{"id":"bf424c94-0b13-48e0-9eb6-2f8d34300fa8","arxiv_id":"2505.13832","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Partial cyclic deconvolution still recovers useful phase information and the dynamic wavefield power in pulsar scintillation data, and a new scintillation-bandwidth-based cyclic merit better ranks sources for the technique.","lead":"Radio pulses from pulsars arrive smeared and shimmering because interstellar gas bends the signals, and a signal-processing method called cyclic spectroscopy can recover some hidden structure even when a full recovery is impossible. The paper shows this partial recovery still improves pulsar timing and interstellar studies, and it proposes a new \"cyclic merit 2.0\" score to help observatories pick the best pulsars and frequency bands.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Partial-phase-retrieval claim rests on visual inspection only; no ground-truth or simulated validation distinguishes genuine ISM phase recovery from a S/N-dependent artifact of pycyc.","rationale":"The reader's strongest claim is that pulsar timing arrays can extract useful phase information from observations outside the full-deconvolution regime. The reader's weakest assumption is the cyclic merit 2.0 calibration: its definition is absent, the threshold is derived from a power-law fit on the same dataset, and error bars are missing from Fig. 18. I agree this is a genuine problem, and it alone justifies a conditional verdict. However, for the central scientific claim, a more load-bearing premise is that the wavefield products in Section 4 actually contain partial phase information about the ISM transfer function. The paper's evidence is qualitative; the one quantitative comparison (Fig. 9) is insensitive to pixel-level fidelity. There is no simulated ground truth and no independent phase measurement in a partial-deconvolution observation. The observed frequency trend in Fig. 7 is predicted by cyclic merit but also by pure S/N selection: a phase-randomized reconstruction of the same data would likely reproduce the visual appearance, with coherent phase where the amplitude is large and random phase where it is not. The paper itself acknowledges a related limitation (the real component resembling the dynamic spectrum and the imaginary component being small), which reinforces the need for a controlled test. I therefore recommend keeping the verdict CONDITIONAL, with an added condition requiring simulation-based validation of the phase-retrieval interpretation. This is a distinct concern from the reader's, hence partial agreement.","tokens_in":23243,"tokens_out":8699,"duration_ms":81026,"concrete_test":"Run an end-to-end simulation in the partial-deconvolution regime with a known transfer function. Choose a scattering screen with a prescribed secondary spectrum (e.g., a few discrete images, as in Fig. 10), synthesize the true dynamic wavefield H_true(ν,t), generate the corresponding cyclic spectra at the same channelization, subintegration time, and per-channel S/N as the GBT L-band observations (Table 1), and process them with the same pycyc pipeline. Then (1) compute the normalized coherence C(ν,t) = |⟨H_rec H_true^*⟩|² / (⟨|H_rec|²⟩⟨|H_true|²⟩) averaged over scintillation timescales, and compare it with the coherence obtained from a noise-only (phase-randomized amplitude-matched) reconstruction; (2) compute the pixel-level normalized residual between |H_rec|² and |H_true|² over the band.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim—that cyclic spectroscopy outside the full-deconvolution regime yields useful partial phase retrieval—is supported primarily by visual interpretation of recovered wavefields and wavefield powers (Sec. 4.2, Figs. 6–8). No comparison to a known transfer function is reported, and no end-to-end simulation is presented. In the partial-deconvolution regime the true H(ν,t) is unknown, so one cannot tell from these data whether the coherent phase bands in Fig. 7 are genuine partial recovery of the ISM transfer-function phase or merely regions of high S/N where the iterative deconvolution (pycyc) returns a stable but arbitrary phase. The trend 'more complete phase retrieval closer to the full-deconvolution regime' (Fig. 7) is also degenerate: at lower observing frequency both S/N per scintle and scattering delay rise, so the same trend would appear even if the recovered phase were pure noise weighted by cyclic merit. The companion claim of 'complete signal recovery' in the dynamic wavefield power (Fig. 8) is supported only by visual similarity and by power-law fits of scintillation bandwidth vs. frequency (Fig. 9); those fits are insensitive to pixel-level distortions (overall normalization, smearing, or S/N-dependent bias), so they do not establish that the wavefield power is a faithful image of the dynamic spectrum. Because the partial-phase-retrieval interpretation underlies all of the claimed benefits over Fourier spectroscopy (secondary wavefield fragments, phase-informed scintillometry), the central claim is underdetermined without a quantitative validation against a known truth.