tessore (2018) to the quad lens B0128+437 using VLBI-resolved subcomponents in images A, B, C, and D. It derives convergence ratios f_j and reduced shear components g_j on image and subcomponent scales, constructs PixeLens free-form and Lensmodel parametric reconstructions, and performs a Fundamental Surface of Quads (FSQ) symmetry analysis. The authors conclude that elliptically symmetric models are too simplistic to characterise the asymmetric mass distribution of B0128, and that milli-arcsecond-scale mass-density gradients cannot be excluded from the limited overlap of subcomponent-scale local lens properties.","tokens_in":23440,"tokens_out":4495,"duration_ms":43039,"significance":"The paper has clear strengths: the analysis is blinded through an independent mediator, alternative subcomponent matchings are systematically explored in Table 4 and Appendix B, observed flux ratios are held back as a consistency check rather than used as constraints, and the model-independent code is publicly available. If the conclusions stand, the work provides an efficient route to local lens properties on sub-component scales and a concrete example where free-form reconstructions are needed. The evidential weight is nevertheless moderate, because the central conclusion is conditional on the adopted subcomponent labelling and on several assumed positional uncertainties.","major_comments":[{"comment":"The abstract states that 'only 40% of the small-scale subcomponent local lens properties overlap within the 1-sigma confidence bounds', while Section 8.1 states that the subcomponent-level local lens properties 'only overlap in 50% of the cases within their 1-sigma confidence bounds' and that the same degree of agreement is found against the Lensmodel reconstruction. This quantitative statement is used to support the milli-arcsecond gradient conclusion, so the manuscript must specify which comparison the 40% refers to and make the abstract consistent with the body.","section":"Abstract and Section 8.1"},{"comment":"The choice of the Biggs et al. (2004) labelling as the only viable matching is made because alternative matchings fail to produce the assumed relative parities of J_C and J_D. The parity pattern itself (A and C same parity, D opposite) is stated as an assumption in Section 3, not derived from the VLBI morphology or from an independent parity measurement. Since the PixeLens, Lensmodel, and FSQ analyses all inherit this labelling, the central conclusion that elliptically symmetric models are too simplistic is not model-independent with respect to the matching. Please test parity-consistent alternative matchings, including permutations that move the inner subcomponent 2, and show whether any such matching can be accommodated by an elliptical potential with external shear.","section":"Section 4.2.1, Table 4, Appendix B"},{"comment":"Section 5.2 says that the radio data provide an astrometric precision of 0.01 milliarcsecond, while Section 3 and Table 2 state an assumed uncertainty of 0.1 mas for subcomponent positions in images A, C, and D and 1 mas for image B. The Lensmodel rms values in Table 6 (0.0005-0.0028 arcsec) are judged against this assumed uncertainty. The factor-of-ten discrepancy changes the assessment of whether simple parametric models are ruled out, and the positional error budget used for the Lensmodel comparison must be corrected and justified.","section":"Section 5.2 and Table 6"},{"comment":"The three subcomponent positions of image B are read off Figure 6 of Biggs et al. (2004) with an assumed 1 mas uncertainty, and the four-image results in Table 5 as well as the ABCD PixeLens and Lensmodel comparisons depend on these values. The 3-mas robustness test in Appendix C changes the most likely values of f_B, g_B, and J_B substantially; for example, f_B changes from -16.75 in the fiducial case to -2.50 or -0.72 depending on configuration. The paper should state explicitly whether these shifts are within the quoted confidence bounds and whether the parity-related conclusions for image B are robust to the assumed B-position uncertainty.","section":"Section 3, Table 2, and Appendix C"}],"minor_comments":[{"comment":"Section 8.2 contains an unremoved author note: 'does that sound better? for me, it's strange to produce something within some error' after the statement about Lensmodel being unable to reproduce images within the astrometric precision. This note must be removed and the sentence rephrased as a finished claim.","section":"Section 8.2"},{"comment":"There are typographical errors: 'astromentric' in Section 5.2 and 'illustrateed' in Section 8.2 should be corrected to 'astrometric' and 'illustrated'.","section":"Sections 5.2 and 8.2"},{"comment":"The PixeLens rms uncertainties in Table 7 are computed as the dispersion between 20 sets of 10 models, but the text does not state whether the quoted local lens properties are means over the 200 accepted models or over the 20 averaged sets; please clarify this in the table caption or text.","section":"Table 7 and Section 6.2"}],"recommendation":"major_revision","confidential_remarks":"The unremoved author note in Section 8.2 and the abstract/body mismatch on the 40% vs. 50% overlap should be fixed before publication. The main scientific risk is the parity-based selection of the subcomponent labelling; I would encourage the editor to ensure that the revision includes an explicit test of parity-consistent alternative matchings, since the central claim depends on that choice."