{"id":"1056cdad-0bea-407a-b14a-d2608651325f","arxiv_id":"2512.12097","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Spatial-symmetry enforcement makes the saGSpD operator pool non-universal; ADAPT-VQE emulations then map exactly which symmetry must be enforced to avoid variational collapse.","lead":"This paper proves that a gate-efficient quantum-chemistry ansatz — singlet spin-adapted singles plus perfect-pairing doubles — is not universal once molecular point-group symmetry is enforced, and shows in variational-quantum-eigensolver emulations when symmetry-breaking operator pools collapse to the wrong state. The payoff is a decision rule: enforce at least one distinguishing symmetry when target states cross, and the specific one when they differ in only one symmetry.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Two-irrep non-universality proof is incomplete: no exhaustive Lie-algebra closure argument rules out A[1] with Γ_P=Γ_Q≠Γ_R=Γ_S.","rationale":"The reader's weakest assumption is the same one I identify: the two-irrep generalization relies on an enumerated commutator analysis that is not proven exhaustive. My stress-test sharpens this concern by noting that the displayed commutators in Supplement S1 have symmetry restrictions corresponding to the row-4 pattern, not the row-2 pattern claimed absent, and that a row-2 A[0] is in fact generated by S22 under relabeling. This makes the missing closure argument concrete rather than merely 'no inductive proof.' The ≥4-irrep theorem is rigorous via Eq. (16), and the numerical H6/STO-6G stagnation with exactly two predicted unreachable CSFs provides strong empirical support for the two-irrep case. However, empirical support does not replace the missing proof, especially because the abstract claims the result for all non-C1 point groups, and the H6 example itself uses only two effectively active irreps. The right verdict remains CONDITIONAL: accept the core result for ≥4-irrep groups, require a completed or explicitly scoped proof for two-irrep groups before claiming full generality, and encourage release of the QForte implementation for independent numerical checks. My recommendation UNCHANGED reflects that the reader's conditional verdict already captures this concern.","tokens_in":26295,"tokens_out":23783,"duration_ms":194753,"concrete_test":"For a minimal two-irrep model (two A and two B spatial orbitals, 8 spin-orbitals), explicitly construct the totally symmetric saGSpD generators in the occupation-number basis and compute the Lie algebra they generate by repeated commutators until closure (finite-dimensional, embedded in gl(16)). Then test whether A[1]^{RS}_{PQ} with Γ_P=Γ_Q≠Γ_R=Γ_S lies in the span, and compare the algebra dimension with the full spin- and symmetry-adapted Lie algebra. If row-2 A[1] is in the span, the paper's two-irrep non-universality claim is false; if it is not, the claim survives this case but still needs a general proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"For point groups with ≥4 irreps, the parity invariant Π_I of Eq. (16) rigorously shows that A[0,1]^{RS}_{PQ} with four distinct irreps is outside the saGSpD Lie algebra; that part is solid. The load-bearing soft spot is the claimed generalization to two-irrep groups (Ci, C2, Cs). The paper's argument rests on the assertion that no nested commutator of totally symmetric saGSpD generators can produce A[1]^{RS}_{PQ} with Γ_P=Γ_Q≠Γ_R=Γ_S. The support is the Supplement S1 enumeration up to triply nested commutators plus a structural number-operator argument. But that enumeration mainly displays commutators whose single-excitation restrictions are Γ_P=Γ_S and Γ_Q=Γ_R (the row-4 pattern of Table I), not the row-2 pattern Γ_P=Γ_Q≠Γ_R=Γ_S targeted by the theorem. In fact, S22, [[A^{QQ}_{PP}, A^R_Q], A^S_P] = A[0]^{QR}_{PS}, with a relabeling Q→R, R→S, P→P, S→Q, generates A[0]^{RS}_{PQ} with Γ_P=Γ_Q≠Γ_R=Γ_S. Thus the in-algebra A[0] needed to disentangle a √3A[1]−A[0] combination is available; whether some commutator of saGSpD elements yields the corresponding A[1] or a √3A[1]−A[0] combination for row 2 is not settled. No inductive proof, no dimension count, and no invariant separating row-2 A[1] from the generated algebra is provided. Since the abstract claims non-universality for all point groups other than C1, and since the H6 stagnation evidence depends on the two-irrep case (D2h with only Ag/B1u active), this gap is load-bearing: if a higher-order commutator generates row-2 A[1], the two-irrep theorem is false, not merely unproven.