{"id":"c3c3af36-b93d-4f0c-92fb-464264b3ec73","arxiv_id":"2507.05801","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Parabolic solutions of the planar n-body problem whose normalized shape converges to an isolated central configuration have no infinite spin: the shape converges to a definite central configuration.","lead":"An escaping or colliding cluster of gravitating bodies cannot keep spinning forever: this paper proves that once its normalized shape converges to an isolated central configuration, the rotation angle converges to a definite limit. The result extends a 2025 theorem of Moeckel and Montgomery from total collisions to parabolic solutions and to partial collisions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.13, the new time-dependent shadowing theorem used to enter the invariant manifold M, is false as stated; the proof never shows the constructed solution lies in N.","rationale":"The reader's conditional focused on the degenerate-CC lemmas (Lemmas 3.2-3.4) and the transfer of the Lojasiewicz inequality to a C^k center manifold. That is a legitimate concern, but I find a more basic gap: the new shadowing theorem used in both the nondegenerate and degenerate cases is false in the stated generality. The counterexample above is a smooth vector field with g=0 and an invariant line N={x=0}; it satisfies every hypothesis of Theorem 1.13 and violates the conclusion. The proof of Theorem 1.13 has a non-sequitur: after constructing a fixed point z of (5.4), the authors assert that y=x*+z is a solution contained in N, but the integral equation was derived by assuming y in N; satisfying it is necessary, not sufficient. This is not a cosmetic gap: in the counterexample z=0 is a fixed point although y=x* is not in N. Because this theorem is the only way the paper obtains a shadowing solution on M, Proposition 3.1 is unsupported as written, and the main theorems inherit that gap. The verdict should remain CONDITIONAL: the paper can be accepted only after Theorem 1.13 is repaired (or Proposition 3.1 is replaced by a direct stable-foliation argument on M) and after the degenerate-CC lemmas are independently verified.","tokens_in":15283,"tokens_out":29064,"duration_ms":373143,"concrete_test":"Run the counterexample in Section 5: for f(x,y)=(-x^3,-y), g=0, N={x=0}, verify that all hypotheses of Theorem 1.13 hold for x(t)=(2(t+1))^{-1/2}, y=0, and that no solution in N is exponentially close; this settles that the theorem is false as stated. Then check whether the corrected hypothesis (N contains the spectral subspace of eigenvalues with real part >= -beta) holds for M={u=0} in (2.9), and if so, redo the fixed-point argument with the projection onto M explicitly imposed.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The main results rely on Proposition 3.1, whose proof is literally 'using Theorem 1.13' with N=M={u=0}. As stated, however, Theorem 1.13 is false. Counterexample: take f(x,y)=(-x^3,-y), g=0, N={x=0}, and A=Df(0)=diag(0,-1), so beta=1. The solution x(t)=(2(t+1))^{-1/2}, y(t)=0 satisfies all displayed hypotheses, including x(t)->0 and e^{alpha t}|g|=0 for every alpha<beta. But every solution in N is (0, C e^{-t}); the distance in the x-coordinate is (2(t+1))^{-1/2}, which is not exponentially small, so no exponential shadow exists. The defect in Section 5 is that the fixed point z of (5.4) is only shown to make y=x*+z solve y'=A y+h(y); the derivation of (5.4) used y in N as a necessary condition, but the contraction argument never enforces membership in N. Indeed z=0 is a fixed point in the counterexample even though y=x* is not in N. Since Theorem 1.13 is the only bridge to the invariant manifold M, Proposition 3.1 and hence Theorems 1.8 and 1.14 are not established by the present proof. A repair needs either a spectral hypothesis on N (e.g. N contains the modes with Re lambda > -beta, plus a proof that the fixed point lies in N) or a direct shadowing construction on M.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies infinite spin for parabolic and collision solutions in the planar n-body problem. The authors introduce a modified McGehee coordinate system in (2.8), derive a non-autonomous blown-up system (2.9), and prove Theorem 2.5 stating that, when the reduced normalized configuration converges to an isolated central configuration, the blown-up solution converges to an equilibrium. They then state a general time-dependent shadowing theorem (Theorem 1.13) and use it, via Proposition 3.1, to shadow the parabolic orbit by a solution on the invariant manifold {u=0}. The remaining argument follows Moeckel and Montgomery [16]: on the invariant manifold, a center-manifold reduction and a Lojasiewicz-type