{"id":"ae3ed32c-0b9a-4e95-875a-15003695bd14","arxiv_id":"2411.19319","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For rooted tree posets, the additive closure of zero-dimensional persistent homology is of finite type, with indecomposables given by reduced rooted tree modules, and decomposition is computable in quadratic time.","lead":"For data filtered over tree-shaped index sets, the paper proves that the zero-dimensional persistent homology can always be decomposed into a finite list of building blocks, and gives a quadratic-time algorithm to find them. It matters because it extends the tractability of classical one-dimensional persistence to a useful non-linear family of posets.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 4.1's key implication (3)⇒(1) is not proven as written: the quantifier in the given argument is wrong, and this step underlies the elder rule and hence the finite-type classification.","rationale":"I read the paper in good faith and checked the places where the reader's conditional verdict pointed. The finiteness of reduced rooted tree quivers in Corollary C is indeed asserted in one sentence, but the induction is straightforward: over Qi the reduced rooted trees form a finite set (by induction on the size of Qi), and a reduced tree over Q is a tuple of antichains in these finite posets, so there are finitely many isomorphism classes. Lemma 3.5 is stated without proof, but Σ and fib are plainly inverse up to natural isomorphism because every vertex of a set-valued functor on a rooted tree quiver has a unique successor path to the root. The algorithms in Section 5 are consistent with the structural results. The real soft spot I found is different: the proof of Proposition 4.1(3)⇒(1) has a quantifier error. The statement is true, and a correct proof can be recovered by a column-sum compatibility argument at the edge from each branch root to the global root, but that argument is not in the paper. Since Proposition 4.1 underpins the elder rule, Lemma 4.5, Theorem 4.6, and hence Corollary C, the written proof of the central claim is incomplete at a load-bearing point. This does not make the mathematics false, but it does justify the reader's CONDITIONAL verdict. I therefore recommend leaving the verdict unchanged; the concern is a proof gap to be repaired, not a counterexample to the classification.","tokens_in":22994,"tokens_out":26850,"duration_ms":250606,"concrete_test":"Prove (3)⇒(1) of Proposition 4.1 rigorously: for a morphism ϕ:kS→kT nonzero at the root, write λ=ϕ_σ and let M_i be the root-scalar matrix of the restriction to Qi. Compatibility at the edge from the root of Qi to the root of Q gives η_i^T M_i 1 = λ for each column? More precisely, for each source branch j, Σ_n (M_i)_{n j} = λ. If λ≠0, every column of every M_i is nonzero, so each branch S^j_i has at least one target with a nonzero root scalar; then apply induction. Verify this column-sum identity in all characteristics; if it fails for some field, the elder rule and Theorem B would be in doubt.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Proposition 4.1 asserts that S≼Q T iff there is a morphism kS→kT nonzero at the root. The proof of (3)⇒(1) states: 'Since ϕ is non-zero at the root, for every 1≤i≤k, there exist 1≤j≤ℓi and 1≤n≤mi such that ϕi,j,n is non-zero at the root of Qi.' But Definition 2.14 requires, for every i and every j, the existence of some n with S^j_i ≼ T^n_i. The written argument only produces, for each i, one pair (j,n), so it does not establish the required implication. This is not merely cosmetic: a block of the root matrix could be zero for one source branch while another branch makes the global root map nonzero, and compatibility along the edge from the branch root to the global root is what forces every column to contribute. Concretely, for a fixed branch i, let M_i be the matrix of root scalars of the restricted morphism on Qi. Compatibility at the edge from the root of Qi to the root of Q forces every column sum of M_i to equal the global root scalar λ. If λ≠0, every column has a nonzero entry, giving for each source branch j some target n with a nonzero root scalar; induction then proves S^j_i ≼ T^n_i. This column-sum argument is absent, and the proof printed in the paper is invalid as written. Because Proposition 4.1 is used in the elder rule (Proposition 4.3), in Lemma 4.5, and in Theorem 4.6, the central classification Corollary C depends on this keystone implication.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the linear representations of a rooted tree quiver Q obtained by applying zero-dimensional persistent homology to Q-indexed filtrations of topological spaces (equivalently, set-valued functors). The main results are: Theorem A characterizes the essential image repH0(Q) as the finite direct sums of linearized rooted tree quivers over Q, and its additive closure as the finite direct sums of rooted tree modules; Theorem B shows every rooted tree module over Q decomposes as a direct sum of reduced rooted tree modules; Corollary C concludes that add(repH0(Q)) is of finite type, with indecomposables precisely the reduced rooted tree modules; and Theorem D provides quadratic-time algorithms for the decomposition. The proofs are built on an inductive description of rooted tree quivers over Q, a