{"id":"f8137d46-a854-4974-8539-0fa7decb78a3","arxiv_id":"2506.14183","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Under hypotheses d=ℵ_n plus weak diamonds, the nth derived limit of a natural inverse system of abelian groups is nonzero, and the second derived limit is nonzero in the Miller and Mitchell models.","lead":"This paper proves new results about higher derived limits, a set-theoretic tool used to study strong homology and condensed mathematics. It shows these limits do not vanish in several standard models of set theory, including the Miller model, disproving a recent conjecture.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 5.5 rests on an unspecified uniform coding of trivializations as branches into 2^{<ω_{n+1}}; without coherent local codes the weak-diamond diagonalization in Theorem 5.1 does not follow.","rationale":"Reading in good faith, the paper's central claim is Theorem 5.1, and its proof is Lemma 5.5. The algebraic content of Claim 5.6 checks out: the computations of dΥ in the n=1 and n>1 cases are correct, and the contradiction with Θ being nontrivial below y∧g is valid. The genuinely insecure point is the penultimate paragraph of Lemma 5.5, where the proof switches from the construction of Φ^s to the weak-diamond diagonalization with only 'for any reasonable choice of encoding attempts at trivialization'. This is precisely the load-bearing step: w♢(S) supplies a function F on 2^{<κ}, and the proof must interpret arbitrary binary sequences as 'attempts at trivialization' in such a way that a genuine global trivialization yields a single branch b whose initial segments are the codes of its restrictions. Such a coding is not automatic. Since the chain A_α is only required to be increasing and continuous, with |A_α|=ℵ_n for every α<ω_{n+1}, there is no reason that initial segments of an ordinal enumeration of the union align with the stages A_α; A_0 already has size ℵ_n while 0 has size 0. Without alignment, b↾α may not decode to the restriction of the trivialization, and the final contradiction cannot be drawn. In the intended application (Lemma 5.8, S=S_n^{n-1}) one has |α|=|A_α^{n-1}| and elements of S are limit ordinals, so a coherent bookkeeping of bijections should be constructible; but the lemma as stated does not include these hypotheses, and the paper does not supply the construction. Thus the concern is real and load-bearing, but repairable. I agree with the reader's weakest assumption. I also note Corollary 4.5 asserts without proof that the Miller, Laver, Mathias, Cohen, and Hechler models satisfy the hypotheses of Theorem 3.1; this is secondary to Theorem 5.1 but should be addressed. No machine-checked verification or reproducible code is provided. For these reasons the conditional verdict should stand.","tokens_in":14638,"tokens_out":31649,"duration_ms":316856,"concrete_test":"Fix the parameter of Lemma 5.5 to its use in Lemma 5.8 (S=S_n^{n-1}, |A_α|=ℵ_{n-1} for α<ω_n). Construct, by recursion, bijections e_α:α→(A_α)^{n-1}×ω for α∈S such that e_β = e_α↾β whenever β<α are in S, using continuity of A_α at limits and the gap between consecutive elements of S. Then verify that for any trivialization Ψ of Φ^h, there is a branch b with b↾α decoding Ψ↾(A_α)^{n-1}. If this construction fails, the diagonalization in Lemma 5.5 is unjustified; if it succeeds, the proof should state it explicitly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central proof of Theorem 5.1 is Lemma 5.5. Its final paragraph invokes w♢(S) after saying 'For any reasonable choice of encoding attempts at trivialization'. This is not a proof. To use w♢, each binary sequence b of length α∈S must be interpreted as a trivialization of the (n+1)-coherent family Φ^s (|s|=α), and a global trivialization of Φ^h must supply a branch b∈2^{ω_{n+1}} such that b↾α is exactly the code of the restriction of that trivialization to A_α, for stationarily many α. The existence of such a coding is nontrivial: the chain A_α is not required to satisfy |A_α|=|α| (in the applications A_0 already has size ℵ_n while |0|=0), so initial segments of an ordinal-length enumeration of the union need not correspond to stages of the chain. The lemma also does not state that elements of S are limit ordinals of the right cofinality, which is what would permit a coherent transfinite bookkeeping of bijections e_α:α→(A_α)^n×ω. If no uniform coherent coding exists, F cannot be defined as claimed and the contradiction in the last paragraph fails. The intended uses (S=S_i^{i-1}) may satisfy the needed closure, but as written Lemma 5.5 is incomplete.