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes trade-offs among symmetry adaptation, universality, and gate efficiency in operator pools for variational quantum eigensolvers, focusing on the singlet spin-adapted singles plus perfect-pairing doubles (saGSpD) pool. The central theoretical claim is that enforcing spatial symmetry in saGSpD, for molecules with non-trivial point-group symmetry, makes the pool non-universal: certain singlet spin-adapted double excitations cannot be generated by the Lie algebra of the totally symmetric saGSpD pool. The paper proves this rigorously for elementary Abelian 2-groups with at least four irreps using a parity invariant Π_I (Eq. 16), and argues the same for two-irrep groups (Ci, C2, Cs) using commutator enumeration. The numerical part performs ADAPT-VQE simulations with four pools (GSD, saGSpD-full, symmetry-adapted saGSpD, pDint0) in three physically motivated scenarios: a global ground state (H6), crossing states differing in multiple symmetries (CH2), and crossing states differing in one symmetry (BeH2 for spin, BO for spatial symmetry). The resulting guidelines concern when symmetry-breaking pools are safe, when one distinguishing symmetry suffices, and when a specific symmetry must be enforced.","tokens_in":26704,"tokens_out":17164,"duration_ms":156538,"significance":"If the full theorem is correct, the paper identifies a fundamental and practically relevant limitation of a widely used gate-efficient ansatz pool, and the numerical scenarios provide a useful benchmark map for symmetry-related variational collapse. The ≥4-irrep part of the proof is clean and convincing: the parity invariant is preserved by every saGSpD generator and hence by the whole Lie algebra, so parity-flipping double excitations are provably unreachable. The paper also gives concrete, reproducible numerical evidence (CSF overlaps, symmetry weights, potential energy curves) across four molecular systems. The main weakness is that the two-irrep case, which is needed for the full force of the abstract and for the H6 stagnation demonstration, is supported by an enumerated commutator analysis rather than an exhaustive closure argument. The numerical simulations are extensive and the paper is clearly written, but the central theorem as stated is not fully proved.","major_comments":[{"comment":"The proof of non-universality for two-irrep point groups (Ci, C2, Cs) is incomplete. The argument shows one family of nested commutators that produces A[1]RS_PQ only with the irrep restrictions Γ_P=Γ_S and Γ_Q=Γ_R (row-1 pattern), and then asserts that A[1]RS_PQ with Γ_P=Γ_Q≠Γ_R=Γ_S (row-2 pattern) is absent. Supplement S1 explicitly states that the enumeration goes only 'up to triply nested commutators' and provides no inductive closure proof, dimension count, or invariant separating row-2 A[1] from the generated algebra. In fact, Eq. (S22) shows [[A^QQ_PP, A^R_Q], A^S_P] = A[0]^QR_PS, which by relabeling gives an in-algebra A[0]^RS_PQ with exactly the row-2 irrep pattern. Thus the algebra contains the A[0] needed to disentangle a √3A[1]−A[0] combination, and whether some higher-order commutator yields the corresponding A[1] or the combination for row 2 is not settled. This gap is load-","section":"Section III, Eqs. (17)-(18) and Supplement S1"},{"comment":"The paper asserts that saGSpD-full—the saGSpD pool without spatial-symmetry enforcement—is universal ('when non-totally symmetric operators are incorporated into it, the saGSpD pool is universal'), and Table II lists it as universal. No proof or reference for this assertion is provided. This matters because the numerical taxonomy and the safety guidelines for 'symmetry-breaking but universal pools' rely on saGSpD-full being universal, and its failure in the BO case (Sec. V.C) is interpreted as symmetry breaking rather than as possible non-universality. Please either supply a proof of universality (e.g., showing the Lie algebra generated by all spin-adapted singles and perfect-pairing doubles closes on the full symmetry-adapted saGSD algebra) or cite a reference that establishes it.","section":"Section III and Table II"}],"minor_comments":[{"comment":"The abstract states that the saGSpD pool is non-universal for 'point groups other than C1', but the proof is restricted to elementary Abelian 2-groups (Ci, C2, Cs, D2, C2v, C2h, D2h). The argument for perfect-pairing doubles being totally symmetric by nature relies on the 2-group property Γ⊗Γ = 1. The abstract and conclusions should be qualified to this class, or the proof should be extended to cover cyclic and non-Abelian groups that appear in molecular point-group practice.","section":"Abstract and Section III"},{"comment":"The proportionality in Eq. (18) is stated without the constant; it would be clearer to display the exact factor (1/√2 or 