{"id":"9835c7ce-bd69-4bac-91ba-0fd66a9b3775","arxiv_id":"2508.21276","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every finite entropy sum in a local QFT is a linear combination of complement entropy differences, mutual informations of non-adjacent regions, and tripartite informations.","lead":"This paper classifies all linear combinations of entanglement entropies whose ultraviolet divergences cancel, showing they reduce to three basic information-theoretic quantities. The classification uses Fourier analysis on the Boolean cube and may give a systematic way to define finite entanglement data in quantum field theory.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Classification's physical applicability rests on an unproven locality assumption for divergences (Eqs. (2),(13)); if divergences have nonlocal or configuration-dependent contributions, finite entropy sums need not match T_G.","rationale":"The reader's weakest assumption—that divergences are local functionals of boundary and corner geometry—is the same load-bearing concern I identify. The mathematical core of the paper, Theorem 3.7 and Theorem 4.3, appears correct and is honestly presented as a statement about functions on a Boolean cube, not as a directly proven QFT theorem. The physical input is the assumption that the cancellation conditions (2) and (13) faithfully encode divergence cancellation. The author explicitly disclaims proving this, so the concern is acknowledged in the text rather than hidden. The internal issues noted by the reader (Proposition 3.12 for n=2, sign error in Lemma 3.4) are real but peripheral: they affect an explicit basis construction and a lemma proof, not the statement or proof of the main classification theorems. A concrete numerical test in a tractable QFT can probe whether the locality assumption holds at least in one nontrivial setting; this is the most direct way to see whether the concern lands. Since the reader already assigned CONDITIONAL and my read does not suggest a different verdict, I recommend UNCHANGED.","tokens_in":17081,"tokens_out":10447,"duration_ms":107551,"concrete_test":"In a 2D free-scalar lattice, place four regions so that three meet at a single point and the fourth is disjoint. Compute the regulated entanglement entropies of all unions on L×L lattices, extract the L→∞ divergent parts, and test the paper's predictions: (i) I(A,B,C) is finite for a triple with no fourth region at the intersection; (ii) I(A,B) diverges if A and B touch only at a point. If (i) diverges, Eq. (13) is not sufficient; if (ii) is finite, it is not necessary. Failing either would undermine the physical applicability of the classification; passing both would support the locality assumption in this model.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 3.7/4.3 is a clean combinatorial statement about ker(E) and its hypergraph analogue, and the proof appears sound aside from fixable issues (Proposition 3.12 fails for n=2; Lemma 3.4 has a sign error). The truly load-bearing step is identifying 'all divergences cancel' with the per-boundary and per-corner sum rules (2) and (13). This identification assumes (a) divergences are local functionals of boundary geometry, so cancellation can be checked component-by-component, and (b) for each corner, the divergence of a region equals that of its complementary corner. The author explicitly disclaims proving necessity or sufficiency for any particular QFT: 'we make no attempt to prove that the conditions we start with are necessary and/or sufficient for divergence cancellation in any particular quantum field theory.' Should either assumption fail in a real QFT—e.g., nonlocal UV divergences or corner coefficients sensitive to global shape—the space of genuinely finite entropy sums would differ from T_G/T_H, and the physical claim in the abstract would not follow from the theorems.