{"id":"cbc4d078-5f30-4b27-a9b5-b47f3b701656","arxiv_id":"2502.07708","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Asymptotically stable nonlinear systems admit global linearizing coordinates, smoothly off the equilibrium in every dimension except 5, where existence is equivalent to the smooth 4D Poincaré conjecture.","lead":"This mathematics paper proves that any system with a stable equilibrium point can be smoothly straightened into a simple linear system, as long as the space has any dimension other than 5. The result also links the dimension-5 case to a famous unsolved problem in topology, the smooth Poincaré conjecture in four dimensions.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified; the central construction is sound and the external dependencies (Wilson, Fathi–Pageault, Smale, Perelman, Freedman) appear to cover the stated hypotheses.","rationale":"The reader's verdict is ACCEPT with low correctness risk, and my stress-test concurs. I checked the internal logic: the reduction to A=-I is valid; the product decomposition B\\{x*} ≅ R×L holds because V is a strict Lyapunov function; the classification steps for L are correct because L bounds a contractible manifold; the n=5 equivalence is correctly handled in Proposition 1; and the Hirsch construction in the converse is standard. The only premise that could invalidate the continuous-case theorem is Wilson's theorem as applied to continuous flows, but the citation [FP19, Sec. 6] is specifically to a modern treatment of smoothing Lyapunov functions for continuous dynamics, so the risk is low. I therefore see no reason to change the reader's verdict.","tokens_in":7517,"tokens_out":33793,"duration_ms":309442,"concrete_test":"Check the statements of [Wil69, Thm 3.2] and [FP19, Sec. 6] to confirm that a proper strict C∞ Lyapunov function with contractible sublevel sets and boundary homotopy-equivalent to S^{n-1} is established for asymptotically stable equilibria of complete, uniquely integrable continuous vector fields on C∞ manifolds. If the theorem is stated only for C^1 vector fields, re-run the proof of Theorem 2 with a locally Lipschitz approximation or add the needed hypothesis.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I read the proof of Theorem 2 in good faith and could not find an internal gap. The construction needs a proper strict C∞ Lyapunov function whose compact sublevel set is contractible with boundary homotopy-equivalent to S^{n-1}; the paper cites [Wil69, Thm 3.2] and [FP19, Sec. 6] for this. The use of Smale for n≥6 is legitimate: removing a ball from the contractible sublevel set W=V^{-1}([0,c]) yields an h-cobordism between L and S^{n-1}, so the h-cobordism theorem gives a C∞ diffeomorphism; exotic spheres cannot arise as boundaries of contractible manifolds in dimensions ≥6. The n=5 conditional is handled by Proposition 1 via Hirsch's construction. The one residual risk is external: whether the cited level-set/smoothing theorems are stated for complete uniquely integrable continuous vector fields rather than only C^1 fields. The authors cite [FP19, Sec. 6] for the continuous case, which appears to be the right reference, but this is the only premise worth re-checking.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a global linearization theorem for asymptotically stable equilibria of nonhyperbolic systems. Theorem 1 gives a local topological conjugacy to a Hurwitz linear flow, with a C^k diffeomorphism away from the equilibrium when n≠5; Theorem 2 extends this to the whole basin of attraction for complete flows, and Proposition 1 shows that the missing n=5 C^k case is equivalent to the 4-dimensional smooth Poincaré conjecture. The proofs use a proper smooth strict Lyapunov function, the identification of the punctured basin with R×L, the classification of homotopy spheres, and Hirsch's construction.","tokens_in":7748,"tokens_out":26771,"duration_ms":234592,"significance":"The result is a natural and significant extension of the Hartman-Grobman theorem, resolving the global nonhyperbolic case under minimal regularity assumptions and with essentially optimal smoothness. The proof is elegant and modular, and the paper is honest about the dimensional obstruction: it proves the n=5 smooth case is equivalent to the 4-dimensional smooth Poincaré conjecture rather than claiming it. The main arguments are transparent derivations from classical external results (Wilson, Fathi–Pageault, Smale, Perelman, Freedman, Hirsch), with no fitted parameters or self-referential assumptions.","major_comments":[],"minor_comments":[{"comment":"The classification of the boundary L as diffeomorphic to S^{n-1} lists n=2,3,4,≥6 but omits n=1; the n=1 case is trivial because L is a two-point 0-manifold, and this should be