{"id":"1c91b079-0112-4726-9ef6-54e782dec478","arxiv_id":"2412.10223","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":1,"one_line_summary":"A purported proof that some sparse diagonal Hamiltonians cannot be locally represented by eigenbasis permutation, undermined by an invalid lemma and unsubstantiated claims.","lead":"This paper claims to prove that certain sparse, non-local quantum Hamiltonians cannot always be rewritten as local ones by permuting their eigenvalues. The proof is presented as a combinatorial impossibility result, but key steps are flawed and the accompanying astrophysical speculation is unsupported.","discovery_kind":"unclear","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma A.12 is false: the coefficient of g in (a+bg)c is the identity coefficient of bc, not 'bc', and an explicit image counterexample kills the node bound. The refutation of Conjecture A.1 is unsupported.","rationale":"The paper's central claim is that for every constant locality m there exist sparse non-local diagonal forms with no m-local eigenbasis-permuted representative. The proof route is: Lemma A.9 integrality, Lemma A.10 per-image support bound A_m, Lemma A.11 per-image node bound B_m, Lemma A.13 collective node bound D_m via repeated Lemma A.12, Lemma A.14 count of possible images G_m, and finally Theorem A.15 via injectivity of Psi. Lemma A.12 is the only step that transfers nodes from individual images to a product image; without it the argument that a subgroup S of size exceeding G_m forces a repeated image has no counting bound. The counterexample above is not at the edge: it satisfies exactly the hypotheses of Lemma A.12 and contradicts its conclusion on a 2-qubit group ring. Therefore the proof of D_m, and hence Theorem A.15, is not merely incomplete but wrong. The reader's verdict of REJECT with high correctness risk is appropriate; our independent reading identifies the same load-bearing assumption. Because Lemma A.12 is stated as a displayed claim and used verbatim in Lemma A.13, this is an internal mathematical inconsistency, not a disagreement with external consensus. The probabilistic transition of Eq. 24 and the astrophysical discussion are secondary; the combinatorial refutation already fails at its core.","tokens_in":11110,"tokens_out":13436,"duration_ms":154412,"concrete_test":"Verify Lemma A.12 against the 2-qubit counterexample: set G=Z_2^2, g=(1,0), a=0, b=(0,1), c=1, pi=id, J=(1,1). Compute (a+bg)c = z and check its x-coefficient; it is 0 while all lemma hypotheses hold. Then attempt to re-derive Lemma A.13 with any corrected version of Lemma A.12; absent such a derivation, the D_m bound cannot be used. Optionally enumerate all pi,J for n<=4 to confirm counterexamples are not isolated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma A.12 is invalid. Expanding (a+bg)c gives ac + bgc. Since a and c have no g-terms, ac has no g-term, and the coefficient of g in bgc is the coefficient of the identity in bc, not 'bc' as the proof states. Nor does nonzeroness of bc force that identity coefficient to be nonzero. In C[Z_2^2], take g=x, a=0, b=y, c=1. Then a,b,c have no x-terms, b is nonzero, c=Psi(0) is an image, and a+bg = yx = z = Psi(z) for pi=id, so a+bg is also an image. Yet (a+bg)c = z has x-coefficient 0, directly contradicting the lemma's conclusion that g is a node. Lemma A.13 applies Lemma A.12 repeatedly to conclude each 'blue' node lies in nodes(Psi(J)); with Lemma A.12 false, the bound |nodes(S)| <= B_m^3 is unsupported. Since Lemma A.14 and the final contradiction in Theorem A.15 depend on D_m, the central claim R \\ R'' nonempty is not established. This is an internal coefficient-computation error, not a matter of interpretation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to refute a 'Quasiparticle Locality Conjecture' (Conjecture A.1) by showing that for each constant locality m there exist n-qubit diagonal forms with a constant number of Pauli terms that cannot be mapped to an m-local diagonal form by any eigenbasis permutation. The proof introduces an injective homomorphism Ψ from Z_2^n into a group ring, bounds the number of localized images of subgroups via a sequence of constants A_m, B_m, D_m, E_m, G_m, and derives a contradiction with injectivity. The paper also discusses a speculative probabilistic transition for random sparse diagonal forms and connects the bound G_m to black-hole entropies.","tokens_in":11406,"tokens_out":32638,"duration_ms":357018,"significance":"If correct, the refutation of Conjecture A.1 would be a noteworthy negative result for the program of variational Hamiltonian diagonalization, and the group-ring counting approach is a reasonable strategy. The paper is explicit about the structure