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies linear combinations S = Σ c_σ S(A_σ) of entanglement entropies for the unions of elementary regions in a spatial partition. It encodes the adjacency structure of the regions in a graph G, represents a coefficient vector by a function T on subsets of vertices, and characterizes the subspace TG of coefficients for which all boundary divergences cancel (condition (2)). The main theorems are: Theorem 3.3, an orthogonal Fourier basis for TG; Theorem 3.7, stating that TG is spanned by T0 (empty-set coefficient), complement-entropy differences S(A)−S(Ac), mutual informations of non-adjacent pairs, and tripartite informations of disjoint triples; and Theorem 4.3, the hypergraph generalization including higher-codimension corner divergences via conditions (13). Proposition 4.8 and Proposition 4.12 describe which information functions survive the stronger conditions and show that, in general, they do not span TH. The author explicitly disclaims proving necessity or sufficiency of the divergence-cancellation conditions for any particular QFT, and includes a short section on algebraic definitions and a note on AI use.","tokens_in":17348,"tokens_out":23291,"duration_ms":235562,"significance":"The mathematical core is clean and, if correct, gives a complete combinatorial classification of coefficient spaces for finite entropy sums. The Fourier/Möbius methods on the Boolean cube are natural for this problem; Theorem 3.7 is an appealing structural result that goes beyond the standard examples. The paper is transparent about its assumptions: the classification is relative to locality and complementarity conditions in Eqs. (2) and (13), and the author reports computer checks for all graphs up to seven vertices, which supports the conjectured form of Theorem 3.7. No parameters are fitted; the proofs are self-contained. The main value is the rigorous characterization of the coefficient-space problem; the physical interpretation is conditional on widely believed but unproved properties of QFT divergences.","major_comments":[{"comment":"The proposed explicit basis is false for n=2. If G has the single edge {1,2}, the listed set {T0, T1, T2} has T2 = −T1, so it spans only a 2-dimensional space, while dim TG = 3; [12]−[∅] is in TG but not in the span. If G is edgeless, the list {T0,T1,T2,T12} again has rank 3 because T1 and T2 are dependent, while dim TG = 4. The counting in the proof double-counts the dependent type-2 functions and omits T_V = [V]−[∅]. A separate n=2 case, or replacing one T_i by T_V, is needed.","section":"§3.2 (Prop. 3.12)"},{"comment":"The displayed equality has the wrong sign. From the definitions, (ĉ_e, T) = (1/2)(χ_e − χ_∅, T) = 2^{n−1}(T̂(e) − T̂(∅)), while (ET)_e = 2^{n−1}(T̂(∅) − T̂(e)). Thus (ET)_e = −(ĉ_e, T). The kernel conclusion in Theorem 3.3 is unaffected because only the vanishing condition matters, but the lemma as stated is incorrect and should be fixed.","section":"§3.1 (Lemma 3.4)"},{"comment":"The proof that (g_{a,b}, χ_σ) is nonzero only for σ ⊂ I is incomplete. The sentence about cancellation for i ∈ σ does not address the case i ∉ σ, and the conclusion does not follow from the pairwise cancellation argument alone. The intended statement is correct: writing τ = (a∪φ) or (b∪φ) with φ ⊆ V\\I gives a factorization with the factor Σ_{φ⊆V\\I} (−1)^{x_{σ_c}·x_φ}, which vanishes unless σ contains no vertex outside I. Please replace the argument with this explicit factorization.","section":"§4.1 (Lemma 4.2)"},{"comment":"The physical framing overstates the reach of the theorems. The paper proves statements about coefficient spaces satisfying the postulated cancellation conditions (2) and (13), and the author correctly disclaims proving necessity or sufficiency for any particular QFT. However, the abstract's opening 'we show that all such quantities...' is unconditional in tone. If actual QFT divergences have nonlocal pieces, or if corner divergences depend on more than the local corner and its complement, the true finite-sum space would differ from TG/TH. I recommend stating the assumption explicitly in the abstract and in Section 1 as a hypothesis of the classification, not merely as a caveat in the introduction.","section":"§1 / Abstract"}],"minor_comments":[{"comment":"The abstract contains the typo 'collections of for these regions'.","section":"Abstract"},{"comment":"The dimension count is stated inconsistently: the abstract and bullet list say 2^n − B − 1, while Theorem 3.3 and the proof strategy note that the space TG, including the empty-set coefficient, has dimension 2^n − |E|. Please clarify which space is being counted.","section":"Intro / §3.1"},{"comment":"In the sentence 'this is a set of 2^{|σ|−1}−1 equations for 2^{|σ|−1} unknowns', the symbol σ is reused for the intersection I. Use |I| for clarity.","section":"§4.1 (Lemma 4.2)"},{"comment":"In the orthogonality proof, the factor is written as a sum over (x_α)_i = ±1, but the characteristic vectors take values 0 and 1. The argument is correct with values 0 and 1; the notation should be adjusted.","section":"§3.1 (Prop. 3.2)"}],"recommendation":"major_revision","confidential_remarks":"The central combinatorial theorems appear correct and the paper is honest about its assumptions and AI use. The false n=2 case, the sign error, and the proof gap in Lemma 4.2 are all fixable, so I do not recommend rejection. The main concern to communicate to the authors is to make the conditional nature of the physical claim prominent, since the title and abstract otherwise suggest a stronger QFT statement than the theorems establish."