{"id":"c65ba807-1bba-4d12-b9da-6309f0ea4c29","arxiv_id":"1908.09115","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For normed spaces with separable duals, weak-topology spaces are sequentially homeomorphic exactly when their classes of closed bounded weak subsets coincide; for hyperplane-isomorphic Banach spaces, the compact-convergence topologies are homeomorphic under the same condition.","lead":"This paper gives a classification rule for when two normed vector spaces equipped with the weak topology are sequentially homeomorphic. It also shows that, under mild extra assumptions, the stronger topology of compact convergence is homeomorphic to the product of the weak unit ball with the space R∞.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Reflexive case of Theorem 1's only-if direction assumes the conclusion; the flagged non-reflexive neighborhood step is actually valid.","rationale":"Most of the classification scheme — the C-injective technology, the use of Banach-Dieudonné, and the applications to c and R∞ — is coherent, and the 'if' direction of Theorem 1 follows from Propositions 5–7 together with the cited Uniqueness Theorem. The paper uses standard tools and has no free parameters; the main risk is concentrated in the only-if direction. I rechecked the specific assertion on which the Reader based the conditional verdict and found that it is in fact true by a standard diagonal argument for first-countable subsets of direct limits of closed towers. Therefore the Reader's stated weakest assumption is not the real soft spot. The soft spot is the reflexive case: the text explicitly invokes W(Y)=W(X) in the middle of proving W(X)=W(Y). For the conclusion to go through one needs an additional, unstated fact about homeomorphisms of s-topologies preserving the ball tower. This is a genuine proof gap, but it does not show the theorem is false; it shows the proof is incomplete. The appropriate verdict remains CONDITIONAL, with the condition being a repaired justification of the reflexive case. This is why I mark agreement as disagree while keeping the final verdict unchanged.","tokens_in":9142,"tokens_out":33722,"duration_ms":379735,"concrete_test":"Try to prove the missing bornological preservation lemma for the homeomorphisms between (X,s) and (Y,s): for every n there is m with h(nB_X)⊂mB_Y. If it holds, Case 1 follows and the 'W(Y)=W(X)' phrase is a typo. If it fails, exhibit a homeomorphism h with Y reflexive and a closed bounded weak set A⊂X such that h(A) is not relatively compact in (Y,s); such an example would show the compactness step is not justified and the reflexive case requires a different argument.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The step flagged in the Reader's verdict — in the non-reflexive case of the only-if proof of Theorem 1 — is actually valid. If a closed, first-countable subset A of (Y,s)=lim nBY had a point a with no neighborhood contained in any nBY, a countable decreasing local base (U_k) would give x_k∈U_k∩(A\\kB_Y); since S={x_k} meets each mB_Y in a finite set and finite sets are closed in the weak topology on mB_Y, V=(Y,s)\\S is an s-open neighborhood of a containing no U_k, contradicting first-countability. The genuine soft spot is the reflexive case: the proof says 'each space from the class W(Y)=W(X) is compact' and concludes that A and A′ are compact. This invokes W(X)=W(Y), the very equality being proved. From A∈W(X) and reflexivity of Y one gets no compactness of A without knowing that the homeomorphism h:(X,s)→(Y,s) preserves the tower bornology, i.e., h(nB_X)⊂mB_Y; no such preservation lemma is stated or proved. Thus the only-if direction is not fully supported for reflexive Y.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the weak topology on normed spaces and the finer topology c of uniform convergence on compact subsets of the dual, in the setting of spaces with separable dual. Its main result, Theorem 1, asserts that for normed spaces X and Y with separable duals, the spaces (X, weak) and (Y, weak) are sequentially homeomorphic if and only if the classes W(X) and W(Y) of spaces homeomorphic to closed bounded subsets of the respective weak topologies coincide. Theorem 2 gives an analogous equivalence between c-topology homeomorphisms and equality of W(X) for Banach spaces isomorphic to their hyperplanes. The proofs are built on a theory of C-injective pairs and spaces, with Theorems 3 and 4 describing pairs of double-dual spaces and products with R^∞. The manuscript also contains Proposition 1, noting that (X,c) and (X,weak) are sequentially homeomorphic but not homeomorphic.","tokens_in":9360,"tokens_out":6905,"duration_ms":76478,"significance":"If the results