{"id":"8de1592d-fac3-4902-925d-a6ad6c2cb841","arxiv_id":"1908.06651","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The new NAOMI adaptive optics system improves single-mode fiber injection by +60% for GRAVITY and +130% for PIONIER and stabilizes fringe tracking at the VLTI Auxiliary Telescopes.","lead":"NAOMI is a new adaptive optics system installed on the four 1.8 m Auxiliary Telescopes of the Very Large Telescope Interferometer. The paper reports that it roughly doubles the light injected into single-mode instruments and stabilizes fringe tracking in poor seeing. If confirmed, it makes the array more usable and extends its limiting magnitudes.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Sec. 4.2's +60%/+130% transmission gains are computed from before/after QC1 histograms without normalizing seeing; the paper itself notes the NAOMI epoch had better median seeing. The qualitative AO benefit is supported by Fig.","rationale":"The reader's ACCEPT verdict with moderate confidence is appropriate. The central claim is that deploying NAOMI improves single-mode coupling and stabilizes fringe tracking. The evidence is convergent rather than resting on one controlled experiment: test-bench transfer functions and Strehl-versus-flux curves (Sec. 3), NCPA measurements to IRIS (Fig. 13), piston conversion measurements (Fig. 12), on-sky injection stability histograms (Figs. 14-15), residual-versus-stability correlation (Fig. 16), the seeing/wind resilience comparison (Fig. 18), and the transmission comparison (Fig. 17). The weakest link is the transmission comparison in Sec. 4.2, because the two epochs are not normalized for atmospheric and operational conditions. The paper itself acknowledges that the NAOMI period may have had better median seeing, which shows that the comparison is sensitive to exactly this uncontrolled variable. However, Fig. 18 provides a seeing-stratified comparison showing the same qualitative benefit, so the central conclusion does not stand or fall on the unnormalized percentages. The concrete test would settle whether the +60%/+130% numbers are robust or epoch-specific. If they are not robust, the main scientific narrative of improved injection and more stable fringe tracking remains supported, only with less precise quantitative claims; therefore the verdict should remain unchanged.","tokens_in":15846,"tokens_out":5073,"duration_ms":57542,"concrete_test":"Recompute Fig. 17 from QC1 with each transmission record tagged by contemporaneous Paranal DIMM seeing and wind speed. Restrict both epochs to identical conditions (e.g., seeing 0.8-1.0 arcsec, wind 3-9 m/s, airmass <1.5, R<12) and to the same target list if possible; regenerate the cumulative distributions and derive bootstrap confidence intervals for the median-transmission ratio. If the stratified gains remain close to +60% and +130%, the concern is resolved; if they collapse below +30%, the headline numbers are atmospheric artifacts rather than a pure AO effect.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing assumption is that Sec. 4.2's before/after comparison isolates the NAOMI hardware change. The STRAP epoch (2018 Jan 1 to Sep 6) and NAOMI epoch (2018 Nov 17 onward, excluding a commissioning window) are compared as cumulative transmission histograms with no conditioning on the joint distribution of seeing, wind, airmass, target R magnitude, or instrument calibration. The paper itself flags that the NAOMI period may have had better median effective seeing (0.9\" vs 1.1\") to explain why the observed gain is below test-bench predictions. That admission shows the metric is sensitive to atmospheric sampling, and it cuts both ways: if NAOMI was used in relatively good conditions, the percentage gain attributed to AO may be underestimated or overestimated depending on how seeing improves STRAP relative to NAOMI; without renormalization the sign is undetermined. No per-point error bars are given on the cumulative distributions, so even the precision of the +60%/+130% values is unclear. This does not sink the paper's central qualitative claim, because Fig. 18 compares NAOMI and STRAP at fixed seeing/wind and shows the same pattern, and the improvement direction is independently predicted by Tatulli et al. (2010). It does mean the two headline percentages in Sec. 4.2 should be read as epoch-dependent descriptive statistics rather than controlled measurements of AO gain.