{"id":"8f589a75-5a83-4038-91f8-54e054dede4f","arxiv_id":"2505.01710","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A new Python tool and template set show that the 40 km/s SDSS versus Hectospec redshift offset is caused by template velocity offsets, and calibrating templates to synthetic rest-frame spectra removes the offset.","lead":"RVSNUpy is a Python package that measures galaxy redshifts by comparing observed spectra to rest-frame templates with inverse-variance weighted cross-correlation. The paper shows that a known 40 km/s offset between SDSS and MMT/Hectospec redshifts comes mainly from differences in the template spectra, not from the spectra themselves, and builds 'universal' templates to remove it.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The universal-template rest-frame calibration is anchored to FSPS and then validated against FSPS-based SDSS catalog redshifts, leaving the absolute zero-point untested; the relative template-offset explanation is otherwise well supported.","rationale":"The reader's weakest_assumption correctly identifies the FSPS anchor as the least secure part of the paper. I agree that the universal-template calibration is validated only against SDSS catalog redshifts that themselves used FSPS-based templates, making the absolute rest-frame calibration circular. However, this concern is not the load-bearing condition for the reader's strongest_claim, which is that template offsets cause the inter-survey offset: that claim relies on relative template offsets, and any common FSPS zero-point would cancel in the SDSS-versus-Hectospec comparison. The evidence in Figures 4, 5, and 9 is strong for the relative statement. The FSPS-anchor concern becomes load-bearing only for the broader claims in the abstract and Section 6.3 that the universal templates are 'carefully calibrated to the rest frame' and that RVSNUpy provides accurate absolute redshifts. Because the paper's headline contribution includes this universal calibration, the conditional verdict is appropriate. I would not change the reader's verdict, but I want to sharpen the distinction: the template-offset explanation is well supported, while the absolute zero-point of the universal templates remains an untested assumption.","tokens_in":22628,"tokens_out":12097,"duration_ms":126504,"concrete_test":"Use RVSNUpy with the universal templates to measure redshifts for a few hundred galaxies that also have independent systemic velocities from H I 21-cm (e.g., ALFALFA) or from high-resolution Ca ii triplet spectroscopy with absolute wavelength calibration. Compute the median of c(z_RVSNUpy - z_independent)/(1+z_independent) with a bootstrap uncertainty. If the median deviates from zero by more than ~10 km/s, the FSPS rest-frame anchor is biased and the universal templates are not absolutely calibrated; if it is consistent with zero, the concern is retired.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's elimination argument—observed SDSS and Hectospec spectra agree (Fig. 4), the measurement tool does not change the result (Fig. 5), and swapping templates moves the offset in the expected direction (Fig. 9)—is convincing that template zero-point differences, not the observed spectra or the algorithm, drive the ~40 km/s SDSS-versus-Hectospec offset. The weaker link is the follow-on claim that the 'universal template set' is carefully calibrated to the rest frame (Sec 6.3). The anchor is the FSPS synthetic SSP spectra, whose own rest-frame zero-point is assumed but never independently tested. The validation is also partly circular: zSDSScat was produced by Redmonster using FSPS-based galaxy templates, so the agreement of RVSNUpy universal-template redshifts with zSDSScat at -2.8 km/s is largely a self-consistency check that the universal templates reproduce the FSPS/Redmonster zero-point. If FSPS spectra carry a velocity zero-point offset, the universal templates inherit it, and the 'carefully calibrated to the rest frame' claim, together with the absolute accuracy of RVSNUpy redshifts for future surveys, would be off by that amount. This does not invalidate the relative claim that using a common template set removes the inter-survey offset; it leaves the absolute calibration unproven.