preorder ≼_Q, and an explicit elder rule (Proposition 4.3).","tokens_in":23278,"tokens_out":18472,"duration_ms":153535,"significance":"If the results hold, the paper makes a valuable contribution to both representation theory and persistence theory: it identifies a natural subcategory of representations of rooted tree quivers—those arising from zero-dimensional persistent homology—that is of finite type even though the ambient category rep(Q) is generally wild, and it provides a concrete quadratic-time decomposition algorithm. The paper redevelops rather than black-boxes Kinser's theory, proves the elder rule by an explicit isomorphism, and gives correctness proofs by invariant for the algorithms. The finite-type classification and the algorithmic results are concrete and falsifiable, and the connection to merge-tree morphisms in Section 6 indicates useful applications. The main caveats are two proof gaps identified below, both repairable.","major_comments":[{"comment":"The proof as printed does not establish the implication: it shows only that for each branch i there exist j and n with φ_{i,j,n} nonzero at the root of Q_i, whereas Definition 2.14 requires, for every i and every j, the existence of some n with S^j_i ≼_{Q_i} T^n_i. The missing step is a column-sum argument at the edge σ_i→σ: since the structure maps from σ_i to σ are sums of identity maps, the compatibility condition forces every column of the root matrix of φ at σ_i to have sum equal to the nonzero root scalar λ, hence each column has a nonzero entry. With that argument, induction applies to every j. Because this implication is used in Proposition 4.3, Lemma 4.5, Theorem 4.6, and Theorem A(2), the proof must be corrected.","section":"Proposition 4.1, proof of (3)⇒(1)"},{"comment":"The finite-type conclusion depends on the assertion that there are finitely many isomorphism classes of reduced rooted tree quivers over a fixed Q, but the proof is a single sentence referring to Definition 2.15 and induction. Please spell out the induction: for each vertex x of Q, the fiber of a reduced T at x is, for each child Q_i of x, an antichain in the finite poset of reduced rooted tree quivers over Q_i (finite by induction), and the height of any T is bounded by |Q|; this gives the required finiteness. Without this step the 'finite type' claim in Corollary C is unsupported.","section":"Corollary C"}],"minor_comments":[{"comment":"The phrase 'join x∨y (i.e., greatest lower bound)' is incorrect: the join is the least upper bound; the greatest lower bound is the meet.","section":"Lemma 2.3"},{"comment":"In the second bullet, the lists are denoted N•_1,...,N•_n, but Q has k branches Q_1,...,Q_k; the index should be k.","section":"Definition 2.18"},{"comment":"The sentence 'where d∈N is such that, and note that, if succ^d(x) is the root of Q' is garbled; it should say 'where d is the unique integer such that succ^d(x) is the root of Q'.","section":"Proof of Theorem A(2)"},{"comment":"In the paragraph checking condition (2), the sentence 'If x has no predecessors, then this tree quiver is the trivial rooted tree quiver, and condition (1) is met' appears to refer to condition (2); please correct the cross-reference.","section":"Proposition 5.1, correctness proof"},{"comment":"The reference to 'Definition 2.1' for the preorder ≼ should be to Definition 2.14.","section":"Proposition 5.1, invariant condition (3) proof"},{"comment":"The set T^{ℓ+1}_0 is used at ℓ=maxℓ, where it is not defined; clarify that it is empty in that case or adjust the loop bounds.","section":"Algorithm 3, lines 14–16"}],"recommendation":"major_revision","confidential_remarks":"The paper is well within the scope of the journal and the central claims are very likely correct. The two issues identified—the invalid quantifier step in Proposition 4.1(3)⇒(1) and the compressed finiteness proof in Corollary C—are repairable and do not suggest deeper flaws. I recommend major revision primarily to force the repair of the keystone implication, which is used throughout Section 4."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a real result, and the main theorems are correct. The paper shows that for a rooted tree quiver Q, the additive closure of the essential image of zero-dimensional persistent homology is of finite type, with indecomposables exactly the reduced rooted tree modules (Corollary C), and it gives a quadratic-time decomposition algorithm. That is genuinely new in persistence theory, where non-linearly indexed posets are usually wild. The algebraic characterization in Theorem A and the decomposition result in Theorem B are proved from scratch, not assumed; Kinser's work appears as definitional background and comparison, which is appropriate.\n\nWhat the paper does well: the categorical setup is clean, the linearization functor is described inductively, and the elder rule (Prop 4.3) is proven with an explicit isomorphism. The algorithms come with correctness invariants and complexity bounds that look right. The merge-tree and restriction applications in Section 6 are sensible and modest.