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The useful core here is the linear-algebraic classification of entropy sums whose divergences cancel, given a graph or hypergraph of region adjacencies. Theorems 3.3 and 3.7 give clean Fourier and \"information\" bases for the space T_G, and the dimension count 2^n - |E| - 1 (or +1 depending on the empty-set convention) matches the cut-map kernel. The Möbius argument for Theorem 3.7 checks out; it's a genuinely nice decomposition into complement entropy differences, mutual informations, and tripartite informations. Theorem 4.3 extends the same logic to higher-order intersections with an explicit Fourier basis. The author is careful to state that the results are combinatorial, not QFT theorems, and disclaims proving necessity or sufficiency of the cancellation conditions for any particular theory. That honesty is welcome, because the entire physical payoff hangs on the locality of divergences encoded in Eqs. (2) and (13). If a real QFT had nonlocal UV divergences, the classification would not capture the genuinely finite sums. But that is not a flaw in the math; it is an explicitly flagged assumption.\n\nSoft spots are minor and fixable. Lemma 3.4 has a sign error in the displayed equality: the left side is 2^{n-1}(\\hat T(∅)-\\hat T(e)), while the inner product with \\hat χ_e is the negative of that. The kernel characterization is unaffected. Proposition 3.12's proposed basis degenerates for n=2: T_i and T_j are dependent when the universe has two elements, so the stated set is not independent. Theorem 3.7 survives, but the basis enumeration needs a small-n caveat. I would also want the author to double-check the wording of Proposition 4.6; the two cases appear conflated in the statement, though the proof is clear enough.\n\nWho is this for? Anyone defining or using finite entropy combinations — topological entanglement entropy, holographic mutual information, lattice corner-divergence cancellations — will find the basis results useful. It is a solid, readable contribution to the combinatorics of entanglement entropy. It does not resolve a deep physical puzzle, but it organizes a known space of quantities and gives a rigorous description. I would send it to peer review; a good referee can catch the small errors and the author will fix them. I would also bring it to a reading group if the group cares about entanglement entropy structure.","headline":"Clean combinatorial classification of finite entropy sums via cut conditions; the physical payoff rests on an explicitly flagged locality assumption, and the small proof blemishes are fixable.","tokens_in":17791,"tokens_out":4936,"would_cite":true,"duration_ms":45003,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper classifies every divergence-free entropy sum in local QFT as a combination of three information-theoretic primitives and provides explicit bases and dimension formulas.","keywords":["entanglement entropy","divergence cancellation","mutual information","tripartite information","Boolean cube","Mobius inversion","hypergraph","quantum field theory"],"falsifier":"Enumerate all graphs on up to seven vertices and check by exact linear algebra whether the four function classes in Definition 3.6 span the kernel of the cut map E; an independent reproduction confirming the author's reported check would support Theorem 3.7, while any counterexample would refute it. On the physical side, compute the regulated entanglement entropy of two adjacent regions sharing a non-cuspy corner in a lattice free scalar and take the continuum limit: if I(A,B) remains finite, then the higher-codimension cancellation condition (13) is not necessary for finiteness in that theory","tokens_in":1969,"feed_emoji":"🧩","tokens_out":2418,"duration_ms":113558,"temperature":0.7,"pith_summary":"The paper establishes a complete structural answer to a practical question in quantum field theory: which linear combinations of subsystem entropies avoid the ultraviolet