{"id":"c77836e9-45ab-4ae4-aefa-a978130cb379","arxiv_id":"2411.09110","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":5,"one_line_summary":"A multi-spacecraft positioning method that uses a stochastic-contraction-based uncertainty ellipsoid and a coverage-overlap cost function to maximize visual information during interstellar object flybys.","lead":"This paper proposes a way to position a swarm of spacecraft around an interstellar object so that, taken together, their cameras capture as much of the object as possible despite large uncertainty in its position. The method first computes a ball where the object is likely to be, then spreads the spacecraft around that ball to maximize visual coverage and minimize redundant images.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1's probability bound (Eq. 7) does not follow from the proof: Markov applied to the quadratic Lyapunov function gives D^2, not D, in the denominator.","rationale":"The paper has two advertised contributions: (i) a formal probabilistic encounter bound via hierarchical stochastic contraction (Theorem 1/Eq. 7), and (ii) an information-cost optimization for deputy positions around the resulting uncertainty ellipsoid. Of these, (i) is the load-bearing foundation: Eq. (8) converts the bound into an ellipsoid, and all simulation setups use that ellipsoid. My check of the proof's final step shows the bound cannot be obtained from the preceding E[Vec] estimate by Markov's inequality unless an additional estimate of E‖x − x_d‖ or a D^2 denominator is supplied. This is an internal inconsistency, not a disagreement with consensus; it makes the 'formal probabilistic guarantees' claim unproven as stated. The cost function concerns raised by the reader (unit-inconsistent κ − ε, unclear aggregation of ε) are real, but they affect the secondary optimization layer and are more easily repaired; the Theorem 1 step is more fundamental because if the ellipsoid is not probabilistically guaranteed, the optimization target loses its meaning. I therefore recommend rejecting, or requiring major revision of, the paper. Credit: the hierarchical contraction framework is a plausible direction, and the paper openly acknowledges several limitations; however, the central theorem's proof gap is decisive. The reader's stated weakest assumption (S not quantified) is related but distinct; my concern is that even within S, Eq. (7) does not follow.","tokens_in":14702,"tokens_out":15401,"duration_ms":219787,"concrete_test":"Independently re-derive the final step of Theorem 1: from E[Vec(t)] ≤ (E[Vec(0)] + ζ̄_I(t)) e^{-2α_s t} + c_s, use Markov's inequality on the event {Vec ≥ m_c D^2}, which is implied by {‖x − x_d‖ ≥ D}. If the resulting bound has denominator m_c D^2 (or requires an additional bound on E‖x − x_d‖) rather than D m, Eq. (7) is false as stated. This algebraic check settles the concern without new simulations.","verdict_should_be":"REJECT","load_bearing_attack":"The central probabilistic guarantee, Theorem 1/Eq. (7), is not a valid consequence of the proof as written. The proof bounds E[Vec(t)], where Vec(t) = ∫_0^1 (∂_μ q_c^T M_c ∂_μ q_c + λ ∂_μ q_e^T M_e ∂_μ q_e) dμ is a quadratic incremental Lyapunov function. For the event {‖x(t) − x_d(t)‖ ≥ D}, the only available lower bound is Vec ≥ m_c ‖x − x_d‖^2 (Cauchy–Schwarz on the path integral from x_d to x), so Markov's inequality gives P[‖x − x_d‖ ≥ D] ≤ E[Vec] / (m_c D^2). The paper instead divides by D m, which is not derived from any stated estimate of E[‖x − x_d‖] and is dimensionally inconsistent if the state carries units. A scalar example (a random variable with P(|e|=0.5)=0.9, E[e^2]=0.225, m=1, D=0.5) has E[Vec]/(D m) = 0.45 while the probability is 0.9, showing the D-linear Markov step is false in general. Since Eq. (8) builds the uncertainty ellipsoid on this bound, the formal probabilistic anchor of the multi-spacecraft positioning framework is unproven as stated; at minimum the denominator should be D^2 m_c and the proof re-worked.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a multi-spacecraft positioning framework for interstellar object (ISO) encounters. It first derives a probabilistic bound on the terminal position error of a chief spacecraft using stochastic contraction theory (Theorem 1), constructs from that bound an uncertainty ellipsoid around the ISO's expected terminal position, and then optimizes deputy spacecraft terminal positions to maximize visual coverage of points of interest while minimizing overlap, using the cost function I = κ_total − ε (Eq. 9). Numerical simulations are presented for a single spacecraft's probability of viewing an ISO and for determining an optimal number of spacecraft. The central claims are that Theorem 1 provides formal probabilistic guarantees on the ISO encounter and that the proposed optimization locally maximizes visual information.","tokens_in":15009,"tokens_out":4901,"duration_ms":51306,"significance":"If the results were correct, the framework would