of the attempted proof and makes a concrete falsifiable claim (R \\ R'' non-empty). However, the significance is undercut by the fact that the proof as written contains a false claim in a central lemma and a key lemma is only sketched informally.","major_comments":[{"comment":"The proof of Lemma A.12 is incorrect. Expanding (a+bg)c gives ac + (bc)g, since g commutes with the group-ring elements; the coefficient of the group element g in (a+bg)c is the coefficient of the identity in bc, not 'bc' as claimed. Nonzeroness of bc does not imply that this coefficient is nonzero; for example, with n=2, g=x, a=0, b=y, c=1, the product is xy, whose coefficient of x is 0 even though b, c, and a+bg satisfy the stated hypotheses. The lemma's conclusion (g is a node) happens to hold in that example because xy uses node x, but the proof's inference is invalid. The argument can be repaired by observing that bc ≠ 0 implies (bc)g has nonzero support on elements with the g-bit set, while ac has support on elements with the g-bit clear, so g is a node. As written, however, the proof of this load-bearing lemma is not valid, and Lemma A.13 relies on it repeatedly.","section":"Appendix B.4, Lemma A.12"},{"comment":"The proof of Lemma A.13 is a sketch rather than a rigorous proof. It depends on an informal red/blue node construction and an accompanying figure, with undefined operations such as 'permute columns', 'copied in', and 'we lose some potential red nodes'. There is no formal definition of the construction and no detailed proof that the product Ψ(J_{i1})···Ψ(J_{il}) contains all blue nodes. Since Lemma A.13 supplies the bound D_m used in Lemma A.14 and in the final contradiction, the main theorem requires a complete and rigorous proof of this statement.","section":"Appendix B.4, Lemma A.13"},{"comment":"The formula for G_m stated in Eq. (22) (and repeated in Appendix E) does not follow from the preceding bounds. Using A_m = 2^{2m-2}, B_m = m A_m, D_m = B_m^3, and E_m = D_m^m, together with |A_{1/2^{m-1}}| = 2^m+1, one obtains G_m = (2^m+1)^{E_m} = (2^m+1)^{m^{3m} 64^{m^2-m}}. The printed expression (2m+1)^{m(3m)^{64(m^2-m)}} has a different base and a different exponent. This discrepancy must be resolved, because the stated size of G_m is used in the definition of the set R and in the quantitative claims of the paper.","section":"Lemma A.14 and Eq. (22)"}],"minor_comments":[{"comment":"The sentence 'Our hypothesize suggests a sharp transition' is ungrammatical and should read 'Our hypothesis suggests a sharp transition'.","section":"Abstract"},{"comment":"Equation (24) is introduced as a hypothesis with no derivation or numerical evidence; the text should explicitly state that this is speculative and separate from the mathematical proof, rather than treating the transition as an established finding.","section":"Section III, Eq. (24)"},{"comment":"The final line of the proof contains a typo: 'Thus bc ̸= 0n a g is a node' should read 'Thus bc ≠ 0 and g is a node.'","section":"Appendix B.4, Lemma A.12 proof"},{"comment":"The condition NNZPSW(D) ≥ 2^{ceil(log2(Gm))} is not motivated; if the intent is to consider forms with at least G_m terms, this should be written directly as NNZPSW(D) ≥ G_m.","section":"Section II, definition of R"},{"comment":"The notation 2m+1 appears where 2^m+1 is evidently intended (e.g., Lemma A.14 and Eq. (22)); all superscripts and bases should be checked carefully in a revision.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The manuscript contains a potentially interesting negative result, but the proof is not in publishable form. The central issue is not the overall strategy but the rigor and correctness of the written proof: Lemma A.12's proof contains a false statement, Lemma A.13 is only sketched, and the final constant G_m is internally inconsistent. These are fixable in principle, so I do not recommend outright rejection, but the authors must supply a complete, correct proof before the paper can be considered further. The speculative astrophysical discussion should also be clearly separated from the mathematical content."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper takes a real question—can every sparse non-local diagonal Pauli operator be mapped to an m-local diagonal form by permuting eigenvalues?—and attacks it with a combinatorial counting argument in a group ring. The question is natural for variational Hamiltonian diagonalization, and the encoding of permutations via the injective map Psi is a nice idea. The early bounds in the appendix (A_m, B_m) are straightforward and fine.\n\nThe central proof does not survive contact with the algebra. Lemma A.12 asserts that the coefficient of a density-one element g in (a+bg)c is \"just bc.