{"id":"fea3d0f3-73cc-41ca-b907-bf79c09fc62d","arxiv_id":"2501.10978","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Serial MultiView iteratively rotates a phase plane through time to calibrate atmospheric errors in VLBI astrometry, matching conventional MultiView at 5 GHz with reduced calibration sampling and automated phase ambiguity correction.","lead":"This paper proposes a new way to calibrate Very Long Baseline Interferometry (VLBI) data called serial MultiView, which tracks the atmosphere's phase errors as a rotating plane instead of fitting a new plane every observing cycle. It claims comparable accuracy to the established MultiView method at 5 GHz with less calibration overhead and automated phase-wrap correction, making high-precision astrometry faster and easier.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The <10 µas RA error claim is unsupported: the sMV−cMV offset difference is 9 ± 42 µas (1σ), so the measured precision is ~4× worse than claimed.","rationale":"The reader's verdict is CONDITIONAL, and this pass identifies the same core weakness in the quantitative claim: the <10 µas RA error is not established by the reported uncertainties. The reader's weakest_assumption field names the anchored-plane model, which is a related but distinct concern; the statistical insufficiency is more load-bearing because it directly targets the abstract's verification statement. The concrete test would settle it by computing the proper uncertainty of the offset difference and checking whether the reference-point offset is actually constant across epochs. If the test confirms σ_δ ≈ 50 µas, the claim must be revised to 'consistent with zero at ~40 µas' or supported with additional epochs. Therefore the conditional verdict remains appropriate.","tokens_in":19053,"tokens_out":4672,"duration_ms":46484,"concrete_test":"Recompute the RA offset difference in Table 4 with full error propagation using all four epoch uncertainties: σ_δ = sqrt(σ_sMV_B1^2 + σ_sMV_B2^2 + σ_cMV_B1^2 + σ_cMV_B2^2). If σ_δ is ≈50 µas, the 9 µas difference cannot support a <10 µas error claim. Additionally, compare the sMV−cMV offsets between B1 and B2 (Table 2: −59 vs −68 µas); if their difference is not significant relative to the combined epoch uncertainties, the constant-reference-offset assumption in Appendix B is untestable with current data.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central quantitative claim is that sMV achieves <10 µas error in RA, based on Table 4. There, the sMV position offset between B1 and B2 is 452 µas and cMV's is 461 µas, giving a difference of −9 µas. But the formal uncertainties are 31.0 µas (sMV) and 28.7 µas (cMV). The uncertainty of the difference is sqrt(31.0^2 + 28.7^2) ≈ 42 µas, so the measured difference is 9 ± 42 µas. This is consistent with zero but also with errors up to ~80 µas at 2σ; it cannot support a bound of <10 µas. Moreover, the individual-epoch differences (Table 2) are −59 µas (B1) and −68 µas (B2), ~6× larger than the claimed error, and are attributed to a reference-point offset (Appendix B) that is assumed constant between epochs but never independently measured. Thus the headline claim rests on subtracting two noisy quantities and on an untested constancy assumption.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes serial MultiView (sMV), a new realization of the VLBI MultiView calibration technique in which the atmospheric phase plane is rotated iteratively using the time series of secondary-calibrator residual phases, rather than being re-fitted from a full cycle of calibrator scans. The method includes a recursive automatic phase-ambiguity detector, optional Kalman and low-pass smoothing, and a released Python tool. The authors demonstrate sMV on VLBA 5 GHz observations of the radio star V1859 Ori over two epochs, comparing phase-referencing, conventional MultiView (cMV), and sMV, including simulated reduced calibrator sampling sequences. The central quantitative claim is that sMV achieves better than 10 microarcsecond error in RA relative to cMV, with reduced calibration overhead and automatic phase-wrap handling.","tokens_in":19284,"tokens_out":7715,"duration_ms":85552,"significance":"If the accuracy claim were established, sMV would be a practically valuable contribution: it shortens the calibrator observing cycle, automates the phase-ambiguity correction that is currently a manual burden in cMV, and is accompanied by open-source code and a reproducible pipeline. The idea of using time-domain continuity to rotate an anchored phase