{"id":"51ceb04d-ee59-4635-81c1-08183f9a5e1c","arxiv_id":"2506.06692","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"In-orbit Crab observations show that Resolve's energy resolution at 6 keV degrades with high count rates in neighboring pixels and that a nearest-neighbor coincidence cut restores it, while energy-scale offsets of about 1 eV between bright and faint pixels can masquerade as Doppler shifts.","lead":"XRISM's Resolve microcalorimeter changes its energy scale and loses energy resolution when observing bright sources; the resolution loss can be undone with a nearest-neighbor coincidence cut. The team mapped these effects with Crab observations and warns that eV-scale pixel-to-pixel offsets mimic roughly 50 km/s velocity shifts in bright point-source spectra.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The quantitative slope in Eq. 1 depends on the pixel-GTI-corrected rate axis, which Sec. 3.3 admits may be distorted by pile-up dead time and secondary-pulse false detections; a rate-dependent bias would change a=0.109±0.014 even though the after-cut zero slope is qualitatively robust.","rationale":"This is not an objection to the paper's core demonstration. The after-cut zero slope is strong independent evidence that the degradation is removed by the coincidence cut, and the paper is appropriately cautious in stating the validity range (<500 mCrab) and in acknowledging residual count-rate distortions. The concern is about the numerical calibration and the transferability of Eq. 1: the paper's own Sec. 3.3 flags the exact mechanism that could bias the x-axis, so a skeptical reader needs a check that the fitted slope is stable against that bias. The reader's weakest_assumption identifies this same point; I agree it is the most load-bearing assumption for the quantitative claim. The GX 13+1 intrinsic-velocity caveat is real but it concerns the application section rather than the calibration claim in the strongest_claim; it does not alter this assessment. Verdict remains CONDITIONAL/UNCHANGED.","tokens_in":13102,"tokens_out":7261,"duration_ms":82851,"concrete_test":"Redo the Fig. 6 fits using the FPGA-triggered event-candidate rate (the \"proxy for true count rate\" shown in the lower panel of Fig. 2) as the x-axis instead of the pixel-GTI-corrected processed rate. If the before-cut slope remains 0.109±0.014 within errors and the after-cut slope remains 0.00±0.02, the residual dead-time/pile-up bias is not load-bearing. If the slopes shift by more than the quoted 1σ uncertainties, Eq. 1 and the energy-offset trend in Fig. 5 should be refit with the FPGA-based rate or the validity range restricted to rates where both estimators agree.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The x-axis of Figures 5, 6, and 8 is the pixel-GTI-corrected count rate, formed by dividing the PSP-processed rate by the live-time fraction (Appendix 1). This corrects only for PSP-overflow dead time. Section 3.3 explicitly concedes: \"other mechanisms may remain to distort the count rate, such as dead time due to pile-up effects and false detection of secondary pulses, which is beyond the scope of this paper.\" If those mechanisms are rate-dependent, x is not the true incident rate on the neighboring pixel. Then the before-cut slope a=0.109±0.014 in Eq. 1 is a biased calibration of cross-talk degradation, and predictions made from it for other bright sources inherit the bias. The after-cut result a=0.00±0.02 is less vulnerable because the restored FWHM is flat in whatever x is used, so the qualitative conclusion that the cut restores ground-test resolution survives. The load-bearing numerical content is therefore the before-cut slope, and it rests on an uncorrected axis.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports in-orbit measurements of high-count-rate effects on the XRISM/Resolve microcalorimeter, using Crab Nebula observations at five offset positions with continuous 55Fe illumination. The authors quantify two main effects at 6 keV: an energy-scale shift that becomes increasingly negative at high count rates, and an energy-resolution degradation that scales with the pixel-GTI-corrected count rate and is attributed to electrical cross-talk from neighboring pixels. They show that a nearest-neighbor coincidence cut removes the resolution degradation, changing the fitted FWHM-versus-rate slope from a = 0.109 ± 0.014 eV/(cts/s/pix) to a = 0.00 ± 0.02 