are correct, they provide a complete sequential classification of weak topologies on normed spaces with separable dual, reducing the problem to the class W(X), and they connect the c-topology to products of metrizable spaces with R^∞. The overall architectural strategy is coherent and uses standard infinite-dimensional topology tools. I note that the non-reflexive neighborhood step flagged in the reader's report is in fact valid: a closed first-countable subset of the direct limit (Y,s) cannot accumulate at infinity, because otherwise one could remove a closed discrete sequence and contradict first-countability. However, the reflexive case of Theorem 1 contains a circular argument, and the uniqueness theorem for C-injective pairs is not proved in the manuscript. These are load-bearing gaps that require repair before the central claims can be accepted.","major_comments":[{"comment":"The proof of the only-if direction in the reflexive case states: 'Then each space from the class W(Y)=W(X) is compact.' This uses the equality W(X)=W(Y), which is exactly what is being proved in that part of the argument. From A ∈ W(X) and reflexivity of Y one only obtains that A is a closed bounded subset of (X,weak); without knowing that the homeomorphism h:(X,s)→(Y,s) preserves boundedness, there is no reason that the homeomorphic image A' is bounded in Y, hence no reason that A' is weakly compact. The proof needs an additional argument showing that h maps bounded closed subsets of (X,s) to bounded subsets of (Y,s), or some other non-circular route to A' ∈ W(Y). This gap is load-bearing because the whole forward implication of Theorem 1 depends on it.","section":"Theorem 1, proof, reflexive case"},{"comment":"Theorem 5 is the uniqueness statement for C-injective pairs and spaces, and it is used as the key step in the 'if' parts of Theorems 1, 3, and 4. The proof is only a reference: 'Repeating arguments of [24] ... one may easily prove.' Since Theorem 5 is foundational for the paper's conclusions, the argument should be supplied in the manuscript or, at minimum, an explicit statement with hypotheses should be quoted from [24], [5], or [23]. As written, a reader cannot verify this load-bearing uniqueness claim from the manuscript.","section":"Section 3, Theorem 5"},{"comment":"The second implication of Theorem 2 depends on Corollary 2, which in turn requires the [0,1]-stability of W(X**,X) for Banach spaces isomorphic to their hyperplanes. This stability is quoted from [7] without proof. This is an external theorem rather than an error, but the dependence should be stated clearly in the proof and the exact result in [7] should be cited with a theorem number, since the validity of Corollary 2 for all such spaces is essential to the classification claim.","section":"Theorem 2, second implication"}],"minor_comments":[{"comment":"The manuscript contains many typographical errors, including 'seqeuntially', 'wek', 'clsoed', 'A\"', 'hyperpane', and 'separale duals'; these should be corrected.","section":"Throughout"},{"comment":"The phrase 'if and only of' appears in the abstract and should read 'if and only if'.","section":"Abstract and Theorem 1 statement"},{"comment":"The text says 'the spaces (X,weak) and (X,weak) are sequentially homeomorphic'; the second space should presumably be (Y,weak).","section":"Proof of Theorem 2, first paragraph"},{"comment":"The class W(X**,X) is defined using pairs (K, K∩X) with K compact in X**_c, but the sentence says 'equivalently, of the second dual space X** endowed with the *-weak topology.' This equivalence is true for compact subsets because the c-topology agrees with the *-weak topology on bounded sets, but the justification should be given explicitly rather than parenthetically.","section":"Definition of W(X**,X), Section 1"},{"comment":"The proof says 'Thus it is legal to apply Proposition 3' after verifying the hypotheses, but the phrase 'W(X**,X)-universal' should be 'W(X**,X)-injective' in the final sentence for clarity.","section":"Proposition 6 proof"}],"recommendation":"major_revision","confidential_remarks":"The paper depends heavily on previous works by the same author, especially [4], [5], [7], and [8]. The editor may wish to confirm that the overlap with [7] is sufficiently transparent and that the claimed new results are not already contained in those papers. The circular step in the reflexive case of Theorem 1 is the main mathematical obstruction; if it cannot be repaired, the central classification theorem would need to be reformulated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Three things. First, the headline result is real: for normed spaces with separable duals, (X,weak) and (Y,weak) are sequentially