{"id":"020f2ef7-f471-4dc3-8f11-02d6ef585ae5","arxiv_id":"2505.03549","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Ultraviolet color excesses for about 12 million stars are derived from GALEX and UVOT photometry combined with Gaia-based stellar parameters, and the resulting R_V map conflicts with infrared-optical R_V maps.","lead":"This paper derives ultraviolet color excesses for about 12 million stars, roughly ten times more than before, by combining Gaia-based stellar parameters with GALEX and UVOT photometry, and builds ultraviolet extinction maps covering up to two-thirds of the sky. It also reports that extinction-shape values derived from the ultraviolet disagree with infrared-optical values, suggesting the standard single-parameter dust extinction law is incomplete.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Section 3.1 blue-edge zero point is shared by every color excess and R_V map; a 0.33 mag FUV calibration residual is large enough to mimic the Section 5.2 anti-correlation, so the single-parameter-law conclusion is not yet supported.","rationale":"The reader identified the Section 3.1 blue-edge zero point as the weakest assumption, and I agree: it is the single link through which every color excess, extinction map, CER, and R_V value is derived. The internal consistency checks (agreement with S21, G19, and cross-survey dispersions) are genuinely useful, but they are not absolute calibrations because S21 shares the same blue-edge construction. The Section 3.3 FUV residual of 0.33 mag versus PARSEC is the most concrete quantitative red flag: if the zero-point offsets are real, they shift all E_UV values and can change the slope of E_UV versus EGBP,GRP, especially at the low EGBP,GRP values selected for the R_V map. That makes the central scientific claim in Section 5.2, that UV-derived R_V shows 'opposite trends' to IR/optical, currently under-supported. This does not invalidate the data-product contribution: the catalog is large, internally consistent, and likely useful once the zero-point question is settled. The reader's CONDITIONAL verdict remains appropriate.","tokens_in":14348,"tokens_out":4398,"duration_ms":48861,"concrete_test":"Recompute the Section 3.1-5.2 pipeline after applying the PARSEC-anchored zero-point residuals from Section 3.3 to the FUV/NUV/UVOT intrinsic colors, then re-derive the Figure 18 R_V correlation; if the anti-correlation disappears, the single-parameter-law claim is not supported. As a cheaper check, exclude stars with activity indicators and recompute the intrinsic colors to see whether E_UV shifts by more than the quoted 0.21-0.30 mag uncertainties.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's catalog and the Section 5.2 R_V comparison rest on the Section 3.1 zero-point assumption: the bluest 5% of stars in each Teff/[Fe/H]/log g bin, selected by optical GBP-GRP after 3-sigma clipping, are treated as zero/low-extinction, and their median ultraviolet color is then adopted as the intrinsic UV color for that bin. This single shared zero point enters every ENUV,GBP, EFUV,GBP, Euvw1,GBP, Euvm2,GBP, Euvw2,GBP and every ratio E_UV/EGBP,GRP used in the R_V fit. If a bin contains no genuinely dust-free star, or if the optically bluest stars are systematically UV-bright (chromospheric activity, binarity, variability) or UV-red (residual dust, effective-wavelength shift), all derived excesses inherit a bin-dependent offset. The paper's own Section 3.3 validation exposes this: a 0.33 mag median offset between the FUV intrinsic colors and PARSEC, plus large cool-star residuals in all UV bands. Because S21 uses the same blue-edge method, agreement with S21 does not validate the absolute zero point. For a star with EGBP,GRP near the 0.1 mag threshold used for the R_V map, a 0.1-0.3 mag UV zero-point error is of the same order as the signal and can invert the fitted UV/optical ratio, plausibly producing the anti-correlation in Figure 18. The headline physical conclusion is therefore not yet distinguishable from a calibration artifact in the blue-edge zero point.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript presents a large ultraviolet color-excess catalog for the GALEX NUV/FUV and Swift/UVOT uvw1, uvm2, and uvw2 bands. Using stellar parameters from LAMOST, GALAH, APOGEE, and Zhang et al. (2023) XP-spectra catalog, the authors apply the blue-edge method to define intrinsic UV colors as functions of Teff, [Fe/H], and log g, then derive color excesses for 11.6 million NUV, 65,531 FUV, and 336,633/137,739/253,271 UVOT sources. They construct HEALPix extinction maps, compare them with Green et al. (2019) and Sun et al. (2021a), and finally derive an R_V map from UV/optical color-excess ratios. The resulting R_V distribution shows noticeable differences and even anti-correlations with infrared/optical R_V maps (ZYC23, ZG25), leading to the conclusion that a single-parameter R_V extinction law is insufficient across