stated for completeness.","section":"Section 3, proof of Theorem 2"},{"comment":"This proposition is stated without a proof; the text says the proof of Theorem 2 can be repeated verbatim, but for a formal paper either a proof or an explicit label as a sketch should be supplied.","section":"Additional Comments, Proposition 2"},{"comment":"The truncation function ψ is described only as equal to 1 on U0 and zero outside a compact set; to guarantee that x* is asymptotically stable for ψf with an open basin containing U0, ψ should be positive on an intermediate neighborhood, and this should be clarified.","section":"Section 3, proof of Theorem 1"},{"comment":"The application of Smale's theorem for n≥6 is terse; adding a sentence explaining that deleting a small ball from the contractible sublevel set V^{-1}([0,c]) yields an h-cobordism between L and S^{n-1} would make the argument self-contained.","section":"Section 3, proof of Theorem 2"},{"comment":"The phrase 'the C^k statement of Theorem 1 (or Theorem 2) is true for n=5' should be defined explicitly, since the main theorems exclude n=5 by hypothesis.","section":"Proposition 1"},{"comment":"The abstract and title contain typographical artifacts ('GLOBAL LINEARIZA TION', 'ST ABLE'); the manuscript should be copy-edited.","section":"Abstract and title"},{"comment":"Since the existence of a proper strict C∞ Lyapunov function for a merely continuous flow is the only non-classical external input, it would be helpful to quote the precise statement from [FP19, Sec. 6] that covers this case.","section":"Section 3, proof of Theorem 2"}],"recommendation":"minor_revision","confidential_remarks":"I found no internal gap in the proof. The only caveat is external: the argument depends on the cited Lyapunov-smoothing theorems being applicable to complete continuous uniquely integrable flows; [FP19, Sec. 6] appears to be the right reference, but an explicit quotation would remove all doubt. The paper is a good fit for the journal and the result is worthy of publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, this is a genuinely complete proof of something people have waved at for years: global Hartman-Grobman linearization for asymptotically stable, nonhyperbolic equilibria. Topological conjugacy on the whole basin in every dimension; C^k conjugacy away from the equilibrium in every dimension except 5; and an exact equivalence between the n=5 smooth case and the 4D smooth Poincaré conjecture. I read the proofs with the stress-test notes in hand and could not find an internal gap.\n\nWhat is new: the paper fills in the sketch from Grüne–Sontag–Wirth, extends it to all dimensions for k=0 using Perelman, and handles n≠5 for C^k. The equivalence with the 4D smooth Poincaré conjecture is new and is a nice standalone observation. The structure is clean: Theorem 2 gives global linearization via a proper strict Lyapunov function; the level set is a homotopy sphere, classified by the appropriate external topology; the flow gives a global section; and Theorem 1 follows by a cutoff argument. The proposition relating n=5 to the Poincaré conjecture uses Hirsch's construction and goes through cleanly. The paper also notes Koopman eigenfunction consequences, which will make it useful to the applied DMD/Koopman crowd.\n\nSoft spots, in proportion. The main external dependency is Wilson's theorem guaranteeing a proper strict C∞ Lyapunov function with level sets homotopy-equivalent to spheres for complete uniquely integrable continuous vector fields. The citations look right—Wilson's papers plus Fathi–Pageault—but this is the one premise I would want a referee to verify against the original statements, since the theorem is stated for merely continuous fields. If that reference is misapplied, the continuous case would need repair; the C^1 case would survive. Also, the n=5 result is conditional on an open conjecture, so a careless reader might overstate what is proved; the paper itself is clear about this. Minor, but the extra Proposition 2 in the additional comments adds side content that the abstract doesn't advertise; fine in a journal, worth noting.