{"id":"ecf765f2-cc94-495e-a911-f178db4d3146","arxiv_id":"2509.06092","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Capture is guaranteed exactly when the attacker-target Apollonius circle lies inside the newly defined sensable region, which yields speed thresholds for guaranteed capture or escape.","lead":"A three-agent pursuit game where a slow sensor must keep a target inside its sensing range so a fast attacker can intercept it. The paper gives geometric conditions and speed thresholds that decide between guaranteed capture and guaranteed escape.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 5's printed quadratic has the wrong sign on the D·Pθ terms: with the paper's own Section V-D parameters it gives roots (0.1416,0.3956), not the reported (0.1765,0.4896), so the sharper escape threshold is unsupported as stated.","rationale":"The reader's stated weakest assumption (passive, non-reacting target) is real but is part of the paper's explicit problem definition, so I do not treat it as the most load-bearing internal risk. The most concrete correctness risk is in Theorem 5, one of the four advertised contributions. The printed coefficients (69)-(71) do not follow from the paper's own Eq. (58) with the correct angle convention, and they do not reproduce the roots computed in Section V-D. Re-deriving with the correct law-of-cosines angle gives plus-sign coefficients, which match the reported roots. Thus the theorem as stated cannot be used to obtain the claimed sharper escape bound. This is fixable and does not affect Lemma 2 or Theorems 2-3, so the reader's conditional verdict is retained.","tokens_in":19891,"tokens_out":27455,"duration_ms":304936,"concrete_test":"Re-derive Theorem 5's quadratic from the law of cosines in triangle C_Ap0-T0-Ttf: write R_Ap² = α²d² + b² - 2αd·b·cos(θAT0-θST0) (not with π - ...), substitute α=µ²/(1-µ²), b=(R-dST0)/(1-ν), µ=v/vA, ν=vS/v, and expand. If the result has plus signs (1+2DP+D² and 1+DP) rather than the printed minus signs, the printed theorem is wrong. Also recompute the roots of the printed quadratic with the Section V-D parameters; if they are 0.1416 and 0.3956 rather than 0.1765 and 0.4896, the numerical section is using a different formula than the one stated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's fourth headline contribution is the sharper escape-speed threshold vT⋆ of Theorem 5. The printed coefficients (Eqs. 69-71) are inconsistent with the derivation and with the paper's own numerical example. In the law-of-cosines step (Eq. 58), the angle at T0 between T0C_Ap and T0Ttf is θAT0-θST0, not π-(θAT0-θST0); using the correct angle gives a cross term with the opposite sign. The resulting quadratic is (1+2DPθ+D²)v² - 2vS(1+DPθ)v + (vS² - D²vA²) = 0, with D=(R-dST0)/dAT0 and Pθ=cos(θAT0-θST0). The printed version instead has 1-2DPθ+D² and -2vS(1-DPθ). With the Section V-D data (S0=(0,0), A0=(3,3), T0=(1.5,0.5), vS=0.3, vA=1, R=2, dST0=1.5811, dAT0=2.9155, θAT0=-120.96°, θST0=18.43°), the printed coefficients give roots 0.1416 and 0.3956, whereas the paper reports 0.1765 and 0.4896. The reported roots are exactly reproduced by the corrected plus-sign coefficients. Thus Eq. (57), as printed, does not yield the advertised vT⋆; the 'sharper upper bound' needs either a corrected formula or a corrected derivation. This does not invalidate Lemma 2 or Theorems 2-3, but it is a load-bearing error for one of the four stated contributions.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper considers a planar three-agent pursuit-evasion game between a cooperative sensor-attacker team and a passive (constant-heading) target. The sensor is slower than the target, which is in turn slower than the attacker; the sensor has limited sensing radius R, and the attacker can only rely on the sensor's measurements. The target wins if it escapes the sensing radius before being intercepted; the team wins if the attacker intercepts the target while it remains within the sensing disc. The authors propose an optimal sensor heading that maximizes the time before escape (Lemma 1), an intercept heading for the attacker based on Apollonius-circle geometry (Eq. (45)), and a 'sensable region' S0 of target positions reachable before escape (Definition 1). The central result, Lemma 2, states that capture is guaranteed if and only if the initial Apollonius circle A0 is fully contained in S0. Theorems 2 and 3 give respective target-speed thresholds guaranteeing capture and escape for arbitrary initial orientations. Theorem 4 asserts the existence of a critical speed separating the two regimes, and Theorem 5 