IR/optical/UV.","tokens_in":14564,"tokens_out":4771,"duration_ms":48038,"significance":"If the catalog and R_V conclusion hold, this is a substantial community resource: the NUV color-excess sample is nearly an order of magnitude larger than Sun et al. (2021a), the UV extinction maps cover large sky fractions, and the cross-survey comparisons (Figures 7 and 9) and optical comparison with Green et al. (2019) (Figure 8) indicate good internal consistency and precision. The machine-readable catalog and maps are likely to be useful for many extinction-law and dust studies. However, the headline physical conclusion about the failure of a single-parameter R_V law rests on the blue-edge zero point of the intrinsic UV colors, which is a shared assumption for every derived color excess and R_V value. The manuscript's own validation shows a 0.33 mag median FUV offset against PARSEC and systematic cool-star residuals in all UV bands, so the R_V anti-correlation is not yet demonstrated to be physical rather than a calibration artifact.","major_comments":[{"comment":"The blue-edge zero point is the single shared anchor for all color excesses and for the R_V map in Section 5.2. Because Sun et al. (2021a) uses the same method, the agreement with S21 in Section 4.3 (median differences ~0.02 mag) does not validate the absolute zero point. Section 3.3 reports a 0.33 mag median offset between the derived FUV intrinsic colors and PARSEC, and Figures 4 and 5 show large cool-star residuals in all UV bands. At the E(B-V) > 0.1 mag threshold used to construct the R_V map, a 0.1-0.3 mag zero-point error is of the same order as the signal and can invert the fitted UV/optical color-excess ratio, plausibly producing the anti-correlation in Figure 18. Please demonstrate that the R_V anti-correlation survives plausible zero-point corrections, for example by recalibrating the intrinsic colors to PARSEC or to a sample with independently confirmed low extinction, and by propagating the PARSEC residual dispersions into the R_V fits.","section":"Section 3.1 and Section 3.3"},{"comment":"The anti-correlation between the UV-based R_V and the ZYC23/ZG25 R_V maps is the sole evidence for the conclusion that a single-parameter R_V extinction law is insufficient. The manuscript does not provide a statistical significance test of this anti-correlation, nor a null test. Please add, e.g., a Spearman rank correlation coefficient with uncertainty, and a perturbation test in which a constant or Teff-dependent offset is added to the UV color excess and the R_V fit is repeated. If the anti-correlation is robust to such perturbations, the conclusion is supported; if it disappears, it is likely a calibration artifact. In addition, report the E(B-V) and distance distributions of the cross-matched samples used for Figure 17 to rule out selection effects from the photometric-depth differences between UV and IR/optical surveys.","section":"Section 5.2 and Figure 18"},{"comment":"The PARSEC comparison shows a systematic Teff-dependent trend in all UV bands, with PARSEC being noticeably redder for cooler stars (larger C0) and with the largest discrepancy in the FUV band. Because the R_V analysis in Section 5.2 is binned by Teff and uses color-excess ratios, a Teff-dependent zero-point error will map into a spatial pattern of R_V whenever the stellar population or the extinction distribution varies spatially. The manuscript reports dispersions but does not quantify how these residuals propagate into the derived CERs and R_V maps. Please either apply a Teff-dependent correction to the intrinsic colors based on the PARSEC comparison, or model the propagation of these residuals into the R_V values and show that the Section 5.2 anti-correlation remains significant after this propagation.","section":"Section 3.3 and Figures 4-5"}],"minor_comments":[{"comment":"The label 'wum2' in the middle-right panel should be 'uvm2'.","section":"Figure 2 caption"},{"comment":"The caption refers to 'ZG24' but the paper cites Zhang & Green (2025) as ZG25; please correct the citation to avoid confusion.","section":"Figure 17 caption"},{"comment":"The term 'CECE diagram' is undefined; please write 'color excess–color excess diagram' on first use.","section":"Section 5.1"},{"comment":"Please clarify why the final selected combined catalogs have 93,230,392 sources in the GBP/GRP band (Section 2.3) while Section 4.1 reports 92,142,820 values of EGBP,GRP; the difference is not explained.","section":"Section 2.3 and Section 4.1"},{"comment":"The acknowledgements thank the referee, which is unconventional in a submitted manuscript; this sentence should be removed.","section":"Acknowledgements"},{"comment":"The phrase 'tenfold increase from previous results' refers to Sun et al. (2021a); please name the baseline explicitly in the abstract for clarity.