{"id":"0375ed93-e93b-4afe-9a52-c8bd2fa1535a","arxiv_id":"2501.11212","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Including returning radiation in relxill does not significantly alter the best-fit parameters of three Galactic black holes.","lead":"X-rays that bounce back onto a black hole's disk were added to a standard spectral model. Fitting three well-known black holes with and without this effect gives nearly identical spin and corona height values.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Null result may be an artifact of relxill's returning-radiation implementation, which only changes the emissivity profile; the paper's own Section V says this approximation is least accurate for exactly the high-spin, low-corona sources studied.","rationale":"The manuscript is careful and transparent about its approximations, and the fits are standard. The reader's CONDITIONAL verdict is appropriate. The load-bearing concern is not an internal inconsistency but the limited physical scope of the test: relxill's returning-radiation flag modifies only the emissivity profile, leaving the reflected spectrum computed for a Comptonized incident continuum. Since the sources were selected to be in exactly the regime where Ref. [43] says this approximation is worst (high spin, low coronal height), the null result may reflect the incompleteness of the model rather than the unimportance of returning radiation. The paper explicitly acknowledges this possibility, so it is a stated limitation, not a hidden flaw. The proposed mock-observation test with the full Mirzaev et al. model would settle whether the approximation hides a real parameter bias. If biases appear, the broad claim should be narrowed; if not, the current result is robust. Boundary-pinned parameters (h=2 rg, a*=0.998) and relatively broad 90% intervals for Rf add to the conditional nature but are secondary. No change to the reader's verdict is needed.","tokens_in":13435,"tokens_out":4931,"duration_ms":44861,"concrete_test":"Generate synthetic NuSTAR spectra with the full returning-radiation model of Mirzaev et al. (2024, ApJ 976, 229) using input parameters matching the best-fit of Swift J1658.2-4242 (a*=0.998, h=2 rg, Gamma=1.65, AFe=0.5, Rf=1, log xi=3.2, log N=15, i=64 deg, NH=18e22 cm^-2), then fit the mock spectra with relxilllpCp returnrad=0 and returnrad=1 using the same pipeline. If recovered a*, h, or Rf deviate from the input by more than the 90% statistical errors, the null result is an artifact of the emissivity-only implementation; if both fits recover the inputs within errors, the approximation is adequate. Repeating for one additional source would confirm.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that returning radiation has no significant impact on measured parameters for these three sources, but the test only toggles returnrad in relxilllpCp. As the authors state in Section V, 'in relxill the effect of the returning radiation is considered only in the emissivity profile ... and the non-relativistic reflection spectrum is still calculated assuming that the incident radiation is a Comptonized spectrum.' The returning radiation incident on the disk is actually a reflection spectrum, not a Comptonized power law, and for low coronal height and high spin the returning component is a large fraction of the illuminating flux. The authors cite Ref. [43] showing the emissivity-only approximation is 'not supposed to work well for very high values of the black hole spin parameters and low values of the coronal height' — precisely the regime of Swift J1658 (h=2 rg, a*=0.998) and MAXI J1535 (h=2 rg, a*=0.93). The observed agreement between returnrad=0 and 1 may therefore only show that the partial implementation misses the dominant effect; a fuller model that reprocesses returning radiation into the reflected continuum (Refs. [43,44]) could shift a*, h, AFe, and Rf. The final sentence of Section V concedes: 'It is still possible that with that model we can obtain different estimates of some parameters.' Thus the broad title-level conclusion is not established for the full physical effect.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper fits high-quality NuSTAR (and Swift for one source) spectra of three Galactic black holes—GX 339-4, Swift J1658.2-4242, and MAXI J1535-571—using the lamppost model relxilllpCp, first with the returning-radiation switch off (returnrad=0) and then on (returnrad=1). For each source the best-fit parameters (spin, coronal height, iron abundance, reflection fraction, etc.) are compared with 90% confidence intervals. The central finding is that the parameters are statistically consistent between the two versions, even for sources with high spin and low coronal height, where returning radiation is expected to be most important. The paper also discusses the approximation in the relxill implementation, noting that only the emissivity profile is modified and the non-relativistic reflection spectrum is still computed assuming a Comptonized incident spectrum.","tokens_in":13706,"tokens_out":5820,"duration_ms":56943,"significance":"If read as a test of the current relxill implementation, the result is useful: it shows that, for these spectra, toggling the emissivity-only returning-radiation correction produces no