divergences that plague individual entanglement entropies? It shows that, assuming divergences are local functionals of boundaries and corners, every such combination is built from just three information-theoretic primitives—complement entropy differences, mutual informations of non-adjacent regions, and tripartite informations of disjoint triples—plus a harmless empty-set constant. The proof translates divergence-cancellation conditions into graph-theoretic statements about functions on subsets, then uses Fourier transforms on the Boolean cube and Mobius inversion to find explicit bases and dimension formulas. The upshot is a complete classification: for n regions with a given adjacency graph, the finite sums form a vector space of dimension 2^n - B - 1, and with corner divergences included the dimension is 2^n - |IE| - 1. If the underlying locality assumption holds in a given QFT, then any finite entropy observable is a combination of these known primitives.","feed_headline":"Every finite entropy sum is built from three primitives","feed_subtitle":"A complete basis: complement-entropy differences, mutual information of separated regions, and tripartite information.","key_machinery":"The load-bearing object is the cut map E, which sends a function on subsets of vertices to the net coefficient of each boundary component in the corresponding entropy sum; the space of divergence-free sums is exactly ker E. The key identity expresses each edge component of E(T) as 2^{n-1}(T_hat(empty) - T_hat(edge)), where T_hat are Fourier coefficients on the Boolean cube (real-valued functions on subsets of a finite set). This converts divergence cancellation into equality of Fourier coefficients and yields explicit orthogonal bases. For the information-theoretic basis, the Mobius transform on the partially ordered set of subsets is used to prove that the span of the four building blocks i","core_discovery":"The paper proves that the space of entropy sums in which boundary divergences cancel is generated by exactly four building blocks: the function assigning a constant to the empty set, the complement-entropy difference S(A)-S(A^c), the mutual information I(A,B) for non-adjacent regions, and the tripartite information I(A,B,C) for disjoint triples (Theorem 3.7). For a spatial slice divided into n regions, the codimension-one-safe sums form a vector space of dimension 2^n - B - 1, where B is the number of adjacent pairs. When higher-codimension corner divergences are included, a hypergraph version of the same Fourier argument gives an explicit orthogonal basis indexed by even-order subsets of in","pith_inferences":["Editorial extension: if the locality assumption survives lattice tests, the classification gives a practical dictionary—any candidate finite entanglement measure in a fixed spatial partition should be checked against this basis before being treated as genuinely new.","Editorial extension: because the proof is purely combinatorial, finiteness of an entropy sum can be certified by evaluating Fourier coefficients, an O(2^n) calculation on the adjacency graph that requires no regulator-dependent computation.","Editorial extension: the n-qubit ground-state reformulation suggests that divergence-free entropy sums form the ground-state space of a frustration-free local Hamiltonian; whether these states have simple tensor-network representations is a natural next step not explored in the paper.","Editorial extension: for holographic states, combining this basis with known inequalities such as monogamy of mutual information could constrain the full space of finite entropy data; proving analogous bounds on arbitrary basis elements would be a testable extension of the paper's claims."],"forward_implications":["Any finite entropy sum for a fixed n-region partition is uniquely expressible through complement entropy differences, mutual informations between non-adjacent regions, and tripartite informations of disjoint triples, so no other independent finite combination types exist.","The dimension formula 2^n - B - 1 for codimension-one-safe sums and 2^n - |IE| - 1 for fully safe sums turns the search for finite combinations into finite linear algebra determined entirely by the adjacency graph or hypergraph.","Mutual information is generically finite only when the two regions share no boundary intersection at any codimension, and tripartite information is generically finite only when the three regions do not meet together with a fourth region, except in