plane is a natural and potentially useful extension of MultiView, and the demonstration that one-third calibrator sampling can yield similar image quality is encouraging. However, the validation as presented is not yet sufficient to support the headline error bound: it rests on one target, two epochs, and a benchmark (cMV) whose own uncertainties and shared systematics are not propagated. The strengths—released code, a concrete algorithmic description, and a real VLBA demonstration—are real, but the quantitative claims need substantial revision.","major_comments":[{"comment":"The headline claim of \"<10 µas error in RA\" is not supported by the reported statistics. Table 4 lists the sMV minus cMV inter-epoch offset difference as −9 µas in RA, while the formal uncertainties are 31.0 µas (sMV) and 28.7 µas (cMV); combining them gives 9 ± 42 µas, so at 2σ the difference is consistent with errors of order 80 µas. The per-epoch sMV−cMV offsets are −59 µas (B1) and −68 µas (B2) in Tables 2 and 3, which are much larger than 10 µas and are attributed to a reference-point offset rather than independently measured. The manuscript should report the difference with its combined uncertainty and explicitly distinguish consistency with cMV from an absolute error bound; the abstract should not claim sub-10 µas accuracy on this basis.","section":"Abstract; Section 3, Table 4"},{"comment":"The validation treats cMV as the true stellar motion without propagating its uncertainty or considering shared systematic errors. Section 3 says \"we consider its random error to be small,\" but the cMV formal uncertainties in Table 4 (28.7 µas in RA) and in Tables 2–3 (13.6–18.0 µas per epoch) are comparable to the claimed 9 µas. Because sMV and cMV use the same calibrators and the same atmospheric plane model, common systematic errors (e.g., non-planar tropospheric/ionospheric structure or calibrator position errors) would not appear in the sMV−cMV difference. The Gaia comparison is a useful partial independent check, but the two MultiView methods both deviate from the Gaia prediction by roughly 140 µas (Table 4), so the data do not establish sub-10 µas absolute accuracy.","section":"Section 3; Table 4"},{"comment":"The assumption that the sMV−cMV reference-point offset is constant between epochs is load-bearing but untested. Appendix B shows that sMV corresponds to a plane constrained to pass through the origin (z = ax + by) while cMV fits a free plane (z = ax + by + c), and Section 3 argues that the resulting offset cancels in the B2−B1 difference. The measured offsets are −59 µas (B1) and −68 µas (B2) in RA, which cancel to −9 µas; however, no independent estimate of the reference-point offset is provided. If the offset varies because of calibrator position errors, source structure, or core shift, the Table 4 \"random error\" of sMV would be biased. The authors should either measure the reference-point offset (e.g., with more epochs or additional sources) or rephrase the result as an epoch-difference consistency check rather than an accuracy verification.","section":"Section 3; Appendix B; Tables 2–3"},{"comment":"The automatic ambiguity detection relies on several free parameters—loss weight w, tree depth n, pruning thresholds for plane inclination and rotational angular velocity, and the Kalman/low-pass smoothing factors—but the manuscript does not report how they were chosen, nor any sensitivity or robustness tests. The only demonstration is a single target at two epochs, so it is unclear whether the claimed automatic phase-wrap correction and reduced calibrator sampling rate generalize to other geometries, frequencies, or atmospheric conditions. A parameter sensitivity study, or at least a clear statement of default values and their influence on the results, is needed to support the algorithmic claims.","section":"Sections 2.2–2.3; Eq. (2)"}],"minor_comments":[{"comment":"The \"Uncertainty\" column in Table 4 is not defined in the caption or text; it is unclear whether these values are propagated from Tables 2–3, and they do not equal the naive quadrature of the per-epoch σRA values shown there.","section":"Table 4"},{"comment":"The flagged sequences are produced by flagging existing scans rather than by observing a genuinely shortened schedule, and the sMV flagged run keeps a small initialization segment from the original sequence; the text should more clearly distinguish these simulated overhead reductions from ones demonstrated by actual scheduling.","section":"Section 3; Figure 6"},{"comment":"The Gaia comparison