inequality yield finite arclength, whence no infinite spin (Theorem 1.8). The same scheme is applied to partial collision solutions (Theorem 1.14). The paper also proves Proposition 1.6, giving asymptotic decay of the subsystem energy and convergence of the normalized configuration to the set of central configurations.","tokens_in":15407,"tokens_out":9043,"duration_ms":105856,"significance":"If the main results were established, they would be a natural and significant extension of Moeckel and Montgomery's total-collision theorem [16] to complete and partial parabolic solutions and to partial collisions, under the same isolated-central-configuration hypothesis. The modified McGehee coordinates, the treatment of the time-dependent terms coming from the rest of the system, and the explicit estimates in Lemmas 2.1 and 6.1 are useful contributions. However, the central shadowing theorem on which the proofs rely is false as stated, and the Lojasiewicz transfer in the degenerate case is only asserted, not proved. The main theorems are therefore not established by the present manuscript; a corrected or replaced shadowing argument would be required.","major_comments":[{"comment":"Theorem 1.13 is false as stated. Counterexample: take f(x,y)=(-x^3,-y), g=0, N={x=0}. Then A=Df(0)=diag(0,-1), β=1, and x(t)=((2(t+1))^{-1/2},0) satisfies x(t)→0 and e^{αt}|g|=0 for every α<1. But every solution in N has the form (0,C e^{-t}), so its distance to x(t) in the x-coordinate is ~(2t)^{-1/2}, which is not exponentially small. Consequently no exponential shadow in N exists. The defect in the proof is at the step after (5.4): the fixed point z of Λ only shows that y=x*+z solves the equation y'=Ay+h(y); the derivation of (5.4) used the representation (5.2), which is necessary for solutions in N but not sufficient. In the counterexample z=0 is a fixed point even though y=x* is not in N. Since Proposition 3.1 invokes Theorem 1.13 with N=M={u=0}, the exponential shadowing onto M is not established, and Theorems 1.8 and 1.14 are not proven by the present argument.","section":"§5, Theorem 1.13 (Eq. (5.4))"},{"comment":"The degenerate-CC half of the proof is not actually carried out. Lemma 3.3 asserts the Lojasiewicz-type inequality |∇˜W(x)|^2 ≥ |W(x)-W(0)|^α, and the text says the proof is 'completely the same' as Lemmas 4.3 and 4.4 of [16]. But W(x)=V_k(x,φ(x)) is built from the center-manifold graph φ, which is only C^k, whereas the Lojasiewicz inequality in [10] is for analytic functions; no argument shows that the restricted potential inherits the needed inequality. In addition, the constant k=2/v(0)>0 has the opposite sign from the corresponding constant in [16], and Lemma 3.4 claims the same proof goes through after replacing W by -W, but the reduction from the equation x'=k∇W(x)+γ(x) with k>0 to the gradient-flow argument in [16] is only asserted. Since Lemma 3.4 is what yields finite arclength in the degenerate case, Theorem 1.8 is incomplete for degenerate isolated central configurations.","section":"§3, Lemma 3.3 and Lemma 3.4"},{"comment":"The proof of Theorem 1.14 is a sketch at the crucial point: after obtaining a result 'similar to Proposition 3.1' via Theorem 1.13, the text says the theorem follows by 'an argument similar to those given in Section 3'. The collision case has v0<0 and the equation r'=rv, so the local center manifold and the sign of the gradient constant in the restriction to M differ from the parabolic case. The degenerate-CC Lojasiewicz step needs to be written out for this setting. Together with the failure of Theorem 1.13, this means Theorem 1.14 is not established as a theorem with a complete proof.","section":"§4, Theorem 1.14"}],"minor_comments":[{"comment":"In the displayed computation of d/dt(∂L/∂ω), the last term is written as '˙µω'; this appears to be a typo for '˙µB(s)' (with the appropriate transposes), since the preceding expression is r^2A(s)ω+μB(s). As printed the term is dimensionally inconsistent.","section":"§2, Eq. (2.1)"},{"comment":"The symbol k is used both for a subset of {1,...,n} and for its cardinality, e.g. 