1/2) and specify the assumed occupancies (P doubly occupied, S unoccupied) explicitly in the equation or immediately before it.","section":"Eq. (18)"},{"comment":"The condition 'ξ = Q if R = P, otherwise ξ = R' in the header of Table I is cryptic. Please define ξ in the caption or footnote, and consider giving an example for each of the five rows.","section":"Table I"},{"comment":"The saGSpD-full pool is defined only as 'the spatial-symmetry-violating variant'; the specific extension of ADAPT-VQE with tuples of two and three non-totally symmetric singles should be described more explicitly, including why these tuples are sufficient for universality and how the closed-form expression of Ref. 48 is used.","section":"Section IV"},{"comment":"The sentence 'At that limit, all four states are degenerate and any linear combination of them is an eigenstate' is correct, but the preceding discussion of the 3.0 Å H6 result would benefit from a quantitative tie-in to the orthogonal CSF weights shown in Fig. S9, since the text references but does not interpret the figure in the main body.","section":"Section V.A"}],"recommendation":"major_revision","confidential_remarks":"The paper is from a group with strong publication record in this area, and the numerical study is carefully done. The main reason for major revision is the incomplete proof for the two-irrep case; this is a central claim and is not a mere presentation issue. If the authors can supply an exhaustive Lie-algebra closure argument or an invariant for the row-2 A[1] operators, the paper would be a valuable contribution. I also encourage them to address the asserted universality of saGSpD-full, as it underpins the interpretation of several numerical experiments. The scope mismatch between the abstract and the Abelian 2-group theorem should also be corrected."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the spatial-symmetry non-universality of saGSpD is real for point groups with at least four irreps. The parity-invariant argument in Eq. (16) is clean and new, and the known literature (QNP, DISCO-VQE, tUPS) didn't enforce point-group symmetry, so the negative result for those groups is genuinely novel.\n\nFor two-irrep groups (Ci, C2, Cs), the abstract overclaims. The supplement enumerates commutators up to triple nesting and then asserts that the row-2 A[1] operators are absent, but there is no closure argument, dimension count, or invariant to rule out higher-order routes. I checked the stress-test concern: the relabeled S22 commutator does produce an in-algebra A[0] with the row-2 pattern using only totally symmetric singles, so the ingredients to disentangle a row-2 √3A1−A0 combination are present. The paper simply doesn't show whether any higher commutator supplies that combination. That leaves the two-irrep case open, not contradicted. And it matters: the H6/STO-6G stagnation — the paper's cleanest numerical evidence — effectively sits in the two-irrep case (D2h with Ag and B1u active), so the numerics corroborate precisely the unproven part of the theorem.\n\nWhat the paper does well: the three-scenario framework is a genuinely useful design guide, and the numerical examples are honest — they report the failed moment-correction attempt and the fortuitous BeH2 success, and they acknowledge the STO-6G limitation. The citation pattern is fine; the negative claim isn't derived from the parent pool's published performance.\n\nSoft spots, in proportion: the two-irrep proof gap is the real one. After that, the absence of shipped code/data makes the stagnation and collapse claims hard to check independently, and the numerical conclusions rest on four minimal-basis molecules. Both are minor relative to the proof gap.\n\nWho it's for: ADAPT-VQE practitioners and anyone choosing symmetry-adapted pools. I'd send it to peer review and ask the authors to either complete the two-irrep proof or scope the theorem to ≥4 irreps and label the rest as conjectural. Either fixes the current mismatch between abstract and proof.","headline":"The ≥4-irrep non-universality theorem is solid and new, but the advertised two-irrep generalization is asserted rather than proven; the paper deserves refereeing and a requested revision either completing or scoping that claim.","tokens_in":27307,"tokens_out":5539,"would_cite":true,"duration_ms":50592,"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 gate-efficient pool saGSpD (singlet spin-adapted singles plus perfect-pairing doubles) is proven non-universal once point-group symmetry is enforced; simulations map when symmetry-breaking pools collapse to the wrong state.","keywords":["symmetry-adapted operator pools","ADAPT-VQE","universality","point-group