{"id":"caf7b7b6-887e-4cc4-8e49-e54d8b71233a","arxiv_id":"2511.22593","paper_version":3,"verdict":"REJECT","confidence":"HIGH","novelty_score":3.0,"correctness_risk":"high","formal_verification":"none","parameter_count":1,"one_line_summary":"The proposed rank test does not determine the Lie algebra generated by a Pauli pool, so the claimed polynomial MCP construction is not established.","lead":"This paper claims a fast polynomial-time way to build the smallest sets of quantum-circuit building blocks (Lie-algebra generator pools) for molecular simulations, and uses those pools in VQE benchmarks. The core rank-based proof is missing and the central theorem is false: a simple two-qubit example shows different generator sets with the same rank generate different quantum algebras.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 19's rank criterion is false: identical Γ-rank and cardinality do not determine the generated Lie algebra, as shown by a rank-4 pair generating su(4) vs su(2)⊕su(2).","rationale":"The central mathematical contribution of the paper is Theorem 19, which states that a bracket-independent Pauli set is an MCP for a target Lie algebra iff its Γ-matrix has the same rank and cardinality as a reference MCP. This is the load-bearing step for the claimed O(N^3) verification algorithm and for the assertion that MCP construction is provable and polynomial. The proofs are not in the main text; the authors defer them to a supplementary file, so the statement itself must be tested directly. The reader's counterexample is correct: the two 2-qubit sets A and B have identical size and Γ-rank but generate different Lie algebras (su(4) versus su(2)⊕su(2)). Since the rank/size invariant does not determine the DLA, Theorem 19 is false as stated. This is an internal inconsistency, not a disagreement with an external consensus. Additionally, the paper itself flags that the target rank 2N−4 for the symmetry-preserving subalgebra is only numerically observed, further weakening the practical completeness guarantee. The numerical results for MB-ADAPT-VQE and NI-DUCC-VQE may retain heuristic value, but the foundational claim of a provable polynomial MCP construction does not survive. I agree with the reader's REJECT verdict, and no change to that verdict is needed.","tokens_in":15711,"tokens_out":12066,"duration_ms":97896,"concrete_test":"Run a small script that: (1) constructs Γ_A and Γ_B over F2 for A={X1,Z1,X1X2,Z2} and B={X1,Z1,X2,Z2}; (2) verifies both are symmetric, zero-diagonal, size 4, and have rank 4; (3) computes the real Lie algebra generated by each set by iterating commutators of Pauli strings until linear closure; (4) reports the dimension of each DLA. If dim(⟨A⟩_Lie)=15 and dim(⟨B⟩_Lie)=6, Theorem 19 is refuted. If the dimensions were equal, the rank criterion would survive this test, but the expected failure would confirm that Γ-rank alone cannot certify MCP status.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is Theorem 19: a bracket-independent Pauli set A is an MCP for a Lie algebra g iff rank(Γ_A) and |A| match a reference MCP. This requires that Γ-rank (and hence congruence class) determines the generated Lie algebra. The reader's counterexample is decisive: A={X1,Z1,X1X2,Z2} and B={X1,Z1,X2,Z2} both have size 4 and rank(Γ)=4 over F2, yet A generates su(4) (dim 15) while B generates su(2)⊕su(2) (dim 6). Both sets are bracket-independent and satisfy the theorem's hypotheses, but they cannot be MCPs for the same g. Therefore the rank-based criterion fails as a completeness check for a prescribed algebra. The main text defers all proofs to a supplementary file, and even the algorithm's termination condition rank=2N−4 is explicitly admitted to be a numerical observation, not a proven statement (Section II D). These are not peripheral gaps: the polynomial, provable MCP construction and the removal of the exponential bottleneck rest entirely on the false rank invariant.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper claims a polynomial-time criterion for verifying Minimal Complete Pools (MCPs) of Pauli-string generators for a prescribed Lie algebra. The central result, Theorem 19, states that a bracket-independent Pauli set A' is an MCP for the same Lie algebra g as a reference MCP A if and only if rank(Γ_A)=rank(Γ_A') and |A|=|A'|, where Γ is the anti-commutation adjacency matrix over F_2. The authors use this to construct symmetry-preserving pools for ADAPT-VQE and NI-DUCC-VQE, reporting numerical benchmarks on LiH, H6, H8, and H2O. The claimed contribution is to replace exponential MCP construction with O(N^3) rank