'k = {1,...,k}'. This is a common abuse of notation, but it should be flagged explicitly to avoid confusion in the asymptotic estimates.","section":"§2, Definition 1.2"},{"comment":"The sentence 'Since -v0/2 is an eigenvalue, when β ≤ 1/2 v0' is tautological: because -v0/2 is an eigenvalue, the spectral gap β automatically satisfies β ≤ v0/2. The wording should be cleaned up.","section":"§3, proof of Proposition 3.1"},{"comment":"The extension x*(t)=x(0) for t<0 makes x* only piecewise differentiable at t=0; the variation-of-constants formula (5.3) is valid in the integral sense, but the lack of differentiability at 0 and the resulting distributional term φ(t) should be discussed more carefully.","section":"§5, extension of x(t)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a serious and clearly written attempt to extend Moeckel and Montgomery [16] to parabolic and partial collision solutions, and the new coordinates and estimates are potentially useful. However, the central shadowing theorem is false as stated, and the degenerate Lojasiewicz step is not proved. I recommend major revision rather than rejection because a corrected shadowing argument or a direct construction on the invariant manifold may be within reach; if the authors cannot supply such a repair, the main theorems should not be considered proven."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper should not be accepted as is, because Theorem 1.13—the bridge to the invariant manifold—is false in the stated generality. The counterexample f(x,y)=(-x^3,-y), g=0, N={x=0} satisfies every hypothesis: A has eigenvalues 0 and -1, so β=1; the solution x(t)=(2(t+1))^{-1/2}, y(t)=0 tends to 0 and makes g identically zero. But every solution in N is (0, Ce^{-t}), and the distance in the x-coordinate decays like t^{-1/2}, so no exponential shadow exists. The defect in the proof is that (5.4) is only a necessary condition for a solution in N; the contraction argument on Zη never enforces y=x*+z to actually lie in N. Since Proposition 3.1 invokes Theorem 1.13 directly, Theorems 1.8 and 1.14 are not established by the present argument.\n\nThis is a pity, because the paper does real work before that. The modified McGehee coordinate (2.8) is well chosen for parabolic motion, and the time-dependent perturbation estimates in Lemma 2.1 and Corollary 2.2 are clean and correct. I verified the main cancellations in Lemma 6.1 and the comparison argument in Theorem 2.5; those parts hold up. The paper also carefully documents why Saari's earlier claimed proofs of the parabolic no-spin result are incomplete, and that is genuinely useful context.\n\nThe reader's secondary worry also stands: Lemmas 3.2–3.4 are imported from Moeckel–Montgomery without proof, including the sign change k=2/v(0)>0 and the Lojasiewicz inequality on a C^k center manifold. That transfer is not automatic, but it is a smaller issue than the false shadowing theorem.\n\nBottom line: the intended strategy is plausible and most of the parabolic analysis is sound conditional on a repaired shadowing lemma. As written, the main theorems are unproved. I would still send it to a serious referee—the problem is important and the error is subtle—but the referee should demand either a spectral hypothesis on N plus a proof that the fixed point lies in N, or a direct shadowing construction on M. It is also a good reading-group paper: the failure of a necessary-condition fixed point to preserve a manifold constraint is a useful cautionary tale.","headline":"Theorem 1.13, the shadowing lemma used to reach the invariant manifold, is false as stated, and the main theorems rely on it directly; the paper needs major repair but is worth referee time.","tokens_in":16205,"tokens_out":4282,"would_cite":false,"duration_ms":46488,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["70F10","70F16","37D10"],"pacs":[],"model":"deepseek-v4-flash","headline":"Infinite spin is ruled out for parabolic and partial-collision n-body orbits when the shape tends to an isolated central configuration.","keywords":["infinite spin","parabolic solutions","collision solutions","central configurations","planar n-body problem","McGehee coordinates","Lojasiewicz inequality","shadowing theorem"],"falsifier":"Exhibit a degenerate central configuration of a planar n-body subsystem, compute the local center-manifold graph, and check whether the restricted potential $W$ satisfies $|\\tilde\\nabla W(x)|^2 \\ge |W(x)-W(0)|^\\alpha$ for some $1<\\alpha<2$ near the origin; if a sequence $x\\to 0$ satisfies $|\\tilde\\nabla W(x)|^2 < |W(x)-W(0)|^\\alpha$, the finite-arclength lemma cannot be invoked and the degenerate-case proof collapses. Alternatively, a numerical integration producing a $k$-parabolic or $k$-collision solution that meets the isolated-CC hypothesis but whose rotation angle does not converge would directly refute the theorem.","tokens_in":14831,"feed_emoji":"🌌","tokens_out":7474,"duration_ms":90430,"temperature":0.7,"pith_summary":"This paper takes on the infinite spin problem in the planar n-body problem: a cluster's normalized shape may settle toward a central configuration while the cluster keeps rotating forever. The authors establish that this cannot happen for k-parabolic solutions, where a subsystem's mutual distances grow like $t^{2/3}$, nor