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The catalog and maps are likely to be a valuable community resource, and the internal validation is largely convincing. The main risk is the interpretation of the UV-derived R_V map: the anti-correlation with IR/optical R_V could easily be produced by the blue-edge zero point, especially given the 0.33 mag FUV offset against PARSEC. I suggest directing the authors to treat the R_V conclusion as provisional unless they can show robustness to zero-point perturbations; otherwise, the paper would be stronger if it presented the color-excess catalog as the primary contribution and moved the R_V comparison to a 'preliminary evidence' status. Also note the unconventional referee-thanking line in the acknowledgements."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe paper (Yang et al., arXiv:2505.03549) is a data-product paper that takes Sun et al. (2021a)'s blue-edge plus random forest pipeline and applies it to the 220-million-star parameter set from Zhang et al. (2023), plus spectroscopic surveys, yielding 11.6 million NUV color excesses (ten times S21) and the first large samples in UVOT uvw1/uvm2/uvw2. That is a real expansion, and the internal validation is decent: cross-survey dispersions of 0.09–0.27 mag, median agreement with S21 to 0.02 mag, and an optical map that tracks Green et al. (2019) at 0.07 mag scatter. The authors also extend the method to giants, which is non-trivial. If the catalog is released, it will be a useful resource for extinction studies and high-latitude dust work.\n\nThe soft spots are mostly about the interpretation, not the catalog assembly. The load-bearing assumption is the blue-edge zero point (Section 3.1): the bluest 5% in each Teff/[Fe/H]/log g bin are treated as unreddened, and their median UV color becomes the intrinsic color. Every color excess and every RV in Section 5 inherits that zero point. The paper's own validation shows a 0.33 mag median offset between FUV intrinsic colors and PARSEC, plus notable cool-star residuals in all UV bands. Agreement with S21 does not rescue the absolute zero point, because S21 uses the same blue-edge assumption. That matters for the headline claim: the RV anti-correlation with IR/optical maps (Figure 18) could plausibly be a bin-dependent zero-point artifact rather than evidence that a single-parameter extinction law fails. No error bars or significance tests are given for the RV maps, and the relevant analysis is deferred. The authors themselves note the RV comparison is preliminary; the abstract and conclusions overstate it.\n\nA few smaller issues: there is an unexplained count discrepancy between the final selected combined catalogs in Section 2.3 (e.g., 11,915,943 NUV sources) and the derived color-excess counts in Section 4.1 (11,624,802), and the data/code are not released. Both should be cleaned up before publication.\n\nOverall: the catalog work deserves a serious referee and, probably, publication after release of the data products. The scientific claim about RV and the UV extinction law needs either much more careful treatment of the zero-point systematics or to be moved to a later paper. A working referee would do the field a service by forcing that distinction.","headline":"A genuinely useful, much larger UV color-excess catalog, but the headline claim that a single-parameter RV law fails is not yet supported by the calibration.","tokens_in":15309,"tokens_out":4143,"would_cite":true,"duration_ms":35187,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The authors derive ultraviolet color excesses for 11.6 million stars and find that the extinction parameter R_V from the ultraviolet disagrees with infrared-optical maps, evidence against a one-parameter dust law.","keywords":["ultraviolet extinction","color excess","interstellar dust","extinction law","GALEX","UVOT","Gaia","blue-edge method"],"falsifier":"Compare the calibration's zero point against an independent dust-free sample: take stars in high-latitude sightlines with negligible dust column (e.g., confirmed by very low far-infrared dust emission) and check whether their derived ultraviolet color excesses scatter around zero; a systematic positive offset would show the bluest-5% samples are still reddened. A sharper test targets the $R_V$ claim: toward a set of well-studied clouds where the extinction curve has been measured directly from paired spectra of reddened and unreddened stars, the paper's ultraviolet color excesses predict an $R_V$ that should either agree or disagree with the infrared-optical value; the FUV band, whose intrinsic colors show a 0.33 mag median offset from stellar models, is the most sensitive place to look for this discrepancy.","tokens_in":13983,"feed_emoji":"🌌","tokens_out":17441,"duration_ms":142799,"temperature":0.7,"pith_summary":"This paper aims to establish that ultraviolet color excesses can be measured for tens of millions of stars by combining Gaia-based stellar parameters with GALEX and UVOT photometry, and that these ultraviolet data expose a problem with the standard one-parameter extinction law. Working from the blue-edge assumption that the bluest stars in narrow