{"id":"dbf03830-409a-4883-8dc8-b6a204ff33b0","arxiv_id":"2501.11822","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Adding eKVN's short-baseline coverage to the EHT significantly improves reconstructed images of the M87* jet and stabilizes results against missing telescopes.","lead":"This paper simulates what happens when South Korea's extended KVN telescope network joins the Event Horizon Telescope in imaging the black hole M87*. Adding the eKVN's short baselines cuts image reconstruction noise by roughly half and makes the faint jet emission much more recoverable, including when other telescopes are lost to bad weather.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Headline gain rests on a single ground-truth realization and ground-truth-informed imaging choices; robustness across source models, weather, and flux normalization is untested.","rationale":"The reader's weakest assumption is essentially the same issue I would flag: a single simulated ground truth and ground-truth-informed calibration. I agree with CONDITIONAL. The paper is transparent about its procedure and the direction of the effect is credible; the missing piece is an ensemble or sensitivity analysis showing the factor-of-two noise reduction and ρjet_NX gain are not artifacts of the chosen model/realization. A blinded multi-model test would settle this. Because the authors already disclose the relevant limitations and the qualitative conclusion is well supported by the short-baseline argument, I would not change the verdict.","tokens_in":15796,"tokens_out":5332,"duration_ms":59721,"concrete_test":"Run the identical ngehtsim→eht-imaging pipeline on a small pre-registered ensemble: at least 3–5 distinct M87* ground truths (different GRMHD snapshots and at least one different simulation family, varying jet brightness/morphology) × 2–3 independent weather/SEFD realizations each. Freeze the imaging hyperparameters before seeing the data (use the current fiducial set, or a fixed grid chosen on a training model), and replace the ground-truth total-flux rescaling with a zero-spacing/short-baseline flux estimate. Report the distribution of Δρjet_NX and the r.m.s ratio. If improvements remain positive across the ensemble, the central claim is robust; if any case degrades, the claim needs morphological/weather qualifications.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline claim—r.m.s. difference against ground truth halves for the jet and ρjet_NX rises from 0.92 to 0.98—is established from one synthetic realization: the Chael et al. (2019) GRMHD model observed on April 6, 2024 under one ngehtsim weather/SEFD draw. The reconstruction is not blind. Section 2.2 selects the RML hyperparameters by maximizing ρNX to that same ground truth, and rescales total flux to the known ground-truth flux before self-calibration; Section 3 then measures all improvements against the same ground truth. Both steps inject information that real EHT+eKVN observations will not have. The direction of the result is physically plausible—short baselines do constrain large-scale jet emission—but the magnitude, and in an unfavorable morphology/weather realization possibly the sign, of the reported gains is unquantified. No uncertainty estimates, multiple models, or independent weather realizations are given, so the paper cannot distinguish an array property from a property of this one model/parameter combination.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an end-to-end imaging simulation for M87* with the 2022-era EHT and with the addition of the four telescopes of the extended KVN. Synthetic observations at 86 and 230 GHz are generated with ngehtsim from a Chael et al. (2019) GRMHD image, reconstructed with eht-imaging RML, and evaluated via normalized cross-correlation against the ground truth, with separate ring/jet masks, jackknifed station loss, and 86+230 GHz multi-frequency synthesis with spectral-index recovery. The central claim is that eKVN's six short baselines below 1 Gλ substantially improve jet-structure recovery, reduce residual noise, and make the array more robust to station loss.","tokens_in":15938,"tokens_out":4704,"duration_ms":50928,"significance":"If the reported gains are genuine, the paper provides quantitative support for a concrete, near-term array addition to the EHT: eKVN operates at 230 GHz, is already partially equipped, and helps fill a known short-baseline gap. The simulation pipeline uses public, community-standard tools (ngehtsim, eht-imaging), a realistic 2024 array schedule, and transparent uv-coverage and jackknife diagnostics. The qualitative conclusion that short baselines help large-scale jet and spectral-index recovery is physically plausible and is well supported by the uv-coverage comparison. However, all headline numbers are in-sample estimates derived from a single ground-truth model and evaluation choices that use that model's known flux and morphology; the magnitudes, and possibly even the sign, of the improvement for real observations are not yet established.","major_comments":[{"comment":"The quantitative comparison is not blind. Section 2.2 states that the fiducial RML hyperparameters are chosen by maximizing ρNX to the groundtruth image, and Appendix B chooses the 8 μas blurring kernel as the one that maximizes ρNX between the reconstructed and groundtruth images. Sections 3.1 and 3.2 then report ρNX and r.m.s. differences computed against