claims a sharper, orientation-dependent escape-speed threshold obtained by solving a quadratic. Numerical sections illustrate the thresholds and trajectories.","tokens_in":20375,"tokens_out":17374,"duration_ms":179889,"significance":"If the central geometric characterization is correct, the paper is a useful contribution to the cooperative pursuit-evasion literature. Lemma 2 and Theorems 2 and 3 are explicit, parameter-free conditions, derived from transparent geometry, and the numerical examples are consistent with those results. The sensable-region/containment viewpoint is clean and likely to be reused. The paper does not rely on curve fitting or hidden normalization; the thresholds are algebraic consequences of the geometry. However, the fourth contribution (Theorem 5) contains a concrete sign error, and the relationship between the computed threshold and the exact critical speed of Theorem 4 is conflated. These issues affect the manuscript's headline claims and require correction. The main framework is sound and the errors appear fixable within the scope of the paper.","major_comments":[{"comment":"The printed quadratic is inconsistent with the derivation and with the paper's own numerical example. In Eq. (58), the angle at T0 in triangle ΔC_Ap^0 T0 Ttf is the angle between the vectors T0→C_Ap^0 and T0→Ttf. Since C_Ap^0 lies on the ray from A0 through T0, this angle is θ_AT0−θ_ST0 (for γ_T=θ_ST0), not π−(θ_AT0−θ_ST0). Using the printed sign gives coefficients a,b with the wrong signs in Eqs. (69)–(70). With the Section V-D parameters (S0=(0,0), A0=(3,3), T0=(1.5,0.5), vS=0.3, vA=1, R=2), the printed quadratic yields roots 0.1416 and 0.3956, whereas the paper reports 0.1765 and 0.4896. The reported roots are reproduced by the corrected coefficients a=1+2DPθ+D², b=−2vS(1+DPθ), c=vS²−D²vA². Eq. (60) also has the sign of the μPθ term reversed; the correct solutions have μPθ±√(1−μ²Qθ²). Since Theorem 5 is one of the four stated contributions, the formulas must be corrected and the numer","section":"Section IV, Theorem 5 / Eqs. (58)–(71)"},{"comment":"The symbol v_{T⋆} is used for two different objects. Theorem 4 asserts the existence of an exact critical speed with capture for v_T≤v_{T⋆} and escape for v_T>v_{T⋆}. Section V-D then calls the Theorem 5 root 'the valid critical speed.' However, Remark 7 states that when v_T<v_{T⋆} (the value from Theorem 5), capture cannot be concluded because the true critical speed may be lower. This directly contradicts Theorem 4(2) if the symbols denote the same quantity. The proof of Theorem 4 does not identify the exact tangency speed, and the Theorem 5 computation is conditional on the assumption that tangency occurs at the minimum-escape point. The paper must either prove that the Theorem 5 root is the exact critical speed of Theorem 4, or clearly present it as a sufficient escape threshold and revise Theorem 4 and the Conclusion accordingly.","section":"Section IV, Theorems 4 and 5 / Remark 7"},{"comment":"Even apart from the sign error, the theorem is not established as stated. The proof assumes that the tangency point of A0 and S0 coincides with the point of minimum escape distance; the sentence before Theorem 5 says 'by assuming that the tangency point coincides with the point of minimum escape distance.' No argument is given that this assumption holds for the given initial geometry, and Remark 5 acknowledges that multiple tangency points may occur. The proof also does not show that the computed root is < v̄T for all admissible parameters, nor that v_T>v_{T⋆} implies A0⊄S0 without the min-escape-point assumption. The numerical example is consistent with the corrected algebra, but the theorem's existence claim over the parameter domain requires a proof or a more carefully bounded statement.","section":"Section IV, Theorem 5 proof"}],"minor_comments":[{"comment":"The attacker strategy is described as 'proportional navigation,' but Eq. (45) is a geometric intercept heading (essentially pure/parallel pursuit to the predicted intercept point). Please use terminology consistent with the actual guidance law.","section":"Abstract and Section I"},{"comment":"The notation v_T (target speed and a threshold), v̄_T, and v_{T⋆} is easy to confuse. Consider using distinct symbols, e.g., v_T^-, v_T^+, v_T^*, for the three thresholds.","section":"Throughout"},{"comment":"The 'if and only if' in Lemma 2 is for passive (open-loop) target headings. Since Assumption 1 nevertheless gives the target full knowledge of the team's strategies, a reactive target would invalidate the sufficiency direction. This limitation should