{"id":"3018be88-a265-4b04-838b-87b8eec50bcf","arxiv_id":"1908.07713","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Blid maps, a replacement for smooth bump functions, extend local smooth maps to global maps on spaces like C[0,1] and support a weakened bump-function condition in linearization theorems.","lead":"This paper introduces 'blid' maps, which localize smooth maps in infinite-dimensional spaces where ordinary smooth bump functions do not exist. It shows how to extend locally defined maps to the whole space and applies the method to Borel's lemma, cohomological equations, and linearization.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.3's proof never constructs the modified global map and differentiates f=DF−Λ, which need not be differentiable; as written, the linearization application is not established.","rationale":"The reader's weakest assumption focuses on the smoothness and boundedness of the explicit blid constructions in Example 2.7 and Lemmas 3.7–3.11, which are asserted with a reference to [BR1] but not demonstrated in this preprint. That is a legitimate deferred-proof concern. However, the more immediate and load-bearing problem is internal to the proof of Theorem 4.3: the object \\tilde f(x)=f(δH(x/δ)) is not shown to be the derivative of any global map, and the proof bounds D\\tilde f although f is only α-Hölder and need not be differentiable. This affects the paper's central advertised application, namely replacing bump functions by blid maps in the Zhang–Lu–Zhang linearization theorem. The extension mechanism itself, Theorem 2.3, is a one-line composition and is sound assuming the blid maps exist, but the linearization application is not established as written. A repair seems plausible by defining the modified perturbation through φ∘(δH(·/δ)), and for this reason I do not move the verdict away from the reader's CONDITIONAL; the preprint should not be accepted until this step is written correctly. My agreement with the reader is partial: the deferred blid-smoothness proofs are a real concern, but the specific internal gap in Theorem 4.3 is different and more pressing.","tokens_in":10759,"tokens_out":16506,"duration_ms":157358,"concrete_test":"Rewrite the proof of Theorem 4.3 with \\tilde F(x)=Λx+φ(δH(x/δ)), where φ=F−Λ, and check three steps: (i) \\tilde F agrees with F on a neighborhood of 0; (ii) D\\tilde F−Λ = f(δH(x/δ))DH(x/δ); (iii) the two ZLZ estimates in (7.6) follow from the corresponding estimates for f using only the boundedness of H and DH, with no differentiability of f. If these steps succeed, the earlier objection is resolved; if the derivation requires f to be differentiable or supplies a different bound, Theorem 4.3's proof is not correct as written.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem 4.3 does not actually produce the object required by the ZLZ condition (7.6). It sets f=DF−Λ and defines \\tilde f(x)=f(δH(x/δ)). To invoke ZLZ, one needs a global map \\tilde F such that D\\tilde F−Λ = \\tilde f, but \\tilde f is never shown to be a derivative. If one takes the natural modification \\tilde F(x)=Λx+φ(δH(x/δ)), with φ=F−Λ, then D\\tilde F−Λ = f(δH(x/δ))DH(x/δ), not f(δH(x/δ)); the factor DH is essential and is missing from the paper's definition. The proof then bounds sup ||D\\tilde f(x)||, which requires differentiability of f; the theorem assumes only that DF is α-Hölder, so f need not be differentiable. Thus the displayed estimates verify a different object from the one needed for (7.6), and even the object defined may not be differentiable. This is not merely a typo: the proof never constructs the modified C^1 diffeomorphism whose estimates would imply the linearization conclusion. Unless this step is repaired by working directly with φ∘(δH(·/δ)) and estimating f(δH(·/δ))DH(·/δ), Theorem 4.3 is unsupported by the text.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces 'blid maps' (global bounded local identity maps) as a replacement for smooth bump functions in infinite-dimensional spaces. It claims (Theorem 2.3) that any C^q germ on a Banach space possessing a C^q blid map has a global C^q representative, with a bounded-derivative version when the blid has bounded derivatives. Section 3 extends the idea to linear topological spaces using bounded and compact differentiability, and gives examples on C^q(R), C^∞[0,1], and C^∞(R). Section 4 states a Borel lemma and a cohomological-equation theorem, both deferred to a companion paper, and gives a proof of a differentiable-linearization theorem (Theorem 4.3) intended to replace the smooth-bump assumption in the Zhang-Lu-Zhang theorem by the existence of a blid map. The paper concludes with additional examples and open problems.","tokens_in":11029,"tokens_out":11850,"duration_ms":109703,"significance":"The blid-map concept is simple and potentially useful: if the extension theorem and the examples are correct, it provides a localization tool for