the same blurred ground truth. Consequently, the 0.92→0.98 ρjet_NX gain and the factor-of-two r.m.s. reduction are optimized with respect to the test image; they quantify best-case in-sample fidelity, not predictive image fidelity for an unknown M87* realization. The spectral-index regularizers (l2=10, TV=50) are selected the same way. The qualitative uv-coverage argument survives, but the quantitative headline numbers need a blinded evaluation, for example by fixing hyperparameters a priori or selecting them on a training model and then applying them to a held-out model or snapshot.","section":"Section 2.2 and Appendix B"},{"comment":"The reconstruction uses information that real EHT+eKVN observations will not have. Section 2.2 rescales the total flux before self-calibration using the known groundtruth flux density, and Section 3.2 estimates the residual noise σs outside a source region that is 'guided by the groundtruth image.' Because the jet is faint and its absolute flux is weakly constrained by closure-only imaging, the reported reduction of residual noise and the spectral-index reliability partly reflect this injected information. The authors should rerun at least one comparison without groundtruth-based flux rescaling, or with the flux estimated from the data itself (for example from short-baseline or zero-spacing constraints), and restate the conclusions accordingly.","section":"Section 2.2 and Section 3.2"},{"comment":"All quantitative claims rest on one ground-truth image (Chael et al. 2019), one observing date (April 6, 2024), one ngehtsim weather/SEFD draw, and one noise realization. No uncertainty estimates accompany the reported ρNX and r.m.s. values, and no test shows that the eKVN gain persists for different jet morphologies, source flux scales, or weather conditions. Since the paper's Discussion emphasizes stability and monitoring, it should include at least a small ensemble with multiple GRMHD snapshots or another M87* model and independent weather realizations, reporting the distributions of the fidelity metrics and of the ring parameters.","section":"Sections 2.1, 3.1, and 3.2"}],"minor_comments":[{"comment":"The sign of the regularizer term appears to be wrong: in a minimization of J(I), positive penalty terms should be added, not subtracted, unless S_R is defined with an unconventional sign.","section":"Equation (1)"},{"comment":"The caption contains a typo: 'telesctopes' should be 'telescopes.'","section":"Figure 1 caption"},{"comment":"The count of 'six more baselines shorter than 1 Gλ' is used to support the main argument, but it appears only in the Discussion; stating this count in Section 2.1 alongside the uv-coverage discussion would make the argument easier to follow.","section":"Section 4"},{"comment":"Figure A.1 is described as 'Same as Figure 4 (top)', but it shows only 230 GHz images and adds the KYS+KPC configuration; the caption should state exactly which panels are reproduced.","section":"Appendix A"},{"comment":"The paper would benefit from a data and software availability statement specifying ngehtsim and eht-imaging versions and whether the synthetic data and imaging scripts are publicly available.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a simulation forecast for a planned array addition, so the non-blind evaluation is more defensible than in a measurement paper, but the authors should not present quantitative improvements as predictions without a robustness or ensemble test. I would not reject: the qualitative uv-coverage argument is strong, and the requested tests are feasible with the existing pipeline. The paper fits the journal's scope well."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe one thing to know: this is a careful simulation study with a physically sensible conclusion. Adding eKVN's short baselines to EHT does improve recovery of M87*'s jet in the synthetic data, and the paper's qualitative case is solid. Don't accept the headline number—the factor-of-two rms reduction—at face value, though. It's measured against the same ground-truth image that was used to tune the imaging parameters, so it's an optimistic estimate.\n\nWhat's new: this is the first quantitative look at the full EHT+eKVN array at 86/230 GHz with jackknife tests and multi-frequency synthesis. The authors use ngehtsim with realistic weather/SEFDs and eht-imaging for RML reconstruction. The jackknife tests are useful: they show eKVN compensates for missing EHT stations, especially for short baselines. The appendix on partial eKVN participation (KYS+KPC only) is honest and informative—single-frequency 230 GHz gains are limited, but the 86 GHz data from all four eKVN sites in MFS imaging nearly recovers full-eKVN performance.\n\nSoft spots, in order of importance. First, one ground-truth model (Chael et al. 2019) and one weather realization. No second source model, no different weather draw, no flux-scale variation. Second, the RML hyperparameters are selected by maximizing rho_NX against that ground truth, and the 8 uas blurring kernel for the effective resolution is chosen the same way. The spectral-index regularizers (l2=10, TV=50) are also picked for highest rho_NX. Third, total flux is rescaled to the known ground-truth flux before self-calibration. The authors are transparent about this, and it's standard in simulation studies, but it means the absolute jet flux recovery is better than what real observations would give. These issues don't undermine the qualitative direction. Short baselines do constrain large-scale emission, and the uv-coverage comparison shows the gap is real. But the paper should present the factor-of-two as model-dependent and should test at least one more morphology and one more weather realization. Uncertainty estimates on the rms ratio would also help.