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"Woillez et al. present NAOMI, a new adaptive optics module deployed on the four 1.8 m Auxiliary Telescopes of the VLTI, replacing the previous STRAP tip-tilt system. The paper describes the system design (ALPAO DM241 deformable mirror, Shack-Hartmann wavefront sensor, SPARTA-light real-time controller), the acquisition and calibration procedures, and the laboratory and on-sky performance. It reports transfer-function agreement with a delay model built from independently measured delay contributions, Strehl and residual-tilt performance versus WFS flux, low piston conversion measured with the GRAVITY fringe tracker, and NCPA below about 100 nm. The paper then evaluates the impact on VLTI instruments: a new P5%/P95% injection-stability metric shows reduced flux dropouts and lower, more stable fringe-tracking residuals for GRAVITY, and before/after QC1 comparisons show +60% and +130% average transmission gains for GRAVITY (K band) and PIONIER (H band), respectively. The authors also report improved resilience to degraded seeing and low-wind dome seeing, increased limiting magnitudes, and a possible extension to J band.","tokens_in":16109,"tokens_out":6536,"duration_ms":67026,"significance":"If the results hold, NAOMI is a substantial upgrade for VLTI operations: it roughly doubles or more than doubles the usable starlight for the main near-infrared single-mode instruments, stabilizes fringe tracking, and opens previously unusable atmospheric conditions to scientific observations. The paper's strengths are its multi-pronged evidence: the transfer-function model is validated against independently measured delays; the laboratory Strehl predictions in Fig. 11 are independent of the on-sky data; the piston-conversion measurements were made with an external fringe tracker; NCPA were measured all the way to the IRIS sensor; and the qualitative injection-stability improvement is corroborated by Fig. 18, which conditions on seeing and wind. The main quantitative transmission claim in Sec. 4.2, however, is an epoch comparison that is not normalized for atmospheric conditions, and the paper itself flags a better median seeing in the NAOMI period. This does not invalidate the central qualitative conclusion, but it means the +60% and +130% figures must be presented with appropriate caveats or rederived from a conditioned comparison.","major_comments":[{"comment":"The +60% and +130% transmission gains are presented as the effect of NAOMI, but they are computed from cumulative QC1 histograms over two epochs without normalizing for the joint distribution of seeing, wind, airmass, target R magnitude, or instrument calibration. The paper itself invokes a better median effective seeing in the NAOMI period (0.9 arcsec versus 1.1 arcsec) to explain why the observed gains fall below the test-bench predictions, which is an explicit admission that the metric is sensitive to atmospheric sampling; without renormalization the sign of the resulting bias on the quoted percentages is undetermined. No per-point error bars are given for the cumulative distributions, so the precision of the +60% and +130% values is unclear. I recommend either renormalizing both epochs to a common seeing distribution or clearly labeling the percentages as epoch-dependent descriptive statistics, and moving the quantitative AO-benefit statement to the partially seeing/wind-conditioned comparison in Fig. 18.","section":"Sec. 4.2, Fig. 17"},{"comment":"The claim that the transmission improvement holds 'on the condition that the R-band magnitude remains in the high Strehl regime (R < 12 mag)' is not directly supported by the shown data, because Fig. 17 does not indicate whether the QC1 histograms were restricted to R < 12. Please state the exact selection criteria applied to the GRAVITY and PIONIER samples, and if the histograms include fainter targets, show the R < 12 subset or quantify the resulting selection effect.","section":"Sec. 4.2, Fig. 17"}],"minor_comments":[{"comment":"The annotation of Eq. (1) is typeset in a garbled way, with the delay contributions and the loop-frequency labels misaligned; please present this as a table or with clearer alignment so that the sum T = Tc + TFT/2 + TRO + TRTC + TDM is readable.","section":"Sec. 3.1, Eq. (1)"},{"comment":"The figure caption states that 20% of open-dome nights with wind speeds above 9 m/s are excluded from the fringe-tracking-residual comparison; this exclusion should also be stated in the main text where the residual correlation is discussed.","section":"Sec. 4.1, Fig. 16"},{"comment":"The statement that the limiting magnitudes of PIONIER and GRAVITY are increased by +1 mag (to K = 9 mag and H = 9 mag) appears to be a derived estimate rather than a directly measured quantity; please clarify whether this is inferred from the transmission and Strehl-stability results and what uncertainty it carries.","section":"Sec. 4.2, conclusions"},{"comment":"The 'effective seeing' quantity used in Fig. 18 should be defined in the text; the introduction's footnote defines AT-level seeing, but the figure's x-axis label alone is not sufficient for a reader to reproduce the selection.","section":"Sec. 4.3, Fig. 18"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a solid instrumentation paper with convincing qualitative on-sky evidence and careful laboratory characterization. The only substantive issue is the uncontrolled before/after comparison in Sec. 4.2, which affects the headline transmission numbers. I would advise the editor that the paper can be accepted after the authors either renormalize the QC1 comparison or explicitly reframe the percentages as epoch-dependent statistics; the qualitative benefit of NAOMI is well supported by Fig. 18 and the theoretical predictions of Tatulli et al. (2010)."