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents RVSNUpy, an open-source Python package that measures spectroscopic redshifts by inverse-variance-weighted cross-correlation in real space, following Kelson et al. (2003). The mathematical framework, implementation steps, template requirements, and quality metrics (χ2_eff and r-value) are described in detail. The package is tested on FSPS synthetic SSP spectra and on roughly 6000 HectoMAP galaxies with both SDSS and MMT/Hectospec spectra. Using four observed-spectrum/template combinations, the authors show that the approximately 40 km/s SDSS-versus-Hectospec offset follows the template set rather than the observed spectra or the measurement tool. They then construct a 'universal template' set by shifting three empirical SDSS templates in wavelength to match FSPS SSP spectra, and show that RVSNUpy with these templates reproduces SDSS catalog redshifts at -2.8 km/s for SDSS spectra and -3.1 km/s for Hectospec spectra. The paper concludes that template zero-point offsets are the source of the inter-survey offset and that RVSNUpy is suitable for current and future large spectroscopic surveys.","tokens_in":22811,"tokens_out":5649,"duration_ms":58690,"significance":"If the central claim holds, the paper makes a concrete and useful contribution: it identifies the origin of a known redshift offset between SDSS and MMT/Hectospec measurements, provides an open-source tool with a clean four-way elimination experiment in Figure 9, and demonstrates that a common template set removes the largest part of the inter-survey offset. The four-combination test is a good falsifiable design, and the relative conclusion that template zero-point, not observed spectra or tool choice, sets the offset is well supported. The weaker point is the absolute rest-frame calibration: the universal templates are anchored to FSPS synthetic spectra and validated against a catalog measured with FSPS-based templates, so the absolute zero-point is not independently established. This limits the strength of the 'carefully calibrated to the rest frame' and future-survey accuracy claims, but it does not undermine the relative template-offset result.","major_comments":[{"comment":"The universal-template rest-frame calibration is anchored by shifting the SDSS templates to match FSPS SSP spectra, and the validation target zSDSScat is itself produced by Redmonster using FSPS-based galaxy templates. The -2.8 km/s agreement between z_SDSS,utemp and zSDSScat is therefore to a substantial degree a self-consistency check: if the FSPS spectra carry a velocity zero-point offset, the universal templates inherit it and the absolute calibration is untested. I request either an independent rest-frame anchor (for example, high signal-to-noise stellar spectra with known radial velocities from Gaia, or telluric/asteroid features) or a clearly qualified claim such as 'calibrated to the FSPS rest-frame convention' in the abstract and Section 6.3. This does not affect the relative conclusion from Figure 9 that template choice sets the inter-survey offset, but it does affect the 'carefully calibrated to the rest frame' and future-survey accuracy statements.","section":"Section 6.3, Figures 7 and 10"},{"comment":"The Hectospec universal-template validation shows a median offset of -3.1 km/s but a 1σ scatter of 38.6 km/s and 307/5504 (5.6%) outliers, while the SDSS comparison shows 17.3 km/s scatter and 1.1% outliers. The outlier census attributes 58% of the Hectospec outliers to 'intrinsic differences between the spectra', which is not a template effect. The paper does not state an a priori quantitative success criterion for the claim that the universal templates yield 'homogeneous redshifts' (for example, a required median offset, scatter, or outlier fraction relative to the quoted uncertainties). I ask for a stated success criterion and an error budget that separates template zero-point contributions, spectral-mismatch contributions, and measurement noise, so that the central performance claim is falsifiable.","section":"Section 6.3, Figure 10"},{"comment":"The procedure for constructing the universal templates is underspecified. The text states that 'we shifted three SDSS templates along the wavelength direction to match the FSPS templates,' but it does not state whether each template was shifted by its own measured offset from Figure 7 or by a common value, how the offsets in Figure 7 were estimated (for example, by Gaussian fitting to a cross-correlation peak), and what uncertainty in the adopted shift is propagated into the final redshift error budget. Because the entire absolute calibration rests on these shifts, the reproducibility and uncertainty of the construction should be documented explicitly.","section":"Section 6.3, construction of universal templates"}],"minor_comments":[{"comment":"The propagated uncertainty expression appears to contain duplications: the (δz)_fit term is added both inside and outside the square root, and the