be stated in the main text more prominently, not only in the introduction.","section":"Assumption 1 / Lemma 2"},{"comment":"Remark 4 claims a first-order sensitivity result without a derivation. Either provide the short calculation or label the statement as an observation.","section":"Remark 4"},{"comment":"The sentence 'by Theorem 2, this containment persists for all future times' is stronger than what Theorem 2 states. Theorem 2 guarantees capture under the speed threshold; the persistence claim needs a proof or rewording.","section":"Section V-B"}],"recommendation":"major_revision","confidential_remarks":"The skeptical report's sign check is correct; I verified it against the paper's own numerical example. The main geometric framework (Lemma 2, Theorems 2 and 3) is sound and worth publishing after revision, but the Theorem 5 formulas and the conflation with Theorem 4's critical speed need to be fixed before the paper can be accepted. I recommend major revision, not rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper is worth a careful read. The core idea—a sensor and attacker cooperating under the constraint that the target must stay inside the sensor's disc—is a genuine new formulation in the pursuit-evasion line, and the if-and-only-if capture condition in Lemma 2 is the right way to frame it. The sensable region is a new construct, and the speed thresholds in Theorems 2 and 3 follow cleanly from the containment geometry. The paper does honest work: the derivations are explicit, the simulations match, and Remark 7 openly acknowledges that Theorem 5 is only sufficient, not necessary.\n\nThe main problem is Theorem 5. As printed, the quadratic in Eq. (57) with coefficients (69)–(71) has the wrong sign on the D·Pθ terms. Running their own Section V-D numbers gives roots 0.1416 and 0.3956, not the reported 0.1765 and 0.4896. The corrected sign—which comes from using the correct included angle θAT0−θST0 in the law of cosines at Eq. (58)—reproduces their reported values exactly. So the sharper escape threshold is conceptually right, but the theorem statement as printed is wrong. That needs fixing before anyone builds on it.\n\nTwo smaller issues. Definition 2 gives the Apollonius circle as ||P A||/||P T|| = µ, but Eqs. (40)–(41) (and the attacker's optimal intercept reasoning) use ||P T||/||P A|| = µ. The formulas are what get used, so the definition should be corrected to match. And the passive-target assumption (Assumption 1) is a real limitation: the thresholds rely on the target sticking to a constant heading. The paper states this plainly, which I credit, but it means the results are about open-loop targets, not reactive ones.\n\nNone of this sinks the central argument. Lemma 2 and Theorems 2–3 are the main contributions, and they appear sound. The Theorem 5 issue is load-bearing for one of the four advertised contributions, but it is a sign slip, not a conceptual failure. A careful referee would catch it.\n\nThis paper is for the pursuit-evasion and guidance subfield. It deserves peer review; the fix is modest. I would engage with it.\n\nBest.","headline":"Core containment result and Theorems 2–3 are sound and worth building on; Theorem 5's printed quadratic has a sign error, though the corrected version matches the paper's own numbers.","tokens_in":20792,"tokens_out":9102,"would_cite":true,"duration_ms":87863,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["91A24","49N75"],"pacs":[],"model":"deepseek-v4-flash","headline":"The attacker can guarantee capture of a constant-heading target if and only if the Apollonius circle between attacker and target is fully contained in the sensor's sensable region; speed thresholds below and above this condition decide the","keywords":["pursuit-evasion game","game of kind","sensable region","Apollonius circle","sensing constraints","open-loop target","speed threshold","cooperative pursuit"],"falsifier":"Simulate an engagement in which the target is allowed to change its heading once at an arbitrary time t > 0, rather than fixing its heading at t = 0. A single such trajectory that escapes from an initial configuration where the Apollonius circle is fully contained in the sensable region would refute the necessity of the containment condition for guaranteed capture.","tokens_in":19836,"feed_emoji":"🎯","tokens_out":3897,"duration_ms":44451,"temperature":0.7,"pith_summary":"This paper studies a three-agent pursuit game: a slow sensor with a finite sensing radius, a fast attacker that depends on the sensor for target information, and a target that is faster than the sensor but slower than the