offer a principled, precomputable way to position a spacecraft swarm under large state uncertainty, a genuinely useful capability for ISO flyby missions. The paper builds on established stochastic contraction machinery and attempts to connect a formal probabilistic bound to a concrete coverage optimization. However, the proof of the main theorem contains a fundamental error that invalidates the probability bound, and the simulation results do not support the claimed optimality conclusions. Because these issues affect the paper's central contributions, the significance of the work as it stands is not established.","major_comments":[{"comment":"The proof of Theorem 1 incorrectly applies Markov's inequality. From the definitions in the proof, Vec(t) is a quadratic incremental Lyapunov function, and the lower bound on Vec(t) in terms of the state error is Vec(t) ≥ m_c ‖x(t) − x_d(t)‖², not a bound linear in the error. Consequently Markov's inequality gives P[‖x(t) − x_d(t)‖ ≥ D] ≤ E[Vec(t)] / (m_c D²), not the claimed E[V(0)]e^(−2αs t) + cs + e^(−2αs t) ζ̄I(t) over D m. The denominator D m is dimensionally inconsistent if the state carries units, and the paper provides no alternative derivation. A scalar example makes the failure concrete: for a random variable with P(|e|=0.5)=0.9, E[e²]=0.225, m=1, D=0.5, the claimed bound gives 0.45, which is less than the true probability 0.9. Since Eq. (8) and the entire uncertainty ellipsoid construction rely on (7), the main probabilistic guarantee of the paper is unproven as stated.","section":"§3, Theorem 1 and Eq. (7)"},{"comment":"The information cost I = κ_total − ε mixes quantities with incompatible units: κ_total is an angular measure of field-of-view overlap (in radians or degrees), while ε is a distance from a point to the FOV cone axis (in length units). Without normalization or dimensionless weighting, the cost is not invariant under changes of units, and the trade-off between overlap minimization and coverage maximization is undefined. This makes the claim that minimizing I 'locally maximizes visual information' not well-posed. The paper should either define a normalized, physically motivated cost or justify the implicit weighting.","section":"§4, Eq. (9) and cost function formulation"},{"comment":"The text states that five spacecraft is the optimal number, 'producing the lowest information cost in the simulations'. This is contradicted by the reported data: Table 5 shows the average negative information cost is 96.6 for three spacecraft and 96.1 for five, so five does not give the lowest cost; Table 6 shows the mean fraction of POIs viewed is 98.6% for seven spacecraft and 96.7% for six, both higher than the 92.7% for five. Additionally, Trial 2 of Table 4 reports 100% coverage with four spacecraft. The empirical basis for the claimed optimality is therefore not supported by the paper's own tables.","section":"§5, Tables 4–6 and the 'optimal number of spacecraft' claim"},{"comment":"The uncertainty sphere radii used in the simulations (50, 500, 1000, and 100 units) are treated as free inputs rather than being derived from the bound (7) or (8). The empirical success probabilities in Tables 2 and 3 are not compared with the right-hand side of (8), so the paper never validates the formal guarantee it derives. For the claimed 'formal probabilistic guarantees' to be credible, the simulation should show that the empirical encounter probability is at least the lower bound from (8) for the same D, or should otherwise explain why the bound is not applied.","section":"§3 and §5, connection between the theoretical bound and simulations"}],"minor_comments":[{"comment":"The remark states that when two spacecraft have the same orientation θ, the formula gives κ = 0, and then adds a small perturbation to θ. In fact, substituting θ_i = θ_j into the definition κ = max(0, κ_end − κ_start) yields κ = 2ν, which is nonzero. The remark's premise is incorrect, and the proposed perturbation is unnecessary as a remedy for the described problem.","section":"§4, Remark 2"},{"comment":"The text says the probability of viewing the ISO decreases as the uncertainty sphere size increases, but Table 3 shows nonmonotonic behavior: the probability is 25.0% for radius 50, 16.7% for radius 500, and 25.0% for radius 1000. The discussion acknowledges nonlinearity and initial-condition dependence, but the original general claim should be qualified more carefully.","section":"§5, Table 3 and surrounding text"},{"comment":"The virtual system (5b) is not clearly tied to the actual estimation dynamics: setting μ = 0 is claimed to give (1a), but (5b) with μ = 0 still contains the ℓ_d(h(q_e1,q_e2,t)−h(œ,x,t)) term and no noise, which does not match (1a) as written. The relationship between q_e and the true/estimated states should be clarified.","section":"§3, proof of Theorem 1"},{"comment":"There are several typographical and notational issues, including the repeated 'It ˆo' (Ito) rendering, the garbled '∥ℓa( ˆœ, ˆx,t)∥2 F' in the assumptions of Theorem 1, and the inconsistent use of x_d and p_d. A careful proofreading pass is needed.","section":"Throughout"}],"recommendation":"reject","confidential_remarks":"The core theoretical result (Theorem 1) has a demonstrably incorrect Markov step, and the empirical claims about the optimal number of spacecraft are contradicted by the paper's own tables. These are load-bearing issues that cannot be fixed without a substantial re-derivation and new simulations. The manuscript may contain a useful idea, but in its current form it should not be accepted. I recommend rejection, while noting that a corrected version with a valid probability bound and properly validated simulations might be reconsidered."