\n\nI would send this to a serious referee. The argument is original, carefully built on external results, and the conditional equivalence is a strong contribution. My own verdict is accept after checking Wilson; if I were the editor I would ask the referee to specifically confirm that reference.","headline":"Global Hartman-Grobman for asymptotically stable nonhyperbolic equilibria, complete in all dimensions for topological conjugacy and tied to the 4D smooth Poincaré conjecture in dimension 5, with sound proofs worth refereeing.","tokens_in":8260,"tokens_out":2770,"would_cite":true,"duration_ms":25317,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37C10","37C15","37C75"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that asymptotically stable equilibria of continuous flows admit global linearizing homeomorphisms on the whole basin of attraction, and that the conjugacy is smooth away from the equilibrium in every dimension except 5…","keywords":["Hartman-Grobman theorem","nonhyperbolic linearization","asymptotic stability","Lyapunov functions","basin of attraction","Koopman eigenfunctions","smooth Poincaré conjecture"],"falsifier":"Take a $C^\\infty$ homotopy 4-sphere $L$ that is not diffeomorphic to $S^4$, build a $C^\\infty$ function $V\\colon S^5\\to[0,1]$ having $L$ as a regular level set, and run the gradient flow $-\\nabla V$. If that system admits a $C^1$ linearizing conjugacy on a neighborhood of the sink, the paper's equivalence claim is false; proving no such conjugacy for every such $L$ would confirm that the dimension-5 gap is exactly the smooth Poincaré conjecture. A failure of the underlying level-set theorem—for instance a continuous uniquely integrable flow whose regular level sets are not homotopy equivalent to spheres—would also break the construction.","tokens_in":7341,"feed_emoji":"📉","tokens_out":12761,"duration_ms":121960,"temperature":0.7,"pith_summary":"The paper proves that the Hartman-Grobman theorem, which classically linearizes dynamics only around hyperbolic equilibria, extends to the opposite regime: an equilibrium that is merely asymptotically stable, with no hyperbolicity assumption, can still be globally linearized. For a complete continuous vector field whose trajectories are unique, there is a homeomorphism from the entire basin of attraction to $\\mathbb{R}^n$ that converts the nonlinear flow into the linear contraction $\\dot y = -y$. If the flow is $C^k$ and the dimension is not $5$, the homeomorphism is actually a $C^k$-diffeomorphism away from the equilibrium. In dimension $5$, the $C^k$ statement is shown to be true if and only if the 4-dimensional smooth Poincaré conjecture is true, so the last missing regularity is tied to an open problem in topology.","feed_headline":"Stable nonlinear flows linearize on the whole basin","feed_subtitle":"Every asymptotically stable system is conjugate to ẏ = −y; dimension 5 waits on the smooth Poincaré conjecture.","key_machinery":"The load-bearing mechanism is the flow-box coordinate system built from a regular level set $L=V^{-1}(c)$ of a proper $C^\\infty$ strict Lyapunov function $V$ on the basin. Every trajectory except the equilibrium crosses $L$ exactly once and transversely, so the basin minus the equilibrium is homeomorphic (and $C^k$-diffeomorphic, when the flow is $C^k$) to $\\mathbb{R}\\times L$ via the map $(t,x)\\mapsto \\Phi_t(x)$. Because $L$ is homotopy equivalent to $S^{n-1}$, and because the low- and high-dimensional classification theorems make $L$ diffeomorphic to the standard sphere for every $n\\ne 5$, one can identify $L$ with $S^{n-1}$ and write the coordinate $h(x)=e^{\\tau(x)}P(\\rho(x))$, where $\\tau$ is the arrival time and $\\rho$ the crossing point. This single formula carries the argument: the linear flow on $\\mathbb{R}^n$ is simply the product of exponential decay in $\\mathbb{R}$ and constant motion along rays, so the conjugacy is exact and global.","core_discovery":"For a complete uniquely integrable continuous vector field on an $n$-dimensional smooth manifold with an asymptotically stable equilibrium $x_*$ and basin $B$, the paper proves that there is a homeomorphism $h\\colon B \\to \\mathbb{R}^n$ satisfying $\\Phi_t|_B = h^{-1}\\circ e^{At}\\circ h$ for every $t\\in\\mathbb{R}$ and every Hurwitz matrix $A$; in particular $A=-I$ turns the flow into $\\dot y = -y$. If the flow is $C^k$ with $k\\ge 1$ and $n\\ne 5$, the same $h$ restricts to a $C^k$-diffeomorphism $B\\setminus\\{x_*\\}\\to\\mathbb{R}^n\\setminus\\{0\\}$. The paper also proves that the $C^k$ statement in dimension $n=5$ is logically equivalent to the 4-dimensional smooth Poincaré conjecture, and derives the local version as a consequence of the global one. The argument works by picking a proper strict Lyapunov function, using one regular level set as a global cross-section, and converting the time-to-reach-the-section and the position on the section into linear coordinates.","pith_inferences":["If the construction is as robust as it appears, the same time-plus-sphere-coordinate recipe should linearize flows near any compact invariant manifold whose stable and unstable cross-sections are sphere-like, not just single equilibria.","The dimension-5 cliff suggests that numerical experiments claiming smooth linearization in $\\mathbb{R}^5$ cannot settle the theoretical question either way: the obstruction is a global property of 4-manifolds