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nYou should know two things. First, this is the first model-independent measurement of local lens properties on milliarcsecond scale for B0128, and it includes a PixeLens free-form reconstruction using all 12 subcomponents as constraints. The analysis is careful: the model-independent and model-based work was blinded, flux ratios were held back as a consistency check, and the Lensmodel failure at full subcomponent set is documented with rms values. Second, the quantitative claims are conditional on the subcomponent matching, and that matching is the weakest link.\n\nWhat is genuinely new here is the application of Wagner & Tessore (2018) to a galaxy-scale lens with resolved substructure, plus the extension to subcomponent-scale properties using Gaussian-fit endpoints. That is a legitimate extension, not a new formalism. The central qualitative result—smooth elliptical potentials cannot reproduce all 12 subcomponents at the 0.1 mas level—is well supported by the Lensmodel rms, the PixeLens mass maps, and the FSQ deviations. The authors also deserve credit for checking alternative reference images and for being explicit about the broad confidence intervals. The citation pattern is clean; the self-citations are to the method papers and are appropriate.\n\nThe matching issue is real. Section 4.2.1 selects the Biggs et al. labelling because it is the only one of four tested configurations that gives the assumed parities (A and C same, D opposite). But the parity pattern is stated as an assumption in Section 3, not derived from morphology. And the search only swaps labels 1 and 3; subcomponent 2 is never moved. If dust changes brightness ordering, a different permutation, including one that moves the middle subcomponent, could be physical and still parity-consistent. The authors themselves note that the intensity-based labelling of A and D may be wrong. Because Lensmodel, PixeLens, and FSQ all inherit the same labelling, a parity-consistent alternative could shift the f/g values and weaken the comparison. This does not kill the broad conclusion, since prior work already found this lens hard to model, but Table 5 should be read as matching-dependent.\n\nMinor issues: the abstract says 40% overlap while Section 8.1 says 50%; the B subcomponent positions are read from a published figure with an assumed 1 mas error (tested at 3 mas, but still digitized); and Section 8.2 contains an unremoved author note asking whether a phrase sounds better. Those are easy fixes.\n\nThis is a solid case study for people working on galaxy-scale lenses, substructure, and parametric versus free-form modelling. It is not a breakthrough, but it is honest, reproducible, and useful. I would bring it to a reading group and cite it. Send it to peer review; the referee should push for a broader matching search or a morphological justification for fixing label 2 and the parity pattern, and for fixing the inconsistencies above. After that, it should be acceptable.","headline":"Careful, useful application of an existing method to a hard lens; the quantitative results are conditional on a subcomponent matching that is narrower than it should be, but the qualitative conclusion survives.","tokens_in":24018,"tokens_out":5046,"would_cite":true,"duration_ms":51757,"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 B0128+437's resolved milliarcsecond quasar subcomponents rule out smooth elliptical lens models and require free-form mass reconstructions.","keywords":["gravitational lensing: strong","dark matter","methods: analytical","galaxies: individual: B0128+437","galaxies: quasars: general","model-independent lens properties","free-form mass reconstruction","VLBI subcomponents"],"falsifier":"A smooth, elliptically symmetric lens model (or a smooth power-law potential with external shear) that reproduces all 12 subcomponent positions to within about 0.1 milliarcseconds would falsify the paper's central claim; alternatively, multi-band observations that establish a different physical matching of the subcomponents would shift the derived