symmetry","spin-adapted singles","perfect-pairing doubles","Lie algebra","variational collapse"],"falsifier":"Build the Lie algebra numerically for a small two-class system (say a four-orbital model with Cs symmetry): write the symmetry-enforced saGSpD generators as explicit matrices, close the algebra under commutation until the dimension stabilizes, and test whether any triplet-intermediate double excitation A[1]^{RS}_{PQ} with Γ_P = Γ_Q ≠ Γ_R = Γ_S lies in the span — if yes, the two-class claim is false. Or run ADAPT-VQE-saGSpD on H6 with a larger basis such as 6-31G: the theorem predicts the analogous basis states stay orthogonal and chemical accuracy is never reached, while populating those state","tokens_in":26061,"feed_emoji":"⚛️","tokens_out":36280,"duration_ms":280680,"temperature":0.7,"pith_summary":"Quantum simulations of molecules build variational wavefunctions from operator pools, and the most circuit-efficient spin-preserving pool in wide use — singlet spin-adapted singles plus perfect-pairing doubles (saGSpD) — stops being universal the moment molecular point-group symmetry is enforced. The proof runs through a parity invariant: every allowed excitation preserves the even-or-odd count of electrons within each symmetry class of orbitals, so families of double excitations present in the universal parent pool are unreachable, and with them certain exact-state components. The claim is proven outright for point groups with four or more symmetry classes, argued via explicit nested commutators for two-class groups, and confirmed numerically: the adaptive variational eigensolver with the symmetry-enforced pool never attains chemical accuracy for linear H6, and exactly the two predicted symmetry-adapted basis states remain orthogonal to the converged wavefunction. Companion simulations map a practical rule — with the target state the global ground state, universal symmetry-breaking pools eventually self-correct; when crossing states differ in several symmetries, conserving any one of them suffices; when they differ in a single symmetry, only conserving that one prevents variational collapse. The result matters because it makes a genuine three-way trade-off explicit for the most economical pool family: strict symmetry adaptation, universal expressiveness, and shallow circuits cannot be had simultaneously, and benchmarks that report only energies can miss the deficiency.","feed_headline":"Proof: gate-efficient, symmetry-safe chemistry pools aren't universal","feed_subtitle":"The proof identifies exact states a workhorse pool can never reach; simulations map which symmetries must be kept.","key_machinery":"The central object is a parity invariant (Eq. 16): for each orbital symmetry class (irreducible representation), a sign recording whether that class holds an even or odd number of electrons. Spin-adapted singles move one electron within a class and perfect-pairing doubles move two, so every generator — hence every commutator and unitary — preserves every parity. This blocks the double excitations the universal parent pool relies on: four-distinct-class operators are unreachable (a complete proof for groups with four or more classes), and the triplet-intermediate excitation with matched class pairs arises only inside the locked combination √3A[1] − A[0] (two-class groups, by explicit nested c","core_discovery":"The paper's central claim: the saGSpD pool — singlet spin-adapted singles plus perfect-pairing doubles — is not universal once restricted to totally symmetric operators of a non-trivial point group. Every generator preserves the electron-count parity of each orbital symmetry class, hence the whole Lie algebra does, putting many totally symmetric double excitations out of reach: proven for groups with at least four classes, argued via nested commutators for two-class groups. ADAPT-VQE with the enforced pool stagnates on linear H6 without chemical accuracy; the two predicted basis states stay orthogonal throughout. Dropping one symmetry restores universality, but sacrificing the wrong one coll","pith_inferences":["Extension: the parity-invariant argument is a transferable diagnostic — any operator pool whose generators change each symmetry class's electron count by an even amount is confined to a fixed parity sector, so other truncated pools (seniority-truncated or number-operator-weighted excitations) can be screened for expressivity ceilings up front using the same invariant.","The paper leaves open whether the two-class outcome is a theorem: a Lie-algebra dimension calculation for a minimal two-class Fock space would either promote the Cs/Ci/C2 claim to a proof by showing the enumeration