{"id":"882d91a7-31bd-43ee-9e06-0562e440d1c3","arxiv_id":"2411.17341","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"The METIS SCAO final design is predicted to meet its Strehl and contrast requirements in end-to-end simulations, with robustness to combined error sources.","lead":"This paper describes the final design of the adaptive optics system for METIS, the mid-infrared instrument for the 39-meter ELT, including a pyramid wavefront sensor and GPU-based real-time computer. It reports simulated performance: 95.4% Strehl at 3.7 microns and high-contrast imaging reaching the required 3e-5 contrast, assuming the design behaves as modeled.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The performance claims rest on COMPASS's pyramid WFS model faithfully representing the real cryogenic PWFS, and no independent test of that model is yet reported.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: predicted performance and the reconstructor share the same PWFS/VDM model, so model fidelity is the critical unvalidated link. I agree, with a slight sharpening: the issue is not that using the same model is logically improper, but that the paper provides no independent empirical test of that model before claiming requirement compliance. The hardware-side evidence is genuinely strong: the RTC timing demonstration (Sec 5.2.4) and the HRTC-in-the-loop numerical check (Sec 5.2.5) show the real-time pipeline is correct and fast. The HEEPS contrast pipeline also includes a broad set of effects and its 300 ms sampling is stated to be conservative by about a factor of two. None of this, however, anchors the PWFS forward model to the physical pyramid and M4. The planned telescope simulator (Sec 2.8.1) is exactly the right test, but no results are reported. The low-wind analysis (Sec 6.4.2) further demonstrates that the model can predict complete loss of loop closure under plausible conditions, reinforcing that the performance envelope is conditional on model correctness. I therefore keep the reader's CONDITIONAL verdict: the design description is credible and the engineering is mature, but the headline 'will meet' claims require either successful telescope-simulator validation with measured interaction matrices or explicit error bars from model mismatch.","tokens_in":37,"tokens_out":5837,"duration_ms":146150,"concrete_test":"At subsystem AIT, use the telescope simulator and SCAO test cryostat: drive the ALPAO 820 DM with a known set of modes, record PWFS slope maps, and compare them with COMPASS p_metis predictions using the same VDM basis and regularisation. Then close the loop under a known turbulence PSD and compare measured residual wavefront/Strehl with the Sec 6.3 baseline prediction. If the rms interaction-matrix mismatch or residual WFE exceeds the error budget allocation (e.g., predicted 128 nm vs measured >155 nm at nominal conditions), re-fit the reconstructor from measured interaction matrices and recompute the Sec 7.4 contrast curve; only then can the 'all conditions' claim be retained.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Sec 6.1.1 states that the COMPASS module p_metis 'implements our reconstruction scheme using the virtual Fried geometry DM and the regularised MMSE inversion and projection steps.' The same model is therefore used to build the reconstructor and to predict the Strehl and contrast numbers. This is not circularity in a formal sense, but it makes model-form error the one ingredient that has no independent empirical anchor. The robustness analyses in Sec 6.4 vary parameters inside the same forward model; they cannot bound errors arising from real pyramid diffraction, modulation dynamics, detector nonlinearity, or the actual M4 influence functions. The HRTC numerical-correctness test (Sec 5.2.5) verifies that the hardware matches COMPASS, not that COMPASS matches the sky. The admitted low-wind failure in Sec 6.4.2 shows that at least one environmental condition can defeat the simulated system outright; if unmodeled sensor errors are comparable in effect, the claimed L-band margin in Fig. 28 could close. Until the telescope-simulator tests (Sec 2.8.1) validate the p_metis forward model, the statement that the L-band requirement 'will be met' is a simulation-based expectation rather than a demonstrated capability.