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a real attempt at a real problem, but the central probability bound in Theorem 1 has a flawed final step, and the paper's own data don't support the claimed optimal swarm size.\n\nWhat's new and good: the combination of stochastic contraction theory with a multi-spacecraft coverage-overlap cost for interstellar object encounters is genuinely new, at least relative to the cited literature. Prior ISO work is single-spacecraft; prior swarm coverage work assumes known orbits. The idea of precomputing deputy terminal positions offline from an uncertainty-history bound is practical and sensible. The paper also acknowledges some simulation limitations honestly, e.g., that initial conditions inside the uncertainty sphere are not accounted for.\n\nThe soft spots are not cosmetic. First, Theorem 1: the proof bounds E[Vec(t)] and then claims P[||x - x_d|| >= D] <= E[Vec]/(D m) via Markov. But the available lower bound is Vec >= m_c ||x - x_d||^2, so Markov gives D^2 in the denominator, not D·m. The m here also seems to mix the estimator and controller parts, only the controller part lower-bounds the position error. That is a genuine error in the load-bearing guarantee, and Eq. (8), the ellipsoid probability, inherits it. Second, the cost function I = kappa_total - epsilon subtracts an angular overlap measure from a distance measure with no weights; the units don't make sense as written. Third, the claimed 'optimal' five-spacecraft result doesn't follow from Tables 4-6: Table 5 shows the lowest (most negative) information cost at three spacecraft, and coverage keeps rising through seven. The numbers are noisy and based on three trials, so 'five' looks like a misreading. Fourth, the probability bound is never evaluated against the simulation probabilities, so the 'probabilistic guarantee' is never checked.\n\nWho is this for? Aerospace GNC researchers working on ISO missions or, more broadly, on using contraction theory for uncertainty-aware swarm placement. They'll find the formulation useful, but they'd need to redo the theorem and the cost-function design before relying on it.\n\nRecommendation: send to peer review. The problem is real, the combination is new, and the authors are clearly thinking carefully, but the paper needs major revision - the theorem proof first, then the cost function and the simulation analysis.","headline":"A promising combination of stochastic contraction and swarm coverage for interstellar object flybys, but the main theorem's Markov step has a D vs D^2 error and the optimal-swarm-size claim conflicts with their own tables.","tokens_in":15540,"tokens_out":3377,"would_cite":false,"duration_ms":55511,"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":"This paper proposes a multi-spacecraft framework that locally maximizes the visual information gained during an interstellar object flyby, while providing a formal probability bound on the encounter.","keywords":["interstellar objects","multi-spacecraft planning","uncertainty ellipsoid","hierarchical stochastic contraction","probabilistic encounter guarantee","field-of-view coverage","information cost","flyby mission design"],"falsifier":"Re-run the encounter simulations with noise realizations that push the relative state outside that compact region, and compare the empirical fraction of trials where the spacecraft fails to view the ISO to the bound in expression (8); if the empirical failure rate exceeds the bound, the probabilistic guarantee does not actually hold.","tokens_in":14499,"feed_emoji":"🛰️","tokens_out":8868,"duration_ms":78339,"temperature":0.7,"pith_summary":"Interstellar objects such as 'Oumuamua barrel through the solar system once and never return, so a flyby has one chance to gather as much visual data as possible. This paper claims that a small swarm of spacecraft, positioned around an uncertainty ellipsoid in which the object is predicted to lie, can locally maximize that data and keep overlap between spacecraft cameras small. The formal backbone is a contraction-theory bound on the probability that the spacecraft misses the object, which is derived under explicitly approximated control and estimation policies so that onboard computation stays feasible. Numerical