homeomorphic iff W(X)=W(Y). The C-injective machinery is doing the lifting, and the product model B×R^∞ for (X,c) is a concrete payoff. Second, the reader's worry about the non-reflexive neighborhood step is not the problem: if a point in a closed first-countable subset A'⊂(Y,s) had no neighborhood inside any nB_Y, a countable decreasing local base would produce a sequence escaping every finite ball; that sequence's set is closed in (Y,s) since it meets each bounded set finitely, so the complement is an s-open neighborhood missing all base sets—contradiction. That step holds. Third, the real soft spot is the reflexive case of the same only-if direction. The proof says \"each space from the class W(Y)=W(X) is compact\" and concludes A and A' are compact. That invokes the very equality being proved. At that point nothing shows the s-homeomorphism (X,s)→(Y,s) preserves boundedness, so a closed copy of A inside (Y,s) need not be bounded and compactness does not follow. This is load-bearing in the only-if direction as written.\n\nCredit where due: Theorem 1, Corollary 2, and Theorem 2 are genuinely new statements, not in the earlier papers. The reliance on [4],[5],[7],[8] is heavy but those are prior published theorems with stated hypotheses, not fitted parameters. Theorem 5 is deferred to prior argument rather than proved; that is acceptable here, though a referee might ask for the argument to be included. The citation pattern is mostly self-citations, but they are to published work and do not look like an attempt to inflate.\n\nWho this is for: people in infinite-dimensional topology and Banach space geometry. They get a natural classification statement, a reusable C-injective toolkit, and a clean product model. The paper deserves a serious referee. My recommendation: send it to peer review, and require the author to repair the reflexive circularity—either by proving that boundedness is preserved by the sequentially weak homeomorphism, or by a different argument. If that is fixed, the results stand. As written, the main theorem is not fully supported, but the gap looks fixable.","headline":"The paper's main classification for weak topologies with separable duals is new and plausible, but the reflexive case of Theorem 1's only-if direction is circular; the non-reflexive step the reader flagged is actually valid.","tokens_in":9889,"tokens_out":4148,"would_cite":true,"duration_ms":40049,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57N17","57N20","46A20","46A19"],"pacs":[],"model":"deepseek-v4-flash","headline":"For normed spaces with separable duals, the weak topology's sequential homeomorphism type is completely determined by its closed bounded weak subsets, and for spaces isomorphic to their hyperplanes the compact-convergence topology is…","keywords":["weak topology","topology of compact convergence","sequential homeomorphism","infinite-dimensional topology","C-injective pairs","R∞","weak unit ball","separable dual"],"falsifier":"Take a non-reflexive normed space $Y$ with separable dual, such as $c_0$, and seek a closed separable metrizable subspace $A'$ of $(Y,s)$ with a point $a$ whose every closed $s$-neighborhood meets $Y \\setminus nB_Y$ for all $n$; the existence of such a subspace would invalidate the only-if direction of Theorem 1.","tokens_in":8932,"feed_emoji":"♾️","tokens_out":14785,"duration_ms":122162,"temperature":0.7,"pith_summary":"The paper proves that for normed spaces $X, Y$ with separable duals, the spaces $(X,\\mathrm{weak})$ and $(Y,\\mathrm{weak})$ are sequentially homeomorphic if and only if the classes $\\mathcal W(X)$ and $\\mathcal W(Y)$ coincide, where $\\mathcal W(Z)$ is the class of topological spaces homeomorphic to closed bounded subsets of $(Z,\\mathrm{weak})$. It also shows that when $X$ is isomorphic to its hyperplane and has separable dual, the topology $c$ of compact convergence on $X$ is homeomorphic to the product $B \\times \\mathbb{R}^\\infty$, where $B$ is the weak unit ball and $\\mathbb{R}^\\infty$ is the real vector space with countable Hamel basis carrying the strongest locally convex topology; the weak topology on $X$ is then sequentially homeomorphic to the same product. As a consequence, for Banach spaces isomorphic to their hyperplanes with separable duals, the spaces $(X,c)$ and $(Y,c)$ are homeomorphic exactly when $\\mathcal W(X) = \\mathcal W(Y)$. The paper's significance lies in reducing a topological classification problem about highly non-metrizable spaces to a question about metrizable compacta, namely the closed bounded weak subsets.","feed_headline":"Weak topology type is fixed by closed bounded subsets","feed_subtitle":"Closed bounded weak subsets completely determine the