partial-derivative terms for h_c.c. and σ_c.c. may be missing cross terms. Please verify the derivation or state the simplifying assumptions used to obtain Eqs. (10) and (11).","section":"Section 2.2, Eqs. (10)-(11)"},{"comment":"The B-spline knot span of 100 Å and the 75% masked-node merging rule are presented without a sensitivity test. Since these are user-set parameters that affect continuum removal, a brief dependence check (for example, knot spans of 50, 100, and 200 Å) would clarify their impact on redshift precision.","section":"Section 3.2"},{"comment":"Please clarify the wavelength system of the FSPS synthetic templates used in the Section 6.2 comparisons and whether they are converted from vacuum to air before computing the offsets in Figure 7. The air-vacuum difference is about 85 km/s near 6000 Å, and an unstated conversion could contaminate the derived template offsets.","section":"Sections 4 and 6.2"},{"comment":"There are several typographical errors, including 'R VSNUpy' in the abstract, 'RSNUpy' in Section 3.4, 'SSDS' and 'teampltes' in Section 6.2, 'habtemp90' vs. 'habetmp90' in Figure 7, and 'skylines' in Figure 12.","section":"Throughout"},{"comment":"The statement that χ2_eff = 4 'roughly corresponds to a 2σ difference' is not generally true without specifying the number of degrees of freedom; please either justify this calibration or rephrase it as a heuristic threshold.","section":"Section 2.3"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe thing to know: this paper convincingly identifies the culprit behind the ~40 km/s SDSS vs MMT/Hectospec redshift offset. The four-way test in Fig 9 is a clean elimination: swap the templates and the offset moves with them; swap the observed spectra or the tool and it does not. That is a real, useful result for anyone combining redshifts from different surveys.\n\nThe package itself is also a solid piece of work. RVSNUpy is a sensible Python implementation of inverse-variance weighted cross-correlation, with a public repo and Zenodo DOI. The synthetic tests are reasonable, and the code is fast enough for survey use.\n\nThe soft spot is the claim that the universal templates are 'carefully calibrated to the rest frame.' The calibration is done by shifting empirical SDSS templates to match FSPS synthetic spectra. Then the validation compares against the SDSS catalog redshifts, which were themselves produced by Redmonster using FSPS-based templates. That is largely a self-consistency check. If FSPS spectra carry any velocity zero-point, the universal templates inherit it, and the absolute accuracy claim does not follow. This does not hurt the relative offset diagnosis, but it does mean the paper is overstating the absolute calibration.\n\nTwo smaller issues. The outlier fraction for Hectospec is 5.6%, and the scatter is larger than the 'homogeneous' language in the abstract suggests. And the code release lacks a commit hash and figure-reproduction scripts, which makes the published results harder to verify exactly.\n\nBottom line: the paper deserves a serious referee. The template-offset result is likely right and will be cited. The absolute calibration claim needs to be softened or supported with an independent rest-frame anchor. I'd send it out.","headline":"A convincing diagnosis that template offset, not the spectra or the tool, drives the SDSS/Hectospec redshift systematic, but the universal-template absolute calibration is anchored to FSPS and partly self-consistent with SDSS catalog redshifts, so the rest-frame claim is unproven.","tokens_in":23434,"tokens_out":1978,"would_cite":true,"duration_ms":19347,"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 ~40 km/s redshift offset between SDSS and MMT/Hectospec is caused by small velocity offsets in the template spectra, not by the galaxies or the measurement tool.","keywords":["spectroscopic redshifts","cross-correlation","inverse-variance weighting","template spectra calibration","HectoMAP","SDSS","Hectospec","RVSNUpy"],"falsifier":"Take a sample of galaxies with spectra from a third, independently calibrated spectrograph plus SDSS or Hectospec data, measure both with RVSNUpy and the same universal templates, and compare with that survey's pipeline redshifts; if a residual zero-point offset appears that tracks the third survey's reduction templates, the template-offset diagnosis is confirmed, but if the same templates still leave a spectrograph-dependent offset, the claim fails. A more direct check is to measure the absolute wavelength zero point of the FSPS templates with laboratory-calibrated arc spectra or precision stellar radial