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents the final design of the METIS SCAO system and its predicted performance, as of the FDR and entry into the MAIT phase. The authors describe the cryogenic pyramid wavefront sensor, the GPU-based real-time computer, the wavefront control strategy built around a virtual Fried-geometry deformable mirror with regularized MMSE reconstruction and numerical derotation, and the auxiliary control loops for pupil stabilization, differential tip-tilt (QACITS), and NCPA correction with the asymmetric Lyot wavefront sensor. Performance predictions are based on COMPASS end-to-end simulations and the HEEPS high-contrast pipeline, leading to the claims that the top-level requirements will be met: Strehl greater than 93% at 10 microns and greater than 60% at 3.7 microns, and a post-processed 5-sigma contrast of 3e-5 at 5 lambda/D in L-band. The paper also reports real-time hardware tests of the HRTC and a numerical correctness check with COMPASS in the loop.","tokens_in":32496,"tokens_out":8136,"duration_ms":71157,"significance":"If the predicted performance is realized on sky, METIS SCAO would be the first ELT adaptive-optics system to deliver the combination of high Strehl and high contrast required for the exoplanet and disk science cases, and the design innovations (virtual-DM reconstruction, numerical derotation, cryogenic PWFS, COTS GPU RTC) are of broad interest to the ELT AO community. The paper is strong on engineering completeness: it reports a successful real-time performance test of the HRTC (Sec 5.2.4, median 365 microseconds against a 909 microsecond requirement) and a hardware-in-the-loop numerical correctness check with COMPASS (Sec 5.2.5), and it makes the AOSAT analysis toolkit publicly available. However, the central performance claims rest entirely on the fidelity of the COMPASS forward model, which is not yet validated against the physical PWFS or the telescope; the authors themselves document a low-wind condition under which the simulated system fails to close the loop.","major_comments":[{"comment":"The performance predictions underlying the requirement-compliance claims use COMPASS modules (p_metis, p_calibration_DM) that implement the same virtual-Fried-geometry DM and regularized MMSE reconstruction scheme as the real-time controller (Sec 6.1.1). The robustness analyses in Sec 6.4 vary parameters inside this same forward model, so they cannot bound errors arising from physical pyramid diffraction, modulation dynamics, detector nonlinearity, or the true M4 influence functions. The HRTC numerical-correctness check (Sec 5.2.5) verifies consistency between the hardware and COMPASS, not between COMPASS and the sky; the planned telescope-simulator tests (Sec 2.8.1) are not yet reported. The statement in Sec 7.4 that the L-band requirement 'will be met by the RAVC in all conditions' is therefore a simulation-based expectation rather than a demonstrated capability, and should be qualified accordingly in the abstract, Sec 7.4, and Sec 8.","section":"Sec 6.1.1, Sec 5.2.5, Sec 2.8.1, Sec 7.4"},{"comment":"The low-wind analysis reports that with the 0.5 m/s OPD maps of Martins et al. (2022), the system 'immediately locks into a 2π petalling state and remains so' when these conditions prevail at loop start-up. This is an admitted operational failure under a plausible environmental condition, and the paper provides no mitigation that has been shown to avoid it (the higher-regularisation closure phase described in the same section did not prevent the lock-in for the 0.5 m/s case). Since Sec 7.4 claims the L-band requirement is met 'in all conditions', the paper must either exclude low-wind conditions from that claim, quantify the expected on-sky frequency of such conditions at Cerro Armazones, or present a demonstrated mitigation strategy. As written, the 'all conditions' statement is not supported by the presented evidence.","section":"Sec 6.4.2"},{"comment":"The reported performance is obtained by re-optimizing the reconstructor and control parameters (loop frequency, regularisation alpha_rec = alpha_proj, modulation amplitude, and integrator gain) for each guide-star magnitude, seeing condition, and NCPA level (e.g., Table 9). This produces an envelope of best-case performance rather than a single fixed configuration, and the selection of the optimum in the simulations presupposes knowledge of the disturbance condition. The paper should state explicitly which parameter set is assumed in the top-level requirement verification (Sec 6.3.4 and Sec 7.4), and should justify that the on-sky system can identify and apply those optima without prior knowledge of the disturbance conditions. Without this, the contrast margin shown in Fig. 28 may not be representative of routine operation.","section":"Sec 6.4.1, Table 9, Sec 6.3.1-6.3.3, Sec 7.4"},{"comment":"The 'superhero' combined-error simulation lists 'Pyramid angle scatter TBD' as one of the scaled error terms, so the amplitude of this term is unspecified. With this value undefined, the robustness result at scaling factor 1 is not fully defined, and the claim in Sec 8 that the contrast requirement