is described only briefly; the manuscript should state whether a known systematic VLBI–Gaia frame offset is expected and how the ~140 µas discrepancy between MultiView and the Gaia prediction is interpreted relative to the claimed sub-10 µas accuracy.","section":"Section 3, Gaia comparison"},{"comment":"The analytical derivation in Appendix B assumes all calibrators are collinear, while the actual observation uses a two-dimensional calibrator arrangement; the text should note that the equations are illustrative of the reference-point mechanism rather than a full 2D error propagation.","section":"Appendix B"},{"comment":"There are minor typographical issues, including a double period after \"future work\" in Section 4.1 and the column header \"Sunres\" in Table 1, which should be written as \"S_unres\" or defined in the note.","section":"Section 4.1; Table 1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript has a useful algorithmic contribution and a real VLBA demonstration, but the central error claim is overstated relative to the statistics presented. In revision, the authors should either reanalyze the data with a properly propagated uncertainty budget or substantially soften the claims in the abstract and Section 3. The use of cMV as a benchmark is not itself disqualifying, but the statistical treatment must treat cMV as having finite uncertainty and shared systematics. I would be willing to review a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the iterative rotating phase-plane and the ternary-tree ambiguity resolution are genuinely new, the code and data are public, and the cMV comparison on two epochs is a reasonable first validation. But the abstract's '<10 µas error in RA' is overreach. Table 4 gives sMV minus cMV of -9 µas with a combined uncertainty near 42 µas, so the measurement is 9 ± 42, not a bound under 10. The per-epoch offsets are -59 and -68 µas, and the paper explains them as a constant reference-point offset (Appendix B) without independently measuring that offset or testing its constancy. So the central quantitative claim is weaker than stated.\n\nWhat is good: the algorithm is clearly described and differs from earlier MultiView work; it does not require all calibrators within each cycle, which is a real operational gain at higher frequencies. The automated phase-wrap detection is a useful engineering contribution. The comparison with cMV, PR, and Gaia is honest in showing where things agree and disagree, and the citation pattern is solid—the relevant MultiView and inverse MultiView literature is cited and the novelty is not oversold relative to it. Shipping code and data is genuinely helpful.\n\nSoft spots beyond the error claim: the benchmark is cMV, developed partly by the same group, so it is not independent; the Gaia comparison helps but is not used to anchor the sMV accuracy claim. The demonstration is one target across two epochs, and the scheduling-gain section is a simulation based on flagging, not a real observation. Free parameters (loss weight, tree depth, pruning thresholds, smoothing) are described but get no sensitivity analysis.\n\nThat said, the central idea is plausible and the evidence is consistent with sMV being comparable to cMV, not worse. The authors just cannot yet support the ten-microarcsecond headline. I would send this to review, with a request to reframe the claim and add a sensitivity and robustness analysis.","headline":"A plausible new MultiView variant with public code, but the headline <10 µas error is not supported by the formal uncertainties; worth refereeing with a requested revision.","tokens_in":19821,"tokens_out":1824,"would_cite":true,"duration_ms":20808,"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":"A serial MultiView calibration scheme achieves the astrometric accuracy of conventional MultiView, with RA error below 10 microarcseconds.","keywords":["radio astrometry","very long baseline interferometry","MultiView","serial MultiView","phase-referencing","atmospheric propagation errors","phase ambiguity correction","tropospheric turbulence"],"falsifier":"Take a third epoch of the same target and calibrators, run the flagged one-third-sampling sequence through sMV and cMV, and measure the sMV-minus-cMV offset in both coordinates; if the offset changes by more than the roughly 25 microarcsecond level across sessions despite unchanged catalog positions, the anchored-plane assumption is biased. A cleaner test is to inject a simulated non-planar turbulence screen with known curvature into the inversion