spaces such as C^q[0,1] that lack smooth bump functions, and it would improve the Zhang-Lu-Zhang linearization theorem by removing the smooth-bump hypothesis. The composition mechanism F=f∘H is transparent, and the concrete formulas in Examples 2.5-2.7 and Lemmas 3.7-3.11 are plausible. However, the central extension theorem and the Borel lemma are stated without proof and deferred to the authors' own prior work, and the proof of Theorem 4.3 has a serious structural gap. The paper is therefore not yet acceptable in its current form, though the underlying ideas appear worth pursuing.","major_comments":[{"comment":"The proof of Theorem 4.3 does not establish the hypotheses of the Zhang-Lu-Zhang theorem. Setting f=DF-Λ, the proof defines \\tilde f(x)=f(δH(x/δ)) and then verifies estimates for \\tilde f. But condition (7.6) in [ZLZ] is a condition on the difference D\\tilde F-Λ for some global map \\tilde F, and no such \\tilde F is constructed. The natural candidate \\tilde F(x)=Λx+φ(δH(x/δ)) with φ=F-Λ gives D\\tilde F(x)-Λ = f(δH(x/δ))DH(x/δ), not f(δH(x/δ)); the factor DH(x/δ) is absent from \\tilde f and from the estimates. In addition, the displayed estimates bound ||D\\tilde f(x)||, whereas (7.6) as reproduced in (4) bounds ||DF(x)-Λ|| = ||f(x)||; the function f need not be differentiable (the theorem assumes only that DF is α-Hölder), so D\\tilde f may not exist. Thus the proof verifies a different object from the one needed, and Theorem 4.3 is not supported by the text as written. The argument can perhaps be repaired by working directly with φ(δH(x/δ)) and estimating f(δH(x/δ))DH(x/δ), but that repair is not present.","section":"Section 4.3, Theorem 4.3"},{"comment":"The manuscript's central extension mechanism is not proved here. Theorem 2.3 is stated with a reference to [BR1] and no proof, yet it is the basis for the claimed applications; Borel Lemma 4.1 is explicitly said to be proved in [BR1], and Theorem 4.2 is deferred to [BR1] as well. The one-line composition F=f∘H would prove the first statement of Theorem 2.3 if the definition of blid map were precisely fixed, but the boundedness assertion and the existence of the required blid maps for the examples are not demonstrated in this paper. Since these results are load-bearing for the paper's advertised contributions, the manuscript should either include their proofs or state exactly which hypotheses from [BR1] are being used and why the reader can rely on them.","section":"Theorem 2.3; Theorems 4.1 and 4.2"},{"comment":"The blid-property examples in Section 3 are asserted rather than proved. In Lemmas 3.7, 3.9, and 3.11, formulas for H (or H_k) are displayed and it is stated without verification that each is a differentiable local identity with the stated boundedness; these lemmas are the only support for Corollaries 3.8, 3.10, and 3.12. The proof of Proposition 3.6 is also a sketch: the displayed inequality (1) is garbled, and the step from monotonicity of the norms to d(H_c(x),0)1 − lnc/ ln 2(1)' does not display equation (1); the equation number should be attached to the displayed inequality.","section":"Proof of Proposition 3.6"},{"comment":"The phrase 'the space X=C^q(R) of all smooth functions' is imprecise for finite q; it should say 'q times continuously differentiable functions'.","section":"Section 3.1"}],"recommendation":"major_revision","confidential_remarks":"The paper leans heavily on the authors' own prior papers for its central results; if the companion papers are not easily accessible to readers, the self-containedness problem is worse. The misplaced references [1]-[6] suggest a template artifact that should be corrected. The gap in Theorem 4.3 is the most serious issue and should be resolved before the manuscript can be accepted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear Colleague,\n\nQuick take: this is an announcement-style paper for the authors' own blid-map method, not a self-contained account. The genuinely new pieces are the explicit blid constructions for C^q(R), C^∞[0,1], C^∞(R), and the claim that the Zhang-Lu-Zhang linearization theorem can replace smooth bump functions by a differentiable blid map. The core idea — compose a germ with a blid map to get a global representative — is simple and sound, and the concrete examples look credible. If the blid method works, it fills a real gap for infinite-dimensional spaces without smooth bump functions, so the concept deserves attention.\n\nCredit where due: the authors are upfront that the load-bearing results (Theorem 2.3, Borel Lemma 4.1, Theorem 4.2) are proved elsewhere, in [BR1]/[BR2]. That makes the paper a useful roadmap, though not a verification. The explicit formulas in Section 3, even with deferred smoothness checks, are a valuable starting point.