special cuspy-corner cases.","The same characterization is equivalent to describing the ground-state space of a certain n-qubit Hamiltonian, giving a quantum-information reformulation of divergence-free entropy sums.","Several finite entropy combinations can be defined algebraically through relative entropies and the split property, without ever introducing a regulator; a fully general algebraic definition of tripartite information remains open."],"supporting_citations":[{"why":"Establishes the physical premise that subsystem entropies for regions with boundaries are UV divergent, motivating the cancellation conditions.","marker":"[2–4]"},{"why":"Documents corner and higher-codimension entanglement divergences that the hypergraph constraints in Eq. (13) are designed to cancel.","marker":"[8–19]"},{"why":"Shows that sufficiently cuspy corners may lack the divergence, delimiting when a combination outside T_H can still be finite.","marker":"[20]"},{"why":"Supplies the Boolean-cube Fourier basis used to compute ker E and prove the orthogonal-basis Theorem 3.3.","marker":"[26]"},{"why":"Supplies the Mobius transform and inversion used to prove that the four information-theoretic function classes span T_G (Theorem 3.7).","marker":"[27, 28]"},{"why":"Gives the topological-entanglement-entropy context in which tripartite information appears as a physically important finite combination.","marker":"[21, 22]"},{"why":"Shows holographic mutual information is monogamous, a physical property of tripartite information used as motivation for treating it as a basic quantity.","marker":"[23]"}],"fun_headline_variants":["Finite entropy sums: just three primitive building blocks","All finite entropy combos come from three primitives","Three primitives generate every finite entropy sum in QFT","Entropy divergences cancel: three primitives suffice","QFT entropy sums that stay finite reduce to three types"],"cache_read_input_tokens":19584,"weakest_assumption_plain":"The classification assumes that UV divergences in regulated entanglement entropies are local, state-independent functions of boundary geometry and intersections, so that checking cancellation boundary component by boundary component and corner by corner is both necessary and sufficient; the paper explicitly does not prove this locality condition for any specific quantum field theory.","fun_headline_variants_meta":{"raw":{"variants":["Finite entropy sums: just three primitive building blocks","All finite entropy combos come from three primitives","Three primitives generate every finite entropy sum in QFT","Entropy divergences cancel: three primitives suffice","QFT entropy sums that stay finite reduce to three types"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000246,"raw_usage":{"total_tokens":1375,"prompt_tokens":741,"completion_tokens":634,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":485,"completion_tokens_details":{"reasoning_tokens":556}},"tokens_in":485,"tokens_out":634,"duration_ms":7145,"temperature":1.0,"reasoning_tokens":556,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T14:27:37.969991+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Enumerate all graphs on up to seven vertices and check by exact linear algebra whether the four function classes in Definition 3.6 span the kernel of the cut map E; an independent reproduction confirming the author's reported check would support Theorem 3.7, while any counterexample would refute it. On the physical side, compute the regulated entanglement entropy of two adjacent regions sharing a non-cuspy corner in a lattice free scalar and take the continuum limit: if I(A,B) remains finite, then the higher-codimension cancellation condition (13) is not necessary for finiteness in that theory","supporting_citations":[{"cited_title":"Generalizing the entanglement entropy of singular regions in conformal field theories","cited_arxiv_id":"1904.11495","evidence_quote":"Shows that sufficiently cuspy corners may lack the divergence, delimiting when a combination outside T_H can still be finite."},{"cited_title":"O’Donnell, Analysis of Boolean Functions","cited_arxiv_id":null,"evidence_quote":"Supplies the Boolean-cube Fourier basis used to compute ker E and prove the orthogonal-basis Theorem 3.3."},{"cited_title":"Hayden, M","cited_arxiv_id":null,"evidence_quote":"Shows holographic mutual information is monogamous, a physical property of tripartite information used as motivation for treating it as a basic quantity."}],"review_version":1}