velocities; a nonzero offset would falsify the paper's rest-frame anchor.","tokens_in":22343,"feed_emoji":"🔭","tokens_out":10088,"duration_ms":94038,"temperature":0.7,"pith_summary":"The paper presents RVSNUpy, a Python package that measures spectroscopic redshifts by cross-correlating observed spectra with rest-frame template spectra, weighting each pixel by inverse variance so noisy features carry little weight. Its central empirical claim is that the roughly 40 km/s offset between SDSS and MMT/Hectospec redshifts for the same galaxies does not come from the observed spectra or from different measurement software; it comes from small radial-velocity offsets baked into the template spectra each survey uses. The evidence is a sample of ~6000 HectoMAP galaxies with spectra from both spectrographs: cross-correlating either survey's spectra with a common template set removes the offset, while using the other survey's templates reintroduces it. To make the fix portable, the authors build a universal template set by shifting three SDSS galaxy templates to the rest frame using FSPS synthetic spectra, and show that RVSNUpy with these templates reproduces SDSS catalog redshifts for both SDSS and Hectospec spectra within a few km/s. If the claim is right, a single calibrated template set lets any future survey produce redshifts that can be combined across instruments without an inter-survey zero-point correction.","feed_headline":"One calibrated template set erases the 40 km/s survey offset","feed_subtitle":"RVSNUpy cross-correlates spectra in real space; the offset lives in the templates, not the galaxies or the tool.","key_machinery":"The load-bearing object is the rest-frame template spectrum. RVSNUpy computes $C.C.(z') = \\sum_i M_i (\\delta G_i/G'_i)^{-2} (G_i/G'_i - 1)(T_i(z')/T'_i(z') - 1)$, an inverse-variance weighted cross-correlation in real space that downweights noisy pixels and yields a Gaussian peak whose mean gives the redshift and whose width, combined with the r-value reliability statistic, gives the uncertainty and a reliability flag. The diagnosis that carries the paper's main claim is the template calibration test: templates themselves are treated as spectra and cross-correlated against the FSPS synthetic rest-frame set, exposing velocity zero-point offsets of 10–70 km/s. The universal template set, three SDSS galaxy templates shifted along the wavelength axis to match FSPS templates, is the instrument that turns the diagnosis into a practical fix, because cross-correlating any input spectrum against that single set removes the spectrograph-dependent offset.","core_discovery":"RVSNUpy recovers redshifts by shifting rest-frame template spectra across a logarithmic wavelength grid and computing an inverse-variance weighted cross-correlation, with continua removed by B-spline fits and the peak located by Gaussian fitting. Tested on synthetic single stellar population spectra, it reproduces input redshifts within ~20 km/s, with small systematic offsets that trace template calibration. On the HectoMAP sample, RVSNUpy redshifts from SDSS spectra with SDSS templates agree with SDSS catalog values (10.9 ± 12.6 km/s), while Hectospec spectra with Hectospec templates are offset by −37.3 ± 30.7 km/s, matching the offset reported in earlier HectoMAP work. Direct cross-correlation of the SDSS and Hectospec spectra themselves shows no such offset (−3.2 ± 45.1 km/s), and using FSPS templates on SDSS spectra nearly reproduces the catalog, ruling out the observed spectra and the tool as the source. Cross-correlating templates against FSPS synthetic spectra shows the Hectospec absorption templates are redshifted by tens of km/s relative to the rest frame, while SDSS templates are off by 10–20 km/s; accordingly, a universal set built from SDSS templates shifted to match FSPS yields −2.8 ± 17.3 km/s for SDSS spectra and −3.1 ± 38.6 km/s for Hectospec spectra relative to the SDSS catalog. The paper concludes that template zero-point offsets are the primary source of the inter-survey redshift offset and that one calibrated template set yields homogeneous redshifts across spectrographs.","pith_inferences":["Beyond the paper: the same template-offset test can be applied to any pair of surveys; if the diagnosis is general, an inter-survey offset should reappear whenever overlapping spectra are reduced with differently calibrated template sets, and disappear when a common set is used.","The universal templates are anchored to FSPS synthetic spectra, so the calibration is only as absolute as FSPS's own velocity zero point; a future absolute wavelength calibration of the synthetic