sequential homeomorphism type.","key_machinery":"The load-bearing objects are the invariant $\\mathcal W(X)$ — the class of topological spaces homeomorphic to closed bounded subsets of $(X,\\mathrm{weak})$ — and the $\\mathcal C$-injective pair: a pair $(X,Y)$ equipped with the direct limit topology of a tower $X_1 \\subset X_2 \\subset \\cdots$ of closed subspaces, each piece $(X_n, X_n \\cap Y)$ belonging to a fixed class $\\mathcal C$ of pairs, such that every closed embedding of a piece $(B, B \\cap C)$ of a pair $(K,C) \\in \\mathcal C$ into $X_n$ extends to a closed embedding of all of $K$ into some $X_m$ with the correct preimage of $Y$. The topology of compact convergence $(X,c)$ is shown to be such a direct limit, over the tower of weak-star compact balls of the second dual; the strongest topology $s$ agreeing with the weak topology on bounded sets is the direct limit over the balls $nB$. The space $\\mathbb{R}^\\infty$ — the real vector space with countable Hamel basis and the strongest locally convex topology — is the canonical unbounded factor: for $X$ isomorphic to its hyperplane, $(X,c)$ is homeomorphic to $B \\times \\mathbb{R}^\\infty$. A uniqueness theorem for $\\mathcal C$-injective pairs carries the classification: any two $\\mathcal C$-injective pairs belonging to the same class $\\mathcal C$ are homeomorphic.","core_discovery":"For a normed space $Z$ with separable dual, let $\\mathcal W(Z)$ denote the class of all topological spaces homeomorphic to closed bounded subsets of $(Z,\\mathrm{weak})$. The paper's central theorem states that for normed spaces $X$ and $Y$ with separable duals, $(X,\\mathrm{weak})$ and $(Y,\\mathrm{weak})$ are sequentially homeomorphic if and only if $\\mathcal W(X) = \\mathcal W(Y)$. A companion result identifies the topology of compact convergence: if $X$ is a normed space isomorphic to its hyperplane and has separable dual, then $(X,c)$ is homeomorphic to $B \\times \\mathbb{R}^\\infty$, where $B$ is the weak unit ball of $X$ and $\\mathbb{R}^\\infty$ is the countable-dimensional locally convex model space; consequently $(X,\\mathrm{weak})$ is sequentially homeomorphic to $B \\times \\mathbb{R}^\\infty$. The proof proceeds by introducing a class of '$\\mathcal C$-injective pairs', showing that $(X,c)$ and $(X,s)$ — the latter being the direct limit of the balls $nB$ with the strongest topology agreeing with the weak topology on bounded sets — are $\\mathcal C$-injective for the appropriate classes, and then invoking a uniqueness theorem: any two pairs injective with respect to the same class of metrizable pairs are homeomorphic.","pith_inferences":["The author leaves implicit that the sequential equivalence in Theorem 1 likely extends to any pair of 'sequentially weak' topologies, since Corollary 1 already states this; the direct-limit space $(X,s)$ shows the sequential type is a property of the bounded pieces alone.","A testable extension would drop separability of the dual: if the weak-star ball is no longer metrizable, the $\\mathcal C$-injective machinery for metrizable pairs may need a broader class, and the $\\mathbb{R}^\\infty$ model may fail; the paper offers no evidence either way.","The unproved neighborhood assertion in the non-reflexive case suggests a possible repair: replace 'closed neighborhood in $A'$ contained in $nB$' by a weaker localization, such as neighborhoods whose weak closure is compact, which might still force $A' \\in \\mathcal W(Y)$.","Because $(X,c)$ is homeomorphic to $B \\times \\mathbb{R}^\\infty$ for hyperplane-isomorphic $X$, the topological classification of such spaces reduces entirely to the classification of weak unit balls, making Theorem 2 effectively a renorming-free version of the earlier weak-unit-ball classification."],"forward_implications":["If $\\mathcal W(X) = \\mathcal W(Y)$ for normed spaces with separable duals, then $(X,\\mathrm{weak})$ and $(Y,\\mathrm{weak})$ are sequentially homeomorphic; bounded closed weak pieces completely determine the sequential weak type.","Conversely, any sequential homeomorphism of weak topologies forces equality of the classes $\\mathcal W(X)$ and $\\mathcal W(Y)$, making $\\mathcal W$ a complete invariant for the weak topology's sequential type.","For Banach spaces isomorphic to their hyperplanes with separable duals, the compact-convergence topologies $(X,c)$ and $(Y,c)$ are homeomorphic exactly when $\\mathcal W(X) = \\mathcal W(Y)$, giving a full classification of the $c$-topology in that class.","For such spaces $X$, the space $(X,c)$ is homeomorphic to $B \\times \\mathbb{R}^\\infty$ and $(X,\\mathrm{weak})$ is sequentially homeomorphic to $B \\times \\mathbb{R}^\\infty$, so the unbounded part is always