evaluation.","tokens_in":16083,"tokens_out":24690,"duration_ms":204995,"significance":"If Theorem 19 were correct, the paper would provide a substantial advance: exponential verification of complete operator pools would be replaced by a polynomial rank computation, with direct consequences for ADAPT-VQE and fixed-ansatz methods. The numerical study is extensive, and the batched-ADAPT-VQE idea is reasonable. However, the central mathematical claim is false. The failure is not a minor technicality: it invalidates the rank-based completeness criterion, the termination condition of the construction algorithm, and the 'provable' status claimed for the method. The numerical benchmarks, while interesting, cannot substitute for a valid completeness guarantee.","major_comments":[{"comment":"The rank criterion is false. Let A={X1,Z1,X1X2,Z2} and B={X1,Z1,X2,Z2} on two qubits. Both sets are bracket-independent, have |A|=|B|=4, and a direct F2 computation gives rank(Γ_A)=rank(Γ_B)=4. Direct bracket closure of A yields the 10 Pauli strings {X1,Z1,X1X2,Z2,Y1,X1Y2,Y1X2,Z1X2,Y1Y2,Z1Y2}, which are closed under commutation, so DLA(A) is 10-dimensional; DLA(B)=su(2)⊕su(2) is 6-dimensional. Take g=DLA(A). Then A is an MCP for g: no 3-element Pauli set can generate a 10-dimensional algebra, since with 3 generators every nested commutator is a product of a subset of the three generators, giving at most 7 nonzero Pauli strings. B satisfies the rank/cardinality hypotheses of Theorem 19 but is not complete for g. Thus the 'if' direction of Theorem 19 is false.","section":"Section II C, Theorem 19"},{"comment":"The mechanism behind the false theorem is that arbitrary congruence is not a valid contraction sequence. Equation (7) with e_i+e_j describes a contraction only when P_i and P_j anti-commute; when they commute, the contraction produces the zero operator, not the Pauli product used in the formula. The cited F2 classification of symmetric zero-diagonal matrices is a classification of all congruence classes, but not every invertible congruence transformation can be decomposed into the elementary moves that correspond to allowed contractions. Since Γ_A and Γ_B in the counterexample are congruent (same size and rank) yet generate different DLAs, congruence of Γ does not preserve the generated Lie algebra. Therefore Theorem 18 and its corollary Theorem 19 cannot be correct as stated.","section":"Section II C, Propositions 13/15 and Theorem 18"},{"comment":"The algorithm's termination condition in the Methods section is rank(Γ_A)=2N−4, and Section II D explicitly states this is 'currently numerical, as a formal proof and detailed characterization of this Lie subalgebra are beyond the scope of this work.' Thus even if Theorems 18–19 were true, the algorithm's output would not be a proven MCP for the prescribed symmetry-preserving algebra; it would rely on an empirical rank target. Since Theorem 19 is false, the numerical convergence in Figures 2–4 cannot provide evidence of mathematically guaranteed completeness. The abstract's claim of 'mathematically verified generator pools' and Table I's 'provable' entry are therefore unsupported.","section":"Section IV, Operator pool construction; Section II D"}],"minor_comments":[{"comment":"The text states both that a minimum of 2N+1 elements is required to generate su(2^N) and that the theoretical minimum MCP size is 2n−2 operators. These statements are inconsistent; please clarify which quantity is intended and under which conditions.","section":"Introduction, paragraph 3"},{"comment":"Typos: 'bracket-independant' should be 'bracket-independent', and 'foward' should be 'forward'.","section":"Definition 16 and Remark 12"},{"comment":"The line 'rank of Γ_A ∈ R^{|A|×|A|}' should read Γ_A ∈ F_2^{|A|×|A|}; the matrix is over F_2, not R.","section":"Section IV, Operator pool construction"}],"recommendation":"reject","confidential_remarks":"The counterexample in Major Comment 1 is decisive and directly falsifies the paper's central theorem. No local revision can repair the rank-based completeness criterion. The numerical benchmarks are extensive, but they do not establish the advertised polynomial, provable MCP construction. If the authors can find additional invariants (e.g., connectedness or component structure of the anti-commutation graph) and prove a correct criterion, that would be a different submission."