attacker. The target is assumed to be passive, committing to a single constant heading at the start of the game. The authors derive an optimal sensor strategy that maximizes how long the target stays inside the sensing radius, and an attacker strategy that intercepts the target in minimum time along Apollonius-circle geometry. Their central claim is that capture is guaranteed exactly when the attacker's Apollonius circle lies entirely within the 'sensable region' — the set of positions the target can reach before escaping the sensor's range. From this geometric containment condition they produce explicit target-speed thresholds: below one speed capture is guaranteed for any initial geometry, above another escape is guaranteed, and in between a critical speed exists at which the two regions first become tangent.","feed_headline":"Target caught iff attacker's circle fits inside sensor's reach","feed_subtitle":"Geometric containment decides the game; two speed thresholds bracket the outcome.","key_machinery":"Two geometric objects carry the argument. The sensable region S_t is the set of all points the target can reach by moving at constant speed on some constant heading during the interval before it exits the sensor's radius, assuming the sensor follows its time-maximizing heading. The Apollonius circle A_t is the locus of points that the attacker and target can reach simultaneously, given their speed ratio μ. The load-bearing condition is containment: A_0 ⊂ S_0 means that whatever constant heading the target chooses, it cannot reach the boundary of sensing before the attacker reaches it. Theorems 2 and 3 translate this containment into explicit inequalities on the target speed, and Theorem 5 re","core_discovery":"The paper's core claim is that the outcome of the sensor-attacker-target engagement is decided by a single containment check at the initial time. Under the proposed sensor and attacker strategies, capture of a constant-heading target is guaranteed if and only if the Apollonius circle between the attacker and the target is fully contained in the sensable region. This iff condition is Lemma 2. From it, the paper derives speed-based criteria: if the target's speed is below v_T = ((R - d_ST0) v_A + d_AT0 v_S) / (d_AT0 + R - d_ST0), capture is guaranteed from any initial orientation; if the speed is above v_T = ((R - d_ST0) v_A + d_AT0 v_S) / (d_AT0 - R + d_ST0), there exists a constant heading g","pith_inferences":["If the target were allowed to change heading even once after observing the sensor's commitment, the 'if and only if' containment condition would likely degrade to a sufficient-only condition: the intercept geometry in Lemma 2 is built on a straight-line target, so an adaptive target could invalidate the capture guarantee.","The containment condition could be turned into a real-time decision rule: before committing, the sensor-attacker team computes whether A_0 ⊂ S_0; if not, it knows escape is possible and can instead maneuver to reshape the sensable region before the engagement truly begins.","The same Apollonius-vs-sensable-region comparison may generalize to multiple sensors whose sensing discs overlap, with the team-level sensable region playing the role of S_0 and the attacker's capture set playing the role of A_0.","A direct numerical test of Theorem 4 would be to sweep the target speed through the interval [v_T, v_T] and locate the smallest speed at which A_0 first intersects S_0, verifying that it matches the root of the quadratic in Theorem 5 for each fixed initial orientation."],"forward_implications":["If the Apollonius circle is initially contained in the sensable region, no constant-heading target can escape: the attacker intercepts it before the target exits the sensor's sensing radius.","A target speed below the closed-form threshold v_T guarantees capture regardless of the initial positions and headings of the three agents.","A target speed above the closed-form threshold v_T guarantees existence of a constant heading that lets the target escape, again for arbitrary initial geometry.","For a fixed initial orientation, the quadratic-derived threshold v_T⋆ gives a sharper escape speed than the geometry-independent bound, and the true critical speed v_T⋆ lies between v_T and v_T.","At the critical speed the Apollonius circle and the sensable region are tangent; increasing the target speed beyond that point makes the two regions intersect and opens an escape route."],"supporting_citations":[{"why":"Supplies the two-phase optimal-control framework for maximizing observation time of a faster fixed-course target, which