{"id":"b9ea3dd8-b5cf-43f5-b290-271cba2eb2a0","arxiv_id":"2607.28816","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"Multi-star fine guidance is simulated to cut roll-induced focal-plane image motion by 31.8% for SuperBIT and 77.4% for a GigaBIT-like geometry, after flight data showed roll leakage is common.","lead":"Using flight data from the 2023 SuperBIT balloon telescope, this paper shows that residual image motion is often driven by telescope roll that the fine-guidance mirror cannot correct, and simulates how using two guide stars instead of one could reduce that motion by 32–77%. The result is a concrete argument for multi-star fine guidance on the planned GigaBIT wide-field balloon telescope.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Empirical FSC gain discrepancy (3.60 vs 58.39) undermines calibration of simulated multi-star reductions; GigaBIT extrapolation depends on unverified mechanism","rationale":"The reader's verdict correctly identifies the empirical gain as the weakest assumption. I agree and add that the Table 3 value 58.39 is internally inconsistent with the stated 3.60, which makes the simulation's calibration opaque. This is load-bearing because the paper's novelty is the quantitative prediction for GigaBIT; if the gain is wrong or the mechanism is flexure rather than FSM projection, the multi-star benefit may not scale. The qualitative finding that passive FSC motion correlates with RSC is well supported by flight data (Fig. 4, n=2348), and the multi-star concept is physically plausible, so the paper deserves CONDITIONAL rather than REJECT. A re-run with the correct gain is cheap and decisive.","tokens_in":16838,"tokens_out":9096,"duration_ms":89902,"concrete_test":"Re-run the estimator comparison in §4.1 and §4.2 with the empirical FSC gain set to 3.60 (the mean |k_fit| from §3.4.1, Eq. 23) instead of 58.39 (Table 3), holding all other parameters fixed. If the science-field RMS values or percentage reductions in Table 4 shift by more than ~10%, the headline quantitative claim is not robust to the internal parameter discrepancy. In addition, inspect the simulation code to determine which value was actually used.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's quantitative central claim—31.8% and 77.4% reductions in roll-induced science-field image motion (Table 4)—rests on a simulation parameter called 'Empirical FSC gain' set to 58.39 in Table 3. Section 3.4.1, however, derives a flight-calibrated coupling coefficient k_fit with mean |k_fit|=3.60 (Fig. 7) and states that this mean is used to scale the simulated model (end of §3.4.1). The factor 58.39 appears nowhere in the text and is ~16× larger than 3.60; no conversion or definition is given. If the simulations were actually run with 58.39, the injected roll amplitudes (1″ and 4″) would produce FSC centroid motions ~16× larger than the flight data, so the absolute RMS values (0.786″, 0.536″, 2.343″) and the percentage reductions would be miscalibrated, even if the sign of the effect is preserved.\n\nBeyond this arithmetic inconsistency, the empirical gain k_fit≈3.60 is 27× larger than the rigid-body geometric prediction k_geo≈0.135 (Eq. 21). The paper attributes the excess to 'field-dependent tip-tilt projection' but explicitly lists 'torsional flexure and other unmodeled roll dynamics' as alternatives. The simulation injects roll as a rigid-body rotation of the input-ray bundle (Eqs. 16–17) and then multiplies by a scalar gain; this linear scaling only captures roll-driven FSM projection effects. If a large part of the amplification is instead physical flexure—motion of the FSC relative to the science camera that correlates with RSC but is not corrected by the FSM—then multi-star estimation cannot reduce that component, and the theory would not transfer to GigaBIT. The GigaBIT results (Table 4) further assume the same gain holds after scaling ℓ from 40.6 to 150 mm and d from 180 to 610 mm, with no independent validation. Thus the 77.4% GigaBIT figure is the least secure quantitative claim in the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes roll leakage in balloon-borne telescope fine-guidance systems using SuperBIT 2023 flight data and a closed-loop simulation framework. Flight-data analysis shows that passive focal-plane star-camera (FSC) centroid motion correlates strongly with independent roll star camera (RSC) measurements in most two-guide-star science exposures, supporting the claim that boresight roll commonly leaks onto the focal plane after two-axis FSM correction. The simulation combines optical ray tracing, asynchronous FSC measurements, a weighted multi-star estimator, and FSM control. For a roll-only disturbance, single-star guidance stabilizes the active guide star but leaves field-dependent residual motion, while multi-star estimation reduces average science-field RMS image motion by 31.8% for the SuperBIT geometry and 77.4% for a representative GigaBIT geometry (Table 4). The authors recommend multi-star fine guidance for GigaBIT, but note that pitch/cross-pitch disturbances, gyroscope fusion, and the actual GigaBIT focal-plane layout are deferred to future work.","tokens_in":17335,"tokens_out":1993,"duration_ms":21800,"significance":"If the results hold, the paper makes a useful contribution to balloon-borne telescope pointing: it documents a previously underexplored systematic effect (roll leakage) using a substantial