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First reaction: the paper's main claim survives the stress-test. NAOMI clearly improves injection stability and fringe tracking on the VLTI ATs. The Sec. 4.2 before/after comparison is the one real soft spot, but it does not sink the paper because Fig. 18 and the Tatulli et al. prediction independently support the qualitative direction.\n\nWhat is actually new: first high-order AO system on the ATs. Several clever engineering details are worth knowing about: piston-free modes defined over the full DM aperture to avoid exciting a 500 Hz membrane resonance; on-sky DM gain calibration by injecting coma modulation into the closed loop; the pupil registration loop based on sub-aperture flux balance; and the P5%/P95% metric for injection stability, which is a simple, sensible statistic for fringe-tracking robustness. The transfer function measurements match a delay model built from independently measured components, which is good practice. Lab Strehl and residual tilt are tabulated as a function of flux and R magnitude, and on-sky piston conversion factors measured with the GRAVITY fringe tracker are consistent with an apodization model. NCPA to the IRIS sensor are presented. For an instrumentation paper, this is a thorough, honest performance characterization. The paper also openly reports equipment issues (creep, CIC higher than spec, DM gain decorrelation) and the workarounds, which is refreshing.\n\nThe soft spot is exactly where the stress-test points: Sec. 4.2 compares cumulative transmission histograms of STRAP (Jan-Sep 2018) and NAOMI (Nov 2018 onward) without conditioning on seeing, wind, airmass, or target magnitude. The paper itself notes the NAOMI period may have had better median effective seeing (0.9\" vs 1.1\") to explain why the observed gain is below test-bench predictions. That means the headline +60%/+130% figures are epoch-dependent descriptive statistics, not controlled measurements of AO gain. There are no per-point error bars on the histograms either. This is a real limitation, but it is not fatal: Fig. 18 shows the same improvement at fixed seeing and wind bins, and the direction of the effect was predicted by Tatulli et al. (2010). So the qualitative conclusion—AO improves injection and fringe-tracking stability and extends usable conditions—is robust. The size of the gain in a properly controlled experiment might differ from the quoted percentages, and it would be good to see a reanalysis with matched atmospheric conditions or a multivariate regression.\n\nBottom line: this paper is for anyone using or building VLTI instrumentation, and for the AO-for-interferometry community. It deserves a serious referee and publication; my own verdict would be accept with a request to reframe the Sec. 4.2 percentages as descriptive and, if feasible, add a seeing-matched comparison. If I were the editor, I would send it to peer review.","headline":"A solid, honest AO instrumentation paper whose qualitative results hold up; the headline Sec. 4.2 transmission gains are epoch-dependent descriptive statistics, not controlled measurements.","tokens_in":17041,"tokens_out":2514,"would_cite":true,"duration_ms":23132,"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":"Equipping the VLTI's 1.8 m Auxiliary Telescopes with a 14-mode adaptive optics system, NAOMI, raises average starlight injected into the K-band instrument GRAVITY by +60% and into the H-band instrument PIONIER by +130%, and makes fringe…","keywords":["adaptive optics","optical interferometry","fringe tracking","single-mode fibre coupling","VLTI Auxiliary Telescopes","NAOMI","Strehl ratio","injection stability"],"falsifier":"Recompute the GRAVITY and PIONIER transmission gain using only observations from both epochs within the same narrow effective-seeing bin, for example 0.8 to 1.0 arcsec, with the same $R<12$ mag target set; if the +60% and +130% advantages shrink or disappear inside matched seeing bins, the headline improvement is an epoch effect rather than an AO effect.","tokens_in":15652,"feed_emoji":"🔭","tokens_out":16902,"duration_ms":154472,"temperature":0.7,"pith_summary":"This paper reports that replacing the tip-tilt-only stabilisation of the 1.8 m Auxiliary Telescopes of the VLTI (the Very Large Telescope Interferometer) with the dedicated adaptive optics system NAOMI substantially increases