\n\nWho is this for? EHT/ngEHT array planning, VLBI imaging methodologists, and anyone doing monitoring campaigns for M87* or Sgr A*. It deserves a serious referee. I'd send it to review with a request for a robustness section rather than desk reject. If the authors add multiple realizations and show the gain direction is stable, it becomes a solid reference for the eKVN contribution.\n\nRecommendation: engage with it; ask for the robustness tests in revision.","headline":"Solid simulation study: eKVN short baselines improve EHT jet reconstruction for M87*, but the headline gain is tuned to the single ground-truth model and needs robustness tests before being quoted.","tokens_in":16618,"tokens_out":2714,"would_cite":true,"duration_ms":28178,"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":"Adding the four-telescope eKVN array to the EHT would roughly halve residual noise in recovered M87* jet images, simulations show.","keywords":["event horizon telescope","M87*","black hole shadow","relativistic jet","very long baseline interferometry","short baseline coverage","multi-frequency synthesis","image reconstruction"],"falsifier":"Rerun the same synthetic-observation pipeline on a different GRMHD snapshot or a different weather realization: if $\\rho_{\\rm NX}^{\\rm jet}$ does not rise from about $0.92$ to $0.98$ when eKVN is added, the gain is model-specific. A data-side check is to flag the KYS station in real April 2024 EHT observations of M87* and measure whether jet-region residual noise increases by the roughly factor of two the simulations predict.","tokens_in":15557,"feed_emoji":"🔭","tokens_out":13555,"duration_ms":117238,"temperature":0.7,"pith_summary":"This paper asks what the Event Horizon Telescope gains when the four-telescope South Korean array eKVN joins it, using synthetic observations of a simulated M87* that includes both the black-hole shadow and an extended jet. The central claim is that the six extra baselines shorter than 1 gigawavelength contributed by eKVN fill the EHT's short-spacing gap, so the faint large-scale jet is recovered with roughly half the residual error, with the jet-region normalized cross-correlation $\\rho_{\\rm NX}^{\\rm jet}$ rising from $0.92$ to $0.98$. This matters because imaging the shadow and the jet simultaneously is the direct observational route to testing how the jet is launched, and because the short baselines also compensate when one or two EHT stations are lost to weather. A second result is that simultaneous 86 and 230 GHz imaging with eKVN improves the recovered spectral-index map along the jet, not just the images at each frequency.","feed_headline":"Adding four Korean dishes to EHT halves M87* jet-image noise","feed_subtitle":"Simulations show South Korea's short-baseline dishes lift M87* jet fidelity and replace lost EHT stations.","key_machinery":"The machinery is short-baseline uv coverage—the set of baseline lengths and orientations the array samples. eKVN is a four 21 m dish array in South Korea with baselines of roughly 130–500 km; at 230 GHz those provide 74–370 Mλ spacings, sensitivity to angular scales of about 2.1–0.5 mas, and at 86 GHz they supply the array's shortest spacings. The EHT alone has only two baselines below 1 Gλ, leaving a gap near 0.7 Gλ. Image reconstruction proceeds by regularized maximum likelihood, first using gain-independent closure quantities, then self-calibration, then visibility amplitudes; multi-frequency synthesis fits a spectral-index map with smoothness regularizers. The short baselines constrain the Fourier components that carry the diffuse jet, which is why their absence shows up as residual noise rather than as a missing ring.","core_discovery":"On the paper's own terms, the discovery is that the EHT's short-baseline deficit—not angular resolution or long-baseline sensitivity—is what limits jet recovery, and eKVN removes that deficit. With eKVN added to the full 2022-era EHT array, the root-mean-square difference between the ground-truth and reconstructed image becomes a half for the jet region, $\\rho_{\\rm NX}^{\\rm jet}$ improves from $0.92$ to $0.98$, and the log-scale correlation improves markedly. Jackknife tests that flag one or two EHT stations show the gain is largest exactly where the EHT alone is weakest: when Chile, Europe, or the US-mainland stations are removed, eKVN keeps the jet visible while EHT-only jet similarity drops to $0.76$–$0.78$. In multi-frequency synthesis at 86+230 GHz, eKVN removes diffuse residual emission outside the true source structure, so the spectral-index map's high-fidelity region traces the jet rather than noise, and the ring ellipticity is recovered more stably.","pith_inferences":["If the gain scales with the number of baselines below about 0.7 Gλ, then any future short-spacing addition to the EHT or ngEHT—not only the Korean sites—should produce comparable jet-recovery improvements; this is testable by repeating the simulation with a different short-baseline station.","Because all metrics are computed against one ground-truth model, the