is met 'even when all known sources of error are applied in combination' cannot be verified. The authors should either quantify this term or state the sensitivity of the result to it, particularly because the VDM/PWFS model fidelity is already a concern.","section":"Table 10, Sec 6.4.3, Sec 8"}],"minor_comments":[{"comment":"The reference 'Boné et al. (2024, in prep.)' should be updated with a published reference or removed from the citation list.","section":"Sec 2.5"},{"comment":"The sentence 'The simulation the impact, we used time-varying OPD maps' should read 'To simulate the impact, we used time-varying OPD maps'.","section":"Sec 6.4.2"},{"comment":"The row 'Total + 20% margin 156.9' appears inconsistent with the preceding 'Total 153.9'; please clarify how the 20% margin is applied to arrive at 156.9 nm.","section":"Table 11"},{"comment":"The caption states 'Contrast at 3.8 m'; the wavelength unit should be '3.8 µm'.","section":"Fig. 18 caption"},{"comment":"Sec 7.2 states the baseline HCI simulation uses a 1-hour ADI sequence with 300 ms sampling, while Sec 7.3.1 reports '10-min SCAO sequences' for the NCPA study; please clarify whether these are separate runs or sub-segments of the 1-hour sequence.","section":"Sec 7.2 and Sec 7.3.1"},{"comment":"The text states that in the NCPA experiment 'the controller was a pure integrator', but Eq. (6) and Table 8 describe a proportional-integral controller with both P and I gains; please clarify the controller structure used for the results in Table 9 and Fig. 21.","section":"Sec 6.4.1"}],"recommendation":"major_revision","confidential_remarks":"This is a comprehensive system-design paper that fits the scope of Experimental Astronomy. The main risk is that the performance requirements are stated as met when the supporting evidence is entirely simulation-based and the forward model is unvalidated; the low-wind failure in Sec 6.4.2 and the per-condition parameter optimization in Table 9 are concrete issues that need to be addressed in a revision. I would not reject the paper, but I would ask the authors to temper the 'will be met' language and to provide a clear statement of which conditions are covered by the requirement-compliance claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a thorough, mature final-design paper for the METIS SCAO system, and it deserves to be taken seriously. The headline numbers—95.4% Strehl at 3.7 µm, the 3e-5 L-band contrast—are predictions from the COMPASS end-to-end simulation chain, not measured values, and the same forward model used to build the reconstructor is used to generate the predicted performance. That does not make the paper circular in a formal sense, but it does mean the predictions are ideal-model results: any error in the PWFS model itself (pyramid diffraction, modulation dynamics, detector nonlinearity, M4 influence functions) is invisible to these simulations.\n\nWhat is genuinely new here is the post-FDR design freeze: the cryogenic pyramid module, the GPU RTC prototype, the control strategy with the virtual-DM reconstruction separated from M4 projection, and the robustness analyses including the low-wind and 'superhero' combined-error cases. The RTC timing requirement is demonstrated on hardware: 300,000 simulated frames processed with a median 365 µs against a 909 µs requirement, and the numerical-correctness check shows the HRTC matches COMPASS in closed loop. That is real, reproducible engineering evidence and the strongest part of the paper.\n\nThe soft spots are proportionate. The robustness analyses vary parameters inside the same forward model, so they cannot bound model-form error. Table 9 shows control parameters (modulation, regularisation, gain) re-optimized for each NCPA level, which introduces a fitting element; the paper is transparent about this, but the margins should not be read too literally. The low-wind section is honest: at 0.5 m/s the system locks into a 2π petal state and cannot run. The statement in Sec 7.4 that the L-band requirement 'will be met' is therefore a simulation-based expectation, not a demonstrated capability. The authors do flag the uncertainty in the low-wind OPD maps, and the telescope-simulator tests are the right next step.