for k-collision solutions near collision time, as long as the reduced normalized configuration converges to an isolated central configuration. This extends a recent no-infinite-spin theorem for total collisions to both complete and partially parabolic solutions, and generalizes it to partial collisions. The proof works by transforming the parabolic problem into a time-dependent perturbation of the total-collision system and then showing the perturbed orbit is exponentially shadowed by an orbit on an invariant manifold, where the known finite-arclength argument applies.","feed_headline":"No infinite spin in parabolic or partial-collision n-body orbits","feed_subtitle":"When a cluster's shape converges to an isolated central configuration, its rotation angle settles instead of spinning forever.","key_machinery":"The load-bearing construction is a modified McGehee coordinate system for parabolic motion, defined by $u=r^{-1/2}$, $v=\\sqrt{r}\\,\\rho$, $w=r^{3/2}\\,\\omega$, and $d\\tau=r^{-3/2}\\,dt$. In these coordinates the equations of motion become a non-autonomous system whose time-dependent terms $u^2P(\\tau)$ and $u^2Q(\\tau)$ are bounded and decay rapidly, and the parabolic solution converges to an isolated equilibrium $p_0=(0,v_0,s_0,0)$. The supporting Theorem 1.13 is a shadowing result for smooth time-dependent systems: if a solution converges to an equilibrium of the autonomous part and the perturbation decays exponentially at every rate below the spectral gap, then that solution is exponentially close to a solution lying on the invariant submanifold $N$. This reduces the parabolic and partial-collision problems to the autonomous dynamics on $M$, where the Lojasiewicz-inequality argument already used in the total-collision case yields finite Fubini--Study arclength and hence convergence of the rotation angle.","core_discovery":"The paper's central claim is Theorem 1.8: if $q(t)$ is a $k$-parabolic solution and the reduced, normalized relative configuration $[q^c_k(t)]$ in $S_k/SO(2)$ converges to an isolated central configuration, then the normalized relative configuration $q^c_k(t)$ converges to a particular central configuration in $\\mathcal{C}_k$, so the rotation angle converges and there is no infinite spin. Theorem 1.14 makes the same assertion for $k$-collision solutions as $t\\to T$, thereby extending the total-collision result to partial collisions. The mechanism is a rescaling $u=r^{-1/2}$, $v=\\sqrt{r}\\,\\rho$, $w=r^{3/2}\\,\\omega$, $d\\tau=r^{-3/2}\\,dt$, in which the parabolic solution approaches an isolated equilibrium and the influence of outside masses appears only as small time-dependent terms. A new shadowing theorem then places the true orbit exponentially close to a solution contained in the invariant submanifold $M=\\{u=0\\}$, where the dynamics coincide with those of the total-collision problem; finite Fubini--Study arclength on that manifold forces the rotation to settle.","pith_inferences":["The same reduction may work in the spatial three-dimensional n-body problem, where the quotient by $SO(3)$ replaces $\\mathbb{CP}^{k-2}$; the shadowing theorem itself is dimension-agnostic.","If the Lojasiewicz-type inequality holds on the center manifold, the argument likely yields quantitative decay rates for the shape's approach to the central configuration, not merely convergence.","Because Theorem 1.13 only requires exponential decay of the perturbation, a plausible extension is to nearly parabolic clusters in hierarchical systems where the separation from other bodies grows faster than linearly, as long as the induced forcing still decays exponentially in the rescaled time.","The most fragile step, the degenerate-case Lojasiewicz inequality on the center manifold, is stated without derivation; testing it numerically on concrete degenerate central configurations would be a meaningful check of the proof route."],"forward_implications":["For any $k$-parabolic solution whose reduced shape limit is an isolated central configuration, the normalized shape has a definite limit rather than a limit circle traced out by continuing rotation.","The no-infinite-spin conclusion now covers partial collisions as well as total collisions, so a subcluster undergoing collision also settles into a definite rotation angle under the same isolated-CC hypothesis.","The proof of Proposition 1.6 shows that parabolicity alone forces the subsystem energy to decay like $O(t^{-5/3})$ and the shape to accumulate on the central-configuration set, without the extra global bound $R(t)=O(t)$.","The shadowing theorem is a general statement about time-dependent perturbations of autonomous systems with an invariant submanifold and an equilibrium, so