\" That is wrong: the expansion is ac + bgc, so the coefficient of g is the coefficient of the identity in bc, not bc itself. The counterexample in the stress-test note is decisive: in C[Z_2^2], take a=0, b=y, c=1, g=x; then a+bg = yx = z = Psi(z) and c=Psi(0), yet (a+bg)c = z has x-coefficient zero. Lemma A.12 is false, and Lemma A.13 leans on it directly. Without the D_m bound, the contradiction in Theorem A.15 evaporates. The claimed refutation of Conjecture A.1 is not established.\n\nOther soft spots: the conjecture being refuted is coined in this paper, so the \"refutation\" is of a self-created target; the question is still interesting, but the headline is weaker than it looks. The probabilistic transition in Eq. 24 is a bare hypothesis with an undefined parameter w, yet the abstract and conclusion treat it as a result. The astrophysical connection to neutron-star/black-hole entropy is speculation. The expression for G_m also appears inconsistent between Section II and Appendix E, though the typesetting makes it hard to tell.\n\nWho gets value from this? Researchers working on VHD or diagonal-form locality might find the question worth knowing, and the group-ring machinery is a legitimate tool to bring to it. But with a false lemma at the core, there is no theorem here. A serious referee would send it back for major revision; I would not spend referee time on it as is.\n\nRecommendation: desk-reject for now. If the authors can repair Lemma A.12 and re-derive the bound, the paper might become a useful contribution; until then, the central claim is unsupported.","headline":"Interesting question and a clever group-ring setup, but Lemma A.12 is plainly false, so the central no-go theorem is unsupported.","tokens_in":11880,"tokens_out":5271,"would_cite":false,"duration_ms":56533,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68","20C05"],"pacs":["03.67.-a"],"model":"deepseek-v4-flash","headline":"The paper proves that some sparse non-local diagonal Hamiltonians cannot be mapped to local diagonal form by any eigenbasis permutation.","keywords":["eigenbasis permutations","Pauli product diagonal operators","quantum Hamiltonian locality","group ring","Fourier transform","quasiparticle locality conjecture","Bekenstein-Hawking entropy","sparse diagonal forms"],"falsifier":"A concrete calculation of the group-ring product $(a+b g)c$ for explicit images of $\\Psi$ on a small bit-vector space would settle it: if the coefficient of $g$ includes a contribution from the $a$-term, then the node bound $D_m$ is unsupported and the contradiction in Theorem A.15 collapses.","tokens_in":10879,"feed_emoji":"⚛️","tokens_out":13901,"duration_ms":126873,"temperature":0.7,"pith_summary":"The paper asks whether every sparse diagonal Hamiltonian — one with only a few non-zero Pauli-product terms — can be transformed into a local diagonal form, where each term touches at most $m$ qubits, simply by permuting its eigenvalue ordering. It answers no: it proves that the \"Quasiparticle Locality Conjecture\" is false, by showing there exist sparse non-local diagonal forms that remain non-local under every possible permutation of their eigenvalues. The proof establishes an explicit lower bound $G_m$ on the number of non-zero terms such a localized form would need, then picks a set of eigenvalues larger than that bound; since the relevant map from bit-vectors to group-ring elements is injective, this forces a contradiction. Because $G_m$ grows enormously fast, the counterexamples only appear for systems with an astronomical number of qubits, and the paper discusses the resulting gap between theory and practical feasibility.","feed_headline":"Sparse non-local Hamiltonians cannot always be localized","feed_subtitle":"A proof shows eigenvalue shuffling can't turn every sparse diagonal quantum form into a local one.","key_machinery":"The central object is an injective group-ring homomorphism $\\Psi: \\mathbb{Z}_2^n \\to (\\mathbb{C}\\mathbb{Z}_2^n)^*$, defined for a permutation $\\pi$ by sending each bit-vector $J$ to the full Fourier-transform element $F_J = \\sum_g a^J_g g$, with $a^J_g = (1/2^n)\\sum_x (-1)^{x\\cdot g + \\pi(x)\\cdot J}$. This map converts the question of whether a set of bit-vectors can be localized into a counting question: each localized image has at most $A_m$ terms, at most $B_m$ nodes, and at most $D_m$ nodes collectively, and the number of distinct locality-$m$ group-ring elements is bounded by $E_m$, leading to the bound $G_m$. Because $\\Psi$ is injective, any subgroup of $\\mathbb{Z}_2^n$ larger than $G_m$ cannot be represented; choosing such a subgroup produces the counterexample.","core_discovery":"The central claim is a proof by contradiction: for any fixed locality $m$, there exists an $n$-qubit diagonal operator whose Pauli-string representation is sparse (a constant number of non-zero terms, independent of $n$) that cannot be written as an $m$-local diagonal operator after any eigenbasis permutation. The proof bounds the number of possible locality-$m$ group-ring elements by $G_m = (2m+1)^{m(3m)}64^{m^2-m}$. It then chooses a subgroup $S$ of the bit-vector space $\\mathbb{Z}_2^n$ whose size exceeds that bound; if every element of $S$ could be localized, the injective map $\\Psi$ would have to send more than $G_m$ distinct inputs into a set of at most $G_m$ outputs, which is impossible. The contradiction shows the set $R \\setminus R''$ is non-empty, where $R$ is the set of diagonal forms with at least $2^{\\lceil \\log_2 G_m \\rceil}$ non-zero terms and $R''$ is the set of forms that are eigenspectrum-equivalent to an $m$-local form. The proof is purely combinatorial and does not depend on the physical dynamics of the Hamiltonian.","pith_inferences":["If the theorem is right, then the notion of \"local form\" is not a harmless gauge choice for sparse diagonal operators: there are genuinely non-localizable spectra, so algorithms must either increase locality with $n$ or restrict to a subclass of sparse forms.","The probabilistic threshold hypothesis could be tested on small systems: sample random sparse diagonal forms with $k$ terms and measure the fraction that can be localized to $m$-local form; a sharp drop in that fraction at some $k$ would support the transition, though the astrophysical scale attached to $w$ is not derivable from the present proof.","The paper's black-hole entropy comparison is speculative and should not be read as a physical derivation; the coincidence that $G_m$ becomes comparable to stellar black-hole entropies between $m=7$ and $m=8$ follows from exponential growth and does not by itself implicate quantum gravity.","The injective-map counting strategy may generalize to other basis-change questions, such as whether a sparse operator can be made local by a Clifford conjugation, by replacing the group ring with a different algebraic model of the symmetry group."],"forward_implications":["For fixed locality $m$, the set of sparse non-local diagonal forms that can be mapped to $m$-local form is strictly smaller than the set of all sparse diagonal forms; the paper constructs a non-empty difference set $R \\setminus R''$.","Any quantum simulation approach that restricts the diagonal ansatz to a fixed locality cannot be universal for sparse diagonal Hamiltonians, even with unlimited classical freedom to permute eigenvalues.","The obstruction is spectral in nature: it survives any permutation of the eigenvalue list and therefore any relabeling of the computational basis.","Because the required system size $n$ is astronomically large for small $m$, the practical impact on near-term quantum devices is indirect; the theorem is a statement about the scaling limit."],"supporting_citations":[],"fun_headline_variants":["Sparse non-local Hamiltonians resist eigenbasis localization","Eigenbasis permutations fail to localize some sparse operators","Quantum Hamiltonian localization impossible for sparse cases","Refuting the Quasiparticle Locality Conjecture with a proof","Cosmological bound shows sparse Hamiltonians cannot be localized"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that in the group-ring product $(a+b g)c$, the coefficient of the single-node term $g$ is exactly the product $b c$, with no contribution from the product of the $a$ term with $c$.","fun_headline_variants_meta":{"raw":{"variants":["Sparse non-local Hamiltonians resist eigenbasis localization","Eigenbasis permutations fail to localize some sparse operators","Quantum Hamiltonian localization impossible for sparse cases","Refuting the Quasiparticle Locality Conjecture with a proof","Cosmological bound shows sparse Hamiltonians cannot be localized"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000839,"raw_usage":{"total_tokens":3687,"prompt_tokens":1008,"completion_tokens":2679,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":624,"completion_tokens_details":{"reasoning_tokens":2599}},"tokens_in":624,"tokens_out":2679,"duration_ms":22061,"temperature":1.0,"reasoning_tokens":2599,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T16:26:49.349662+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete calculation of the group-ring product $(a+b g)c$ for explicit images of $\\Psi$ on a small bit-vector space would settle it: if the coefficient of $g$ includes a contribution from the $a$-term, then the node bound $D_m$ is unsupported and the contradiction in Theorem A.15 collapses.","supporting_citations":[],"review_version":1}