the starlight delivered to the interferometer's fibre-fed instruments. In the bright-star regime, defined by an R-band magnitude $R<12$, the average injection into the K-band instrument GRAVITY rises by +60% and into the H-band instrument PIONIER by +130%. The gain matters because these fibre-fed instruments accept light through a single optical mode, so their throughput tracks the instantaneous Strehl ratio, the fraction of corrected starlight that stays in the diffraction peak. A stable high Strehl suppresses the flux dropouts that previously broke fringe tracking and forced the array to wait for excellent seeing. With NAOMI, fringe-tracking residuals stay near the 80–150 nm range and the array keeps working through degraded seeing and low-wind dome-seeing nights.","feed_headline":"Adaptive optics doubles starlight for VLTI's small telescopes","feed_subtitle":"A new adaptive-optics system lifts starlight delivery by 60% to 130% and stabilizes fringe tracking.","key_machinery":"The load-bearing element is the NAOMI correction chain together with the metric used to quantify its benefit. The deformable mirror is a 241-actuator device with about 145 actuators inside the 28 mm pupil, installed in the telescope's coudé train; the wavefront sensor is a 12-sub-aperture Shack–Hartmann sensor on an electron-multiplying CCD (EMCCD); and the real-time controller reconstructs 14 Zernike modes, the standard optical aberration basis, at loop rates of 500 to 50 Hz. Two design choices carry the argument: the controlled modes are made piston-free (no net phase shift) over the full mirror aperture, which suppresses a 500 Hz membrane resonance, and a pupil-registration loop keeps the deformable mirror aligned with the wavefront sensor as the telescope rotates. The quantity that connects the hardware to the science is P5%/P95%, the ratio of low-percentile to high-percentile single-mode injection; this is the metric that predicts whether the fringe tracker suffers flux dropouts.","core_discovery":"On the paper's own terms, the central discovery is that a modest high-order AO correction delivered at the telescope converts directly into scientifically usable light for an interferometer. The NAOMI system controls 14 Zernike modes at loop rates of 500, 100, or 50 Hz depending on target brightness, raising the H-band Strehl ratio from roughly 8% to roughly 40% in median seeing. The observable signature is the injection-stability metric P5%/P95% — the ratio of the single-mode flux at the 5th percentile to that at the 95th percentile — which climbs from a median of 0.22 with the old tip-tilt system to 0.58 with NAOMI on the GRAVITY fringe tracker. Consequently the average coupling efficiency increases by +60% in K band and +130% in H band, the limiting magnitudes of GRAVITY and PIONIER improve by one magnitude, and the array becomes usable in seeing conditions where it previously was not.","pith_inferences":["A matched-seeing re-analysis of the same pipeline data, comparing the two epochs only within identical effective-seeing bins, would isolate the AO gain from the epoch-to-epoch seeing difference; until that is done, the exact size of the +60% and +130% figures carries some epoch-dependent uncertainty.","The same coupling physics predicts that a fibre-fed instrument at even shorter wavelengths would show larger relative gains than PIONIER, because partial wavefront correction degrades single-mode coupling more strongly at shorter wavelengths.","Because the $R<12$ mag constraint is a colour-dependent limit, the AO benefit is largest for red targets; this suggests the observable population of the VLTI shifts further toward cool stars than the paper's aggregate numbers make explicit.","The P5%/P95% metric could serve as a standard commissioning benchmark for any adaptive optics system feeding a single-mode interferometer, since it directly maps onto fringe-tracking residual behaviour."],"forward_implications":["The one-magnitude gain in limiting magnitude puts GRAVITY on-axis at $K=9$ mag and PIONIER at $H=9$ mag on the Auxiliary Telescopes, opening fainter science targets to the array.","The GRAVITY fringe tracker can measure phase continuously rather than in bright intervals, which is the precondition for using it to co-phase the mid-infrared MATISSE instrument.","The array keeps its injection stability in effective seeing up to about 1.5 to 1.6 arcsec and in low-wind nights with dome seeing; the paper estimates that an earlier 1.4 arcsec AT-level seeing cutoff corresponds to roughly +15% usable time.","Median-seeing J-band Strehl would rise from about 8% to about 40%, making a J-band extension attractive and adding about +25% angular resolution.","Corrected telescope beams also make non-spatially-filtered instrument concepts worth investigating, potentially enabling higher throughput than single-mode fibre