it can be applied outside the n-body setting whenever the perturbation and spectral-gap conditions hold.","If Smale's finiteness conjecture holds, every normalized central configuration is isolated up to rotation, making the isolated-CC hypothesis the natural generic condition under which the theorems apply."],"supporting_citations":[{"why":"Supplies the total-collision no-infinite-spin theorem and the center-manifold/Lojasiewicz argument that Sections 3 and 4 adapt to parabolic and partial-collision solutions.","marker":"[16]"},{"why":"Provides the Lojasiewicz inequality used in Lemma 3.3 to obtain the estimate $|\\tilde\\nabla W(x)|^2 \\ge |W(x)-W(0)|^\\alpha$ and hence finite arclength.","marker":"[10]"},{"why":"The center-manifold theorem proof that the shadowing theorem (Theorem 1.13) follows, and that the degenerate-case reduction to the center manifold relies on.","marker":"[2]"},{"why":"Marchal and Saari's Corollary 4 gives Proposition 1.6(b), namely that the normalized relative configuration of a k-parabolic solution approaches the set of normalized central configurations.","marker":"[12]"},{"why":"McGehee's original blow-up coordinates, modified in equation (2.8) with a different rescaling appropriate for parabolic motion.","marker":"[13]"}],"fun_headline_variants":["No infinite spin for parabolic and collision solutions","Rotation settles in n-body parabolic and collision orbits","Parabolic and collision n-body solutions avoid infinite spin","Infinite spin ruled out for n-body parabolic and collisions","n-body rotation converges for parabolic and collision motions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof's load-bearing premise is that the Lojasiewicz-type inequality for the restricted potential $W$ still holds after pulling back to the local center manifold in the degenerate case, even though the center-manifold graph is only finitely differentiable and the constant $k=2/v(0)>0$ has the opposite sign from the collision case; if that inequality fails, the finite-arclength conclusion and with it Theorems 1.8 and 1.14 no longer follow.","fun_headline_variants_meta":{"raw":{"variants":["No infinite spin for parabolic and collision solutions","Rotation settles in n-body parabolic and collision orbits","Parabolic and collision n-body solutions avoid infinite spin","Infinite spin ruled out for n-body parabolic and collisions","n-body rotation converges for parabolic and collision motions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000182,"raw_usage":{"total_tokens":1274,"prompt_tokens":873,"completion_tokens":401,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":489,"completion_tokens_details":{"reasoning_tokens":328}},"tokens_in":489,"tokens_out":401,"duration_ms":4655,"temperature":1.0,"reasoning_tokens":328,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:23:15.908460+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a degenerate central configuration of a planar n-body subsystem, compute the local center-manifold graph, and check whether the restricted potential $W$ satisfies $|\\tilde\\nabla W(x)|^2 \\ge |W(x)-W(0)|^\\alpha$ for some $1<\\alpha<2$ near the origin; if a sequence $x\\to 0$ satisfies $|\\tilde\\nabla W(x)|^2 < |W(x)-W(0)|^\\alpha$, the finite-arclength lemma cannot be invoked and the degenerate-case proof collapses. Alternatively, a numerical integration producing a $k$-parabolic or $k$-collision solution that meets the isolated-CC hypothesis but whose rotation angle does not converge would directly refute the theorem.","supporting_citations":[{"cited_title":"Moeckel and R","cited_arxiv_id":null,"evidence_quote":"Supplies the total-collision no-infinite-spin theorem and the center-manifold/Lojasiewicz argument that Sections 3 and 4 adapt to parabolic and partial-collision solutions."},{"cited_title":"L ojasiewicz","cited_arxiv_id":null,"evidence_quote":"Provides the Lojasiewicz inequality used in Lemma 3.3 to obtain the estimate $|\\tilde\\nabla W(x)|^2 \\ge |W(x)-W(0)|^\\alpha$ and hence finite arclength."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The center-manifold theorem proof that the shadowing theorem (Theorem 1.13) follows, and that the degenerate-case reduction to the center manifold relies on."},{"cited_title":"Marchal and D","cited_arxiv_id":null,"evidence_quote":"Marchal and Saari's Corollary 4 gives Proposition 1.6(b), namely that the normalized relative configuration of a k-parabolic solution approaches the set of normalized central configurations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"McGehee's original blow-up coordinates, modified in equation (2.8) with a different rescaling appropriate for parabolic motion."}],"review_version":1}