{"id":"334b2a33-9563-4caf-b6ca-59f382f5746b","arxiv_id":"2505.19772","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"tVHA truncates a Hamiltonian variational ansatz by discarding small non-Coulomb two-body terms, reducing CNOT counts by factors of 2 to 5 on LiH, H2, H4, and CH2 test systems while often staying close to chemical accuracy.","lead":"The paper introduces tVHA, a variational quantum circuit for molecules that keeps only the largest non-Coulomb two-electron terms, cutting circuit size while aiming to preserve accuracy. For quantum chemists and hardware engineers, this is a practical ansatz that could make small molecular VQE experiments feasible on near-term devices at the cost of a tunable accuracy trade-off.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The truncation heuristic rests on an unproven magnitude-to-importance assumption; Eq. (12) provides no per-term variational compensation, and the p≈0.5 plateau is only demonstrated on the same four molecules used to set it.","rationale":"The paper is a useful engineering contribution: tVHA is clearly specified, the code is released, and the circuit-size reductions on four molecules are concrete and reproducible in principle. The reader's CONDITIONAL verdict is appropriate. My stress-test pass looked for a reason to move beyond conditional and did not find one, but it sharpened the central caveat. The load-bearing assumption is not merely that many small terms can collectively contribute; it is that truncating Hγ by integral magnitude preserves the relevant directions of the ansatz given that Eq. (12) supplies only one parameter per Trotter step to the whole Hγ block. That assumption is empirically plausible on the four benchmarks but has no supporting theorem and no validation outside the calibration set, and the p≈0.5 recommendation is derived from those same systems. A random-truncation control would separate the effect of the magnitude ranking from the effect of merely having fewer terms, and a second optimizer budget would separate convergence ease from parameter count. If either test fails, the central claim would need to be weakened from a general NISQ circuit-design principle to a statement about the specific molecules tested.","tokens_in":20608,"tokens_out":11373,"duration_ms":137775,"concrete_test":"Run a random-truncation control on LiH and H4 with two Trotter steps: keep exactly the same number of non-Coulomb two-body terms (including Hermitian-conjugate pairs) as the p=0.5 cutoff, but select those terms uniformly at random from the full Hγ set, preserving Hermitian pairing. Optimize at least 20 random subsets with SBPLX for 1000 evaluations and compare the median and best energies against the magnitude-ordered p=0.5 result. If random subsets reproduce the ~3 mHa accuracy, the magnitude ordering is not the operative mechanism and the redundancy claim needs reformulation; if they fail systematically, the ordering is load-bearing. Repeat the best cases with 2000 and 5000 evaluations to separate optimizer-budget effects from ansatz expressibility.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central efficiency claim depends on the Sec. 3.5 assertion that non-Coulomb two-body terms can be ordered by |g_s| and that terms below the threshold p are redundant. This is not derived and is supported only by the empirical plateaus in LiH, H2, H4, and CH2. The ansatz in Eq. (12) assigns a single variational parameter γ_n per Trotter step to the entire Hγ block, not one parameter per non-Coulomb term. After internal Trotterization, each retained term's rotation angle is proportional to its integral times the shared γ_n, so removing a term changes the fixed generator Hγ and leaves no per-term parameter to ‘adjust’ in compensation, contrary to the wording in Sec. 4.2. Moreover, the p=0 to p≈0.2 comparison in Figs. 3, 6, 9, and 13 shows that many small terms collectively contribute significant correlation energy, so ‘small’ is not equivalent to ‘redundant’ within this ansatz structure. Because the p≈0.5 rule-of-thumb is calibrated on the same four benchmark systems, the generalization of the truncation heuristic to other molecules is the least secure link in the paper's central claim. The fixed 1000-evaluation SBPLX budget additionally confounds the ‘easier convergence’ comparison, but the truncation-redundancy assumption is the load-bearing issue.