factor-of-two noise reduction is best read as a demonstration of mechanism; rerunning with other jet morphologies would show whether the benefit is universal.","The 86-plus-230 GHz result suggests a cheaper path to jet science: the partial-participation case with only two eKVN stations at 230 GHz and all four at 86 GHz already delivers most of the multi-frequency gain, so simultaneous-band capability may matter more than full 230 GHz coverage."],"forward_implications":["EHT+eKVN observations of M87* at 230 GHz should recover the jet alongside the black-hole shadow with about half the residual noise of the EHT alone.","Losing one or two EHT stations—for example in Chile, Europe, or the US mainland—no longer removes the short-baseline coverage needed for the jet, since eKVN restores most of the lost fidelity in jackknife tests.","Simultaneous 86 and 230 GHz imaging with eKVN improves both single-frequency images and gives a spectral-index map whose high-fidelity region follows the jet, whereas EHT-only synthesis leaves diffuse residual emission.","At 86 GHz the eKVN baselines are the shortest in the array, giving a better constraint on the compact flux that separates the bright ring from the faint jet.","For monitoring campaigns and Sgr A* dynamic imaging, the added snapshot uv coverage should improve movie fidelity."],"supporting_citations":[{"why":"Supplies the GRMHD ground-truth image of M87* with shadow and jet that all synthetic observations are based on.","marker":"Chael et al. 2019"},{"why":"Describes the synthetic-observation simulator that generates visibility data with realistic weather, gains, and fringe-detection thresholds.","marker":"Pesce et al. 2024a"},{"why":"Provides the simulator release used to produce the April 6, 2024 synthetic data set.","marker":"Pesce et al. 2024b"},{"why":"Introduces the regularized maximum-likelihood reconstruction code and regularizers used for imaging.","marker":"Chael et al. 2018"},{"why":"Extends the reconstruction approach to multi-frequency synthesis, enabling simultaneous 86/230 GHz images and spectral-index maps.","marker":"Chael et al. 2023"},{"why":"Defines the imaging parameter-search strategy adopted to select fiducial images.","marker":"Event Horizon Telescope Collaboration et al. 2019d"},{"why":"Supplies the VIDA and mF-ring model used to extract ring diameter and ellipticity in the jackknife tests.","marker":"Tiede et al. 2022"},{"why":"Provides the measured 2018 M87* ring diameter used as a comparison for recovered ring parameters.","marker":"Event Horizon Telescope Collaboration et al. 2024"}],"fun_headline_variants":["Korean dishes halve M87* jet noise in EHT simulations","eKVN addition cuts M87* jet image noise in half","Short-baseline Korean dishes boost M87* jet fidelity","eKVN compensates for missing EHT dishes in M87* jet imaging","Simulations: eKVN halves M87* jet reconstruction error"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the single general-relativistic magnetohydrodynamic (GRMHD) simulation used as ground truth faithfully represents the real M87* source, and that rescaling the total flux to that known model before self-calibration is a fair stand-in for real data; if the real jet morphology, flux scale, or weather differs, the reported improvement could shrink.","fun_headline_variants_meta":{"raw":{"variants":["Korean dishes halve M87* jet noise in EHT simulations","eKVN addition cuts M87* jet image noise in half","Short-baseline Korean dishes boost M87* jet fidelity","eKVN compensates for missing EHT dishes in M87* jet imaging","Simulations: eKVN halves M87* jet reconstruction error"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000588,"raw_usage":{"total_tokens":2831,"prompt_tokens":1083,"completion_tokens":1748,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":699,"completion_tokens_details":{"reasoning_tokens":1655}},"tokens_in":699,"tokens_out":1748,"duration_ms":13038,"temperature":1.0,"reasoning_tokens":1655,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T17:49:38.449341+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Rerun the same synthetic-observation pipeline on a different GRMHD snapshot or a different weather realization: if $\\rho_{\\rm NX}^{\\rm jet}$ does not rise from about $0.92$ to $0.98$ when eKVN is added, the gain is model-specific. A data-side check is to flag the KYS station in real April 2024 EHT observations of M87* and measure whether jet-region residual noise increases by the roughly factor of two the simulations predict.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the GRMHD ground-truth image of M87* with shadow and jet that all synthetic observations are based on."},{"cited_title":"A., Johnson , M","cited_arxiv_id":null,"evidence_quote":"Introduces the regularized maximum-likelihood reconstruction code and regularizers used for imaging."},{"cited_title":"W., et al","cited_arxiv_id":null,"evidence_quote":"Extends the reconstruction approach to multi-frequency synthesis, enabling simultaneous 86/230 GHz images and spectral-index maps."},{"cited_title":"E., & Palumbo , D","cited_arxiv_id":null,"evidence_quote":"Supplies the VIDA and mF-ring model used to extract ring diameter and ellipticity in the jackknife tests."},{"cited_title":"2024, , 681, A79","cited_arxiv_id":null,"evidence_quote":"Provides the measured 2018 M87* ring diameter used as a comparison for recovered ring parameters."}],"review_version":1}