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper investigates whether cyclic spectroscopy retains benefits for pulsar observations outside the full-deconvolution regime, where the scintillation bandwidth exceeds the inverse pulse width and complete recovery of the ISM transfer function is not possible. The authors analyze baseband observations of PSR B1937+21 (Arecibo ASP and PUPPI; GBT VEGAS), PSR J1643−1224, and PSR J1903+0327 (GBT), compute the cyclic figure of merit mcyc (Eq. 5) and full-deconvolution regime boundaries (Eq. 3), and extend these estimates to the NANOGrav 12.5-year pulsar sample under GBT and DSA-2000 sensitivities (Figs. 2–4). Comparing data products across regimes, they report that the partial-deconvolution dynamic wavefield contains coherent structure indicative of incomplete but genuine phase retrieval (Figs. 6–7); that dynamic wavefield power recovers the scintillation pattern and yields scintillation-bandwidth power-law fits nearly identical to those from the dynamic spectrum (Figs. 8–9); and that partial recovery of the secondary wavefield is possible, with the θ−θ transform discussed as a complementary approach (Figs. 10–11). Motivated by the null result for PSR J1903+0327—in the full-deconvolution regime yet with no discernible phase recovery—the paper introduces 'cyclic merit 2.0', based on scintillation rather than observing bandwidth, and uses a power-law fit (Fig. 18) to translate the original mcyc ≫ 1 threshold into mcyc,2.0 ≫ 0.05.","tokens_in":23550,"tokens_out":16745,"duration_ms":136877,"significance":"The central claim—that partial cyclic deconvolution still yields useful phase information and that dynamic wavefield power recovers the scintillation pattern in the partial-deconvolution regime—is practically important, since most NANOGrav pulsars are observed outside the full-deconvolution regime. If it holds, pulsar timing arrays could justify adopting cyclic processing for essentially all observations, with real gains in scattering-delay estimation and ISM studies. The survey-style merit and regime-boundary calculations for the full NANOGrav sample, with explicitly stated receiver and sky-temperature assumptions, are a genuinely useful planning resource for the GBT and DSA-2000. The paper's strengths include its detailed data products, its willingness to report the J1903+0327 null result, its explicit acknowledgment that part of the phase-imaginary resemblance is a small-angle effect, and its fair treatment of the θ−θ transform's limitations. The significance is currently capped by the absence of ground-truth or simulated validation of the partial-phase-retrieval claim, and by the under-specified, self-calibrated definition of cyclic merit 2.0.","major_comments":[{"comment":"Cyclic merit 2.0 is never defined by an equation. The text states that it 'behaves identical to the original cyclic merit' but 'relies on scintillation bandwidth rather than observing bandwidth' and that 'S/N now relies on flux density over a single scintle rather than the width of the observing band.' Since the original mcyc in Eq. (5) inherits its S/N from the bandwidth-dependent estimate of Eq. (8), the definition of mcyc,2.0 is ambiguous as written: exactly which bandwidth (full observing band, PFB channel, or scintillation bandwidth) enters the S/N term, and is the effective-pulse-width factor unchanged? Without an explicit formula, the values in Figs. 16–18 and the headline threshold are not reproducible. Please provide the full formula for mcyc,2.0 and state all assumptions used to compute it.","section":"Section 5, definition of cyclic merit 2.0"},{"comment":"The decision threshold mcyc,2.0 ≫ 0.05 is obtained from a power-law fit between mcyc,2.0 and mcyc computed on the same NANOGrav sample with the same S/N model (Eq. 8) used to rank those same pulsars in Figs. 16 and 17. Because mcyc,2.0 is constructed from the same underlying quantities (scattering delay, S/N, pulse widths) as mcyc, the fit in Fig. 18 largely demonstrates self-correlation; it does not establish that mcyc,2.0 