while leaving the intercept at 4.44 eV. They also attribute the low-rate +0.2 eV offset to a ~1 µK effective-temperature difference between fiducial and cleaned data, and they apply the findings to GX 13+1, finding a −1.15 eV inner-outer pixel offset that would mimic a +50 km/s velocity shift. The paper is a calibration study with explicit screening choices and an acknowledged limitation: the count-rate axis is corrected only for PSP-overflow dead time, with pile-up dead time and false secondary-pulse detection stated to be beyond the scope.","tokens_in":13283,"tokens_out":7405,"duration_ms":75824,"significance":"If the quantitative results hold, this is a useful calibration reference for bright-source spectroscopy with XRISM/Resolve and for future microcalorimeter missions. The cleanest result is the before/after comparison of the cross-talk cut: the FWHM slope against count rate changes from 0.109 ± 0.014 to 0.00 ± 0.02 with an unchanged intercept, giving a direct empirical demonstration that the cut restores the resolution. The +0.2 eV orbital-offset explanation is quantitatively tied to a 1 µK effective-temperature change, and the analysis uses public XRISM data and standard ftools with clear screening choices. The main quantitative claim is vulnerable to the acknowledged incompleteness of the count-rate correction, but the qualitative restoration result is robust to that concern.","major_comments":[{"comment":"The independent variable x in Eq. (1) is the pixel-GTI-corrected count rate, obtained by dividing the PSP-processed rate by the live-time fraction. This correction accounts only for PSP-overflow dead time, while Sec. 3.3 states that other mechanisms may remain to distort the count rate, such as dead time due to pile-up effects and false detection of secondary pulses. If those mechanisms are rate-dependent, the fitted slope a = 0.109 ± 0.014 and the energy-offset trends in Figs. 5 and 8 are systematically biased. I request a robustness check using the FPGA-triggered candidate rate, which the paper in Sec. 3.3 identifies as a proxy for the true count rate, or an explicit estimate of the residual dead-time bias; without this, the quantitative calibration in Eq. (1) should be presented as provisional.","section":"Sec. 3.3, Appendix 1; Eq. (1) and Fig. 6"},{"comment":"The text says that the energy resolution degrades with increasing count rates in neighboring pixels, but Eq. (1) defines x only as the pixel-GTI-corrected count rate. Please specify explicitly whether x is the count rate of the pixel whose FWHM is being fitted, the average count rate of its two electrical neighbors, or another aggregate. The figure caption and the text should use identical terminology; without this, the central physical interpretation of the slope cannot be reproduced.","section":"Sec. 4.2, Eq. (1) and Fig. 6"},{"comment":"The before-cut linear fit gives adjusted R^2 = 0.370, meaning that the model explains only about 37% of the variance. The paper does not account for possible correlations among data points, such as the same pixel appearing in multiple observations or pixels sharing a PSP quadrant, and the count-rate axis is treated as error-free. A discussion of the scatter, or a fit that accounts for these correlations, is needed to support the quoted uncertainty of ±0.014 on the slope.","section":"Sec. 4.2, Fig. 6"}],"minor_comments":[{"comment":"There is a typo: 'primarily attributed to to the complex interplay' should read 'primarily attributed to the complex interplay'.","section":"Sec. 5.1"},{"comment":"The description of cross-talk GTI construction during PSP overflow is hard to follow; please clarify how periods of event loss in pixels i−1 and i+1 are excluded when creating the GTI for pixel i, and whether this exclusion is applied symmetrically.","section":"Sec. 3.4"},{"comment":"The correction factor c in Eq. (3) is not defined. Please define it or note that it is of order unity.","section":"Sec. 4.2, Eq. (3)"},{"comment":"The upper panel of Fig. 2 is described as the rate of CPU consumption; please define the units and whether values above unity indicate overflow, to make the connection to the subsequent PSP-overflow discussion immediate.","section":"Sec. 3.3 and Fig. 2"},{"comment":"The claim that the post-cut resolution is 'consistent with ground-based testing' would be easier to verify if the ground-based FWHM value from Mizumoto et al. (2025) were quoted alongside