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies higher derived limits lim^n A of the Mardešić–Prasolov inverse system A indexed by ω^ω. The main theorem (Theorem 5.1) states that if d = ℵ_n and w♢(S_{k+1}^k) holds for all 1 ≤ k < n, then lim^n A ≠ 0, which is presented as a common generalization of results of Bergfalk and of Casarosa–Lambie-Hanson, with the improvement that the weak diamond on ω_1 is no longer needed. The paper also proves lim^2 A ≠ 0 in several standard models (Mitchell, Miller, Laver, Mathias, Cohen, Hechler) and completes the computation of derived limits in the Mitchell model, showing lim^n A ≠ 0 iff n = 0, 2. A further theorem (Theorem 5.9) constructs nontrivial ⊕ G_i-valued n-coherent families under d = ℵ_n. Section 7 establishes preservation of nontrivial 1-coherent families under several classes of forcings.","tokens_in":14935,"tokens_out":14105,"duration_ms":124329,"significance":"If the main arguments are correct, the paper resolves an open question by disproving a conjecture of Bergfalk–Hrušák–Lambie-Hanson about the Miller model, and it provides the first computation of the full pattern of derived limits in the Mitchell model. The removal of w♢_{ω_1} from the hypothesis of the nonvanishing theorem is a genuine strengthening. The paper is careful in many of its constructions, and the preservation results in Section 7 are of independent interest. However, the proof of the central Lemma 5.5 contains a substantial gap concerning the coding of trivializations, and the corollary listing models does not supply the required verifications; these issues need to be addressed before the results can be considered established.","major_comments":[{"comment":"The proof uses the phrase 'For any reasonable choice of encoding attempts at trivialization' and then defines a function F on 2^{<ω_{n+1}} to apply w♢(S). This is not a proof: to apply w♢, one must specify a concrete coding by which a binary sequence b of length α (for α ∈ S) encodes a trivialization of the family Φ_s on A_α, and one must ensure that these codes are coherent under restrictions along the chain A_β (β < α). The lemma as stated allows |A_α| = ℵ_n for every α, while a sequence of length α can only code objects of size |α|; for α < ω_n this is impossible. In the actual application (Lemma 5.8) one has S = S_n^{n-1} and therefore |α| = ℵ_{n-1} = |A_α| for α ∈ S, but this is not part of the lemma and the proof does not describe the bookkeeping of bijections e_α: α → (A_α)^n × ω needed for a coherent coding. Without such a coding, the function F may not exist and the diagonalization argument in the last paragraph fails. Since Lemma 5.5 is the engine of Theorem 5.1, this is a load-bearing gap.","section":"Section 5, Lemma 5.5, final paragraph"},{"comment":"The corollary asserts lim^2 A ≠ 0 in six models (Mitchell, Miller, Laver, Mathias, length-ω_2 Cohen, length-ω_2 Hechler) but contains no verification that the hypotheses of Corollary 4.2 (or any other theorem) hold in these models. For example, for the Miller model the paper does not show that the ground model satisfies ♢(S_1^2), that the relevant forcing has cardinality ℵ_2, or that it preserves the stationarity of all stationary subsets of S_1^2; indeed Miller forcing in L has size ℵ_1, so Corollary 4.2 does not apply directly. Since the Miller model is featured in the abstract as a disproof of a conjecture, the verification must be supplied or precise references given.","section":"Section 4, Corollary 4.5"},{"comment":"The proof of lim^1 A = 0 in the Mitchell model is quite terse. The reflection step 'there is a club C ⊆ κ such that whenever α ∈ C has uncountable cofinality, ˙Φ reflects to an M_α name ˙Φ_α' is stated without proof or precise definition of the structures M_α, and the subsequent argument relies on the assertion 'all reals added by Mitchell forcing are added on the Cohens coordinate', which is neither proved nor referenced. Since this completes the computation of derived limits in the Mitchell model (Theorem 6.1), the proof needs more detail.","section":"Section 6, proof of Corollary 6.4"}],"minor_comments":[{"comment":"The line 'We claim that for some h ∈ 2^{ω_1}' should read h ∈ 2^{ω_{n+1}}; as written, h↾α for α up to ω_{n+1} is not defined.","section":"Section 5, Lemma 5.5"},{"comment":"The third hypothesis uses Φ for the hypothesized n-coherent family, but Φ is already used for the constructed (n+1)-family; in the proof this family is called Θ. Rename in the statement for clarity.","section":"Section 5, Lemma 5.5"},{"comment":"In the successor step, the text defines the extensions on tuples in (A_{α+1})^{n+1} \\ (A_α)^n; the complement should be (A_{α+1})^{n+1} \\ (A_α)^{n+1} for the definition to match the arity of the families.","section":"Section 5, proof of Lemma 5.5"},{"comment":"The labels 'F act 1' and 'F act 2' have extra spaces and should be formatted as 'Fact 1' and 'Fact 2'.","section":"Section 2"},{"comment":"Reference [11] gives the author as 'K. Kenneth'; the correct name is 'K. Kunen'.","section":"References"},{"comment":"In the first sentence, 'inacessible' should be 'inaccessible'.","section":"Section 6"}],"recommendation":"major_revision","confidential_remarks":"The paper's main ideas are promising and likely correct, but the current version has a significant gap in the proof of Lemma 5.5 that is load-bearing for Theorem 5.1, and Corollary 4.5 makes unverified claims about specific models. I recommend major revision rather than rejection, as the issues appear fixable with a more detailed coding argument and explicit model-by-model verifications."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this paper has real, usable results: Theorem 3.1 (diamond on S_1^2 plus d = aleph_2 gives lim^2 A nonzero) drops b from Bergfalk's theorem, and the Mitchell model analysis in Section 6 (lim^n A nonzero iff n = 0, 2) is clean, assuming the Cohen preservation argument checks out. Second, the advertised result, Theorem 5.1 removing weak diamond from Casarosa-Lambie-Hanson, is not proven as written. The issue is Lemma 5.5's final paragraph. It says that for any reasonable encoding, a binary sequence of length alpha codes a trivialization of the family at stage alpha. That only makes sense if |A_alpha| <= |alpha|. The lemma only assumes |A_alpha| = aleph_n, so for alpha < omega_n this is false. In the intended application S = S_n^{n-1}, the ordinals in S do have size aleph_{n-1} and match |A_alpha|, so the lemma is likely repairable. But the statement needs an explicit hypothesis like |alpha| = |A_alpha| for alpha in S, and the proof needs to show the bijections can be chosen coherently. As written, this is a load-bearing gap, not a stylistic hiccup.\n\nWhat is genuinely new: Section 4's corollaries for standard models (Miller, Laver, Mathias, Cohen, Hechler) follow from Theorem 3.1 and are worth having. Section 7's preservation results, especially Lemma 6.2 on Cohen forcing not trivializing nontrivial 1-coherent families, are clean and useful. The recursive construction in Section 3 is clearly presented. The self-citations are only for context; I see no circularity.\n\nSoft spots beyond Lemma 5.5: Corollary 4.5 lists models without verifying their hypotheses, which is a minor presentational issue. The statement of Lemma 5.5 says S subset omega_{n+1} but the application uses S_n^{n-1} subset omega_n; the index shift should be stated explicitly.\n\nWho is this for: set theorists working on higher derived limits, strong homology, and Cech cohomology. It deserves a serious referee, not a desk rejection, but the referee should insist on a complete proof of Lemma 5.5 before the main theorem is accepted. If the coding can be patched, the paper will be a solid contribution.","headline":"Good core results on nonvanishing derived limits, but the paper's headline theorem rests on Lemma 5.5, whose proof as written has an unproven coding step.","tokens_in":15435,"tokens_out":7221,"would_cite":true,"duration_ms":68410,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["03E35","03E05","03E17"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that the higher derived limits of the inverse system $\\mathbf{A}$ do not vanish under the hypotheses $d = \\aleph_n$ and $w\\diamondsuit(S_{k+1}^k)$ for $1 \\le k < n$, and that in the Mitchell model the nonvanishing derived…","keywords":["higher derived limits","coherent families","weak diamond","Mitchell model","Miller model","dominating number","inverse system A","set theory of the reals"],"falsifier":"A model of $d = \\aleph_2$ with $w\\diamondsuit(S_2^1)$ in which every 2-coherent family is trivial (so $\\lim^2 \\mathbf{A} = 0$) would falsify Theorem 5.1 at $n = 2$; equivalently, one could try to find a concrete recursion where the assumed uniform binary encoding of trivialization attempts provably does not exist, which would show that the diagonalization step of Lemma 5.5 fails.","tokens_in":14450,"feed_emoji":"♾️","tokens_out":10328,"duration_ms":97445,"temperature":0.7,"pith_summary":"The paper establishes a general nonvanishing theorem for the higher derived limits of the standard inverse system of abelian groups $\\mathbf{A}$ indexed by $\\omega^\\omega$. It shows that whenever the dominating number $d$ equals $\\aleph_n$ and weak diamond principles $w\\diamondsuit(S_{k+1}^k)$ hold for $1 \\le