predicts deconvolution success on any independent outcome. No known-outcome test (e.g., the observations in Table 2 or a held-out set) is reported. The fit parameters, scatter, and residuals for Fig. 18 are not given, and the criterion as stated ('≫ 0.05') is non-operational: the paper does not state the mcyc,2.0 value corresponding to its own suggestion that mcyc ≳ 10 is sufficient. Given that the metric is motivated by the single J1903+0327 failure case, the 0.05 threshold should either be validated on independent outcomes or explicitly presented as a provisional calibration.","section":"Section 5, Fig. 18"},{"comment":"The central claim that genuine partial phase retrieval occurs rests on visual inspection of the recovered wavefields. In the partial-deconvolution regime the true H(ν,t) is unknown, so the coherent patches in Fig. 6 and the banded phase structure in Fig. 7 cannot be distinguished from regions of high S/N in which the iterative pycyc deconvolution converges to a stable but arbitrary phase. The paper itself notes in Section 4.2 that the phase and imaginary components resemble each other mainly because many recovered phases are near zero, which weakens the claim that non-trivial phase information is being recovered, and Section 4.3 reports non-physical power at negative delays in the partial-regime secondary wavefield. Moreover, the trend of more complete phase retrieval near the full-deconvolution regime (Fig. 7) is degenerate: observing frequency, S/N per scintle, and scattering delay vary together across the sub-bands, so the same trend would be expected even if the recovered phase were S/N-weighted noise. No simulation with an injected phase screen or quantitative phase-error measurement against the full-deconvolution control observation is provided. An end-to-end simulation with a known transfer function is needed to support the headline claim of partial phase retrieval, or the claim should be substantially softened.","section":"Section 4.2, Figs. 6–8"},{"comment":"The claim that the dynamic wavefield power shows 'complete signal recovery' in the partial-deconvolution regime is supported only by visual similarity to the dynamic spectrum (Fig. 8) and by the power-law fits of scintillation bandwidth versus frequency in Fig. 9. Such fits aggregate many scintles and are insensitive to pixel-level errors such as normalization offsets, smearing, or S/N-dependent bias, so they establish similar aggregate scintillation statistics but not faithful image recovery. Please either temper the 'complete signal recovery' wording or add a quantitative fidelity check, such as a cross-correlation or residual statistics between the dynamic spectrum and the wavefield power over the same time-frequency region.","section":"Section 4.2, Fig. 9"}],"minor_comments":[{"comment":"The 'a' note to the PSR J1903+0327 Δνd entry says the value was 'estimated from pulse broadening-inferred scattering delay'; please state explicitly how the conversion was made (presumably via Eq. 6 with C1 = 0.957) so the reader can propagate the uncertainty correctly.","section":"Table 2"},{"comment":"The pycyc iterative deconvolution is treated as a black box; a sentence on the update rule, initialization, number of iterations, and stopping criterion would materially aid reproducibility, particularly since the partial-regime results are interpreted as genuine phase recovery.","section":"Section 3.2"},{"comment":"The approximation Σk k² a_k ≈ (P/We)^{3/2} is stated without derivation; please cite the derivation in Turner et al. (2023) or give a brief justification of the step.","section":"Section 3.4.1, Eq. (7)"},{"comment":"The full- and partial-regime panels use different color scales (2–10 versus 1–6 in arbitrary units); a shared normalization or an explicit statement of the scales would make the claimed 'complete signal recovery' easier for the reader to judge.","section":"Fig. 8"},{"comment":"The NANOGrav survey calculations depend on several assumed constants (C1 = 0.957, filling factor 0.2, and S/N bandwidths of 5/50/100 MHz); a brief sensitivity check of the merit rankings over the stated allowed ranges (e.g., C1 ∈ [0.6, 1.5]) would clarify how robust the conclusions in Figs. 2–4 are.","section":"Section 4.1.2, Figs. 2–4"}],"recommendation":"major_revision","confidential_remarks":"I want to flag two things beyond the formal comments. First, the evidentiary standard for the headline claim is the main risk: the partial-phase-retrieval conclusion is currently an interpretation of iterative-deconvolution output over a handful of observations, and I