the measured intercept b = 4.44 eV.","section":"Sec. 4.2"}],"recommendation":"major_revision","confidential_remarks":"The main issue is the count-rate axis used in the quantitative fits. The authors already acknowledge the residual mechanisms in Sec. 3.3, so the concern is not an internal inconsistency but a load-bearing calibration uncertainty that should be addressed with a robustness check using the FPGA-triggered rate. The ambiguity in Eq. (1) about whether x is the pixel's own rate or the neighbor rate should also be resolved. The paper is otherwise a solid calibration study and fits the scope of the journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Max,\n\nThis is a calibration paper with real results. The in-orbit demonstration is new: they used Crab at five offset positions and the filter-wheel 55Fe lines to measure how the 6 keV energy scale and resolution respond to count rate and neighbor count rate. The key number is the before/after cross-talk cut: FWHM slope vs neighbor rate goes from 0.109±0.014 eV per cts/s/pix to 0.00±0.02, intercept unchanged at 4.44 eV. The cut restores ground-test resolution. That is a clean, convincing result and worth citing.\n\nThe paper does well at separating effects: energy scale shifts are small (sub-eV to ~1 eV at 6 keV), and they tie the low-rate +0.2 eV offset to a 1 microkelvin effective-temperature drop. The data are public, the analysis is careful, and the validity range (below ~500 mCrab, GV closed) is stated.\n\nSoft spots, in proportion. The x-axis concern is real but limited: all trends use the pixel-GTI-corrected count rate, which corrects only for PSP-overflow dead time. Section 3.3 admits pile-up dead time and secondary-pulse false detections could distort the rate further. If those are rate-dependent, the before-cut slope in Eq. 1 is biased. That is a legitimate worry for anyone using the slope to predict degradation for other bright sources. It does not touch the after-cut flatness conclusion, because a flat line is flat in any x-axis. Also, the GX 13+1 test case shows a -1.15 eV offset between inner and outer pixels, but the paper does not test whether some of that could be intrinsic spatial variation in the Cr Ly alpha line centroids. They frame it as a caution for users, not a definitive instrumental correction, so it is a caveat rather than a flaw.\n\nThe paper leans on the companion ground-based paper for the cut definition, but that is standard for a calibration series. Self-citation is appropriate here.\n\nWho is this for: anyone analyzing bright point sources with Resolve, and detector teams planning future microcalorimeter missions. It deserves a serious referee; the main questions a referee should push on are the x-axis calibration and the GX 13+1 attribution.\n\nRecommendation: send it out. It is worth referee time and will be the reference for high-count-rate corrections in XRISM/Resolve.","headline":"Solid in-orbit calibration paper: the cross-talk cut demonstrably restores Resolve's 6 keV resolution to ground-test levels, with two acknowledged caveats (rate-axis fidelity and the GX 13+1 offset attribution) worth referee attention.","tokens_in":13946,"tokens_out":2123,"would_cite":true,"duration_ms":20469,"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":"This paper shows that at 6 keV the XRISM/Resolve microcalorimeter's energy resolution degrades by 0.109 eV per cts s$^{-1}$ pix$^{-1}$ of neighbor count rate, and a nearest-neighbor coincidence cut removes this degradation, restoring…","keywords":["X-ray microcalorimeter","energy resolution","cross-talk cut","count rate effects","energy scale","XRISM","Resolve","Mn K-alpha calibration"],"falsifier":"Take a single pixel whose neighboring pixels are illuminated at a known rate, shuffle the neighbor event times to destroy the $\\pm 25$ ms coincidence structure while preserving the rate, and measure the 6 keV FWHM of the central pixel: if the FWHM still degrades with rate after the shuffle, the degradation is not temporal cross-talk and the cross-talk cut cannot be what restores the resolution.","tokens_in":12847,"feed_emoji":"🔭","tokens_out":11856,"duration_ms":100489,"temperature":0.7,"pith_summary":"Using Crab Nebula observations at five offset positions with continuous $^{55}$Fe calibration illumination, this paper measures how the XRISM/Resolve microcalorimeter responds to count rates well above its design optimum. It establishes