significant change in the inferred parameters. The paper is transparent about the model's limitations and cites the relevant literature on where the approximation should break down. However, because the implementation excludes the spectral reprocessing of the returning radiation, the broader physical conclusion about returning radiation is not established. The study illustrates an important caveat for the reflection-spectroscopy community, but the interpretation in the abstract and title should be scoped more carefully.","major_comments":[{"comment":"The conclusion that returning radiation has no significant impact on the measured parameters is not established for the full physical effect. As the authors state, the relxill model includes the returning radiation only in the emissivity profile and still computes the non-relativistic reflection spectrum assuming that the incident radiation is a Comptonized power law. They also cite Ref. [43] showing this approximation is expected to work poorly for high spin and low coronal height—exactly the regime of Swift J1658.2-4242 and MAXI J1535-571. Therefore the observed agreement between returnrad=0 and returnrad=1 may only demonstrate that the emissivity-only correction is unimportant in the fits, not that the full returning-radiation effect is unimportant. The abstract and title should be explicitly restricted to the relxill implementation, or the interpretation should be revised to avoid overgeneralizing.","section":"Section V, final paragraph; Abstract"},{"comment":"The sensitivity of the test is limited by parameter boundaries. For Swift J1658.2-4242 and MAXI J1535-571, the coronal height is pinned at the lower boundary (h=2 rg), and for Swift J1658.2-4242 and GX 339-4 the spin is pinned at the upper boundary (a*=0.998). The fits therefore cannot explore the strongest light-bending regime, and the statement that these sources 'should maximize the effect' is only partially realized. The paper should quantify the fraction of the relevant parameter space that is excluded by these boundaries and discuss how this truncation affects the strength of the null result.","section":"Tables IV and V; Section IV.B"},{"comment":"The claim of 'no significant difference' rests on overlapping 90% confidence intervals, but for key parameters those intervals are broad or one-sided. For example, a* for Swift J1658.2-4242 is 0.998-0.13, and h for the two low-corona sources has only an upper uncertainty. A more direct quantitative comparison, such as the change in chi-square for a fixed parameter or a confidence level for the parameter shift, would make the sensitivity of the test more explicit and strengthen the conclusion.","section":"Tables IV and V"}],"minor_comments":[{"comment":"There is a typo in the passage 'in the case of a lamppost corona a, h ≲ 5 rg': the stray 'a,' should be removed or replaced, e.g., 'in the case of a lamppost corona h ≲ 5 rg.'","section":"Section I"},{"comment":"The kerrbb normalization is not reported in Table III; providing it would improve reproducibility, since the disk component is part of the spectral model.","section":"Section IV.B and Table III"},{"comment":"The small chi-square differences between the returnrad=0 and returnrad=1 fits (Δχ2 ≈ 1.4-1.7 for comparable dof) are a useful quantitative summary of the null result and could be highlighted in the text or table.","section":"Tables III-V"},{"comment":"The residual panels show data-to-model ratios with a limited vertical scale; using chi-squared residuals might better reveal systematic residual features, particularly around the iron line and Compton hump.","section":"Figures 2-4"}],"recommendation":"major_revision","confidential_remarks":"The paper is honest and clearly discloses the model limitation, so there is no concern about circularity or concealment. The main issue is scope: the title and abstract claim an 'impact of returning radiation' while the test only toggles the emissivity profile in relxill. Given the authors' own acknowledgment that the approximation is expected to fail in the high-spin/low-corona regime, the paper should be revised to restrict the claim to the tested implementation and to discuss the sensitivity limitations more prominently. A major revision is appropriate, but the underlying data analysis and the empirical comparison are sound."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe headline: this is a careful and honest test of the current returning-radiation implementation in relxill, not a general statement about the physics of returning radiation. Read it as a validation of the returnrad switch, and it works. Read it as proof that returning radiation doesn't matter, and it overreaches.\n\nWhat's new: the first comparison of relxilllpCp fits with returnrad=0 vs 1 on real NuSTAR spectra of three Galactic black holes. The analysis is standard and the presentation is transparent. Tables show consistent parameters with overlapping 90% errors, and the authors flag when parameters sit at grid boundaries. They also cite the relevant model papers, including the ones that spell out the approximation's limits.