{"id":"f5a4fb20-a396-4dcc-ac9f-48ce2d542486","arxiv_id":"2502.09705","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"The CIB-CXB cross-power signal previously attributed to exotic high-redshift sources is reproduced by JWST-resolved galaxies, with a significant 6<z<13 component.","lead":"Astronomers compared faint galaxies seen by JWST with the leftover X-ray glow measured by Chandra and found they track each other across the sky, including a first statistical signal from galaxies at redshifts 6 to 13. The result ties part of the cosmic X-ray background to ordinary galaxies and black holes rather than exotic sources.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"High-z X-ray signal depends on unquantified photo-z leakage: low-z flux scattered into the 6<z<13 map can mimic the 5–7σ cross-power because low-z shot noise is ~100× higher; a leak of ~1% dominates the high-z bin.","rationale":"Read in good faith, the paper's central new claim is that JWST-resolved sources at 6<z<13 produce a significant CIB–CXB cross-power signal. The broad CIB auto-power consistency with direct source counts and the agreement of the full-redshift cross-power with earlier work are genuine supporting evidence. However, the high-z bin rests entirely on z-PDF flux weighting, and the unquantified residual leakage from much brighter low-z sources can dominate the high-z map. This is the same weakest assumption identified by the reader. The proposed leakage-matrix or high-purity subsample test would settle whether the 5–7σ high-z signal is astrophysical. The intermediate-bin 4σ shot-noise discrepancy and the 'coherence' versus cross-power wording are secondary issues; they do not change the conditional status of the main claim.","tokens_in":18048,"tokens_out":3670,"duration_ms":42141,"concrete_test":"Construct a leakage matrix from spectroscopically confirmed COSMOS-Web sources: for each photo-z bin i, compute P(z_spec in bin j | z_photo in bin i). Apply the measured leakage rates to the low-z maps and subtract the predicted leaked contribution from the 6<z<13 cross-power; if the corrected S/N drops below ~5, the high-z claim is not robust. Alternatively, rebuild the high-z map using only sources with integrated z-PDF P(6<z<13)>0.9 (or an ODDS cut) and recompute the F277W/F444W cross-power; the claim stands only if the 5.39–7.32σ signal survives.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing concern is unquantified leakage of low-redshift flux into the Δz=[6–13] maps through the Le Phare z-PDFs. In Section 2.2.1, each source is assigned a binned flux F(Δz)=∫p(z)dz·F, so every source contributes to every redshift bin with a weight set by its z-PDF. The paper removes sources flagged as having contaminated photo-z measurements, but it does not measure the residual contamination in the 6–13 bin. The flux contrast between bins is large: for F277W the shot-noise power in the [0–3] bin is 4.86×10^-11 nW^2 m^-4 sr^-1 versus 4.32×10^-13 in the [6–13] bin (Section 3.1). Thus even ~1% leakage of [0–3] flux into the high-z map contributes as much as the entire genuine high-z shot-noise term. A similar leakage into the cross-power can produce apparent CIB–CXB coherence at 6<z<13 even if no z>6 source emits X-rays, because low-z IR flux misallocated to the high-z map is being correlated with low-z CXB fluctuations. The stated removal of contaminated photo-z sources does not quantify the residual wings of the remaining z-PDFs, and the high-z bin is precisely where broad p(z) tails and catastrophic outliers are hardest to calibrate. This directly affects the headline claim of 'first evidence of significant emission of X-rays among the newly discovered population of JWST z>6 galaxies' (Section 5, point 1).","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs CIB fluctuation maps from COSMOS-Web NIRCam F277W and F444W catalogs at magnitudes fainter than the Spitzer limit, splits sources into photometric-redshift bins Δz = [0–3], [3–6], [6–13], and [0–13] using z-PDF flux weighting, and cross-correlates the maps with unresolved Chandra [0.5–2] keV CXB fluctuation maps from COSMOS-Legacy. The authors report significant CIB–CXB cross-power in both filters at 0 < z < 13 (S/N 4.80 and 6.20), and especially in the 6 < z < 13 bin (S/N 7.32 in F277W