and check whether the rotating-plane model recovers the target position or shifts it.","tokens_in":18859,"feed_emoji":"📡","tokens_out":7058,"duration_ms":69782,"temperature":0.7,"pith_summary":"This paper introduces serial MultiView (sMV), a way of calibrating very long baseline interferometry (VLBI) observations that corrects for atmospheric spatial-structure errors without requiring all calibrators to be observed within a single short cycle. Instead of fitting a fresh phase plane every cycle, sMV tracks a phase plane anchored at the primary calibrator and rotates it step by step using the whole time series of calibrator residual phases. The paper claims this preserves the astrometric accuracy of conventional MultiView while shortening observing cycles, and demonstrates on two-epoch 5 GHz observations of a radio star that the RA difference between sMV and conventional MultiView is below 10 microarcseconds, with comparable image quality even when secondary calibrator sampling is cut to a third. If right, sMV makes high-precision differential astrometry feasible at higher frequencies where atmospheric phase fluctuations change faster than a full MultiView cycle can sample.","feed_headline":"Serial MultiView matches MultiView at sub-10 microarcseconds","feed_subtitle":"New phase-plane rotation cuts observing cycles and automates phase-wrap correction for VLBI astrometry.","key_machinery":"The central object is the anchored phase plane: a plane in (RA, DEC, phase) space that always intersects the origin at the primary calibrator, so it has the form $z = ax + by$ rather than the free plane $z = ax + by + c$ used by conventional MultiView. The iteration rotates this plane at each calibrator scan by the minimal angle required to pass through the new residual-phase point, using the Euler-Rodrigues rotation formula; this yields a normal-vector time series that approximates the time-varying atmospheric spatial gradient. Phase ambiguity handling is a recursive ternary-tree search: at each step the algorithm explores future $+2\\pi$, $-2\\pi$, and $0$ wrap options up to a limited depth, prunes branches that exceed thresholds on plane inclination and rotation angular velocity, and picks the branch with the smallest loss defined as a weighted sum of accumulated rotation angle and deviation from the linearly predicted normal vector, marking a scan as an outlier when no branch survives. Smoothing via a Kalman filter during the iteration and a low-pass filter afterwards damps oscillation when calibrators do not lie on a common plane.","core_discovery":"The central claim is that serial MultiView, a new realization of the MultiView technique, attains the same astrometric accuracy as conventional MultiView while requiring a shorter observing cycle and removing the manual step of phase-ambiguity correction. The paper presents sMV as an iterative phase-plane tracker: instead of fitting a free plane to calibrator phases in each observing cycle, it rotates a plane that always passes through the primary calibrator's phase reference point, using the full time series of secondary calibrator residual phases. On two epochs of C-band Very Long Baseline Array observations of the radio star V1859 Ori, sMV produces images and positions nearly identical to conventional MultiView; relative to cMV across epochs, sMV's random error is 9 microarcseconds in RA and 25 microarcseconds in DEC, while phase-referencing alone shows errors of hundreds of microarcseconds. The paper also reports that cutting the secondary-calibrator sampling rate to one third of the full sequence yields nearly unchanged images and positions for both sMV and cMV, implying on-target time can be substantially increased. A systematic reference-point offset between sMV and cMV is identified and shown to be constant between epochs, so it does not affect parallax or proper-motion measurements.","pith_inferences":["If the anchored-plane model is taken literally, sMV's reference point is set by the primary calibrator's a priori position; the constant offset found against cMV could in principle be calibrated by observing the same target with both methods or by interleaving geodetic blocks, converting sMV into a fully absolute astrometric mode.","The recursive wrap-search machinery generalizes beyond MultiView: the same ternary-tree idea could be applied to other phase-referencing strategies or to geodetic VLBI group-delay ambiguity resolution.","A direct high-frequency test would be to run sMV at 8 or 22 GHz with a deliberately shortened cycle and compare against