{"id":"de7c1850-e1cb-4e0c-8cdf-9918c730f74e","arxiv_id":"2411.13746","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"A deformable mirror on the THD2 bench creates a polarization-dependent beam shift of several hundred nanometers, about ten times larger than predicted, degrading coronagraph contrast.","lead":"This paper measures how a deformable mirror on a high-contrast imaging testbed shifts light beams by different amounts depending on their polarization, by up to a few hundred nanometers. This matters because future space telescopes aiming to image Earth-like planets need to control such tiny shifts to reach contrasts of one part in ten billion.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The DM2 shift magnitude rests on the unvalidated linear-drift model in Eq. 10; residual ordered beam motion can be amplified by the 1/(ω−ω⊥) fit and bias the Table 2 values.","rationale":"The reader's weakest assumption is precisely the load-bearing point. The central claim is empirical: DM2 introduces a differential polarization shift, and the only quantitative support is the weak-measurement fitting of q in Eq. 9 after subtracting the drift/prism model in Eq. 10. The paper is honest about the empirical nature of the drift model, but honesty does not remove the risk: if the model misses real beam motion, the 1/(ω−ω⊥) amplifier converts that residual into a fitted shift. The reported 1σ RMS values (70–130 nm) are comparable to the vertical shifts of 240–340 nm, so a systematic at that level could change the conclusions for DM2. The reverse-order scan is a direct, low-cost check: it preserves the polarization physics while changing the time-ordering of any slow drift, so a correctly modeled linear drift will yield the same q, whereas a nonlinear or angle-dependent drift will not. The paper's other limitations (DM1 unmeasured, theoretical discrepancy with GH/IF, non-Gaussian beam) weaken the interpretation but do not threaten the basic attribution as directly as a biased drift subtraction does. The reader already conditioned acceptance on this concern, so I recommend no change to the verdict; the condition should be explicit that Table 2 needs the drift-model validation before the DM2 numbers are used in instrument design.","tokens_in":10375,"tokens_out":7539,"duration_ms":78414,"concrete_test":"For the DM2-only configuration, repeat the 0°→360° analyzer scan in reverse (360°→0°) with the same step schedule and starting conditions, and reduce both data sets with Eqs. 10 and 9. If the fitted H and V shifts for DM2 agree within the reported 1σ RMS (130/70 nm), the linear drift model is adequate and the DM2 attribution is supported; if they disagree by more than that, residual ordered beam motion is biasing Table 2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reported DM2 shifts (555 nm H, 340 nm V) are not directly measured centroids; they are parameters q extracted from the weak-measurement amplification function (Eq. 9) after subtracting the empirical model f(ω,t) = a sin(ω+φ1) + bω(t) + d1 (Eq. 10). The paper explicitly calls the linear drift term an empirical approximation, and the 360° sine term plus linear drift are fitted without independent validation. If the true beam motion contains a slow nonlinear drift or a mechanical mode with an angular signature, the residual after the first fit is not white noise. The second fit amplifies any residual that projects onto 1/(ω−ω⊥) over the fitted interval (|ω−ω⊥| between 0.1° and 15° for small shifts), so a systematic motion of order 50–100 nm could become a spurious contribution to the fitted q. There are no residual plots, no repeat with reversed scan order, and no independent drift monitor (e.g., an idle analyzer position or LOWFS telemetry) to establish that the model is adequate. Because DM2 is the only component with a large measured shift, this is exactly where an unmodeled systematic would change the central conclusion.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports polarization-dependent beam shifts on the THD2 high-contrast imaging bench. It first shows that the dark-hole contrast degrades when the analyzer is rotated away from the polarization state used for wavefront control, and reproduces the morphology with a ~800 nm tip-tilt. It then computes expected Goos-Hänchen and Imbert-Fedorov shifts for the bench mirrors using analytical expressions from Aiello et al., obtaining total shifts of order tens of nanometers. To measure such small shifts, the authors use a weak-measurement scheme with crossed polarizer/analyzer, record centroid positions versus analyzer angle, subtract a fitted 360° sinusoidal prism term and a linear temporal drift, and fit the residual with an amplification function 1/(ω−ω⊥). Table 2 reports shifts of several hundred nanometers for configurations containing DM2 (about 555 nm horizontal and 340 nm vertical for DM2 alone), far exceeding the theoretical predictions, and the paper concludes that DM2 introduces a dominant differential polarization shift.","tokens_in":10689,"tokens_out":6972,"duration_ms":66932,"significance":"If the quantitative result is robust, the paper is relevant to high-contrast instrument polarization error budgets: it identifies a specific pupil-plane deformable mirror as the dominant source of differential polarization shifts, and it demonstrates a bench-level impact on a FQPM coronagraph. The paper also provides a useful explicit comparison between analytical GH/IF predictions and measurements on a complete optical bench. The main strength is the use of weak measurement to amplify subpixel beam shifts; the main weakness is that the central quantitative claim depends on an empirically fitted drift model whose systematic error is not characterized.","major_comments":[{"comment":"The data-reduction model f(ω,t) = a sin(ω+φ1) + bω(t) + d1 is load-bearing for every value in Table 2, but it is not independently validated. The authors explicitly call the linear drift an empirical approximation, and the amplification fit of Eq. (9) is applied to the residuals after subtracting this model. Any unmodeled ordered motion, such as a slowly nonlinear mechanical drift or a vibration mode with an angular signature, can project onto 1/(ω−ω⊥) over the fitted interval and