\n\nWho this is for: anyone working on ELT AO or on METIS science planning. It is a reference document, not a theoretical contribution. I would bring it to a reading group focused on AO instrumentation. It deserves a serious referee: the design description is strong, the hardware evidence is real, and the performance claims, while simulation-bound, are clearly presented and mostly appropriately hedged. If I were editor, I would send it out for review. My main request would be a sensitivity analysis or error bars on the headline Strehl and contrast numbers, plus an explicit statement that these are pre-commissioning predictions awaiting telescope-simulator validation.","headline":"A mature, honest final-design paper for METIS SCAO with real RTC hardware evidence; the performance numbers are solid simulations but should be read as ideal-model predictions, not on-sky guarantees.","tokens_in":33131,"tokens_out":3497,"would_cite":true,"duration_ms":41838,"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":"METIS adaptive optics design predicted to meet top-level Strehl and contrast requirements, including a 5-sigma contrast of 3e-5 at 5 lambda/D in L-band.","keywords":["adaptive optics","single-conjugate adaptive optics","pyramid wavefront sensor","METIS","Extremely Large Telescope","high-contrast imaging","Strehl ratio","real-time control"],"falsifier":"During commissioning, measure the closed-loop Strehl at 3.7 microns and the L-band angular differential imaging contrast at $5\\lambda/D$ under median seeing; if Strehl falls below 60% or the post-processed 5-$\\sigma$ contrast exceeds $3 \\times 10^{-5}$, the paper's central performance claim is contradicted.","tokens_in":1729,"feed_emoji":"🔭","tokens_out":3982,"duration_ms":85287,"temperature":0.7,"pith_summary":"This paper presents the final design of the single-conjugate adaptive optics (SCAO) system for METIS, the mid-infrared instrument for the 39-meter Extremely Large Telescope, and argues from end-to-end simulations that the system will meet its requirements. The claimed performance is a Strehl ratio above 93% at 10 microns and above 60% at 3.7 microns, with a peak of 97.8% and a nominal value of 95.4% at 3.7 microns. It further claims that the L-band high-contrast imaging requirement, a post-processed 5-$\\sigma$ contrast of $3 \\times 10^{-5}$ at $5\\lambda/D$, will be met by the ring-apodized vortex coronagraph in all conditions. The design separates wavefront reconstruction from projection onto the deformable mirror control modes, a choice that makes the system robust to the ELT's fragmented, rotating pupil.","feed_headline":"METIS adaptive optics predicted to hit Strehl and contrast goals","feed_subtitle":"Final SCAO design claims >95% Strehl at 3.7 um and 3e-5 contrast at 5 lambda/D in L-band.","key_machinery":"The load-bearing element is a virtual deformable mirror (VDM): a fixed, Fried-geometry grid of bilinear spline influence functions aligned with the pyramid wavefront sensor's subapertures, onto which the wavefront is first reconstructed using regularized minimum mean square error estimation. Reconstruction is logically separated from a second, regularized projection onto the ELT's M4 deformable mirror control modes, so pupil rotation and misregistration are handled numerically as projection parameters rather than by moving optics. The pyramid wavefront sensor operating in K-band, a GPU-based real-time computer running at up to 1 kHz, and focal-plane wavefront sensing loops for tip-tilt and high-order non-common path aberration correction complete the control architecture.","core_discovery":"The central claim is that the METIS SCAO system, as designed, will deliver diffraction-limited performance at the ELT: Strehl above 93% at 10 microns and above 60% at 3.7 microns, with predicted peak 97.8% and nominal 95.4% at 3.7 microns, plus a post-processed 5-$\\sigma$ contrast of $3 \\times 10^{-5}$ at $5\\lambda/D$ in L-band. The paper supports this with a wavefront error budget, robustness analyses against non-common path aberrations, water-vapor seeing, low-wind petal effects, and a combined 'super-hero' simulation that stacks misaligned mirror segments, missing pupil segments, and unknown pyramid angle errors. It states that the L-band contrast requirement is met by the ring-apodized vortex coronagraph in all conditions, and that the system stays stable with non-common path aberrations up to 300 nm rms, with operation still possible at 400 nm rms.","pith_inferences":["If the virtual-deformable-mirror approach proves as robust on sky as in simulation, the same separation of reconstruction from projection could simplify adaptive optics for other segmented-aperture telescopes by removing the need for opto-mechanical derotation.","The demonstrated tolerance of non-common path aberrations up to 300 nm rms suggests that focal-plane wavefront sensing will be a necessary component of any future mid-infrared high-contrast instrument, not just an optional refinement.","The large gap between predicted Strehl (95.4%) and the requirement (60%) indicates that residual risk lies less in the control law than in unmodeled sensor behavior, so commissioning effort should focus on calibrating the pyramid wavefront sensor