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline is that the main mathematical claim is wrong. Theorem 19 says a bracket-independent Pauli set is an MCP iff its Γ-rank and cardinality match a reference MCP. It isn't. Take A = {X1, Z1, X1X2, Z2} and B = {X1, Z1, X2, Z2}. Both have size 4, both are bracket-independent, and both Γ matrices have rank 4 over F2. But A generates su(4) while B generates su(2)⊕su(2). So rank plus cardinality cannot certify completeness for a prescribed algebra. That is not a peripheral flaw; it eliminates the claimed O(N^3) provable verification and the removal of the exponential bottleneck.\n\nWhat the paper does well: it targets a real problem. Constructing minimal complete pools for ADAPT-VQE is expensive, and a polynomial check would genuinely matter. The MB-ADAPT-VQE variant is a sensible combination of batched ADAPT with reduced pools, and the numerical benchmarks on molecules up to 26 qubits are substantial. The authors also honestly flag that the rank target 2N−4 is a numerical observation, not a proven statement, and they defer all proofs to a supplementary file that is not available. That honesty is appreciated, but it does not rescue the theory.\n\nThe soft spots are proportionate to the central failure. The theorem is not merely unproved; it is refuted by the simplest possible example. The numerical experiments therefore cannot be read as validating a completeness guarantee—they test a heuristic pool construction. The batching results may stand on their own as evidence that larger batches reduce function evaluations, but the paper's core claim of provable, polynomial MCP construction does not survive.\n\nWho is this for? Readers working on adaptive VQE and operator pool design might find the batched ADAPT results useful as a heuristic, and the counterexample is instructive. But the paper should not be cited as a proof that rank-based MCP verification works. It deserves a serious referee because the problem is important and the empirical work is substantial, but the theoretical core needs major repair: either a genuinely correct invariant or a reframing of the whole approach as heuristic.","headline":"The central rank-only MCP criterion is false: a two-qubit counterexample refutes Theorem 19, so the paper's polynomial completeness guarantee does not hold, though the batching heuristics may have some value.","tokens_in":16512,"tokens_out":3457,"would_cite":false,"duration_ms":31732,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["15A63","17B05","81P68"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that the completeness of a Pauli-operator pool for a Lie algebra can be certified in polynomial time by checking the rank of an anti-commutation matrix, replacing exponential searches for minimal generator sets.","keywords":["Lie algebras","Pauli strings","minimal complete pools","anti-commutation graphs","matrix congruence","ADAPT-VQE","variational quantum eigensolver","quantum chemistry"],"falsifier":"For N=2 qubits, compare the four-string pool {X1, Z1, X1X2, Z2} with {X1, Z1, X2, Z2}. Both have Γ-matrices of rank 4, but explicit DLA-basis closure computations show the first generates su(4) while the second generates su(2)⊕su(2). Existence of such same-rank/different-algebra pairs would directly refute Theorem 19.","tokens_in":15634,"feed_emoji":"⚛️","tokens_out":5986,"duration_ms":50037,"temperature":0.7,"pith_summary":"The paper tackles the bottleneck of building minimal complete pools (MCPs) of Pauli operators for variational quantum eigensolvers: previously, verifying that a candidate pool generates the full target Lie algebra required constructing the algebra's basis through nested commutators, a process that scales exponentially with system size. The authors claim a pool is an MCP if and only if its associated anti-commutation matrix Γ_A has the same rank and cardinality as a reference MCP's matrix. Because Γ_A is a symmetric zero-diagonal binary matrix, its congruence class — and hence the generated algebra — is determined by its rank alone. If correct, this turns MCP verification into an O(N^3) rank computation and enables user-defined symmetry-preserving pools that scale linearly with qubit count. The authors demonstrate the method in batched ADAPT-VQE and fixed-ansatz NI-DUCC-VQE on molecules up to 26 qubits, reaching chemical accuracy in strongly correlated systems without the previous exponential pre-computation.","feed_headline":"A single rank test certifies minimal generator pools","feed_subtitle":"The method shrinks operator-pool construction from exponential to O(N^3), unlocking 26-qubit molecule simulations.","key_machinery":"The Γ_A matrix: a symmetric binary matrix whose (i,j) entry is 1 exactly when Pauli strings P_i and P_j anti-commute, and 0 otherwise. Pool contraction — replacing one generator by its commutator with another — is represented as congruence Γ_A → P^T Γ_A P for an invertible