and 5.39 in F444W), which they interpret as evidence of X-ray emission from JWST-discovered z > 6 galaxies. They also fit the large-scale bias of the CIB sources, estimate the coherent CXB flux and derive lower limits on the bias of X-ray sources, and conclude that at least ~94% of the [0.5–2] keV CXB has been resolved.","tokens_in":18402,"tokens_out":4742,"duration_ms":53097,"significance":"If the high-redshift detection is robust, this is a timely and important result: it connects the long-studied CIB–CXB coherence to individually resolved JWST galaxy populations, provides a population-level probe of X-ray emission from z > 6 galaxies, and constrains models of early SMBH seeding and accretion. The paper has real strengths: the shot-noise estimates computed from source counts largely reproduce the directly measured map shot noise, the F444W cross-power is consistent with earlier measurements by Li et al. (2018), the map-making includes a careful flat-fielding treatment of the varying COSMOS-Web exposure depth, and the bias fits are carried out with MCMC and reported with reduced-χ² values. The derived CXB flux and bias estimates are explicitly labeled as estimates, and the authors do not overstate their independence. The central caveat is that the headline high-z signal depends on the unquantified reliability of the Le Phare z-PDF tails, a concern that must be addressed before the population-level interpretation can be accepted.","major_comments":[{"comment":"The map construction assigns to every redshift bin a flux F(Δz)=∫p(z)dz·F for every source, so low-z sources with broad or catastrophic z-PDF tails contribute to the Δz=[6–13] map. The manuscript states that sources with contaminated photo-z measurements are removed, but it does not quantify the residual contamination from the remaining z-PDF wings in the 6–13 bin. This matters quantitatively: in F277W the shot-noise power is 4.86×10^-11 nW² m^-4 sr^-1 in [0–3] versus 4.32×10^-13 nW² m^-4 sr^-1 in [6–13] (§3.1), so even ~1% leakage of [0–3] flux into the high-z map contributes an amount comparable to the entire genuine high-z shot-noise term. A similar leakage into the cross-power can correlate misallocated low-z IR flux with low-z CXB fluctuations and produce an apparent 6<z<13 CIB–CXB coherence without any z>6 X-ray emission. Because the claim in §5 point 1 rests on this signal, I request a quantitative leakage analysis: for example, stack the z-PDFs of all sources, compare the high-z maps with and without sources whose best-fit redshift is below 6, use empirical outlier rates from spectroscopic samples, or otherwise demonstrate that the high-z auto- and cross-power are insensitive to z-PDF tails. Without such a test, the reported S/N values of 7.32 and 5.39 cannot yet be interpreted as evidence of X-ray emission from z>6 galaxies.","section":"§2.2.1 and §5, point 1"},{"comment":"The F277W Δz=[3–6] shot noise estimated from the map, (7.992±0.007)×10^-12 nW² m^-4 sr^-1, differs from the value computed directly from the source counts, 7.964×10^-12 nW² m^-4 sr^-1, at the ~4σ level. This inconsistency is reported without discussion, even though the same maps enter the cross-power spectra and the MCMC bias fits. It indicates an unmodeled systematic in the map-making, source-count weighting, or z-PDF assignment that should be identified and propagated into the derived quantities, since a similar but smaller effect could bias the other redshift bins as well.","section":"§3.1"},{"comment":"The statement that 'approximately 94% of the CXB is resolved' is derived from the coherent CXB flux estimate of 2.64+0.52−0.67 ×10^-13 erg/s/cm²/deg², which is computed from the same auto- and cross-power spectra that are affected by the photo-z leakage concern. If part of the high-z cross-power is actually low-z leakage, the cumulative flux estimate would be biased high, and the 94% conclusion would weaken correspondingly. The 27+13−10% resolution fraction quoted in §5 point 3 is also presented without the caveat that the high-z contribution is the least secure part of this estimate; the discussion should explicitly state how the CXB flux and its uncertainty change under plausible leakage scenarios.","section":"§4.2 and §5, point 3"}],"minor_comments":[{"comment":"The displayed