flight dataset, and it provides a quantitative, architecture-level argument that multi-star fine guidance can substantially mitigate that effect, especially for larger focal planes like GigaBIT. The flight-data correlation analysis (Figures 3-4) is a meaningful empirical result on its own. The simulation framework is clearly described and appears reusable, though the central quantitative percentages depend on a calibration step that is not fully transparent. The qualitative conclusion—single-star fine guidance inherently distributes roll-induced image motion unevenly, and multi-star estimation can re-center the residual rotation—is well motivated and supported by the optical geometry argument.","major_comments":[{"comment":"The simulation calibration is internally inconsistent. The text (§3.4.1, end) states that the mean |k_fit|≈3.60 is used to scale the simulated model, but Table 3 lists 'Empirical FSC gain' as 58.39. No definition, conversion, or derivation of 58.39 is given anywhere. If the simulations actually used 58.39, the injected 1″/4″ roll amplitudes would produce FSC centroid motions roughly 16× larger than flight-calibrated k_fit predicts, making the absolute RMS values (0.786″, 0.536″, 2.343″) and the percentage reductions in Table 4 miscalibrated. The authors must explain the relationship between k_fit and the Table 3 value, and rerun or re-report the results with a quantity that is either fully derived or explicitly justified.","section":"§3.4.1 and Table 3, Eq. (21)-(23)"},{"comment":"The flight-calibrated gain k_fit≈3.60 is ~27× larger than the rigid-body geometric prediction k_geo≈0.135. The manuscript attributes this to 'field-dependent tip-tilt projection' while also listing 'torsional flexure and other unmodeled roll dynamics' as possible causes. The simulation injects roll as a rigid-body rotation of the input-ray bundle (Eqs. 16-17) and then multiplies by a scalar gain. This exactly models roll-driven FSM projection effects, but if a substantial part of the amplification is due to mechanical flexure or common-mode noise that the FSM cannot correct, the simulated multi-star benefit will not transfer to GigaBIT. This concern is load-bearing for the GigaBIT 77.4% claim. Please provide a sensitivity analysis varying the gain over the observed distribution (Fig. 7 shows σ=0.83) or otherwise demonstrate that the qualitative conclusion is robust to the physical interp","section":"§3.4.1, Eq. (21), Figure 7"},{"comment":"The headline results (31.8% and 77.4%) are reported as exact numbers with no uncertainty, despite the model depending on multiple fitted or user-chosen parameters (k_fit, roll amplitudes/frequencies, τ_d, weights, noise, gains, geometry). No Monte Carlo or error propagation is presented. At minimum, the percentage reductions should be accompanied by a spread arising from the observed distribution of k_fit and from reasonable variations in pole parameters (e.g., roll amplitude, relative phase). This is needed to judge whether the SuperBIT 31.8% and GigaBIT 77.4% claims are statistically meaningful and whether they overlap.","section":"§4.1-4.2, Table 4"},{"comment":"The GigaBIT simulation uses a 'representative' scaled SuperBIT geometry (ℓ=150 mm, d=610 mm) but the actual GigaBIT focal plane is annular and more complex. The 77.4% reduction is therefore a projection for an idealized geometry, not a prediction for the real instrument. This is acknowledged in the text, but the abstract and conclusion present 77.4% without this caveat. Please soften the abstract/conclusion or add explicit reporting of the GigaBIT geometry as an idealized scaled case, so the headline number is not over-interpreted.","section":"§4.2, Table 2"}],"minor_comments":[{"comment":"Typo: 'esxtimator' should be 'estimator' in 'age-weighted measurement fusion used by the esxtimator.'","section":"§4.1, p.14"},{"comment":"Typo: 'AAs noted in previous studies' should be 'As noted in previous studies.'","section":"§3.4, p.10"},{"comment":"Typo: 'by field rotation' likely should be 'by field rotation'; also 'by the simple by field rotation assumptions' reads awkwardly and may be missing a word.","section":"§3.4.1, p.11"},{"comment":"Missing space: '∼0.050 arcsec' should be '∼0.050 arcsec' (the LaTeX source has '∼' without a following space).","section":"§1, p.1"},{"comment":"The figure caption reports n=2348 exposures but the right panel splits into n=1379 and n=969, which sum to 2348. Please clarify that this split is by which FSC is passive and that only one exposure per FSC configuration is counted; otherwise the caption could be clearer.","section":"Figure 4"},{"comment":"The 'Empirical FSC gain' parameter is not described in the text. If it is a scaled version of k_fit, the conversion (e.g., pixel-to-arcsecond scaling) should be stated in the table caption or in §3.4.1.","section":"Table 3"}],"recommendation":"major_revision","confidential_remarks":"The paper has a solid empirical core: the SuperBIT passive-FSC/RSC correlation analysis is convincing and is a valuable contribution by itself. The main issue is the simulation calibration: the discrepancy between k_fit≈3.60 and the Table 3 value 58.39 is unexplained and directly affects the headline percentages. Even if that discrepancy is resolved, the simulation's sensitivity to the choice of calibration gain and to the roll-only disturbance model should be quantified. The GigaBIT claim is also based