would want at least one simulation-based validation before this claim becomes citable. Second, the cyclic merit 2.0 section reads like a separate methods paper bolted onto an empirical study; if the metric is not fully specified and validated, I would recommend that the journal require either proper validation or removal to a follow-up paper. The manuscript fits the journal's scope, the data products are detailed, and the J1903+0327 null result plus the θ−θ comparison are honest and useful; all of the concerns above are addressable in a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nQuick take: this paper is worth engaging with, but it needs revision before I would trust its headline metric. What is genuinely new is the empirical demonstration that cyclic spectroscopy in the partial-deconvolution regime still yields informative wavefield products—the dynamic wavefield power tracks the dynamic spectrum, and the secondary wavefield shows partial structure. The NANOGrav survey of full-deconvolution boundaries and cyclic merits is a practical contribution, and the discussion of PSR J1903+0327 is honest and useful.\n\nThe main soft spot is cyclic merit 2.0. The definition is never written down; the paper says it 'relies on scintillation bandwidth rather than observing bandwidth' but gives no formula. The threshold mcyc,2.0 > 0.05 comes from a power-law fit between mcyc and mcyc,2.0 computed on the same NANOGrav sample, and Fig. 18 has no error bars. That is calibration, not independent validation. The source rankings in Figs. 16 and 17 inherit that uncertainty.\n\nThere is also a load-bearing evidential gap in the central empirical claim. The partial-phase-retrieval interpretation in Fig. 7 rests on visual inspection of recovered phases. No end-to-end simulation with a known transfer function is presented, and the observed trend—more complete phase retrieval closer to the full-deconvolution regime—is degenerate with S/N per scintle, since both scattering delay and S/N rise at lower frequency. The authors note the resemblance of the real part to the dynamic spectrum, but that resemblance could also arise from a S/N-dependent artifact of the pycyc deconvolution. I do not think this kills the paper; the dynamic wavefield power comparison in Fig. 8 is suggestive, and the scintillation bandwidth power-law fits in Fig. 9 are consistent. But the partial-recovery interpretation needs a quantitative validation against synthetic data before it is solid.\n\nMinor points: no code or data released, which makes the calibration hard to check; the self-citations are appropriate given the continuity of the project; the prose is clear.\n\nBottom line: a solid, honest engineering survey with one under-supported metric. It deserves peer review, but the referees should push for a real definition of cyclic merit 2.0, error bars on the calibration, and at least one simulation validating partial phase retrieval.","headline":"Worth engaging: the empirical partial-deconvolution demonstration is useful, but cyclic merit 2.0 is a calibrated fit rather than an independent predictor, and the phase-retrieval claim needs simulation support before it is trusted.","tokens_in":24145,"tokens_out":2597,"would_cite":true,"duration_ms":21622,"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":"Cyclic spectroscopy remains useful when full deconvolution is impossible: partial phase retrieval still recovers the scintillation pattern and enables scattering-delay measurements.","keywords":["cyclic spectroscopy","pulsar timing arrays","interstellar medium scintillation","dynamic wavefield","scattering delay","cyclic merit","pulsar timing","radio astronomy methods"],"falsifier":"Compute cyclic merit 2.0 for a large sample and observe the predicted-best and predicted-worst pulsars with sensitive telescopes; if deconvolution success rates do not separate according to the 0.05 threshold, the calibration is falsified. Alternatively, on a single partial-deconvolution observation, compare the dynamic wavefield power pixel-by-pixel with the simultaneously measured dynamic spectrum: if some scintles appear in one data product and not the other, the paper's 'complete signal recovery' claim for the wavefield power would be falsified.","tokens_in":23045,"feed_emoji":"📡","tokens_out":8946,"duration_ms":81324,"temperature":0.7,"pith_summary":"Cyclic spectroscopy is usually judged by whether it can fully deconvolve the interstellar-medium transfer function from a pulsar's pulse. This paper argues that even when full deconvolution is out of reach, the technique still beats