bias the extracted coefficient q, which is then reported as the physical shift. This is especially concerning because DM2 is the only component with a large measured shift, so a systematic at the level of tens of nanometers could change the central conclusion. The paper should show residual plots after the first fit, repeat measurements with reversed or scrambled analyzer angle order, or provide an independent drift monitor (for example, an idle analyzer position) to demonstrate that the model is adequate.","section":"Section 3.2, Eq. (10)"},{"comment":"The reported 1σ RMS values (70 to 130 nm for the largest shifts) appear to be fit uncertainties from the inverse-function linear fit, but they do not include the uncertainty introduced by the first-stage subtraction of Eq. (10) or by the choice of fitting intervals. The fitted interval is changed by defect size (|ω−ω⊥| between 0.1° and 1° for small defects, 0.5° and 15° for large defects), yet there is no stability test of q against the interval bounds or against alternative drift models. Without this, the statistical significance of the 555 nm and 340 nm DM2 shifts is established only under the assumption that the residual noise is white, which is precisely the assumption in question.","section":"Section 3.2, Table 2"},{"comment":"The theoretical predictions in Table 1 assume that all mirrors are silver-coated, but the text notes that the DMs are aluminum-coated, and DM2 is the component later identified as the anomalous contributor. No calculation is shown for the GH/IF shifts of an aluminum-coated mirror at the DM2 incidence angle (6.55°), so the statement that the measured shifts are 'more than ten times smaller' than the predictions is not quantitatively established for the actual coating of DM2. The authors should provide predictions using the appropriate aluminum permittivity, or explicitly justify why the silver coating approximation is valid for the comparison.","section":"Section 2.3, Table 1"}],"minor_comments":[{"comment":"The parameter list below Eq. (10) contains a typo: it reads 'a, d1 b, d1 are parameters of the fit', which should presumably be 'a, φ1, b, d1 are parameters of the fit'.","section":"Section 3.2, Eq. (10)"},{"comment":"The table layout is difficult to parse: the column labels, wavelength entries, and the H/V pairs are not clearly aligned in the text, and it is not immediately clear which wavelength and optical configuration correspond to each measured value. Please reformat the table and clarify in the caption whether the quoted 1σ RMS values include any systematic component.","section":"Table 2"},{"comment":"The figures would benefit from explicit axis labels and error bars on the centroid positions, and from residual plots after subtracting Eq. (10), so the reader can assess whether the drift model leaves any structure as a function of angle.","section":"Figures 8–10"},{"comment":"The predicted angular shifts depend on the beam waist w0 in Eq. (2), but the value used for Table 1 is only vaguely described as the FWHM of the beam at the focal plane. Please state the numerical value of w0 and discuss how the truncated, non-Gaussian bench beam affects the predicted shifts.","section":"Section 2.2"}],"recommendation":"major_revision","confidential_remarks":"This is a conference-style paper whose central quantitative claim -- DM2 produces a ~555 nm horizontal differential polarization shift -- is potentially important but rests on an unvalidated empirical drift model. I do not think rejection is warranted because the polarization-dependent contrast degradation in Figs. 2-3 and the raw weak-measurement curves provide independent qualitative evidence that a real effect exists. However, the paper needs either additional measurements or a much more thorough systematic-error analysis before the specific magnitude and attribution to DM2 can be accepted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear X,\n\nThe headline result is real and worth knowing: a Boston Micromachine DM on the THD2 bench shifts the beam by several hundred nanometers depending on polarization, an order of magnitude above Goos-Hänchen and Imbert-Fedorov predictions. Applying weak measurement to a full optical bench is new in this context, and the paper has internal controls that support the finding: configurations without DM2 show small shifts, configurations with DM2 consistently show large ones. The writing is honest and the limitations are stated clearly.\n\nThe main soft spot is the drift model. To isolate the weak-measurement amplification, they fit and subtract a 360° prismatic term plus a linear drift (Eq. 10), then fit the 1/(ω−ω⊥) amplification. The drift term is openly declared an empirical approximation, and there are no residual plots, no reversed scan orders, and no independent drift monitor. A residual ordered motion could be amplified by the second fit. That said, the control measurements (20 nm with no mirror, 32–60 nm for FM2 alone) show the pipeline does not generate spurious large shifts without DM2. The DM2 shift is ~500 nm with 70–130 nm errors; even a few tens of nm of unmodeled drift would not erase the discrepancy with theory.\n\nThe bigger gap is that DM1 was never measured, so the claim that DM2 is the dominant contributor is provisional, as the authors say. The physical origin is unexplained; two hypotheses are offered, but the theory section and measurement section sit uneasily together. The paper does not try to force them into agreement, which is fine for a measurement paper.\n\nCitations look solid, including the relevant van Holstein et al. beam-shift paper, and the method is described well enough to reproduce. I would send this to a referee, asking for residual plots or an independent drift check and a careful sign-convention statement.\n\nBest","headline":"A convincing weak-measurement demonstration that a deformable mirror introduces large polarization-dependent beam shifts, despite an imperfectly validated drift model.","tokens_in":11182,"tokens_out":3536,"would_cite":false,"duration_ms":34142,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":null,"created_at":"2026-08-12T15:56:45.382083+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":null,"supporting_citations":[],"review_version":1}