models would upgrade the relative agreement into an absolute rest-frame standard.","A testable extension: run RVSNUpy with universal templates on spectra from a third spectrograph, such as DESI or Subaru/PFS, and compare with that survey's pipeline redshifts; residual systematics would identify whether template zero-point offsets remain the dominant term at higher spectral resolution."],"forward_implications":["Using one rest-frame-calibrated template set removes the ~40 km/s systematic offset between SDSS and MMT/Hectospec redshifts, so surveys can combine redshifts without an inter-survey zero-point correction.","RVSNUpy with the universal templates reproduces SDSS catalog redshifts to within a few km/s for both SDSS and Hectospec spectra, with most outliers traced to low signal-to-noise, poor sky subtraction, or genuinely different spectra rather than method bias.","Because the package measures a spectrum in 0.2–0.4 seconds and is model-independent, it is suited to large spectroscopic surveys such as A-SPEC, DESI, 4MOST, and Subaru/PFS.","A rest-frame-calibrated template library becomes a community reference: any survey that adopts it gains redshifts directly comparable to SDSS's without re-running full spectral fitting.","Template velocity offsets of 10–70 km/s are detectable by cross-correlating template spectra against synthetic rest-frame spectra, providing a simple quality check before survey data release."],"supporting_citations":[{"why":"Supplies the inverse-variance weighted real-space cross-correlation method and uncertainty estimation that RVSNUpy is built on.","marker":"Kelson et al. 2003"},{"why":"Provides the theoretical cross-correlation framework, the Gaussian signal assumption, and the r-value used to judge redshift reliability.","marker":"Tonry & Davis 1979"},{"why":"Reports the ~40 km/s offset between Hectospec and SDSS redshifts that this paper sets out to explain.","marker":"Sohn et al. 2021"},{"why":"Defines the HectoMAP test sample of 6009 objects with both SDSS and Hectospec spectra and supplies the earlier RVSAO comparison offset.","marker":"Sohn et al. 2023"},{"why":"Provides the FSPS synthetic stellar-population spectra used as the rest-frame reference for calibrating the universal templates.","marker":"Conroy et al. 2009"},{"why":"Describes Redmonster, the full spectral fitting tool that produced the SDSS catalog redshifts used as the comparison baseline.","marker":"Hutchinson et al. 2016"}],"fun_headline_variants":["Calibrated templates erase 40 km/s offset","One template set wipes out 40 km/s survey gap","Offset traced to template zero-points, not galaxies","RVSNUpy fixes redshift offset with calibrated templates","Universal templates unify SDSS and Hectospec redshifts"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The universal template calibration assumes the FSPS synthetic spectra define the true rest frame; if the synthetic spectra carry a velocity zero-point offset of their own, the universal templates inherit it and the agreement with SDSS catalog redshifts becomes an internal consistency check rather than an absolute wavelength calibration.","fun_headline_variants_meta":{"raw":{"variants":["Calibrated templates erase 40 km/s offset","One template set wipes out 40 km/s survey gap","Offset traced to template zero-points, not galaxies","RVSNUpy fixes redshift offset with calibrated templates","Universal templates unify SDSS and Hectospec redshifts"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000296,"raw_usage":{"total_tokens":1793,"prompt_tokens":1093,"completion_tokens":700,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":709,"completion_tokens_details":{"reasoning_tokens":621}},"tokens_in":709,"tokens_out":700,"duration_ms":6953,"temperature":1.0,"reasoning_tokens":621,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:11:25.500101+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a sample of galaxies with spectra from a third, independently calibrated spectrograph plus SDSS or Hectospec data, measure both with RVSNUpy and the same universal templates, and compare with that survey's pipeline redshifts; if a residual zero-point offset appears that tracks the third survey's reduction templates, the template-offset diagnosis is confirmed, but if the same templates still leave a spectrograph-dependent offset, the claim fails. A more direct check is to measure the absolute wavelength zero point of the FSPS templates with laboratory-calibrated arc spectra or precision stellar radial velocities; a nonzero offset would falsify the paper's rest-frame anchor.","supporting_citations":[],"review_version":1}