\n\nThe soft spot is real, and the stress test is right. In relxill, returning radiation only modifies the emissivity profile; the non-relativistic reflection spectrum is still computed for a Comptonized incident spectrum. But the actual returning radiation is a reflection spectrum, and for the fastest-spinning, lowest-corona sources it is a large fraction of the illuminating flux. The authors themselves say in Section V that this approximation is not expected to work well for exactly the regime they study (h < 3 rg, a* > 0.9), and they note that a slower, more complete model could shift some parameters. So the null result may be an artifact of the partial implementation. They could have made that limitation more prominent in the abstract; as written, \"we do not find any significant difference\" is unqualified, and a casual reader could walk away with too strong a conclusion.\n\nThat said, they deserve credit for stating the limitation in the text. The data work is solid, the fits are reproducible, and the conclusion is appropriately qualified in Section V. It is a useful sanity check for the many relxill users out there.\n\nWho it is for: anyone fitting reflection spectra with lamppost models, especially those measuring spins. It is a moderate-significance empirical result.\n\nMy verdict: send to peer review. It deserves a serious referee, and the right referee will push them to either soften the abstract or add a caveat. I would like the final version to make clear that the test only covers the current relxill implementation.","headline":"Useful as a validation of relxill's returnrad switch, but the null result only covers that implementation, not the full physics of returning radiation.","tokens_in":14288,"tokens_out":2312,"would_cite":true,"duration_ms":22605,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["95.85.Nv","97.60.Lf"],"model":"deepseek-v4-flash","headline":"Adding returning radiation leaves black hole spin fits unchanged","keywords":["X-ray reflection spectroscopy","returning radiation","black hole spin measurement","lamppost corona","relxill","Galactic black holes","NuSTAR"],"falsifier":"Fit the same NuSTAR spectra with a model that treats the returning radiation self-consistently in the non-relativistic reflection spectrum, as in Refs. [43,44], and compare the best-fit $a_*$ and $h$ with the `returnrad = 1` relxill results; a significant shift would show that the emissivity-only implementation hides the effect.","tokens_in":13166,"feed_emoji":"🕳️","tokens_out":7950,"duration_ms":73739,"temperature":0.7,"pith_summary":"X-ray reflection spectroscopy infers black hole spin from the distorted reflection spectrum of the accretion disk. A correction called returning radiation—disk-emitted photons bent back onto the disk by strong gravity—has been added to the relxill model family, but only through the disk emissivity profile. This paper tests whether that correction changes measured parameters by fitting NuSTAR spectra of three fast-spinning Galactic black holes (GX 339–4, Swift J1658.2–4242, and MAXI J1535–571) with the correction switched off and on. In all three cases the best-fit spin, coronal height, iron abundance, and most parameters agree within uncertainties. The authors conclude that, at current data quality and within this model, neglecting returning radiation does not significantly bias parameter estimates, while cautioning that the implementation omits changes to the non-relativistic reflection spectrum.","feed_headline":"Adding returning radiation leaves black hole spin fits unchanged","feed_subtitle":"Three fast-rotating Galactic black holes give consistent spin, corona height, and iron abundance either way.","key_machinery":"The central object is the `relxilllpCp` lamppost reflection model in relxill, with the `returnrad` switch controlling whether the disk emissivity profile includes returning radiation. Returning radiation is the component of disk emission bent back to the disk by strong light bending. The switch changes only the emissivity profile; the non-relativistic reflection spectrum is still computed as if the incident radiation were a Comptonized power law. The paper's comparison is a controlled switch of this flag on three high-quality NuSTAR spectra.","core_discovery":"The paper's central claim is that including the returning radiation in `relxilllpCp` leaves the inferred parameters statistically unchanged for these three Galactic black holes. For GX 339–4 the fit gives $a_* \\approx 0.998$ and $h \\approx 7.5 r_g$; for Swift J1658.2–4242, $a_* > 0.98$ and $h \\approx 2 r_g$; for MAXI J1535–571, $a_* \\approx 0.93$ and $h \\approx 2 r_g$. These include exactly the conditions expected to maximize returning radiation (high spin, low corona), yet the values from fits with `returnrad = 0` and `returnrad = 1` are consistent at 90% confidence. The one systematic shift is a lower best-fit reflection fraction with returning radiation on, still marginally consistent with the off case. The paper does not claim the effect is negligible generally: it stresses that relxill includes the effect only in the emissivity profile and that a fuller treatment may change the estimates.","pith_inferences":["If the full self-consistent treatment changes parameter estimates, then current relxill-based spin catalogs could develop a systematic offset that grows with spin and shrinks with coronal height; that would only be visible in a mock-injection comparison, not