k < n$, the $n$-th derived limit $\\lim^n \\mathbf{A}$ is nonzero. This refines earlier results by removing the hypothesis $b=d$ and by removing the full weak diamond at $\\omega_1$. In the Mitchell model, the paper completes the picture: $\\lim^n \\mathbf{A} \\ne 0$ exactly for $n = 0, 2$. A key corollary is that higher derived limits do not vanish in the Miller model, disproving a published conjecture.","feed_headline":"Second derived limit survives in the Miller and Mitchell models","feed_subtitle":"Under d = ℵ_n plus weak diamond, the n-th derived limit never vanishes; the Miller-model conjecture falls.","key_machinery":"The central object is the inverse system $\\mathbf{A}$ of abelian groups indexed by $\\omega^\\omega$ (with $\\mathbf{A}_x = \\bigoplus_{i<\\omega} \\mathbb{Z}_{x(i)}$ and projection maps), whose $n$-th derived limit vanishes exactly when every $n$-coherent family of functions is $n$-trivial. The paper's engine is a recursion on an internally approaching chain of elementary substructures $M_\\alpha$, combined with the weak diamond principle $w\\diamondsuit(S)$: at each stage, if the current family has a trivialization $\\Psi$, the construction splits into two branches, one using $\\Psi$ and one using $\\Psi + \\Theta$ where $\\Theta$ is an $n$-coherent family nontrivial below $y \\wedge g$, so that no single trivialization can cover both branches. The weak diamond then selects a branch $h$ through the binary tree along which every coded trivialization attempt is diagonalized away. The auxiliary notion ``trivial below $g$'' (with $g$ possibly the constant function $\\omega$) lets the induction climb from $n$ to $n+1$.","core_discovery":"The central discovery is that the nonvanishing of higher derived limits follows from the dominating number alone together with weak diamonds on the stationarity ladder $S_{k+1}^k$, without needing $b=d$ or $w\\diamondsuit_{\\omega_1}$. The proof constructs, for any attempt at a trivializing family, a binary-branching recursion indexed by ordinals; the weak diamond principle guarantees a branch along which every trivialization attempt fails, yielding a nontrivial $(n+1)$-coherent family below a carefully chosen function $g$. As a consequence, in the Mitchell model the derived limits $\\lim^n \\mathbf{A}$ are zero for $n \\neq 0, 2$ and nonzero for $n = 0, 2$, and in models such as the Miller, Laver, Mathias, length-$\\omega_2$ Cohen, and length-$\\omega_2$ Hechler models, $\\lim^2 \\mathbf{A} \\ne 0$, contradicting the conjecture that all higher derived limits vanish in the Miller model.","pith_inferences":["The proof of the key lemma only uses weak diamond to choose one of two branches at stationarily many stages; if a coding-free version of that choice could be made, Theorem 5.1 might hold under weaker guessing principles such as club guessing, which the author does not claim.","Because the group-valued result replaces weak diamonds with the large cardinality of the value group, the paper suggests that the size of the target group is what creates nontrivial coherent families; one could try to transfer this to other inverse systems by enlarging value groups rather than adding diamonds.","The preservation results in Section 7 are stated for 1-coherent families under Cohen, Miller, $\\sigma$-distributive, strongly proper, and $\\omega^\\omega$-bounding forcings; the same methods may extend to $n$-coherent families for $n \\ge 2$ only if a version of ``trivial below $g$'' retains its inductive strength, which the paper leaves open."],"forward_implications":["Under the hypotheses $d = \\aleph_n$ and $w\\diamondsuit(S_{k+1}^k)$ for $1 \\le k < n$, the $n$-th derived limit $\\lim^n \\mathbf{A}$ is nonzero, so every model with these hypotheses contains a nontrivial $n$-coherent family.","In particular, $\\lim^2 \\mathbf{A} \\ne 0$ in the Miller, Mitchell, Laver, Mathias, length-$\\omega_2$ Cohen, and length-$\\omega_2$ Hechler models, since each satisfies the relevant diamond hypotheses after forcing.","The Mitchell model is completely analyzed: $\\lim^n \\mathbf{A} \\ne 0$ if and only if $n = 0, 2$, so its derived-limit pattern matches the proper-forcing-axiom pattern.","The conjecture that all higher derived limits of $\\mathbf{A}$ vanish in the Miller model is false: the second derived limit is provably nonzero there.","With $d = \\aleph_n$, for any nonzero abelian groups $G_i$ indexed by $i < \\omega_n$ there is a nontrivial $n$-coherent family valued in $\\bigoplus_{i<\\omega_n} G_i$, showing that small continuum forces an abundance of nontrivial coherent families."],"supporting_citations":[{"why":"Introduced the inverse system $\\mathbf{A}$ and established that CH gives $\\lim^1 \\mathbf{A} \\ne 0$; this is the object all later results study.","marker":"[10]"},{"why":"Proved the first consistent nonvanishing of $\\lim^2 \\mathbf{A}$ from $b=d=\\aleph_2$ and $\\diamondsuit(S_2^1)$; the theorem that Theorem 3.1 refines.","marker":"[3]"},{"why":"Established simultaneous nonvanishing under $w\\diamondsuit_{\\omega_1}$ and supplied Theorem A(2), which Theorem 5.1 generalizes by removing $w\\diamondsuit_{\\omega_1}$; also the source of the re-proved Theorem A(1) on group-valued families.","marker":"[7]"},{"why":"Gave the template for Theorem 5.1's hypotheses, showing consistency of $d = \\aleph_n$ with $w\\diamondsuit(S_{k+1}^k)$ for $1 \\le k < n$.","marker":"[14]"},{"why":"Weakened the $b=d$ hypothesis for nonvanishing relative to [14]; the starting point for eliminating $b=d$ entirely.","marker":"[6]"},{"why":"Provided the chain construction theorem used to produce 1-coherent families nontrivial below a function $g$ in Lemma 3.4.","marker":"[13]"},{"why":"Constructed models with all higher derived limits vanishing and raised the Miller-model question; this is the conjecture the paper disproves.","marker":"[5]"},{"why":"Used for the standard fact that $\\diamondsuit^+$ implies $\\diamondsuit(S)$ for every stationary $S$, which puts the Mitchell model into the hypotheses of Theorem 3.1.","marker":"[11]"}],"fun_headline_variants":["Weak diamond and dominating number defeat derived limit conjecture","Miller model conjecture refuted: higher derived limits survive","Nonvanishing derived limits via weak diamond and dominating number","Second derived limit survives: Miller model conjecture disproved","Higher derived limits refuse to vanish in the Miller model"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that every possible trivialization of the families built along the recursion can be uniformly encoded as a binary sequence so that the weak diamond principle can diagonalize against them, and the paper does not specify this encoding.","fun_headline_variants_meta":{"raw":{"variants":["Weak diamond and dominating number defeat derived limit conjecture","Miller model conjecture refuted: higher derived limits survive","Nonvanishing derived limits via weak diamond and dominating number","Second derived limit survives: Miller model conjecture disproved","Higher derived limits refuse to vanish in the Miller model"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000183,"raw_usage":{"total_tokens":1281,"prompt_tokens":875,"completion_tokens":406,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":491,"completion_tokens_details":{"reasoning_tokens":330}},"tokens_in":491,"tokens_out":406,"duration_ms":4933,"temperature":1.0,"reasoning_tokens":330,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T00:18:45.638859+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A model of $d = \\aleph_2$ with $w\\diamondsuit(S_2^1)$ in which every 2-coherent family is trivial (so $\\lim^2 \\mathbf{A} = 0$) would falsify Theorem 5.1 at $n = 2$; equivalently, one could try to find a concrete recursion where the assumed uniform binary encoding of trivialization attempts provably does not exist, which would show that the diagonalization step of Lemma 5.5 fails.","supporting_citations":[{"cited_title":"Mardeˇ si´ c and A.V","cited_arxiv_id":null,"evidence_quote":"Introduced the inverse system $\\mathbf{A}$ and established that CH gives $\\lim^1 \\mathbf{A} \\ne 0$; this is the object all later results study."},{"cited_title":"Bergfalk","cited_arxiv_id":null,"evidence_quote":"Proved the first consistent nonvanishing of $\\lim^2 \\mathbf{A}$ from $b=d=\\aleph_2$ and $\\diamondsuit(S_2^1)$; the theorem that Theorem 3.1 refines."},{"cited_title":"Velickovic, A","cited_arxiv_id":null,"evidence_quote":"Gave the template for Theorem 5.1's hypotheses, showing consistency of $d = \\aleph_n$ with $w\\diamondsuit(S_{k+1}^k)$ for $1 \\le k < n$."},{"cited_title":"Todorˇ cev ´ ıc.Walks On Ordinals and Their Characteristics","cited_arxiv_id":null,"evidence_quote":"Provided the chain construction theorem used to produce 1-coherent families nontrivial below a function $g$ in Lemma 3.4."},{"cited_title":"Bergfalk, M","cited_arxiv_id":null,"evidence_quote":"Constructed models with all higher derived limits vanishing and raised the Miller-model question; this is the conjecture the paper disproves."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Used for the standard fact that $\\diamondsuit^+$ implies $\\diamondsuit(S)$ for every stationary $S$, which puts the Mitchell model into the hypotheses of Theorem 3.1."}],"review_version":1}