\n\nNow the soft spots. The first is structural: a paper whose central theorems are all deferred cannot stand alone. The second is more serious: the proof of Theorem 4.3, as written, does not construct the object required by ZLZ. If you take the natural modified map Λx + φ(δH(x/δ)), its derivative differs from the paper's f(δH(x/δ)) by the factor DH(x/δ). That factor is missing from the definition of \\tilde f. Moreover, f is only α-Hölder, so \\tilde f need not be differentiable in the first place; the estimates on D\\tilde f may be estimates of something that doesn't exist. This is not a typo—the proof never builds the C^1 conjugating map whose estimates would imply the linearization. It might well be repairable by working directly with φ∘(δH(·/δ)) and estimating f(δH(·/δ))DH(·/δ), but as it stands Theorem 4.3 is unsupported.\n\nMinor issues: the unresolved 'Section ??' and the bibliography entries about molecular biology and grid computing are clear signs of sloppy manuscript preparation. Proposition 3.6's scaling argument is also a sketch, though it looks plausible.\n\nWho is this for? Someone working on local-to-global extension in infinite-dimensional spaces, or on differentiable linearization, will get the main idea and the examples. But cite [BR1] if you need the theorems; don't rely on this preprint.\n\nMy recommendation: the paper deserves a serious referee because the blid method is potentially important, but I would not accept it in this form. The authors need to either supply a complete proof of Theorem 4.3 or reframe the paper as a survey with the missing details in [BR1]. The gap is load-bearing.\n\nBest,","headline":"Useful blid-map framework, but this preprint is an announcement: the main theorems are deferred to prior papers, and the proof of the ZLZ linearization application has a genuine missing factor that leaves Theorem 4.3 unsupported as written.","tokens_in":11615,"tokens_out":4006,"would_cite":false,"duration_ms":88275,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46T20","37C05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that a C^q germ on a Banach space with a C^q blid map always has a global C^q representative, and derives differentiable linearization on spaces like C^q[0,1] that have no smooth bump functions.","keywords":["blid maps","smooth extension of local maps","bump functions","Banach spaces","linear topological spaces","Borel lemma","differentiable linearization","cohomological equations"],"falsifier":"Compute explicitly the Fréchet derivative of the operator $H$ in Example 2.7 on $C^q[0,1]$. If for some $x$ the derivative fails to exist or is unbounded as a linear operator even on bounded subsets, the claimed $C^\\infty$ blid map is not delivered and the extension theorem's example collapses. A simpler check: test the claimed bound $\\sup_x\\|H(x)\\|_k < a e^k$ in Lemma 3.7 numerically on a sequence of functions with growing derivatives.","tokens_in":10517,"feed_emoji":"","tokens_out":10261,"duration_ms":240124,"temperature":0.7,"pith_summary":"The paper proposes a replacement for smooth bump functions in infinite-dimensional local analysis. It defines a 'blid map'—a globally defined map that equals the identity in a neighborhood of zero and has bounded image—and proves that whenever a Banach space carries a C^q blid map, every C^q germ at zero into any Banach space has a global C^q representative. The same composition idea extends to linear topological spaces under a 'blid-differentiable' property. If the constructions are sound, this covers spaces such as $C^q[0,1]$ that have no smooth bump functions, and it supplies a Borel-type reconstruction of a map from its Taylor jet, global solutions of cohomological equations, and differentiable linearization at hyperbolic fixed points.","feed_headline":"Every C^q germ extends globally when a 'blid' map exists","feed_subtitle":"Unlike bump functions, blid maps work on C^q[0,1], enabling global extension and differentiable linearization.","key_machinery":"The central object is the blid map (bounded local identity): a $C^q$ map $H:X\\to X$ with $H(x)=x$ near $0$ and $\\sup_x\\|H(x)\\|<\\infty$. The basic mechanism is the one-line composition $F=f\\circ H$, which turns a germ into a global map because $H$ pushes the whole space into the germ's domain while behaving as the identity near zero. For non-smooth spaces like $C^q[0,1]$, the paper constructs blid maps by an iterated-integral Taylor formula, e.g. $H(x)(t)=\\sum_{j=0}^{q-1}\\frac{t^j}{j!