on a simplified geometry; it should be framed as an idealized projection. I do not see grounds for rejection—the central mechanism is plausible and the flight data support the existence of roll leakage—but the quantitative claims need substantial clarification and robustness analysis before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things up front. The flight-data result is real: over a thousand SuperBIT exposures with two guide stars show that the passive star camera's residual motion correlates strongly with the roll star camera, and the paper documents this carefully with thresholds and distributions. That is a useful, reusable empirical result for anyone building balloon-borne fine guidance systems. The simulation framework is a reasonable tool, and the qualitative conclusion that multi-star estimation moves the rotation center from the active guide star toward the science field is physically sensible and probably right.\n\nThe problem is the numbers. The text says the simulated model is scaled using the mean measured coupling, |k_fit| = 3.60. Table 3 lists \"Empirical FSC gain\" as 58.39. No conversion is given. Those differ by 16x, and the simulation results are sensitive to this choice because sensor noise is not scaled with the gain. So the 31.8% and 77.4% reductions are not as secure as the abstract implies. They are point values from a single disturbance realization, a simplified two-sinusoid roll model, and a scaled GigaBIT geometry, with no error bars and no sensitivity analysis.\n\nThere is a deeper issue behind the gain. The measured coupling is 27x larger than the rigid-body geometric prediction. The paper speculates that field-dependent tip-tilt projection explains it, but explicitly leaves torsional flexure and other unmodeled roll dynamics on the table. If flexure is a big part of the amplification, multi-star estimation can't fix it, because the FSM cannot correct motion that is not common to the guide-star rays. That makes the GigaBIT extrapolation, which assumes the same gain after scaling the geometry, the weakest claim in the paper.\n\nNone of this kills the paper. The qualitative message—single-guide-star fine guidance leaks roll into the science field, and multi-star weighting reduces that leakage—is well supported. But the quantitative reductions should be described as illustrative, not predictive, until the calibration is fixed and a sensitivity run is done.\n\nThis should go to peer review. The data are new and the question matters for GigaBIT. But the referee should ask for three things: reconcile 3.60 and 58.39, add a sensitivity analysis over the roll amplitude and gain, and soften the claims about the GigaBIT benefit. If you have a student working on balloon pointing, read the flight-data section; skip the headline numbers for now.","headline":"Roll-leakage characterization from SuperBIT flight data is genuinely new, but the headline multi-star improvement percentages rest on a simulation gain that is internally inconsistent (3.60 vs 58.39), and the GigaBIT extrapolation depends on an unverified mechanism, so treat those numbers as provisional.","tokens_in":18068,"tokens_out":5132,"would_cite":true,"duration_ms":52417,"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":"Roll leakage is inherent in single-guide-star fine guidance, and multi-star estimation reduces its image smear by up to 77%.","keywords":["roll leakage","fine guidance system","fast steering mirror","multi-star estimation","balloon-borne telescope","image stabilization","SuperBIT","GigaBIT"],"falsifier":"A decisive test would be to measure the coupling between passive focal-plane star-camera motion and the roll star camera while commanding known fast-steering-mirror tilts during a balloon flight. If the amplification factor k_fit drops to the rigid-body geometric prediction when the FSM is disabled, the leakage is FSM projection and the simulation's scaling is valid; if it persists with the FSM off, the amplification is structural and the simulated transfer to GigaBIT is suspect. Alternatively, fly GigaBIT with switchable single- and multi-star guidance and compare science-field RMS image moti","tokens_in":16777,"feed_emoji":"🔭","tokens_out":4766,"duration_ms":46679,"temperature":0.7,"pith_summary":"This paper establishes that 'roll leakage'—residual image rotation that a two-axis fast steering mirror cannot correct—is a systematic, flight-observed limitation of single-guide-star fine guidance on balloon-borne telescopes. Using 2348 quality science exposures from SuperBIT's 2023 flight, the authors show that passive focal-plane star-camera motion correlates strongly with independent roll measurements in most exposures, proving roll frequently drives residual focal-plane motion. They build a closed-loop simulation of the fine guidance system and show that fusing centroid measurements from two guide stars reduces average science-field image motion by 31.8% for SuperBIT and 77.4% for the larger GigaBIT design, by shifting the effective field-rotation center away from the active guide star toward the science field. The work matters because it identifies a concrete, architected fix for a limitation that grows worse as balloon telescopes widen their fields of view.","feed_headline":"Multi-star guidance cuts roll smear by up to 77%","feed_subtitle":"Flight data show roll leakage is inherent in single-guide-star fine guidance; fusing multiple