bins of temperature, metallicity, and gravity are nearly dust-free, the authors train a random forest to predict intrinsic ultraviolet colors, then subtract those predicted colors from observed ones to obtain color excesses for 11,624,802 stars in the GALEX near-UV band (about ten times more than the previous catalog), 65,531 in the far-UV band, and 336,633, 137,739, and 253,271 stars in the UVOT uvw1, uvm2, and uvw2 bands. These excesses yield ultraviolet extinction maps whose near-UV version covers roughly two-thirds of the sky at 0.4-degree resolution and whose structure matches the optical and infrared dust maps where they overlap. The pointed result is that $R_V$ values derived from the ultraviolet color excesses disagree with $R_V$ maps built from infrared and optical data, with some regions showing opposite trends; the paper takes this as evidence that a single $R_V$ parameter is not enough to describe extinction from the infrared to the ultraviolet.","feed_headline":"11.6 million UV color excesses challenge the one-parameter dust law","feed_subtitle":"The near-UV extinction map spans two-thirds of the sky, and its R_V values often oppose infrared-optical maps.","key_machinery":"The load-bearing machinery is the blue-edge method as a zero-extinction selector, coupled to a random forest regression as an interpolator. For every narrow bin of stellar parameters (100 K in $T_\\mathrm{eff}$, 0.1 dex in [Fe/H], 0.5 dex in $\\log g$), the bluest 5% of stars after 3-$\\sigma$ clipping in the optical bands are declared zero- or low-extinction sources; the median ultraviolet color of those stars in each bin is the representative intrinsic color, and the random forest smooths these representatives into a continuous function of the three parameters. Color excess is then the observed ultraviolet color minus the predicted intrinsic color, and the excesses are gridded into HEALPix extinction maps. The same color-excess ratios feed a forward model that, after correcting for effective-wavelength shifts, matches observed ratios against a standard extinction curve (the F99 law) on a grid of $R_V$ values to assign each star an $R_V$.","core_discovery":"The central discovery, stated on the paper's own terms, is a calibration that makes ultraviolet extinction measurable at scale: an empirical mapping from stellar parameters ($T_\\mathrm{eff}$, [Fe/H], $\\log g$) to intrinsic ultraviolet color indices, built with the blue-edge method and random forest regression, that converts the overlap between Gaia XP stellar parameters and GALEX/UVOT photometry into a tenfold-larger ultraviolet color-excess catalog. Subtracting predicted intrinsic colors from observed colors yields $E_{\\mathrm{NUV},G_\\mathrm{BP}}$ for 11,624,802 stars, $E_{\\mathrm{FUV},G_\\mathrm{BP}}$ for 65,531 stars, and $E_{\\mathrm{uvw1},G_\\mathrm{BP}}$, $E_{\\mathrm{uvm2},G_\\mathrm{BP}}$, $E_{\\mathrm{uvw2},G_\\mathrm{BP}}$ for 336,633, 137,739, and 253,271 stars, with typical uncertainties of 0.21, 0.30, 0.19, 0.24, and 0.21 mag, and from these it builds HEALPix extinction maps whose near-UV version covers about two-thirds of the sky. The decisive result is the $R_V$ comparison: fitting each star's ultraviolet-to-optical color-excess ratios against a standard extinction curve on a grid of $R_V$ values gives a sky map of $R_V$ that differs noticeably from, and in places anti-correlates with, the $R_V$ maps built from infrared and optical data in previous work, while a control $R_V$ map derived from infrared and optical bands with the same pipeline reproduces those maps. The paper reads that split as evidence that a single-parameter $R_V$ extinction law cannot simultaneously describe the infrared, optical, and ultraviolet extinction behavior.","pith_inferences":["Beyond the paper: the anti-correlation between ultraviolet-based and infrared-optical $R_V$ admits a second reading the paper does not fully exclude — that the ultraviolet intrinsic-color zero point drifts with stellar parameters in ways the blue-edge method cannot see; re-deriving the $R_V$ map with the random forest trained only on stars in nearly dust-free high-latitude fields would separate du","Beyond the paper: because FUV shows the largest model offsets (0.33 mag median versus PARSEC) and is the most extinction-sensitive band, the FUV link in the $R_V$ chain is the most fragile; a small systematic there would propagate directly into the reported disagreement, so the FUV color-excess subsample is the natural first target for external validation.","Beyond the paper: the same pipeline applied to future ultraviolet surveys could produce all-sky ultraviolet extinction maps, and the comparison of the UV-based $R_V$ map with spectroscopically measured extinction curves toward individual clouds would give a direct, model-independent test of whether the dust law really needs more than one parameter."],"forward_implications":["The near-UV color-excess catalog (~11.6 million stars) and the two-thirds-sky NUV