experiments with a quasi-realistic empirical population of interstellar objects show that multi-spacecraft systems observe more points of interest than a single spacecraft, with a local optimum around five spacecraft in the tested case. If correct, the method lets mission planners precompute terminal positions offline, freeing the spacecraft to respond autonomously during the encounter.","feed_headline":"Multi-probe swarm maximizes data from interstellar flybys","feed_subtitle":"Each probe is placed around a predicted uncertainty ellipsoid to maximize camera coverage of rare interstellar visitors.","key_machinery":"The load-bearing object is the hierarchical stochastic contraction bound of Theorem 1: a differential Lyapunov function $V_{\\mathrm{ec}} = \\int_0^1 (\\partial_\\mu q_c^\\top M_c \\partial_\\mu q_c + \\lambda\\, \\partial_\\mu q_e^\\top M_e \\partial_\\mu q_e)\\,d\\mu$ over virtual systems that interpolate between target and actual trajectories yields the failure-probability estimate (7), and hence the uncertainty ellipsoid (8) in which the ISO lies with finite probability. The second piece is the information cost $I = \\kappa_{\\mathrm{total}} - \\varepsilon$: $\\varepsilon$ counts how many sampled points of interest fall inside a cone-shaped field of view, and $\\kappa_{\\mathrm{total}}$ measures the angular overlap between any two spacecraft's fields of view. Together they convert the probabilistic encounter statement into a well-posed, non-differentiable optimization over spacecraft terminal positions and camera orientations, which the paper solves with Nelder-Mead.","core_discovery":"The central claim is that the interstellar-object encounter problem splits into two coupled tasks that can be solved with formal guarantees. First, using hierarchical stochastic contraction, the paper turns the large and rapidly changing uncertainty in the object's state into an upper bound on the probability that the chief spacecraft misses its target (Theorem 1); setting the desired relative position to zero makes this an ellipsoid in which the object lies with a known probability. Second, the paper distributes deputy spacecraft around that ellipsoid by minimizing a cost function $I = \\kappa_{\\mathrm{total}} - \\varepsilon$, where the $\\varepsilon$ term rewards the number of visible points of interest in each spacecraft's conal field of view and the $\\kappa_{\\mathrm{total}}$ term penalizes overlap between fields of view. The numerical simulations show that the optimized swarm observes more of the ellipsoid than a single spacecraft, that adding spacecraft eventually saturates because overlap rises, and that the entire positioning can be precomputed offline from an upper bound on the uncertainty history.","pith_inferences":["A natural extension the paper leaves implicit is weighting points of interest by scientific value instead of sampling them uniformly; that would shift the optimal terminal distribution toward the most scientifically informative regions of the ellipsoid.","The 'five spacecraft is optimal' result is an empirical statement tied to the simulated sphere radius and camera model; a likely testable scaling law is that the optimal swarm size grows with the ellipsoid's volume and shrinks with the camera's field-of-view angle.","Because Theorem 1 is local, the practical guarantee is only as good as the size of the compact set over which the approximation bounds hold; checking whether realistic ISO encounter trajectories actually stay inside such a set is an empirical question the paper does not resolve.","The probability bound could be inverted into a sensor-requirement tool: given a required confidence of seeing the ISO, planners could solve for the maximum allowed state uncertainty and thus the needed estimation accuracy."],"forward_implications":["Mission planners can precompute the deputy spacecraft's terminal positions offline using only an upper bound on the ISO's uncertainty history, so the onboard computer is free for other autonomous tasks.","If every deputy is placed around the uncertainty ellipsoid, at least one spacecraft sees the ISO with probability at least the right-hand side of expression (8).","Coverage gains saturate with swarm size: in the paper's simulations the percentage of points of interest viewed grows roughly logarithmically and then flattens, while the information cost rises from overlap, so there is a locally optimal number of spacecraft (five in the tested setup).","The framework extends to stochastic terminal positions by minimizing the expected information cost, so it tolerates additional process noise and state uncertainty without changing the positioning logic.","The information-cost formulation applies equally to other fast flyby targets such as comets or asteroids, since it only needs a sampled uncertainty region and a camera model."],"supporting_citations":[{"why":"Establishes the approximated control policy ua for ISO rendezvous whose error bounds feed into the approximation