convergence ratios and reduced shears, as the paper's own comparison of alternate matchings shows.","tokens_in":1791,"feed_emoji":"🔭","tokens_out":2740,"duration_ms":97419,"temperature":0.7,"pith_summary":"This paper sets out to characterize the galaxy-scale gravitational lens B0128+437 using the milliarcsecond radio subcomponents of its four quasar images. It argues that a model-independent extraction of local lens properties, specifically ratios of scaled mass densities (convergences) and reduced shears at the image positions, together with free-form pixel reconstructions, yields a consistent picture, while smooth elliptical parametric models fail once all subcomponents are included. The upshot is that the lens galaxy's mass distribution is asymmetric on small scales, with possible mass-density gradients on milliarcsecond scales that smooth models cannot capture.","feed_headline":"Smooth elliptical lens models fail quasar lens B0128's subcomponents","feed_subtitle":"Resolved radio subcomponents show the lens mass is asymmetric; simple elliptical fits miss real structure.","key_machinery":"The load-bearing identity is the mass-sheet-invariant relation between multiple images: $f_{ij} = (1-\\kappa_i)/(1-\\kappa_j)$ and $g_i = \\gamma_i/(1-\\kappa_i)$, obtained by linearly mapping three reference points (subcomponent positions, or the endpoints of fitted Gaussian axes) from a reference image to the others. These are the leading-order local lens properties that every valid lens model must share. The comparison machinery has three legs: parametric potentials (softened power-law and boxy power-law potentials with external shear) fitted by a parametric modelling code; a free-form pixel mass reconstruction that averages many regularised solutions; and a model-free check based on the relative polar angles of the quads against the Fundamental Surface of Quads. The argument works by showing that the three legs agree on image scale while only the free-form and local descriptions capture the subcomponent scale.","core_discovery":"B0128+437 is a quadruple-image quasar lens whose three images resolve into three VLBI subcomponents each, while the fourth image is scatter-broadened. The paper's central claim is that no smooth, elliptically symmetric mass model reproduces all subcomponent positions at sub-milliarcsecond precision, because the underlying mass distribution is asymmetric. Using a model-independent mapping of subcomponents, the paper obtains mass-sheet-invariant ratios $f_{ij}=(1-\\kappa_i)/(1-\\kappa_j)$ and reduced shears $g_i=\\gamma_i/(1-\\kappa_i)$ at image and subcomponent scales. These agree within 1-$\\sigma$ with free-form pixel reconstructions at image scale, but only about 40 percent of the subcomponent-scale properties overlap in confidence bounds, so mass-density gradients on milliarcsecond scales cannot be excluded. The conclusion is that elliptical symmetry is too simplistic for this lenticular or late-type galaxy and that high-resolution observations demand flexible free-form models.","pith_inferences":["If this pattern holds for other VLBI-resolved quads, survey pipelines that fit smooth elliptical or power-law lens models will systematically miss or misattribute milliarcsecond-scale mass complexity; quads that deviate from the Fundamental Surface of Quads are cheap screening targets for such asymmetry.","The 40 percent overlap statistic is a coarse diagnostic; a direct next step is to compute the same local lens properties with more than three reference points per image once deeper VLBI imaging resolves additional knots, which should shrink the confidence bounds and either confirm or dissolve the gradient signal.","The paper's interpretation of large external shear as a placeholder for internal small-scale asymmetry is testable by comparing shear estimates at different radii in other resolved lens systems: if the shear angle and magnitude change with scale, smooth-model shear is absorbing substructure rather than measuring a distant perturbing galaxy."],"forward_implications":["Within 1-$\\sigma$, the model-independent image-scale local lens properties agree with the free-form reconstruction values, so the fast local mapping can stand in for expensive global reconstructions on image scales.","Only 40 percent of the subcomponent-scale local lens properties overlap in confidence bounds, so milliarcsecond-scale mass-density gradients remain plausible and smooth models cannot be validated at that scale.","Parametric models fit only when a single subcomponent per image is used; adding all subcomponents degrades the fit, marking the lens mass distribution as asymmetric beyond elliptical symmetry.","The model-free polar-angle test places B0128 at effective ellipticity and shear around 0.25, and such a large shear value likely