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript introduces tVHA, a variant of the Hamiltonian Variational Ansatz in which the molecular Hamiltonian is split into one-body (Hα), Coulomb two-body (Hβ), and non-Coulomb two-body (Hγ) parts, with independent variational parameters per Trotter step. The circuit is built from a truncated Hγ containing only the largest non-Coulomb terms, chosen so that the retained fraction of the total |g| sum equals p. The authors benchmark statevector VQE runs for LiH, H2, H4, and CH2 (active space), comparing energy error with respect to FCI/CASCI, CNOT counts, and parameter counts against UCCSD, UCCSDT, and HEA. The central claims are that tVHA sharply reduces the number of variational parameters, can reduce CNOT counts by factors of roughly 2.2–2.5 at p≈0.5 with moderate accuracy loss, converges more easily than UCCSD/HEA, and is compatible with active-space constructions.","tokens_in":20863,"tokens_out":4567,"duration_ms":48575,"significance":"The paper has the merit of being concrete and reproducible: the statevector simulations, the CNOT and parameter counts, and the released tVHA code give a clear benchmark of what the ansatz does on specific molecules. If the truncation heuristic is reliable beyond the four tested systems, tVHA is a useful NISQ-era ansatz. However, the key transferable rule (p≈0.5 and magnitude-based ordering) is calibrated and validated on the same systems, and the ansatz does not have a per-term variational parameter that could demonstrably compensate for discarded terms. The significance therefore depends on empirical generality that is asserted rather than established.","major_comments":[{"comment":"The statement that omitted non-Coulomb terms are 'redundant' is load-bearing, but it is not derived and is in tension with the ansatz structure. In Eq. (12) each Trotter step has one shared parameter γ_n multiplying the entire truncated Hγ; after internal Trotterization each retained non-Coulomb term's rotation angle is proportional to its own integral times the common γ_n. Removing a term therefore changes the fixed generator and does not create a free parameter that can be readjusted to compensate for the missing term. The wording in Sec. 4.2 ('adjust the parameters of the other terms in the variational approach') is not supported by Eq. (12). Please either introduce per-term variational flexibility, prove a compensation mechanism, or replace 'redundant' with a weaker empirical claim.","section":"3.5 and 4.1, Eq. (12)"},{"comment":"The proposed rule-of-thumb p≈0.5 is obtained from the same four molecules on which the method is then demonstrated, so the generalization claim is circular in a practical sense. Moreover, the p=0 to p≈0.2 portions of all four energy curves show that many individually small terms collectively add significant correlation energy; 'small magnitude' is therefore not equivalent to 'negligible' within this ansatz. A held-out molecule or a formal error bound is required before the truncation heuristic can be presented as a general design principle.","section":"3.5 and Figs. 3, 6, 9, 13"},{"comment":"The 'easier convergence' comparison is confounded by the fixed SBPLX budget of 1000 function evaluations. The text acknowledges that this budget is insufficient for more than about 15 parameters, which directly affects the UCCSD and HEA comparisons in Figs. 3, 6, 9, and 13. A fair test requires either increasing the budget until convergence, reporting energy-vs-evaluations curves, or restricting the convergence claim to the parameter-count reduction (which is supported by the data).","section":"Appendix B"},{"comment":"The attribution of the remaining energy error at p=1 and 5 Trotter steps to optimizer limitations is plausible but unsupported; it could also be an expressibility or initialization deficiency. Please provide evidence, for example by continuing the optimization or using a second optimizer, or state plainly that the cause is undetermined. This matters because the same optimizer limitation is used to explain why tVHA at p=0.5 and two Trotter steps stays 3 mHa above FCI.","section":"4.1"}],"minor_comments":[{"comment":"There are typos: 'expection value' in Sec. 1 and 'conjuction' in Sec. 3.2 should be 'expectation' and 'conjunction'.","section":"1 and 3.2"},{"comment":"The phrase 'Møller-Plesset gates on devices based on Rydberg atoms' appears to be a typo for 'Mølmer–Sørensen gates'; as written it conflates an electronic-structure method with an entangling gate.","section":"4.2, side note"},{"comment":"The captions contain missing closing parentheses in the references to the figures ('see figure 7.' and 'see figure 10.'); these should be corrected.","section":"Figs. 7 and 10"},{"comment":"The symbol T is used both for the total adiabatic time and for the time-ordering operator, which is a source of confusion; consider renaming one of them.","section":"3.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for a quantum-information/quantum-chemistry journal, and the empirical circuit-reduction data are solid and worth preserving. My main concern is that the authors present a truncation heuristic as a general principle when the evidence is limited to four molecules. I would recommend allowing revision rather than rejection, provided the 'redundant' language is replaced or justified and the optimizer-budget issue is addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a solid engineering paper that does what it says—shows a magnitude-based truncation of non-Coulomb two-body terms can cut CNOT counts by 2–5× on small molecules while staying within a few mHa of FCI. The novel piece is the decomposition of the molecular Hamiltonian into one-body, Coulomb, and non-Coulomb parts and truncating the last by integral magnitude. The numerical work is clean: statevector simulations, four molecules (including an active-space CH2), and they ship the code. I give them real credit for that.