star cameras fixes it.","key_machinery":"The central mechanism is 'roll leakage': boresight roll maps to position-dependent focal-plane motion (field rotation), which a two-axis fast steering mirror cannot remove; it only relocates the apparent rotation center to the active guide star, sacrificing the rest of the field. The argument is carried by a closed-loop simulation combining optical ray tracing through the FSM, asynchronous guide-star centroid measurements with zero-order hold, an age-weighted multi-star estimator (weights decay with measurement age), and PI control of the FSM. The empirical anchor is the coupling coefficient k_fit between passive star-camera and roll-star-camera motion, measured from flight data and used to","core_discovery":"The central claim is that roll leakage is an inherent feature of single-guide-star fine-guidance architectures, not an occasional disturbance. The evidence is flight data: across 2348 SuperBIT science exposures with two guide stars, the passive star camera's centroid motion correlates with the roll star camera's independent measurement in the majority of cases (|ρ|>0.5), and the fitted coupling coefficient k_fit exceeds the rigid-body geometric prediction by ~27×, showing amplification that is still roll-driven. The paper further claims that a multi-star estimator, which fuses asynchronous guide-star centroids with age-based weights before commanding the fast steering mirror, reduces average","pith_inferences":["If the ~27× amplification of the coupling coefficient is partly due to unmodeled torsional flexure in the optical assembly rather than FSM projection effects, then stiffening the structure could yield a similar reduction without multi-star fusion; the paper's simulation would then overstate the transferable gain to GigaBIT.","The same roll-leakage argument generalizes to any two-axis fine-steering system, including space observatories; the historical use of two guide stars on Hubble is a precedent that this paper re-derives from flight data.","A direct testable extension: in SuperBIT data, fit the full focal-plane rotation model (rotation center + amplitude) from the passive FSC and science PSF ellipticities; if the inferred rotation center matches the active guide-star location, the leakage mechanism is confirmed.","The 77.4% GigaBIT figure is likely a lower bound, since the simulation uses a scaled SuperBIT geometry with equal nominal weights; optimizing guide-star placement in GigaBIT's actual annular focal plane could improve the outcome further."],"forward_implications":["Adopting multi-star fine guidance becomes a concrete design recommendation for GigaBIT, where single-guide-star control leaves 2.343″ science-field smear and multi-star cuts it to 0.529″.","The benefit of multi-star estimation scales with focal-plane size, so any future wide-field balloon observatory faces the same tradeoff.","The simulation framework provides a reusable tool to test guidance architectures before flight, using SuperBIT's measured disturbance environment.","The age-weighted fusion is a simple proof of concept; weighting by guide-star brightness or proximity to the science field could improve the gains further.","Multi-star guidance complements coarse roll control; improving gimbal roll rejection reduces the same leakage from the other side."],"fun_headline_variants":["Roll leakage inherent in single-star guidance; multi-star fixes it","Multi-star fusion cuts roll smear up to 77%","Single-guide-star roll leak? Multi-star fusion cuts it by 77%","Roll leakage is not a fluke: multi-star guidance trims it"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The simulated multi-star benefit rests on an empirical scaling factor (mean |k_fit|≈3.60, used as 58.39 in the model) that is about 27 times larger than the rigid-body geometric prediction; if this amplification is caused by unmodeled torsional flexure or common-mode noise rather than roll-driven FSM projection effects, the predicted reduction may not transfer from SuperBIT to GigaBIT.","fun_headline_variants_meta":{"raw":{"variants":["Roll leakage inherent in single-star guidance; multi-star fixes it","Multi-star fusion cuts roll smear up to 77%","Single-guide-star roll leak? Multi-star fusion cuts it by 77%","Roll leakage is not a fluke: multi-star guidance trims it"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000611,"raw_usage":{"total_tokens":2705,"prompt_tokens":794,"completion_tokens":1911,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":538,"completion_tokens_details":{"reasoning_tokens":1835}},"tokens_in":538,"tokens_out":1911,"duration_ms":15033,"temperature":1.0,"reasoning_tokens":1835,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T00:18:10.264268+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test would be to measure the coupling between passive focal-plane star-camera motion and the roll star camera while commanding known fast-steering-mirror tilts during a balloon flight. If the amplification factor k_fit drops to the rigid-body geometric prediction when the FSM is disabled, the leakage is FSM projection and the simulation's scaling is valid; if it persists with the FSM off, the amplification is structural and the simulated transfer to GigaBIT is suspect. Alternatively, fly GigaBIT with switchable single- and multi-star guidance and compare science-field RMS image moti","supporting_citations":[],"review_version":1}