conventional spectroscopy: the dynamic wavefield power recovers the complete scintillation pattern, partial phase information appears in the wavefield, and fine frequency resolution makes scattering-delay and scintillation-bandwidth measurements possible for highly scattered pulsars. The authors demonstrate this on baseband observations of three millisecond pulsars, quantify how recovered phase improves as observing frequency approaches the full-deconvolution boundary, and show that useful signal can be extracted before complete phase retrieval is achieved. They also introduce a revised figure of merit, cyclic merit 2.0, which ranks pulsars by per-scintle signal-to-noise rather than observing bandwidth, and calibrate a threshold for expecting deconvolution success. If correct, pulsar timing arrays can obtain interstellar-medium and timing information from observations that fall short of complete phase recovery.","feed_headline":"Partial phase recovery still yields full scintillation patterns","feed_subtitle":"Pulsar timing arrays can measure scattering delays even when full phase retrieval fails.","key_machinery":"The load-bearing object is the cyclic spectrum $S_E(\\nu,\\alpha_k)=\\langle E(\\nu+\\alpha_k/2)E^*(\\nu-\\alpha_k/2)\\rangle$, through which the periodic pulsar signal is resampled at harmonics $\\alpha_k=k/P$ of the pulse period, bypassing the usual time-frequency (Gabor) resolution limit. In the interstellar-medium case the cyclic spectrum factors into the intrinsic pulse transform times a bilinear product of the transfer function $H$, so the transfer function, and hence the dynamic wavefield $H(\\nu,t)$, can be recovered by iterative deconvolution. Full deconvolution requires the scintillation bandwidth to satisfy $\\Delta\\nu_d\\lesssim 1/W_{10}$, where $W_{10}$ is the pulse width at 10% of maximum. In the partial-deconvolution regime the same machinery still produces a dynamic wavefield whose squared modulus is the dynamic wavefield power and whose 2D Fourier transform is the secondary wavefield. The paper's new diagnostic, cyclic merit 2.0, replaces observing bandwidth with scintillation bandwidth in the signal-to-noise estimate, which changes which pulsars appear promising.","core_discovery":"The paper's central claim is that complete phase retrieval is not the dividing line for the usefulness of cyclic spectroscopy. Using observations of three millisecond pulsars in both the full- and partial-deconvolution regimes, it shows that the squared modulus of a partially recovered dynamic wavefield, $|H(\\nu,t)|^2$, reproduces the same scintillation pattern as the dynamic spectrum and yields nearly identical scintillation-bandwidth power-law fits, despite the wavefield phase being only partly recovered. Phase retrieval is found to be continuous: closer to the full-deconvolution boundary, more of the wavefield phase survives as signal structure rather than noise. The paper therefore concludes that partial-deconvolution data should be treated as incomplete rather than incorrect, and that cyclic processing is preferable to Fourier spectroscopy in all regimes when baseband data or a cyclic backend are available. It further claims that a modest cyclic merit can coexist with no observable deconvolution for low-flux, highly scattered pulsars, motivating a new predictor, cyclic merit 2.0, in which signal-to-noise is evaluated over a single scintillation bandwidth rather than the observing band; a power-law calibration maps the old threshold $m_{\\rm cyc}\\gg 1$ to $m_{\\rm cyc,2.0}\\gg 0.05$.","pith_inferences":["The claim that dynamic wavefield power fully mirrors the dynamic spectrum in the partial regime could be quantified with a pixel-by-pixel residual statistic; testing this across many epochs would delimit exactly how complete the 'complete signal recovery' actually is.","A natural prospective test is to pre-select pulsars predicted above the cyclic merit 2.0 threshold of 0.05, observe them, and compare deconvolution success rates against the old merit; this would validate or refute the transferability of the power-law calibration.","The partial-recovery results suggest a hybrid pipeline the paper advocates but does not test: use cyclic spectroscopy to anchor the wavefield amplitude, and apply the $\\theta-\\theta$ transform only where its single-screen, anisotropic-scattering assumption holds, reserving cyclic partial phase for multi-screen lines of sight.","Because the full-deconvolution boundary fluctuates with scintillation bandwidth from epoch to epoch, scheduling decisions could treat cyclic merit as a live observing