two rate-dependent effects at 6 keV: an energy-scale shift that turns negative at high count rates, and an energy-resolution degradation that grows with the count rate in electrically neighboring pixels. The central quantitative result is that the resolution degradation is fully removable: before a nearest-neighbor coincidence cut the FWHM rises with slope $a = 0.109 \\pm 0.014$ eV per cts s$^{-1}$ pix$^{-1}$, and after the cut the slope is $a = 0.00 \\pm 0.02$ with intercept $b = 4.44 \\pm 0.05$ eV, consistent with ground testing. Applied to the bright point source GX 13+1, the inner pixels show a $-1.15^{+0.40}_{-0.25}$ eV offset at 6 keV relative to the outer pixels, which would masquerade as a $+50$ km s$^{-1}$ velocity shift. The paper concludes that users analyzing velocity structures at the tens of km s$^{-1}$ level must account for these effects.","feed_headline":"Cross-talk cut restores Resolve's lost energy resolution","feed_subtitle":"Crab data: the 0.11 eV-per-rate slope drops to zero once neighbor pulses are cut.","key_machinery":"The load-bearing object is the 36-pixel microcalorimeter array itself, in which each pixel has its own thermal link and its own buffer in the Pulse Shape Processor, and in which pairs of pixels that share readout wiring experience mutual electrical cross-talk. When an X-ray hits pixel $i$, a small 'cross-talk child' pulse appears in pixel $i \\pm 1$; if a real event in the neighbor occurs near in time, the two pulses contaminate each other's inferred energies. The paper's central operation is the nearest-neighbor coincidence cut: a time filter that removes events in pixel $i$ that fall within $\\pm 25$ ms of a pulse in pixel $i \\pm 1$ in the same quadrant, while also discarding periods when the neighbor's events were lost to processor overflow so that no such filter could be constructed. The count-rate axis used throughout is the pixel-GTI-corrected rate, the processed rate divided by the pixel's live-time fraction, which is meant to recover the true incident rate. The argument runs through linear fits of FWHM versus that rate, and through the effective-temperature model used to correct the energy scale.","core_discovery":"The paper's central claim is that the high-count-rate degradation of Resolve's energy resolution at 6 keV is caused by untriggered electrical cross-talk from events in neighboring pixels, and that excluding each event whose pixel had a pulse in pixel $i \\pm 1$ within $\\pm 25$ ms (the 'cross-talk cut') removes this degradation entirely. As quantified by a linear fit of FWHM versus pixel-GTI-corrected neighbor count rate, the FWHM before the cut is $y = (0.109\\pm0.014)x + (4.44\\pm0.04)$ eV, and after the cut it is $y = (0.00\\pm0.02)x + (4.44\\pm0.05)$ eV, i.e., the resolution returns to the level measured in ground-based tests. The paper also establishes that the energy scale at 6 keV shifts increasingly negative as the count rate rises, while a small positive offset of about $+0.2$ eV at low count rates is attributed to orbital variation of the electronics and sparse effective-temperature fiducial sampling. For a centrally placed bright point source such as GX 13+1, the brighter inner pixels are measured to be $-1.15^{+0.40}_{-0.25}$ eV below the outer pixels at 6 keV, corresponding to a spurious velocity shift of about $+50$ km s$^{-1}$, while resolution degradation is negligible because the electrical neighbors of the inner pixels are dim.","pith_inferences":["The linear FWHM-versus-neighbor-rate slope measured here could be used to predict cross-talk degradation for future microcalorimeter arrays with similar readout coupling, if the coupling strength scales the same way.","The cross-talk cut's exposure-time cost suggests a selective strategy: apply the cut only to pixels whose neighbor count rates push the FWHM degradation past the scientific requirement.","Because the $^{55}$Fe calibration lines illuminate the array during science observations, the Mn K$\\alpha$ centroids could be used as a continuous in-situ monitor of rate-dependent energy shifts, enabling per-pixel correction without waiting for fiducial points."],"forward_implications":["For observations of bright sources, applying the cross-talk cut restores the energy resolution to the ground-calibrated value, at the cost of losing effective exposure time that grows with neighbor count rate.","The energy-scale offset between bright and faint