the same $\\mathbb{R}^\\infty$ factor and the bounded part carries all the information.","The topologies $c$ and $\\mathrm{weak}$ coincide on bounded sets and are sequentially homeomorphic, yet for infinite-dimensional $X$ with separable dual they are not homeomorphic at all — the compact-convergence topology is stratifiable, the weak topology is not."],"supporting_citations":[{"why":"It introduces the class W(X) and proves the homeomorphism classification of weak unit balls under renormings that make weak and norm topologies agree on the unit sphere, which Theorem 2 builds on.","marker":"[7]"},{"why":"It supplies the standard facts about weak-star compactness and metrizability of balls in spaces with separable duals, and the existence of suitable renormings.","marker":"[19]"},{"why":"It gives the topological characterization of R∞ used to identify the product model with a direct limit of cubes.","marker":"[24]"},{"why":"It provides the Banach-Dieudonné theorem identifying the topology of compact convergence as the strongest topology agreeing with the weak-star topology on bounded sets.","marker":"[25]"},{"why":"It supplies the strong universality of convex sets used to verify that the pairs (X,c) and (X,s) are injective.","marker":"[4]"},{"why":"It demonstrates the non-stratifiability of the weak topology of infinite-dimensional normed spaces, used to show c and weak are not homeomorphic.","marker":"[17]"}],"fun_headline_variants":["Weak topology type fixed by closed bounded sets","Bounded closed subsets determine sequential homeomorphism type","Separable duals: weak type from bounded closed sets","Weak topology classified by closed bounded subsets"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"In the only-if direction of Theorem 1, the author assumes without proof that in the direct-limit topology $(Y,s)$, each point of a closed separable metrizable subspace has a closed neighborhood lying in some finite ball $nB_Y$.","fun_headline_variants_meta":{"raw":{"variants":["Weak topology type fixed by closed bounded sets","Bounded closed subsets determine sequential homeomorphism type","Separable duals: weak type from bounded closed sets","Weak topology classified by closed bounded subsets"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00104,"raw_usage":{"total_tokens":4468,"prompt_tokens":1132,"completion_tokens":3336,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":748,"completion_tokens_details":{"reasoning_tokens":3278}},"tokens_in":748,"tokens_out":3336,"duration_ms":24541,"temperature":1.0,"reasoning_tokens":3278,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:21:53.335389+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a non-reflexive normed space $Y$ with separable dual, such as $c_0$, and seek a closed separable metrizable subspace $A'$ of $(Y,s)$ with a point $a$ whose every closed $s$-neighborhood meets $Y \\setminus nB_Y$ for all $n$; the existence of such a subspace would invalidate the only-if direction of Theorem 1.","supporting_citations":[{"cited_title":"Banakh, On topological classification of weak unit balls in Banach sp aces, Dissert","cited_arxiv_id":null,"evidence_quote":"It introduces the class W(X) and proves the homeomorphism classification of weak unit balls under renormings that make weak and norm topologies agree on the unit sphere, which Theorem 2 builds on."},{"cited_title":"Habala, P","cited_arxiv_id":null,"evidence_quote":"It supplies the standard facts about weak-star compactness and metrizability of balls in spaces with separable duals, and the existence of suitable renormings."},{"cited_title":"Sakai, On R∞ -manifolds and Q∞ -manifolds, Topology Appl","cited_arxiv_id":null,"evidence_quote":"It gives the topological characterization of R∞ used to identify the product model with a direct limit of cubes."},{"cited_title":"Schaefer, Topological vector spaces, The Macmillan Co., New York, (1966), ix+294 pp","cited_arxiv_id":null,"evidence_quote":"It provides the Banach-Dieudonné theorem identifying the topology of compact convergence as the strongest topology agreeing with the weak-star topology on bounded sets."},{"cited_title":"Banakh, Toward a topological classification of convex sets in infinit e-dimensional Frchet spaces , Topology Appl","cited_arxiv_id":null,"evidence_quote":"It supplies the strong universality of convex sets used to verify that the pairs (X,c) and (X,s) are injective."},{"cited_title":"Gartside, Nonstratifiability of topological vector spaces , Topology Appl","cited_arxiv_id":null,"evidence_quote":"It demonstrates the non-stratifiability of the weak topology of infinite-dimensional normed spaces, used to show c and weak are not homeomorphic."}],"review_version":1}