extinction map become a reference resource for correcting ultraviolet photometry and for tracing dust at high Galactic latitudes, where ultraviolet extinction is several times more sensitive than optical.","The UVOT-band color excesses ($uvw1$, $uvm2$, $uvw2$) extend extinction measurements across the 2175 Å feature, so the catalog can constrain variations in the ultraviolet bump, not just the far-ultraviolet rise.","If the $R_V$ disagreement is real, then converting infrared-optical $R_V$ maps into ultraviolet extinction corrections will systematically misestimate ultraviolet reddening, and multi-parameter extinction laws will be required for any analysis spanning UV-to-IR observations.","Because the intrinsic-color calibration now includes giants and stars down to 4500 K in the near-UV, the same pipeline can be reapplied as new photometric surveys appear, with the per-bin blue-edge threshold acting as the main dial between sample size and zero-point fidelity."],"supporting_citations":[{"why":"The prior NUV color-excess catalog and improved blue-edge implementation that this work extends about tenfold and validates against in the NUV and FUV bands.","marker":"Sun et al. (2021a)"},{"why":"Supplies stellar parameters for over 220 million Gaia XP stars, the data source that lets the method scale from about 1 million to about 11.6 million NUV color excesses.","marker":"Zhang et al. (2023)"},{"why":"The extinction curve used in the forward model that assigns each star an $R_V$ from its ultraviolet-to-optical color-excess ratios.","marker":"Fitzpatrick (1999)"},{"why":"The infrared-optical $R_V$ map and forward-modeling approach that the paper's $R_V$ derivation follows and compares against.","marker":"Zhang et al. (2023b)"},{"why":"A second infrared-optical extinction-variation map used as comparison; the UV-based $R_V$ disagrees with both this and the other map.","marker":"Zhang & Green (2025)"},{"why":"Supplies the $R_{G_\\mathrm{BP},G_\\mathrm{RP}}$ conversion and color-excess ratios used in the $R_V$ fitting and in the CER comparison.","marker":"Zhang & Yuan (2023a)"},{"why":"The source of the blue-edge method's binned-bluest-stars logic that anchors the intrinsic-color calibration.","marker":"Jian et al. (2017)"},{"why":"Provides the independent optical/reddening map used to validate the paper's $E_{G_\\mathrm{BP},G_\\mathrm{RP}}$ and the NUV extinction map's large-scale structure.","marker":"Green et al. (2019)"}],"fun_headline_variants":["UV extinction map: 11.6M stars, two-thirds sky, R_V mismatch","Tenfold UV excess data reveals R_V inconsistency across bands","Random forest + blue edge: UV reddening for 11M+ stars","UV extinction law challenged: R_V from UV differs from optical","Sky-wide UV reddening maps hint at multi-parameter dust law"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything downstream rests on the blue-edge premise: within every narrow box of temperature, metallicity, and gravity, the bluest 5% of stars (after optical 3-$\\sigma$ clipping) are genuinely almost dust-free, so their median color is the true intrinsic color; if any box's bluest stars are still reddened, or are biased by chromospheric activity or ultraviolet variability, the intrinsic colors shift and every color excess, map, and $R_V$ value inherits a systematic offset.","fun_headline_variants_meta":{"raw":{"variants":["UV extinction map: 11.6M stars, two-thirds sky, R_V mismatch","Tenfold UV excess data reveals R_V inconsistency across bands","Random forest + blue edge: UV reddening for 11M+ stars","UV extinction law challenged: R_V from UV differs from optical","Sky-wide UV reddening maps hint at multi-parameter dust law"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000365,"raw_usage":{"total_tokens":2140,"prompt_tokens":1294,"completion_tokens":846,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":910,"completion_tokens_details":{"reasoning_tokens":749}},"tokens_in":910,"tokens_out":846,"duration_ms":7405,"temperature":1.0,"reasoning_tokens":749,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:50:15.252748+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the calibration's zero point against an independent dust-free sample: take stars in high-latitude sightlines with negligible dust column (e.g., confirmed by very low far-infrared dust emission) and check whether their derived ultraviolet color excesses scatter around zero; a systematic positive offset would show the bluest-5% samples are still reddened. A sharper test targets the $R_V$ claim: toward a set of well-studied clouds where the extinction curve has been measured directly from paired spectra of reddened and unreddened stars, the paper's ultraviolet color excesses predict an $R_V$ that should either agree or disagree with the infrared-optical value; the FUV band, whose intrinsic colors show a 0.33 mag median offset from stellar models, is the most sensitive place to look for this discrepancy.","supporting_citations":[],"review_version":1}