assumption (4).","marker":"[7]"},{"why":"Provides the contraction theory foundation used to define the contracting parts of the virtual system.","marker":"[22]"},{"why":"Extends contraction theory to stochastic systems, supplying the hierarchical stochastic contraction property used in Theorem 1.","marker":"[23]"},{"why":"Tutorial reference for contraction analysis of learning-based control, invoked for finding metrics Mc and Me.","marker":"[26]"},{"why":"Supplies the navigation evaluation whose empirical uncertainty history bounds the noise term in the ellipsoid construction.","marker":"[9]"},{"why":"Source of the empirical ISO population and uncertainty history used in the numerical simulations.","marker":"[6]"},{"why":"Source of the conal field-of-view approximation used in the information cost.","marker":"[31]"},{"why":"Prior information-based guidance architecture for multi-spacecraft inspection that the cost function extends.","marker":"[14]"},{"why":"Nelder-Mead optimizer used to solve the non-differentiable cost-function minimization.","marker":"[33]"}],"fun_headline_variants":["Swarm around uncertainty ellipsoid boosts ISO imagery","Multi-probe placement maximizes information from flybys","Optimal probe swarm for uncertain interstellar encounters","Ellipsoid-based swarm positioning for ISO data","Probes positioned on uncertainty ellipsoid maximize data"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole probability guarantee collapses if the spacecraft's realized state leaves the compact region in which the approximated control and estimation policies stay within their assumed error bounds and the contraction metrics exist.","fun_headline_variants_meta":{"raw":{"variants":["Swarm around uncertainty ellipsoid boosts ISO imagery","Multi-probe placement maximizes information from flybys","Optimal probe swarm for uncertain interstellar encounters","Ellipsoid-based swarm positioning for ISO data","Probes positioned on uncertainty ellipsoid maximize data"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001006,"raw_usage":{"total_tokens":4276,"prompt_tokens":989,"completion_tokens":3287,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":605,"completion_tokens_details":{"reasoning_tokens":3225}},"tokens_in":605,"tokens_out":3287,"duration_ms":21522,"temperature":1.0,"reasoning_tokens":3225,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:01:11.370251+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the encounter simulations with noise realizations that push the relative state outside that compact region, and compare the empirical fraction of trials where the spacecraft fails to view the ISO to the bound in expression (8); if the empirical failure rate exceeds the bound, the probabilistic guarantee does not actually hold.","supporting_citations":[{"cited_title":"Neural-Rendezvous: Provably-based robust guidance and control to en- counter interstellar objects,","cited_arxiv_id":null,"evidence_quote":"Establishes the approximated control policy ua for ISO rendezvous whose error bounds feed into the approximation assumption (4)."},{"cited_title":"On contraction anal- ysis for nonlinear systems,","cited_arxiv_id":null,"evidence_quote":"Provides the contraction theory foundation used to define the contracting parts of the virtual system."},{"cited_title":"A contrac- tion theory approach to stochastic incremental stability,","cited_arxiv_id":null,"evidence_quote":"Extends contraction theory to stochastic systems, supplying the hierarchical stochastic contraction property used in Theorem 1."},{"cited_title":"Con- traction theory for nonlinear stability analysis and learning-based control: A tutorial overview,","cited_arxiv_id":null,"evidence_quote":"Tutorial reference for contraction analysis of learning-based control, invoked for finding metrics Mc and Me."},{"cited_title":"Navigation evaluation for fast interstellar object fly- bys,","cited_arxiv_id":null,"evidence_quote":"Supplies the navigation evaluation whose empirical uncertainty history bounds the noise term in the ellipsoid construction."},{"cited_title":"Interstellar object accessibility and mission design,","cited_arxiv_id":null,"evidence_quote":"Source of the empirical ISO population and uncertainty history used in the numerical simulations."},{"cited_title":"Periph- eral vision and pattern recognition: A review,","cited_arxiv_id":null,"evidence_quote":"Source of the conal field-of-view approximation used in the information cost."},{"cited_title":"Information-based guidance and control architecture for multi-spacecraft on-orbit inspection,","cited_arxiv_id":null,"evidence_quote":"Prior information-based guidance architecture for multi-spacecraft inspection that the cost function extends."},{"cited_title":"Numerical optimization strate- gies for springback compensation in sheet metal form- ing,","cited_arxiv_id":null,"evidence_quote":"Nelder-Mead optimizer used to solve the non-differentiable cost-function minimization."}],"review_version":1}