and measuring non-common path aberrations on sky.","A direct test of the core modeling assumption would compare the on-sky pyramid sensor interaction matrix and optical gain with the simulated forward model at first light, a validation the paper does not claim to have completed."],"forward_implications":["METIS SCAO can be operated almost as 'pushbutton AO', with only loop frequency and regularization adjusted for faint guide stars or bad seeing.","The limiting guide star magnitude is about 13 in K-band, where a Strehl of 63.6% at 3.7 microns is still expected under median seeing.","Combined realistic errors, including 21 missing M1 segments, M4 misalignment, non-common path aberrations, and pyramid angle errors, add about 40 nm rms wavefront error and leave Strehl at 94.6% at 3.7 microns.","The L-band high-contrast requirement is met by the ring-apodized vortex coronagraph in all simulated conditions, with N-band performance dominated by water-vapor seeing.","The real-time computer prototype processes 300,000 wavefront sensor images with a maximum loop computation time of 410 microseconds, comfortably below the 909 microsecond requirement."],"supporting_citations":[{"why":"Supplies the earlier METIS SCAO design and performance study that this paper updates.","marker":"Hippler et al. (2018)"},{"why":"Defines the wavefront control strategy with separated reconstruction and projection onto M4 modes.","marker":"Correia et al. (2022)"},{"why":"Provides the end-to-end high-contrast imaging simulation pipeline used for the contrast predictions.","marker":"Delacroix et al. (2022)"},{"why":"Models water-vapor seeing and its impact on mid-infrared high-contrast imaging at ELT scale.","marker":"Absil et al. (2022)"},{"why":"Introduces the pyramid wavefront sensor concept on which the SCAO sensing scheme is based.","marker":"Ragazzoni (1996)"},{"why":"Supplies the regularized minimum variance wavefront reconstruction formalism used in the reconstructor.","marker":"Ellerbroek (2002)"},{"why":"Previous SCAO simulations whose performance and methodology are extended to the final design.","marker":"Feldt et al. (2023)"}],"fun_headline_variants":["METIS SCAO on ELT predicts 97.8% Strehl, 3e-5 L-band contrast","ELT's METIS AO: 97.8% Strehl and 3e-5 contrast predicted","METIS AO final design hits 97.8% Strehl, 3e-5 contrast","ELT METIS SCAO: up to 97.8% Strehl, 3e-5 contrast","METIS SCAO on ELT: 97.8% Strehl, 3e-5 contrast at 5 lambda/D"],"cache_read_input_tokens":35200,"weakest_assumption_plain":"The predicted Strehl and contrast numbers rest on end-to-end simulations whose forward model of the pyramid wavefront sensor is also used to build the reconstructor, so if the real sensor's response differs from that model, the predicted performance could be lower.","fun_headline_variants_meta":{"raw":{"variants":["METIS SCAO on ELT predicts 97.8% Strehl, 3e-5 L-band contrast","ELT's METIS AO: 97.8% Strehl and 3e-5 contrast predicted","METIS AO final design hits 97.8% Strehl, 3e-5 contrast","ELT METIS SCAO: up to 97.8% Strehl, 3e-5 contrast","METIS SCAO on ELT: 97.8% Strehl, 3e-5 contrast at 5 lambda/D"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001202,"raw_usage":{"total_tokens":5041,"prompt_tokens":1123,"completion_tokens":3918,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":739,"completion_tokens_details":{"reasoning_tokens":3773}},"tokens_in":739,"tokens_out":3918,"duration_ms":24219,"temperature":1.0,"reasoning_tokens":3773,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:13:07.292985+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"During commissioning, measure the closed-loop Strehl at 3.7 microns and the L-band angular differential imaging contrast at $5\\lambda/D$ under median seeing; if Strehl falls below 60% or the post-processed 5-$\\sigma$ contrast exceeds $3 \\times 10^{-5}$, the paper's central performance claim is contradicted.","supporting_citations":[{"cited_title":"In: Evans, C.J., Bryant, J.J., Motohara, K","cited_arxiv_id":null,"evidence_quote":"Provides the end-to-end high-contrast imaging simulation pipeline used for the contrast predictions."},{"cited_title":"Journal of Modern Optics 43, 289–293 (1996) https://doi.org/10.1080/09500349608232742","cited_arxiv_id":null,"evidence_quote":"Introduces the pyramid wavefront sensor concept on which the SCAO sensing scheme is based."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the regularized minimum variance wavefront reconstruction formalism used in the reconstructor."},{"cited_title":"In: Adaptive Optics for Extremely Large Telescopes, vol","cited_arxiv_id":null,"evidence_quote":"Previous SCAO simulations whose performance and methodology are extended to the final design."}],"review_version":1}