invisible to local computation.","A natural testable extension is to the input-to-state setting, replacing $\\dot x = f(x)$ by $\\dot x = f(x,u)$ and asking whether the same Lyapunov cross-section yields finite-energy gain; the authors explicitly leave this open.","The one-dimensional example $h(x)=e^{-1/(2x^2)}$ for $\\dot x=-x^3$ suggests that the conjugacy need not be Lipschitz at the equilibrium, so algorithms based on derivative information at the fixed point may fail even though global linearizing coordinates exist."],"forward_implications":["Every complete asymptotically stable continuous flow is topologically conjugate, on its entire basin, to the linear system $\\dot y = -y$, so the basin is homeomorphic to $\\mathbb{R}^n$.","For $C^k$ flows in any dimension other than $5$, the linearizing coordinates are $C^k$ away from the equilibrium, preserving all finite-time derivative information up to order $k$ in the linear model.","Choosing $A$ diagonal produces $n$ continuous Koopman eigenfunctions whose joint map is a global homeomorphism from the basin to $\\mathbb{R}^n$, giving existence of minimal-dimensional linearizing observables.","The equivalence with the 4-dimensional smooth Poincaré conjecture means that the remaining dimension-5 smoothness gap cannot be closed by dynamics alone; it is a question of 4-manifold topology.","The results provide existence targets for data-driven algorithms, such as extended dynamic mode decomposition, that seek linearizing coordinates of dimension equal to the state space."],"supporting_citations":[{"why":"Supplies the proper $C^\\infty$ strict Lyapunov function whose regular level sets serve as the global cross-section.","marker":"[Wil69, Thm 3.2]"},{"why":"Establishes that the regular level sets are homotopy equivalent to $S^{n-1}$ and that every trajectory crosses the section exactly once and transversely.","marker":"[Wil67, pp. 326–327]"},{"why":"Provides the smoothing result needed to obtain a $C^\\infty$ Lyapunov function from the continuous setting.","marker":"[FP19, Sec. 6]"},{"why":"Supplies the classification result making the level set diffeomorphic to the sphere in the four-dimensional case.","marker":"[MT07, Cor. 0.2]"},{"why":"Handles dimensions $n\\ge 6$, where the level set is diffeomorphic to the standard sphere by the high-dimensional manifold classification.","marker":"[Sma62, Thm 5.1]"},{"why":"Gives the homeomorphism between the level set and $S^4$ in dimension $5$, where smoothness is unavailable.","marker":"[Fre82, Thm 1.6]"},{"why":"Used in the converse of Proposition 1 to realize any homotopy 4-sphere as a Lyapunov level set, linking dimension 5 to the smooth Poincaré conjecture.","marker":"[Hir65, Thm 2]"}],"fun_headline_variants":["Nonhyperbolic stable systems linearize globally","5D linearization rests on smooth Poincaré conjecture","Hartman-Grobman without hyperbolicity, global basin","Global conjugacy for stable flows, no hyperbolicity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument depends on the theorem that every asymptotically stable equilibrium of a complete uniquely integrable continuous flow admits a proper, strictly decreasing smooth Lyapunov function whose regular level sets are homotopy equivalent to spheres; if that theorem ever failed, the construction of the cross-section and the linearizing map collapses.","fun_headline_variants_meta":{"raw":{"variants":["Nonhyperbolic stable systems linearize globally","5D linearization rests on smooth Poincaré conjecture","Hartman-Grobman without hyperbolicity, global basin","Global conjugacy for stable flows, no hyperbolicity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000547,"raw_usage":{"total_tokens":2594,"prompt_tokens":903,"completion_tokens":1691,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":519,"completion_tokens_details":{"reasoning_tokens":1626}},"tokens_in":519,"tokens_out":1691,"duration_ms":14466,"temperature":1.0,"reasoning_tokens":1626,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T11:50:15.042196+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a $C^\\infty$ homotopy 4-sphere $L$ that is not diffeomorphic to $S^4$, build a $C^\\infty$ function $V\\colon S^5\\to[0,1]$ having $L$ as a regular level set, and run the gradient flow $-\\nabla V$. If that system admits a $C^1$ linearizing conjugacy on a neighborhood of the sink, the paper's equivalence claim is false; proving no such conjugacy for every such $L$ would confirm that the dimension-5 gap is exactly the smooth Poincaré conjecture. A failure of the underlying level-set theorem—for instance a continuous uniquely integrable flow whose regular level sets are not homotopy equivalent to spheres—would also break the construction.","supporting_citations":[],"review_version":1}