underlies the sensor's optimal time-maximizing heading.","marker":"[5]"},{"why":"Uses Apollonius circles to define capture regions against superior evaders, motivating the attacker's geometric intercept strategy.","marker":"[6]"},{"why":"Proves that in many pursuit-evasion games the evader cannot escape beyond the initial Apollonius circle, supporting the containment-based capture argument in Lemma 2.","marker":"[12]"},{"why":"Describes beam-rider guidance, the practical scenario in which an attacker relies on an external sensor for target information.","marker":"[21]"},{"why":"Recent geometric analysis of dominance regions under non-anticipatory information patterns, providing context for the slower-sensor three-agent setting.","marker":"[26]"},{"why":"Pontryagin's Maximum Principle is used to derive the optimal constant sensor heading that maximizes the time the target remains in sensing range.","marker":"[32]"},{"why":"Provides the center and radius formulas for the Apollonius circle that are used throughout the proofs of Theorems 2, 3, and 5.","marker":"[33]"}],"fun_headline_variants":["Capture iff attacker's circle fits inside sensor's sensing zone","Pursuit game: one circle containment check decides all","Sensor-attacker team wins when circle fits in sensable region","Target speed thresholds and circle containment set outcome","Apollonius circle inside sensor's zone means capture"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The target commits to a single straight-line heading at the start of the game and never changes it, even though it knows the sensor's and attacker's strategies; the containment condition and all speed thresholds are derived for this open-loop target.","fun_headline_variants_meta":{"raw":{"variants":["Capture iff attacker's circle fits inside sensor's sensing zone","Pursuit game: one circle containment check decides all","Sensor-attacker team wins when circle fits in sensable region","Target speed thresholds and circle containment set outcome","Apollonius circle inside sensor's zone means capture"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000542,"raw_usage":{"total_tokens":2505,"prompt_tokens":885,"completion_tokens":1620,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":629,"completion_tokens_details":{"reasoning_tokens":1541}},"tokens_in":629,"tokens_out":1620,"duration_ms":14706,"temperature":1.0,"reasoning_tokens":1541,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T04:31:59.614938+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate an engagement in which the target is allowed to change its heading once at an arbitrary time t > 0, rather than fixing its heading at t = 0. A single such trajectory that escapes from an initial configuration where the Apollonius circle is fully contained in the sensable region would refute the necessity of the containment condition for guaranteed capture.","supporting_citations":[{"cited_title":"Surveillance of a faster fixed-course target,","cited_arxiv_id":null,"evidence_quote":"Supplies the two-phase optimal-control framework for maximizing observation time of a faster fixed-course target, which underlies the sensor's optimal time-maximizing heading."},{"cited_title":"A decentralized fuzzy learning algorithm for pursuit-evasion differential games with superior evaders,","cited_arxiv_id":null,"evidence_quote":"Uses Apollonius circles to define capture regions against superior evaders, motivating the attacker's geometric intercept strategy."},{"cited_title":"One Apollonius circle is enough for many pursuit-evasion games,","cited_arxiv_id":null,"evidence_quote":"Proves that in many pursuit-evasion games the evader cannot escape beyond the initial Apollonius circle, supporting the containment-based capture argument in Lemma 2."},{"cited_title":"A beam rider concept for three point aerial rendezvous guidance,","cited_arxiv_id":null,"evidence_quote":"Describes beam-rider guidance, the practical scenario in which an attacker relies on an external sensor for target information."},{"cited_title":"Dominance regions of pursuit-evasion games in non-anticipative information patterns,","cited_arxiv_id":null,"evidence_quote":"Recent geometric analysis of dominance regions under non-anticipatory information patterns, providing context for the slower-sensor three-agent setting."},{"cited_title":"Optimal guidance strategy for the defense of a non-manoeuvrable target in 3-dimensions,","cited_arxiv_id":null,"evidence_quote":"Provides the center and radius formulas for the Apollonius circle that are used throughout the proofs of Theorems 2, 3, and 5."}],"review_version":1}