geodetic-block calibration; the paper's implied prediction is that the shorter sMV cycle yields better coherence than cMV at those frequencies.","The one-third-sampling result suggests an adaptive scheduling rule: use dense calibrator sampling when elevation is low or the atmosphere is turbulent, and sparse sampling at high elevation; the paper mentions this possibility but does not yet test it."],"forward_implications":["sMV matched conventional MultiView in image quality and astrometry on the test source, with RA random error below 10 microarcseconds and DEC error around 25 microarcseconds across two epochs.","Cutting secondary-calibrator sampling to one third of the full sequence still yielded nearly identical images and positions, implying MultiView observing can spend roughly 1.4 to 1.8 times more time on target.","Phase-ambiguity correction becomes automatic in sMV, removing the manual step that conventional MultiView currently requires for reliable astrometry.","Because sMV shortens the time between calibrator observations, MultiView becomes applicable at higher frequencies where the atmospheric coherence time is shorter than a full conventional MultiView cycle.","The systematic reference-point offset between sMV and conventional MultiView is constant between epochs, so parallax and proper-motion measurements are unaffected unless calibrator source structure changes."],"supporting_citations":[{"why":"Defines conventional MultiView and provides the phase-plane interpolation method that sMV compares against.","marker":"Rioja et al. 2017"},{"why":"Discusses phase-ambiguity problems in VLBI phase-referencing, motivating the automatic ambiguity detection.","marker":"Reid 2022"},{"why":"Supplies the Euler-Rodrigues rotation formula used to rotate the phase plane.","marker":"Dai 2015"},{"why":"Provides the Radio Fundamental Catalog used to select the calibrators with a priori positions.","marker":"Petrov & Kovalev 2025"},{"why":"Gaia DR3 supplies the independent stellar-motion prediction that the MultiView results are compared against.","marker":"Gaia Collaboration et al. 2023"},{"why":"Supplies the uncertainty estimation for the Gaia prediction used in the comparison.","marker":"Zhang et al. 2024"},{"why":"Introduces phase-referencing, the baseline technique that MultiView extends.","marker":"Lestrade et al. 1990"}],"fun_headline_variants":["Serial MultiView: same accuracy as MultiView in shorter cycles","New VLBI method cuts observing time while hitting <10 μas","Serial MultiView achieves <10 μas error with fewer calibrator cycles","Faster VLBI astrometry: serial MultiView keeps accuracy high","Serial MultiView reduces cycles and automates phase-wrap fixes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the atmospheric residual phase across the small patch of sky around the target is a flat plane anchored at the primary calibrator that rotates smoothly in time, with non-planar structure, calibrator position errors, and source structure small enough to ignore.","fun_headline_variants_meta":{"raw":{"variants":["Serial MultiView: same accuracy as MultiView in shorter cycles","New VLBI method cuts observing time while hitting <10 μas","Serial MultiView achieves <10 μas error with fewer calibrator cycles","Faster VLBI astrometry: serial MultiView keeps accuracy high","Serial MultiView reduces cycles and automates phase-wrap fixes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000274,"raw_usage":{"total_tokens":1716,"prompt_tokens":1099,"completion_tokens":617,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":715,"completion_tokens_details":{"reasoning_tokens":528}},"tokens_in":715,"tokens_out":617,"duration_ms":6785,"temperature":1.0,"reasoning_tokens":528,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T18:45:39.350081+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a third epoch of the same target and calibrators, run the flagged one-third-sampling sequence through sMV and cMV, and measure the sMV-minus-cMV offset in both coordinates; if the offset changes by more than the roughly 25 microarcsecond level across sessions despite unchanged catalog positions, the anchored-plane assumption is biased. A cleaner test is to inject a simulated non-planar turbulence screen with known curvature into the inversion and check whether the rotating-plane model recovers the target position or shifts it.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Discusses phase-ambiguity problems in VLBI phase-referencing, motivating the automatic ambiguity detection."}],"review_version":1}