\n\nSoft spots, in order of seriousness. First, the proof of (3)⇒(1) in Proposition 4.1 is not written correctly. The printed argument says that a nonzero root morphism gives, for each branch i of Q, some source branch j and target n with a nonzero root entry. But Definition 2.14 requires this for every source branch j. The missing step is compatibility along the arrow from the root of Q_i to the root of Q: the naturality square forces each row (or column, depending on convention) of the root block to sum to the nonzero global scalar, so every source branch has at least one nonzero entry. This is fixable, and the claim is true, but the proof as printed is incomplete. Because Prop 4.1 feeds into the elder rule and Theorem 4.6, it should be corrected.\n\nSecond, the finiteness argument in Corollary C is compressed into one sentence. The induction is routine and the claim is true, but given that finite type is the paper's headline conclusion, the referee should ask for the induction to be written out.\n\nThird, Lemma 3.5 (the equivalence between set-valued functors and disjoint unions of rooted tree quivers over Q) is stated without proof. It is straightforward, but it supports Theorem A(1); a sketch would help. There are also a few minor typos.\n\nWho should read it: persistence theorists, especially people working with merge trees or hierarchical clustering, and quiver representation people interested in tree posets. It deserves a serious referee and, after minor revision, acceptance. I would bring it to reading group.","headline":"Genuine new finite-type result for zero-dimensional persistence over rooted tree posets, with a quadratic decomposition algorithm; the main claims hold, but two proof details need expansion.","tokens_in":23856,"tokens_out":3754,"would_cite":true,"duration_ms":31550,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["16G20","55N31"],"pacs":[],"model":"deepseek-v4-flash","headline":"Tree-indexed zero-dimensional persistence is classifiable","keywords":["zero-dimensional persistent homology","rooted tree quivers","tree modules","finite representation type","elder rule","merge trees","quiver representations","persistence modules"],"falsifier":"Exhibit a finite rooted tree quiver $Q$ and a rooted tree module over $Q$ that is not isomorphic to a direct sum of reduced rooted tree modules, or an infinite family of pairwise non-isomorphic reduced rooted tree quivers over $Q$; either would directly contradict Corollary C.","tokens_in":22750,"feed_emoji":"🌲","tokens_out":6484,"duration_ms":53196,"temperature":0.7,"pith_summary":"Persistent homology is usually tractable because the indexing poset is a line, whose representations are of finite type. This paper shows the same tractability survives when the poset is a rooted tree: the objects that arise as zero-dimensional persistent homology, once closed under direct sums and direct summands, form a category of finite type. The indecomposable objects are exactly the reduced rooted tree modules, and there is a quadratic-time algorithm that decomposes any such representation. That makes tree-indexed clusterings, merge-tree morphisms, and restrictions of multiparameter filtrations amenable to the same decompose-into-features strategy that powers one-dimensional persistence.","feed_headline":"Zero-dimensional persistence over trees is classifiable","feed_subtitle":"Its building blocks are the reduced rooted tree modules, and any such module decomposes in quadratic time.","key_machinery":"The load-bearing construction is the linearization of a rooted tree quiver over $Q$: a morphism $f: T \\to Q$ is turned into a representation $k_T$ by pushing forward the constant representation of $T$, so each vertex of $T$ contributes a basis vector at its image in $Q$. The decomposition is controlled by the elder rule (Proposition 4.3): if two branches glued above the same vertex are comparable in the preorder on rooted tree quivers over $Q$, the smaller branch splits off as a direct summand. Reduced rooted tree quivers are those whose branches form antichains in this preorder; equivalently (Proposition 4.2) they are the ones admitting only the identity endomorphism, and their linearizations are the indecomposables. A gluing operation $G$ assembles rooted tree quivers by adjoining a new root, and representations and morphisms glue the same way.","core_discovery":"Let $Q$ be a finite rooted tree quiver, the quiver analogue of a rooted tree poset. The paper establishes two characterisations. First, the representations obtainable as zero-dimensional persistent homology $\\mathrm{H}_0$ of a $Q$-indexed filtration are precisely the finite direct sums of linearized rooted tree quivers over $Q$ (Theorem A(1)). Second, the additive closure of this class is precisely the category of finite direct sums of rooted tree modules over $Q$ (Theorem A(2)). The main structural result (Theorem B) says every rooted tree module over a rooted tree quiver splits as a direct sum of reduced rooted tree modules, which are indecomposable; hence the additive closure is of finite type and its indecomposables are these reduced modules (Corollary C). The proof runs through an elder rule for the preorder on rooted tree quivers over $Q$, and the same rule yields algorithms that decompose a linearized tree in $O(|T|^2)$ time and the zero-dimensional persistent homology of a $Q$-filtered graph in $O(|G|^2)$ time (Theorem D).","pith_inferences":["Beyond the paper's statements, the multiset of reduced tree summands could serve as a feature vector for tree-indexed