coupling."],"supporting_citations":[{"why":"It supplies the theoretical prediction that higher-order AO improves single-mode coupling under partial correction, which is the model the paper's flux-dropout and transmission measurements are compared against.","marker":"Tatulli et al. 2010"},{"why":"It establishes the criterion (pupil size of about three Fried parameters) showing that tip-tilt correction is cost-optimal only around L band, which is the reason the old STRAP system was inadequate for near-infrared fringe tracking.","marker":"Keen et al. 2001"},{"why":"It describes the GRAVITY K-band instrument whose archived pipeline data provide the transmission and fringe-tracker flux statistics used in the comparison.","marker":"GRAVITY Collaboration et al. 2017"},{"why":"It describes the PIONIER H-band instrument whose transmission statistics support the +130% gain claim.","marker":"Le Bouquin et al. 2011"},{"why":"It documents the GRAVITY fringe tracker and its phase-tracking S/N limit, which the paper uses to interpret the limiting-magnitude and residual improvements.","marker":"Lacour et al. 2019"},{"why":"It identifies low-wind dome-seeing as the failure mode that made otherwise excellent nights unusable, defining the resilience condition that NAOMI is shown to overcome.","marker":"Woillez et al. 2018"},{"why":"It describes the SPARTA-light real-time controller that performs the 500–50 Hz modal reconstruction central to the NAOMI loop.","marker":"Suárez Valles et al. 2012"},{"why":"It describes the Auxiliary Telescopes and their original tip-tilt-only stabilisation, the baseline against which NAOMI is measured.","marker":"Koehler et al. 2002"}],"fun_headline_variants":["NAOMI: AO boosts VLTI aux telescope throughput by 60–130%","VLTI small scopes get adaptive optics: fringe tracking stabilizes","NAOMI lifts VLTI coupling: injection ratio jumps 0.22→0.58","Adaptive optics on VLTIs: Strehl rises from 8% to 40%","NAOMI makes VLTI auxiliaries work in worse seeing"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The before/after transmission comparison treats the old tip-tilt system (STRAP) era and the NAOMI era as directly comparable, even though the two periods have different median seeing (about 1.1 arcsec versus 0.9 arcsec), and the paper does not normalize the quoted gains for the seeing distribution.","fun_headline_variants_meta":{"raw":{"variants":["NAOMI: AO boosts VLTI aux telescope throughput by 60–130%","VLTI small scopes get adaptive optics: fringe tracking stabilizes","NAOMI lifts VLTI coupling: injection ratio jumps 0.22→0.58","Adaptive optics on VLTIs: Strehl rises from 8% to 40%","NAOMI makes VLTI auxiliaries work in worse seeing"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000872,"raw_usage":{"total_tokens":3728,"prompt_tokens":853,"completion_tokens":2875,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":469,"completion_tokens_details":{"reasoning_tokens":2768}},"tokens_in":469,"tokens_out":2875,"duration_ms":22290,"temperature":1.0,"reasoning_tokens":2768,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:37:31.133758+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the GRAVITY and PIONIER transmission gain using only observations from both epochs within the same narrow effective-seeing bin, for example 0.8 to 1.0 arcsec, with the same $R<12$ mag target set; if the +60% and +130% advantages shrink or disappear inside matched seeing bins, the headline improvement is an epoch effect rather than an AO effect.","supporting_citations":[{"cited_title":"P., Chelli, A., & Malbet, F","cited_arxiv_id":null,"evidence_quote":"It supplies the theoretical prediction that higher-order AO improves single-mode coupling under partial correction, which is the model the paper's flux-dropout and transmission measurements are compared against."},{"cited_title":"W., Buscher, D","cited_arxiv_id":null,"evidence_quote":"It establishes the criterion (pupil size of about three Fried parameters) showing that tip-tilt correction is cost-optimal only around L band, which is the reason the old STRAP system was inadequate for near-infrared fringe tracking."},{"cited_title":"2019, A&A, 624, A99 Le Bouquin, J.-B., Berger, J.-P., Beuzit, J.-L., et al","cited_arxiv_id":null,"evidence_quote":"It documents the GRAVITY fringe tracker and its phase-tracking S/N limit, which the paper uses to interpret the limiting-magnitude and residual improvements."},{"cited_title":"2018, in Proc","cited_arxiv_id":null,"evidence_quote":"It identifies low-wind dome-seeing as the failure mode that made otherwise excellent nights unusable, defining the resilience condition that NAOMI is shown to overcome."},{"cited_title":"2002, The Messenger, 110, 21","cited_arxiv_id":null,"evidence_quote":"It describes the Auxiliary Telescopes and their original tip-tilt-only stabilisation, the baseline against which NAOMI is measured."}],"review_version":1}