}h(x^{(j)}(0))x^{(j)}(0)+\\int_0^t dt_1\\cdots\\int_0^{t_{q-1}} h(x^{(q)}(s))x^{(q)}(s)\\,ds$, where $h$ is a real bump function; this operator is what gives $C^q[0,1]$ a smooth blid map despite having no smooth bump function. The same operator family also supports the Borel Lemma and the linearization argument.","core_discovery":"The central claim is that for a Banach space $X$, the existence of a $C^q$ blid map $H$ is sufficient to globalize every $C^q$ germ at $0$. If $f$ is defined on a neighborhood $U$ of $0$ and $H(X)$ lies in $U$, the map $F(x)=f(H(x))$ is a global representative; boundedness of $H$'s derivatives transfers boundedness of $f$'s derivatives. The same argument, with 'differentiable local identity' in place of bounded local identity, proves Proposition 3.5 for linear topological spaces. In Section 4, the paper uses blid maps to prove a Borel Lemma (every sequence of continuous homogeneous polynomials is the Taylor jet of a smooth map) and to replace the smooth-bump-function hypothesis in the differentiable linearization theorem of [ZLZ] by a differentiable blid map with bounded derivative; the corollary gives differentiable linearization on $C^q[0,1]$.","pith_inferences":["Beyond the paper, the iterated-integral construction on $C^q[0,1]$ is likely adaptable to $C^q(M)$ for a compact manifold $M$, using local coordinates and a partition of unity, which would broaden Theorem 2.3's examples beyond intervals.","If smooth blid maps were built on $\\ell^p$ with non-even $p$, the same theorems would immediately give global extension and differentiable linearization on those spaces; the paper leaves this as Question 5.1, so this is an extrapolation, not a claim.","The proof of Theorem 4.3 suggests a testable threshold: a blid map with only first-order bounded derivative suffices for the linearization estimate, so spaces that admit $C^1$ blid maps but not smoother ones may still linearize differentiably."],"forward_implications":["Every $C^q$ germ at $0$ on a Banach space with a $C^q$ blid map extends to a global $C^q$ map, with bounded derivatives when the local representative and blid map have bounded derivatives.","The Borel Lemma holds on any Banach space with a $C^\\infty$ blid map whose derivatives of all orders are bounded: any prescribed sequence of homogeneous polynomials occurs as the Taylor jet of some $C^\\infty$ map.","For a hyperbolic linear automorphism $A$, a formally solvable cohomological equation $g(Ax)-g(x)=f(x)$ has a global $C^\\infty$ solution on spaces with a $C^\\infty$ blid map with bounded derivatives.","Differentiable linearization at a hyperbolic fixed point holds on $C^q[0,1]$: the conjugating homeomorphism is differentiable at the fixed point, with the stated remainder estimate, even though no smooth bump function exists there."],"supporting_citations":[{"why":"Defines C^q blid maps and supplies the Borel Lemma; Theorems 2.3 and 4.1 are stated as consequences of its machinery.","marker":"[BR1]"},{"why":"The differentiable linearization theorem whose smooth-bump-function hypothesis Theorem 4.3 replaces with the blid condition.","marker":"[ZLZ]"},{"why":"Meshkov's criterion shows that many Banach spaces (e.g. l^1, C[0,1]) lack smooth bump functions, motivating the blid replacement.","marker":"[M]"},{"why":"Palis's Lipschitz extension via Lipschitz bump functions is the classical benchmark the paper's smooth extension improves upon.","marker":"[P]"},{"why":"Provides Banach spaces without C^2-extension, which frames the open question whether spaces without differentiable blid maps exist.","marker":"[DH]"},{"why":"First application of blid-type maps to smooth conjugation on Banach spaces, the historical root of the construction.","marker":"[B]"}],"fun_headline_variants":["Blid maps globalize every C^q germ","Smooth extension without bump functions via blid maps","Borel Lemma and linearization with blid maps","From local to global: blid maps in Banach spaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Every example that makes the theorems non-vacuous rests on the asserted smoothness and boundedness of the explicit blid constructions, especially the iterated-integral operator on $C^q[0,1]$ and its cousins in Lemmas 3.7, 3.9, and 3.11; if any of these is not differentiable to the claimed order with bounded derivatives, Theorems 2.3 and 4.3 lose their supporting examples.","fun_headline_variants_meta":{"raw":{"variants":["Blid maps globalize every C^q germ","Smooth extension without bump functions via blid maps","Borel Lemma and linearization with blid maps","From local to global: blid maps in Banach spaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000599,"raw_usage":{"total_tokens":2824,"prompt_tokens":993,"completion_tokens":1831,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":609,"completion_tokens_details":{"reasoning_tokens":1766}},"tokens_in":609,"tokens_out":1831,"duration_ms":13265,"temperature":1.0,"reasoning_tokens":1766,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:58:47.524318+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute explicitly the Fréchet derivative of the operator $H$ in Example 2.7 on $C^q[0,1]$. If for some $x$ the derivative fails to exist or is unbounded as a linear operator even on bounded subsets, the claimed $C^\\infty$ blid map is not delivered and the extension theorem's example collapses. A simpler check: test the claimed bound $\\sup_x\\|H(x)\\|_k < a e^k$ in Lemma 3.7 numerically on a sequence of functions with growing derivatives.","supporting_citations":[],"review_version":1}