metric rather than a static catalog value."],"forward_implications":["Pulsar timing arrays can measure scattering delays for highly scattered pulsars even when full cyclic deconvolution fails, improving noise models and gravitational-wave sensitivity.","Scintillation-bandwidth power-law studies of the interstellar medium can be done from dynamic wavefield power instead of dynamic spectra, with nearly identical measurements and extra harmonic information.","Observing strategies should favor lower frequencies and highly scattered pulsars with narrow scintles, because phase recovery improves continuously toward the full-deconvolution boundary.","Cyclic merit 2.0, once calibrated, lets observers rank pulsars by per-scintle signal-to-noise, exposing sources whose apparent merit is inflated by large scattering delay but whose faint flux per scintle makes deconvolution impractical.","The upcoming availability of a cyclic spectroscopy backend would let observers obtain these benefits without retaining baseband data."],"supporting_citations":[{"why":"Introduces pulsar cyclic spectroscopy and the idea of bypassing the Gabor limit to recover phase information.","marker":"Demorest 2011"},{"why":"Establishes the factorization of the cyclic spectrum into intrinsic pulse and interstellar-medium transfer function, and identifies the dynamic spectrum as the 0th harmonic.","marker":"Walker et al. 2013"},{"why":"Defines the cyclic figure of merit and the full-deconvolution regime criterion used throughout the paper.","marker":"Dolch et al. 2021"},{"why":"Provides the transfer-function reconstruction approach and the approximation of the cyclic-merit radical.","marker":"Turner et al. 2023"},{"why":"Supplies the dynamic and secondary wavefield data products and the scintillometry context the paper builds on.","marker":"Baker et al. 2022"},{"why":"Describes the theta-theta transform whose assumptions the paper contrasts with cyclic spectroscopy.","marker":"Sprenger et al. 2021"},{"why":"Provides the NE2001 electron-density model used to predict scintillation parameters when measurements are missing.","marker":"Cordes & Lazio 2002"},{"why":"Supplies the 9-year pulsar timing survey scintillation measurements that anchor the merit calculations.","marker":"Levin et al. 2016"},{"why":"Supplies the 12.5-year scintillation measurements and the scattering-delay resolution limit that motivate finer channelization.","marker":"Turner et al. 2021"},{"why":"Provides the directly measured scattering delay for PSR J1903+0327 used to interpret the failure of deconvolution despite full-regime conditions.","marker":"Geiger et al. 2025"}],"fun_headline_variants":["Partial wavefield recovery preserves scintillation statistics","Cyclic merit 2.0 predicts pulsar scattering delays","Incomplete phase retrieval still reveals pulsar structure","New metric: cyclic merit over scintillation bandwidth"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The prediction of which pulsars will benefit rests on a power-law calibration between the old and new cyclic merits computed from one pulsar sample; if that mapping does not transfer to other telescopes, observing bands, or sources, the headline threshold of 0.05 and the source rankings would not hold.","fun_headline_variants_meta":{"raw":{"variants":["Partial wavefield recovery preserves scintillation statistics","Cyclic merit 2.0 predicts pulsar scattering delays","Incomplete phase retrieval still reveals pulsar structure","New metric: cyclic merit over scintillation bandwidth"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000213,"raw_usage":{"total_tokens":1488,"prompt_tokens":1079,"completion_tokens":409,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":695,"completion_tokens_details":{"reasoning_tokens":348}},"tokens_in":695,"tokens_out":409,"duration_ms":4235,"temperature":1.0,"reasoning_tokens":348,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:09:25.465872+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute cyclic merit 2.0 for a large sample and observe the predicted-best and predicted-worst pulsars with sensitive telescopes; if deconvolution success rates do not separate according to the 0.05 threshold, the calibration is falsified. Alternatively, on a single partial-deconvolution observation, compare the dynamic wavefield power pixel-by-pixel with the simultaneously measured dynamic spectrum: if some scintles appear in one data product and not the other, the paper's 'complete signal recovery' claim for the wavefield power would be falsified.","supporting_citations":[],"review_version":1}