formulas for the broadband averages ⟨P_IR⟩ and ⟨P_IR,X⟩ are missing the division by the summed inverse variances; as typeset, the expressions appear to be products rather than inverse-variance-weighted averages. Please correct the notation.","section":"§2.3, Eqs. (1)–(2)"},{"comment":"The code name 'SourcExtractor++' should be 'SourceExtractor++' (or 'SE++' as used elsewhere).","section":"§2.2, first paragraph"},{"comment":"The text says the 4.5 μm CIB auto-power spectrum computed in 'Kashlinsky et al. (2012) and Li et al. (2018)' is significantly higher than the JWST-based measurement, but Li et al. (2018) is a cross-power study, not an auto-power measurement. Please correct the citation.","section":"§5, point 4"},{"comment":"The units of the cross-power are given as 'erg/s/cm2 nW/m2/sr'; this should be clearly written as (erg s^-1 cm^-2)(nW m^-2 sr^-1) to avoid ambiguity.","section":"Table 1"},{"comment":"The text says 'we focus our analysis on the latter' (F444W), but the paper then reports results for both F277W and F444W throughout; please clarify the intended statement.","section":"§2.2"},{"comment":"The Soltan-argument equation has garbled exponents and parentheses in the rendered text; please check the typeset equation and define all symbols at first use.","section":"§4.4, Eq. (13)"}],"recommendation":"major_revision","confidential_remarks":"The photo-z leakage issue is the key obstacle: it is a standard and fixable systematic, not a fundamental flaw, but it is load-bearing for the paper's most striking claim. If the authors can provide a quantitative leakage test and show that the high-z signal survives, the paper would be a strong contribution. I also suggest that the editor ask the authors to temper the phrase 'first evidence' in light of existing serendipitous X-ray detections at z>6 (e.g., Bogdán et al. 2024), which the paper itself cites."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the short version. This paper does something new: it builds CIB maps from JWST COSMOS-Web sources and cross-correlates them with the Chandra CXB, splitting by photo-z. The headline is a 7.3σ (F277W) and 5.4σ (F444W) cross-power in the 6<z<13 bin, which the authors interpret as X-ray emission from JWST galaxies, including a z>6 population. The machinery is mostly solid: they flatten depth variations across the field, mask X-ray sources, and the auto-power shot noise matches source-count integrals for the broad bin. The broad-band cross-power is consistent with Li et al. (2018), which gives confidence that the method is working.\n\nBut the high-z result has a weakness that needs to be resolved before I'd quote it. The flux of every source is spread into all redshift bins using Le Phare z-PDFs. So low-z flux leaks into the high-z map with weight set by the PDF tail. The low-z shot noise is ~100 times higher than the high-z shot noise, so a leakage of about one percent is enough to produce a cross-power equal to the entire high-z signal. The authors remove sources with contaminated photo-z, but they don't quantify the residual wings of the remaining PDFs. That's a quantitative statement that can be made, and it's essential for the 6<z<13 claim.\n\nThere are also two smaller issues. The F277W [3-6] shot noise disagrees with the source counts at ~4σ; that inconsistency suggests something in the photo-z weighting or map-making in that bin, and it makes the high-z bin look less secure. And the abstract calls the high-z signal 'coherence,' but the paper never computes the coherence statistic, only the cross-power S/N. Minor wording, but easy to fix.\n\nThe rest of the paper—the bias fits, CXB flux estimates, and the discussion of the unresolved CIB excess—is reasonable. The estimates of CXB flux and b_X are clearly labeled as estimates, not predictions, so no circularity problem.