pixels can masquerade as a velocity shift of tens of km s$^{-1}$, so pixel-by-pixel spectral fitting and offset correction are needed for precision velocity measurements.","Combining spectra from pixels without correcting rate-dependent energy shifts can artificially broaden spectral lines, affecting line-width measurements.","For a point source centered on the array, the cross-talk resolution loss is negligible because the neighbors of the brightest pixels have low count rates, so the cut is unnecessary in that configuration."],"supporting_citations":[{"why":"Defines the cross-talk cut and supplies the ground-based energy-resolution baseline to which the in-orbit post-cut resolution is compared.","marker":"Mizumoto et al. 2025"},{"why":"Describes the Pulse Shape Processor, its 50 cts s$^{-1}$ quadrant$^{-1}$ throughput limit, and the Hp/Mp/Ms/Lp/Ls event grading used for spectral selection.","marker":"Ishisaki et al. 2018"},{"why":"Provides the Mn K$\\alpha$ Lorentzian line complex used to fit the $^{55}$Fe calibration spectra at 6 keV.","marker":"Hölzer et al. 1997"},{"why":"Established the pixel-GTI exposure-time correction for in-orbit microcalorimeter data that yields the pixel-GTI-corrected count rate used throughout this paper.","marker":"Tsujimoto et al. 2018"},{"why":"Supplies the effective-temperature energy-scale correction algorithm and the temperature-to-energy relation used to interpret the low-count-rate positive offset.","marker":"Porter et al. submitted"},{"why":"Provides the 1 Crab flux normalization used to convert count rates into mCrab units for the astrophysical context.","marker":"Willingale et al. 2001"}],"fun_headline_variants":["Cross-talk cut restores Resolve's lost resolution","Neighbor-pulse cut erases high-rate blur in Resolve","Crab data: cross-talk cut fixes Resolve's resolution","Zero slope after cut: Resolve resolution stays sharp"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The count-rate axis assumes that dividing the processed event rate by the pixel live-time fraction recovers the true incident rate, with no additional dead time from pile-up or false secondary pulses; if those effects are significant, the fitted slopes and offsets could be biased.","fun_headline_variants_meta":{"raw":{"variants":["Cross-talk cut restores Resolve's lost resolution","Neighbor-pulse cut erases high-rate blur in Resolve","Crab data: cross-talk cut fixes Resolve's resolution","Zero slope after cut: Resolve resolution stays sharp"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000574,"raw_usage":{"total_tokens":2837,"prompt_tokens":1196,"completion_tokens":1641,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":812,"completion_tokens_details":{"reasoning_tokens":1573}},"tokens_in":812,"tokens_out":1641,"duration_ms":12544,"temperature":1.0,"reasoning_tokens":1573,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T05:52:20.544462+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a single pixel whose neighboring pixels are illuminated at a known rate, shuffle the neighbor event times to destroy the $\\pm 25$ ms coincidence structure while preserving the rate, and measure the 6 keV FWHM of the central pixel: if the FWHM still degrades with rate after the shuffle, the degradation is not temporal cross-talk and the cross-talk cut cannot be what restores the resolution.","supporting_citations":[{"cited_title":"2025, Journal of Astronomical Telescopes, Instruments, and Systems, 11, 042005, 10.1117/1.JATIS.11.4.042005","cited_arxiv_id":null,"evidence_quote":"Defines the cross-talk cut and supplies the ground-based energy-resolution baseline to which the in-orbit post-cut resolution is compared."},{"cited_title":"2018, Journal of Astronomical Telescopes, Instruments, and Systems, 4, 011217, 10.1117/1.JATIS.4.1.011217","cited_arxiv_id":null,"evidence_quote":"Describes the Pulse Shape Processor, its 50 cts s$^{-1}$ quadrant$^{-1}$ throughput limit, and the Hp/Mp/Ms/Lp/Ls event grading used for spectral selection."},{"cited_title":"E., et al","cited_arxiv_id":null,"evidence_quote":"Established the pixel-GTI exposure-time correction for in-orbit microcalorimeter data that yields the pixel-GTI-corrected count rate used throughout this paper."},{"cited_title":"G., et al","cited_arxiv_id":null,"evidence_quote":"Provides the 1 Crab flux normalization used to convert count rates into mCrab units for the astrophysical context."}],"review_version":1}