absorbs non-elliptical complexity rather than indicating a dominant nearby perturbing galaxy.","Together the methods give a consistent reconstruction of B0128 at current precision including the subcomponent structure, without invoking dark matter substructure as the primary cause of the flux anomalies."],"supporting_citations":[{"why":"Supplies the discovery data and image positions for B0128 that anchor the analysis.","marker":"Phillips et al. (2000)"},{"why":"Supplies the VLBI subcomponent positions and flux ratios that drive all reconstructions.","marker":"Biggs et al. (2004)"},{"why":"Provides HST and VLBA observations showing subcomponents and dust reddening that motivate the alternative matchings.","marker":"Norbury (2002)"},{"why":"Gives the Kp-band astrometry and the lens and source redshifts used in the models.","marker":"Lagattuta et al. (2010)"},{"why":"Derives the model-independent local lens property formalism used here.","marker":"Wagner & Tessore (2018)"},{"why":"Established the blinded comparison protocol on a cluster-scale lens that this paper applies to B0128.","marker":"Wagner et al. (2018)"},{"why":"Provides the free-form pixel mass reconstruction method used for the global PixeLens models.","marker":"Saha & Williams (2004)"},{"why":"Provides the parametric Lensmodel fitting code used to obtain model-based local properties.","marker":"Keeton (2001)"},{"why":"Documents the Lensmodel software and potentials used for the parametric reconstructions.","marker":"Keeton (2004)"},{"why":"Defines the Fundamental Surface of Quads used in the relative-polar-angle analysis.","marker":"Woldesenbet & Williams (2012)"}],"fun_headline_variants":["B0128 lens: elliptical models can't match its subcomponents","Asymmetric lens B0128 defies smooth elliptical fits","Milliarcsecond substructure exposes B0128's mass asymmetry","Beyond ellipses: free-form maps needed for lens B0128","B0128 quasar lens: smooth models fail on sub-kpc scales"],"cache_read_input_tokens":26112,"weakest_assumption_plain":"The load-bearing assumption is that the subcomponents labelled 1, 2, and 3 in images A, C, D, and B correspond to the same source components; if dust changes their brightness ordering, a different physical matching could be correct and every derived local lens property would shift.","fun_headline_variants_meta":{"raw":{"variants":["B0128 lens: elliptical models can't match its subcomponents","Asymmetric lens B0128 defies smooth elliptical fits","Milliarcsecond substructure exposes B0128's mass asymmetry","Beyond ellipses: free-form maps needed for lens B0128","B0128 quasar lens: smooth models fail on sub-kpc scales"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00104,"raw_usage":{"total_tokens":4419,"prompt_tokens":1035,"completion_tokens":3384,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":651,"completion_tokens_details":{"reasoning_tokens":3293}},"tokens_in":651,"tokens_out":3384,"duration_ms":23030,"temperature":1.0,"reasoning_tokens":3293,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T05:20:36.189213+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A smooth, elliptically symmetric lens model (or a smooth power-law potential with external shear) that reproduces all 12 subcomponent positions to within about 0.1 milliarcseconds would falsify the paper's central claim; alternatively, multi-band observations that establish a different physical matching of the subcomponents would shift the derived convergence ratios and reduced shears, as the paper's own comparison of alternate matchings shows.","supporting_citations":[{"cited_title":"M., Norbury, M","cited_arxiv_id":null,"evidence_quote":"Supplies the discovery data and image positions for B0128 that anchor the analysis."},{"cited_title":"D., Browne, I","cited_arxiv_id":null,"evidence_quote":"Supplies the VLBI subcomponent positions and flux ratios that drive all reconstructions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides HST and VLBA observations showing subcomponents and dust reddening that motivate the alternative matchings."},{"cited_title":"J., Auger, M","cited_arxiv_id":null,"evidence_quote":"Gives the Kp-band astrometry and the lens and source redshifts used in the models."},{"cited_title":"& Williams, L","cited_arxiv_id":null,"evidence_quote":"Provides the free-form pixel mass reconstruction method used for the global PixeLens models."},{"cited_title":"2004, gravlens 1.06 Software for Gravitational Lensing, 9th edn., uRL: http://www.physics.rutgers.edu/ keeton/gravlens/manual.pdf","cited_arxiv_id":null,"evidence_quote":"Documents the Lensmodel software and potentials used for the parametric reconstructions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the Fundamental Surface of Quads used in the relative-polar-angle analysis."}],"review_version":1}