\n\nBottom line: this deserves a serious referee. The measurement is potentially important, and the photo-z leakage is testable. I'd send it to review with the recommendation that the authors add a quantitative leakage test for the high-z bin and explain the [3-6] shot-noise discrepancy. If those hold up, the result is a real step forward.","headline":"First JWST-resolved CIB-CXB cross-power measurement is interesting, but the 5-7σ high-z signal hinges on unquantified photo-z leakage; a referee should demand a contamination test before that claim is accepted.","tokens_in":19078,"tokens_out":3390,"would_cite":true,"duration_ms":34768,"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":"JWST-resolved galaxies explain the infrared-X-ray background coherence, with a first X-ray signal at z>6.","keywords":["cosmic infrared background","cosmic X-ray background","cross-power spectrum","high-redshift galaxies","JWST","Chandra","active galactic nuclei","large-scale structure"],"falsifier":"Recalculate the 6<z<13 cross-power using only sources with spectroscopic redshifts or with photo-z quality flags demanding single-peaked, narrow z-PDFs; if the high-z signal drops below significance, the result is a photo-z artifact. Alternatively, scramble source positions within the high-z bin while keeping their fluxes and z-PDFs: a persistent cross-power would indicate a masking or map-making systematic rather than real clustering.","tokens_in":17870,"feed_emoji":"🔭","tokens_out":4980,"duration_ms":47336,"temperature":0.7,"pith_summary":"This paper tries to show that the long-studied coherence between the unresolved cosmic infrared background (CIB) and the soft cosmic X-ray background (CXB) comes from ordinary galaxies that JWST can now resolve, and that some of those galaxies live at redshifts above 6. Using JWST's COSMOS-Web survey, the authors build maps of infrared light from sources fainter than the old Spitzer limit and cross-correlate them with Chandra's unresolved [0.5-2] keV X-ray map. They report significant cross-power signals, including a 5-7 sigma signal in the 6<z<13 bin, which they interpret as the first aggregate evidence of X-ray emission from JWST-discovered z>6 galaxies. If correct, the result would let the known CIB-CXB coherence be explained by resolved galaxy populations rather than exotic sources, while still leaving room for a high-redshift accreting black hole population that contributes to the soft X-ray background.","feed_headline":"JWST galaxies show first X-ray signal at z>6","feed_subtitle":"Cross-correlating resolved JWST galaxies with the soft X-ray background closes the CIB-CXB coherence puzzle.","key_machinery":"The machinery is the angular cross-power spectrum $P_{\\rm IR,X}(q)=\\langle\\Delta_{\\rm IR}(q)\\Delta_{\\rm X}^{*}(q)\\rangle$ between CIB fluctuation maps and CXB fluctuation maps, together with the coherence term $C=P^2_{\\rm IR,X}/(P_{\\rm X}P_{\\rm IR})$. CIB maps are built by placing COSMOS-Web sources fainter than $m_{\\rm AB}=25$ as unresolved point sources, with each source's flux split into redshift bins by its photometric redshift probability distribution from Le Phare and weighted by the survey completeness. The clustering component is modeled with Limber's equation using a single fitted bias parameter $\\tilde{b}_{\\rm IR}$, and the CXB flux production is estimated from the cross-power and auto-power via Monte Carlo sampling. This allows the paper to translate measured power into statements about the bias, halo masses, and X-ray flux production of the populations involved.","core_discovery":"The central claim is that sources resolved by JWST emit soft X-rays and are clustered on large spatial scales, so the coherence previously seen between unresolved CIB and CXB fluctuations can be reproduced with ordinary star-forming galaxies and active galactic nuclei. Specifically, the paper reports cross-power spectrum signal-to-noise ratios of 4.80 (F277W) and 6.20 (F444W) over 1-1000 arcseconds in the full 0<z<13 range, and 7.32 and 5.39 in the 6<z<13 bin. This high-z signal is described as the first significant evidence of X-ray emission among the newly discovered JWST z>6 galaxy population. The same measurement pipeline yields CXB flux estimates implying roughly 94% of the [0.5-2] keV CXB is resolved, with an accreted black hole mass density at z=6 of $\\rho_{\\rm acc}\\approx 10^{5.15}\\,M_\\odot\\,\\mathrm{Mpc}^{-3}$ if that emission is entirely from accreting black holes.","pith_inferences":["A direct testable extension would be to repeat the analysis with the z>6 bin split into finer redshift slices or with SED-derived star-formation rates, which would show whether the X-ray signal tracks star formation or black hole accretion.","The