closes, or expose a higher-order commutator that isolated the missing excitation.","A sharper test follows from the paper's own data: at stretched H6 geometries the large error shrinks because several states become degenerate, so energy-error benchmarks systematically underestimate the deficiency; a molecule whose correlation is dominated by the missing excitations should make the pool fail far more catastrophically than the 24–36 mEh seen here.","The paper's fortuitously exact BeH2 case implies that minimum-basis benchmarks of non-universal pools are optimistic by construction — enlarging the basis to add orbitals of the missing symmetry classes should convert that accidental success into a failure, a prediction the authors do not test."],"forward_implications":["Any variational ansatz built from totally symmetric singlet singles plus perfect-pairing doubles cannot reach the exact symmetry-adapted wavefunction for a molecule with non-trivial point-group symmetry: ADAPT-VQE with the enforced pool never attains chemical accuracy for linear H6, and exactly two of the 92 symmetry-adapted basis states remain orthogonal at every geometry.","Universal pools that drop a single symmetry converge to numerically exact energies in the tested cases (GSD and saGSpD-full on H6 and CH2), with symmetry contamination appearing early and self-correcting — but only when the target state is the global minimum or another distinguishing symmetry is retained.","The safe-use rule: when two crossing states differ in several symmetry properties, conserving any one of them suffices to avoid collapse (both GSD and saGSpD-full succeed for CH2's A1–B1 crossing); when they differ in one property only, that property must be conserved — GSD collapses to the triplet in BeH2 and saGSpD-full collapses to the Π² state in BO.","Fully symmetry-adapted universal pools are the most parameter-efficient (pDint0 reaches chemical accuracy in 67 operators for H6) but cost the most per operator — hundreds to thousands of CNOT gates — making the three-way trade-off between universality, symmetry preservation, and gate count explicit.","The paper claims the phenomenon is generic: any quantum algorithm that uses symmetry-violating gates risks symmetry contamination and variational collapse, so the three-scenario rule applies beyond the specific pools simulated here."],"fun_headline_variants":["Symmetry-enforced quantum pool proven non-universal","Gate-efficient symmetry pool can't reach all chemistry states","Proof: singlet-adapted pool fails when spatial symmetry locked","H6 test exposes non-universality of symmetry-adapted pool","Quantum chemistry: symmetry-safe pools sacrifice universality"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"For point groups with only two symmetry classes (Cs, Ci, C2), the theorem that the missing triplet-intermediate excitations can never be generated rests on the assertion that the commutator families enumerated up to third nesting in the supplementary material, together with a structural argument about number-operator strings, exhaust the Lie algebra — stated without an inductive closure proof or a dimension count, so the two-class claim is only as strong as that enumeration.","fun_headline_variants_meta":{"raw":{"variants":["Symmetry-enforced quantum pool proven non-universal","Gate-efficient symmetry pool can't reach all chemistry states","Proof: singlet-adapted pool fails when spatial symmetry locked","H6 test exposes non-universality of symmetry-adapted pool","Quantum chemistry: symmetry-safe pools sacrifice universality"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001805,"raw_usage":{"total_tokens":6976,"prompt_tokens":808,"completion_tokens":6168,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":552,"completion_tokens_details":{"reasoning_tokens":6087}},"tokens_in":552,"tokens_out":6168,"duration_ms":40178,"temperature":1.0,"reasoning_tokens":6087,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T16:44:20.304820+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build the Lie algebra numerically for a small two-class system (say a four-orbital model with Cs symmetry): write the symmetry-enforced saGSpD generators as explicit matrices, close the algebra under commutation until the dimension stabilizes, and test whether any triplet-intermediate double excitation A[1]^{RS}_{PQ} with Γ_P = Γ_Q ≠ Γ_R = Γ_S lies in the span — if yes, the two-class claim is false. Or run ADAPT-VQE-saGSpD on H6 with a larger basis such as 6-31G: the theorem predicts the analogous basis states stay orthogonal and chemical accuracy is never reached, while populating those state","supporting_citations":[],"review_version":1}