clusterings, since the decomposition is unique and computable in quadratic time.","The elder-rule mechanism suggests a template for other posets: whenever a preorder makes linearized branches form antichains after pruning, the same finite-type conclusion may hold.","Since the paper notes that higher-degree homology of a finite poset sees all representations, the finite-type phenomenon is specific to degree zero; this marks a boundary worth testing for other homology functors."],"forward_implications":["Every zero-dimensional persistent homology module indexed by a rooted tree poset has a unique decomposition into reduced rooted tree modules, so the multiset of summands is a well-defined statistic of the filtration.","The decomposition of a linearized rooted tree and of the $\\mathrm{H}_0$ of a filtered graph can be computed in quadratic time, making the classification usable in practice.","Each morphism between merge trees gives a representation of the target merge tree that decomposes by the same algorithm, providing an invariant of the morphism.","Restricting a multi-parameter filtration to any rooted tree subposet yields a $\\mathrm{H}_0$ module that can be fully decomposed, turning a generally wild problem into a tractable one on the restriction."],"supporting_citations":[{"why":"Defines reduced representations of rooted tree quivers and the rank-function theory the paper generalises; its Corollary 19 is reproved as the indecomposability criterion of Theorem 4.6.","marker":"[18]"},{"why":"Introduces tree modules and the coefficient quiver, and supplies the basis argument used in Lemma 2.24.","marker":"[23]"},{"why":"Originates the elder rule for zero-dimensional persistence, which Proposition 4.3 generalises to rooted tree quivers.","marker":"[13]"},{"why":"Motivates the study of the image of H0 by showing that the epimorphic subcategory for grids is still wild; Theorem A is a contrasting positive result.","marker":"[3]"},{"why":"Shows the additive closure of H0 can be of infinite type for the opposite of a rooted tree poset, providing the sharpness contrast for Corollary C.","marker":"[5]"},{"why":"Supplies the gluing construction for representations used in Definition 2.18 and Lemma 2.19.","marker":"[24]"}],"fun_headline_variants":["Finite type classification for 0D persistence on rooted trees","Quadratic-time decomposition for tree-indexed H0","Reduced tree modules classify H0 persistence on trees","Tree posets yield finite type 0D persistence","Zero-dim persistence over trees decomposes in quadratic time"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification rests on the claim that, for a fixed finite rooted tree quiver, there are only finitely many reduced rooted tree quivers over it up to isomorphism; the paper states this follows by induction but leaves the induction implicit.","fun_headline_variants_meta":{"raw":{"variants":["Finite type classification for 0D persistence on rooted trees","Quadratic-time decomposition for tree-indexed H0","Reduced tree modules classify H0 persistence on trees","Tree posets yield finite type 0D persistence","Zero-dim persistence over trees decomposes in quadratic time"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000792,"raw_usage":{"total_tokens":3488,"prompt_tokens":941,"completion_tokens":2547,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":557,"completion_tokens_details":{"reasoning_tokens":2469}},"tokens_in":557,"tokens_out":2547,"duration_ms":16224,"temperature":1.0,"reasoning_tokens":2469,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:20:36.200328+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a finite rooted tree quiver $Q$ and a rooted tree module over $Q$ that is not isomorphic to a direct sum of reduced rooted tree modules, or an infinite family of pairwise non-isomorphic reduced rooted tree quivers over $Q$; either would directly contradict Corollary C.","supporting_citations":[{"cited_title":"Rank functions on rooted tree quivers","cited_arxiv_id":null,"evidence_quote":"Defines reduced representations of rooted tree quivers and the rank-function theory the paper generalises; its Corollary 19 is reproved as the indecomposability criterion of Theorem 4.6."},{"cited_title":"Exceptional modules are tree mod ules","cited_arxiv_id":null,"evidence_quote":"Introduces tree modules and the coefficient quiver, and supplies the basis argument used in Lemma 2.24."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Originates the elder rule for zero-dimensional persistence, which Proposition 4.3 generalises to rooted tree quivers."},{"cited_title":"Botnan, Steffen Oppermann, and Jo han Steen","cited_arxiv_id":null,"evidence_quote":"Motivates the study of the image of H0 by showing that the epimorphic subcategory for grids is still wild; Theorem A is a contrasting positive result."},{"cited_title":"O n the complexity of zero-dimensional multi- parameter persistence, 2020","cited_arxiv_id":null,"evidence_quote":"Shows the additive closure of H0 can be of infinite type for the opposite of a rooted tree poset, providing the sharpness contrast for Corollary C."},{"cited_title":"Distinguished bases of exceptio nal modules","cited_arxiv_id":null,"evidence_quote":"Supplies the gluing construction for representations used in Definition 2.18 and Lemma 2.19."}],"review_version":1}