in the single-source consistency shown here.","Polarimetric observations may be a more sensitive test of returning radiation than the spectral parameters considered here; the paper cites a Cygnus X-1 polarization study showing soft X-ray polarization dominated by returning reflection, so combined spectral-polarimetric fits could reveal the effect even when best-fit spectral parameters do not move.","Because two of the three sources pin the coronal height at the lower model boundary, the parameter space with the largest expected effect is only partially explored; allowing even lower heights or different coronal geometries in the model might bring the effect into view."],"forward_implications":["For GX 339–4, Swift J1658.2–4242, and MAXI J1535–571, spin and coronal height measurements are consistent whether returning radiation is on or off, so existing relxill-based measurements of these sources are not overturned by the correction.","The lower best-fit reflection fraction with returning radiation suggests that population-level reflection fraction estimates may shift slightly, though within current uncertainties.","The fact that the effect stays small in the high-spin, low-corona regime means the omission is unlikely to explain any tension in spin constraints for similar Galactic black holes at NuSTAR quality.","A faster implementation of the full treatment is needed before concluding that the approximation is safe beyond this sample."],"supporting_citations":[{"why":"Supplies the relxill version with the returnrad switch, the model used for both sets of fits.","marker":"[39]"},{"why":"Shows the emissivity-only approximation works poorly for high spin and low coronal height, defining the regime the paper tests.","marker":"[43]"},{"why":"Provides the previous NuSTAR analysis of GX 339–4 whose data and lamppost fit are reused.","marker":"[47]"},{"why":"Establishes Swift J1658.2–4242 as a high-spin, low-corona source from earlier NuSTAR analysis.","marker":"[48]"},{"why":"Gives the earlier NuSTAR analysis of MAXI J1535–571 that characterizes its spin and coronal height.","marker":"[52]"},{"why":"Provides the optimal binning algorithm used to group the spectra before fitting.","marker":"[53]"},{"why":"First study of returning radiation, establishing the effect is important for fast rotators and low coronae.","marker":"[35]"}],"fun_headline_variants":["Returning radiation no match for black hole spin fits","Reflection fits unaffected by returning radiation","Galactic black holes shrug off returning radiation","Spin estimates survive returning radiation test","Black hole parameters unchanged by returning radiation"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The comparison assumes that putting the returning radiation into the emissivity profile only, while keeping the incident spectrum a Comptonized power law, captures the effect; the paper itself notes this approximation is expected to be worse for high spins and low coronal heights, which are exactly the conditions of two of the three sources.","fun_headline_variants_meta":{"raw":{"variants":["Returning radiation no match for black hole spin fits","Reflection fits unaffected by returning radiation","Galactic black holes shrug off returning radiation","Spin estimates survive returning radiation test","Black hole parameters unchanged by returning radiation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000176,"raw_usage":{"total_tokens":1292,"prompt_tokens":953,"completion_tokens":339,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":569,"completion_tokens_details":{"reasoning_tokens":275}},"tokens_in":569,"tokens_out":339,"duration_ms":3768,"temperature":1.0,"reasoning_tokens":275,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T18:30:49.926532+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fit the same NuSTAR spectra with a model that treats the returning radiation self-consistently in the non-relativistic reflection spectrum, as in Refs. [43,44], and compare the best-fit $a_*$ and $h$ with the `returnrad = 1` relxill results; a significant shift would show that the emissivity-only implementation hides the effect.","supporting_citations":[{"cited_title":"The Effect of Returning Radiation on Relativistic Reflection","cited_arxiv_id":"2206.07973","evidence_quote":"Supplies the relxill version with the returnrad switch, the model used for both sets of fits."},{"cited_title":"Towards more accurate synthetic reflection spectra: improving the calculations of returning radiation","cited_arxiv_id":"2401.05582","evidence_quote":"Shows the emissivity-only approximation works poorly for high spin and low coronal height, defining the regime the paper tests."},{"cited_title":"The Hard State of the Highly Absorbed High Inclination Black Hole Binary Candidate Swift J1658.2-4242 Observed by NuSTAR and Swift","cited_arxiv_id":"1805.07705","evidence_quote":"Establishes Swift J1658.2–4242 as a high-spin, low-corona source from earlier NuSTAR analysis."},{"cited_title":"The profile and equivalent width of the X-ray iron emission-line from a disk around a Kerr black hole","cited_arxiv_id":"astro-ph/9704177","evidence_quote":"First study of returning radiation, establishing the effect is important for fast rotators and low coronae."}],"review_version":1}