binary matrix P, so the entire search for a complete pool becomes a statement in linear algebra. Since symmetric zero-diagonal binary matrices are classified up to congruence by rank, rank(Γ_A) becomes a candidate complete invariant for the Lie algebra generated by the pool, provided bracket-independence filters out redundant generators.","core_discovery":"The central claim is Theorem 19: a bracket-independent set of Pauli strings is a minimal complete pool for a given Lie algebra if and only if the rank of its Γ_A matrix equals the rank of the reference MCP's Γ matrix and the cardinalities match. This descends from the MCP Congruence Theorem, which states that a bracket-independent set is an MCP exactly when its Γ_A matrix is congruent to a canonical matrix for the target algebra under an invertible binary transformation. The proof uses the classical classification of symmetric zero-diagonal binary matrices over F2, where congruence classes are fully determined by rank, together with the observation that the graph contraction procedure of poo","pith_inferences":["The paper's termination condition for symmetry-preserving pools — rank(Γ_A)=2N−4 — is presented as a numerical observation; a formal characterization of the 'odd-string even-flip' subalgebra would make the algorithm fully rigorous.","If rank is truly a complete invariant, pool design reduces to a graph classification problem: enumerate all bracket-independent sets whose anti-commutation graph has the reference rank, which could allow systematic exploration of all MCPs for a target algebra.","The congruence-preserves-algebra step (Proposition 13) is the logical hinge; before relying on the rank criterion for large systems, one should stress-test it against small sets with equal rank but different generated algebras.","Testable extension: for small N, enumerate all bracket-independent pools of a given size and rank, compute their DLA closures explicitly, and check whether equal-rank pools always generate isomorphic algebras; any counterexample would bound the validity of the theorem."],"forward_implications":["MCP verification and construction become polynomial-time operations, replacing exponential DLA-closure checks with an O(N^3) rank computation.","Symmetry-preserving pools that respect molecular symmetries can be built and certified at scale, yielding operator pools of size O(N) instead of O(N^4).","Batched ADAPT-VQE using these pools requires far fewer gradient evaluations to reach chemical accuracy, especially in strongly correlated systems such as stretched H6.","The fixed-ansatz NI-DUCC-VQE, previously capped near 14 qubits by pool-construction cost, can now simulate 16- and 26-qubit molecules (H8 and H2O) with modest numbers of function evaluations.","Since the framework is stated for any Lie algebra with a Pauli-string basis, the same rank criterion applies beyond chemistry, including quantum error correction, quantum control, and Hamiltonian simulation."],"fun_headline_variants":["Rank test certifies minimal generator pools in O(N^3)","Polynomial-scaling pool construction for quantum chemistry","Cubic-time pool generation enables 26-qubit simulations","Single rank check yields minimal Lie algebra pools"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The framework assumes that a congruence transformation of the anti-commutation matrix preserves the Lie algebra generated by the pool — if two pools with congruent Γ matrices can generate different algebras, the rank criterion ceases to certify completeness.","fun_headline_variants_meta":{"raw":{"variants":["Rank test certifies minimal generator pools in O(N^3)","Polynomial-scaling pool construction for quantum chemistry","Cubic-time pool generation enables 26-qubit simulations","Single rank check yields minimal Lie algebra pools"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000371,"raw_usage":{"total_tokens":1795,"prompt_tokens":691,"completion_tokens":1104,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":435,"completion_tokens_details":{"reasoning_tokens":1052}},"tokens_in":435,"tokens_out":1104,"duration_ms":8164,"temperature":1.0,"reasoning_tokens":1052,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T19:44:58.610793+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For N=2 qubits, compare the four-string pool {X1, Z1, X1X2, Z2} with {X1, Z1, X2, Z2}. Both have Γ-matrices of rank 4, but explicit DLA-basis closure computations show the first generates su(4) while the second generates su(2)⊕su(2). Existence of such same-rank/different-algebra pairs would directly refute Theorem 19.","supporting_citations":[],"review_version":1}