paper's logic implies that if the high-z X-ray emission is indeed from accreting black holes, the Soltan argument gives a lower bound on the seed black hole mass density at z=6, linking JWST's overmassive black hole discoveries to the cosmic X-ray background.","The fact that the JWST auto-power spectrum lies well below the old Spitzer fluctuation excess suggests the mysterious clustering component may be even fainter or more diffuse than the $m_{\\rm AB}>29$ population, a separation worth probing with deeper JWST mosaics."],"forward_implications":["If the CIB-CXB coherence is explained by resolved galaxies, models invoking primordial black holes or direct-collapse black holes as the dominant source of the large-scale coherence lose a key observational motivation.","The z>6 cross-power signal implies an X-ray emitting galaxy population that Chandra cannot detect individually, which future X-ray missions like AXIS could resolve.","The fitted bias values $\\tilde{b}_{\\rm IR}\\approx 1.03$, $3.21$, and $6.51$ for the $z=0$--$3$, $3$--$6$, and $6$--$13$ bins give a redshift evolution of large-scale structure that can be compared with halo occupation models.","The CXB flux estimate of roughly 94% resolved places a tight upper limit on the unresolved soft X-ray background that any remaining diffuse or exotic component must satisfy."],"supporting_citations":[{"why":"Detected the CIB-CXB coherence that this work updates with resolved JWST sources.","marker":"Cappelluti et al. 2013"},{"why":"Supplied the Chandra CXB fluctuation maps and the previous cross-power baseline used for comparison.","marker":"Li et al. 2018"},{"why":"Defines the COSMOS-Web survey field and catalog from which the resolved CIB maps are built.","marker":"Casey et al. 2023"},{"why":"Provides the model reconstruction of CIB fluctuations from known galaxy populations that the JWST auto-power is compared against.","marker":"Helgason et al. 2012"},{"why":"Measured the unresolved CIB auto-power excess whose origin motivates the high-z investigation.","marker":"Kashlinsky et al. 2012"},{"why":"Measured the resolved and unresolved CXB fractions used to compute the approximately 94% resolved value.","marker":"Cappelluti et al. 2017a"},{"why":"Provides the COSMOS-Web source properties and photometric redshift catalog used in the map-making.","marker":"Shuntov et al. 2024"}],"fun_headline_variants":["First X-ray emission from JWST's z>6 galaxies","JWST galaxies at z>6 detected in soft X-rays","Coherent X-ray signal from JWST galaxies at z>6","JWST resolves X-ray-emitting galaxies at z>6"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the photometric redshift probability distributions correctly assign faint COSMOS-Web sources to the 6<z<13 bin; if low-redshift sources leak into that bin through broad tails or catastrophic photo-z failures, the reported high-z X-ray signal could be produced without any true z>6 X-ray emission.","fun_headline_variants_meta":{"raw":{"variants":["First X-ray emission from JWST's z>6 galaxies","JWST galaxies at z>6 detected in soft X-rays","Coherent X-ray signal from JWST galaxies at z>6","JWST resolves X-ray-emitting galaxies at z>6"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000474,"raw_usage":{"total_tokens":2418,"prompt_tokens":1074,"completion_tokens":1344,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":690,"completion_tokens_details":{"reasoning_tokens":1273}},"tokens_in":690,"tokens_out":1344,"duration_ms":11993,"temperature":1.0,"reasoning_tokens":1273,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T20:45:47.352914+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recalculate the 6<z<13 cross-power using only sources with spectroscopic redshifts or with photo-z quality flags demanding single-peaked, narrow z-PDFs; if the high-z signal drops below significance, the result is a photo-z artifact. Alternatively, scramble source positions within the high-z bin while keeping their fluxes and z-PDFs: a persistent cross-power would indicate a masking or map-making systematic rather than real clustering.","supporting_citations":[{"cited_title":"G., Ashby, M","cited_arxiv_id":null,"evidence_quote":"Measured the unresolved CIB auto-power excess whose origin motivates the high-z investigation."}],"review_version":1}