\n\nThe soft spots are exactly where you'd expect. The central assumption—that non-Coulomb terms below a cumulative magnitude threshold p are redundant—is asserted rather than derived. The plateau around p≈0.5 is shown on the same four molecules used to set the heuristic, so generalization to other systems is not guaranteed. The stress-test point about Eq. (12) is also fair: the ansatz has a single parameter γ_n for the whole Hγ block per Trotter step, not per term, so removing a term doesn't leave a free parameter to 'adjust' for it. The language in Sec. 4.2 about adjusting the other terms is loose; what you actually adjust is the shared scale. And the p=0 to p≈0.2 jumps in energy show that many small terms collectively contribute correlation, so 'small' is not the same as 'redundant' within this ansatz.\n\nThe other significant weakness is the fixed 1000-function-evaluation SBPLX budget. That disadvantages high-parameter ansatze like UCCSD and HEA and muddies the 'easier convergence' claim. The authors know this—they flag it in Appendix B—but it means the comparative statements should be treated cautiously.\n\nOverall, though, the core circuit-reduction claim holds up for the tested systems. The method is a reasonable heuristic, not a rigorous theory, and the paper is honest about that. I'd send it out for peer review. The authors should be asked to (1) run with fairer optimizer budgets (or repeat with multiple seeds), (2) validate on at least one molecule outside the calibration set, and (3) either provide a derivation or soften the 'redundant' language.\n\nWorth a reading group slot.","headline":"A useful circuit-truncation heuristic with reproducible numerics, but the magnitude-to-importance assumption is empirical, not proven, and the convergence comparison is confounded by the fixed optimizer budget.","tokens_in":21443,"tokens_out":3865,"would_cite":true,"duration_ms":41976,"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":"Truncating the smallest non-Coulomb two-electron terms in a variational Hamiltonian ansatz cuts CNOT counts by factors of 2 to 5 on LiH, H2, H4, and CH2, while keeping energies within a few millihartree of full configuration interaction.","keywords":["variational quantum eigensolver","truncated Hamiltonian ansatz","non-Coulomb two-body terms","quantum chemistry","NISQ devices","CNOT count","active space","Jordan-Wigner mapping"],"falsifier":"Run tVHA at $p=0.5$ on a molecule whose non-Coulomb two-electron integrals have no magnitude gap, so the omitted tail contains many terms comparable to the kept ones, and compare against FCI while increasing the number of Trotter steps. If the omitted terms remove a substantial share of correlation energy that neither more Trotter steps nor re-optimization of retained parameters recovers, the redundancy assumption is falsified; a complementary test is to build the circuit from the omitted terms alone and check whether it approaches FCI after optimization.","tokens_in":20393,"feed_emoji":"⚛️","tokens_out":10408,"duration_ms":96380,"temperature":0.7,"pith_summary":"The paper introduces a circuit-design rule for the variational quantum eigensolver: in the Hamiltonian variational ansatz, throw away the non-Coulomb two-electron terms with the smallest integrals. Those terms are the ones that turn into long CNOT ladders and dominate circuit size, and the paper argues that for typical molecules their contribution is largely redundant once larger terms and variational parameters are present. The resulting truncated ansatz (tVHA) uses the full Hamiltonian for energy measurement but builds its circuit from a truncated operator list selected by a threshold $p$, the fraction of total non-Coulomb integral magnitude retained. In simulations of LiH, H2, H4, and CH2, $p\\approx 0.5$ cuts CNOT counts by factors of roughly 2 to 5 compared with full VHA while keeping energies close to FCI, and the smaller parameter count makes VQE convergence easier than with UCCSD or HEA. If this holds, the method gives a cheap, physics-based way to fit quantum chemistry circuits onto near-term hardware.","feed_headline":"Dropping small electron terms shrinks quantum chemistry circuits 2–5×","feed_subtitle":"Truncating the weakest two-electron terms keeps energies near FCI on LiH, H2, H4, and CH2 while slashing CNOT counts.","key_machinery":"The load-bearing object is the truncated non-Coulomb two-body Hamiltonian $H_\\gamma^{\\mathrm{cut}}$, built by rewriting the two-electron integrals in antisymmetrized form $\\tilde{g}_{ijkl}$, sorting the resulting terms by $|\\tilde{g}|$, and keeping the first $s_{\\mathrm{cut}}$ terms such that $p=\\sum_{s=1}^{s_{\\mathrm{cut}}}|g_s^\\gamma|/\\sum_s |g_s^\\gamma|$. Each retained term is a fermionic double-excitation that becomes a parametrized Pauli-string rotation, implemented after Jordan-Wigner mapping as a ladder of CNOT gates surrounding a single-qubit $Z$ rotation; one-body terms become single-qubit rotations and Coulomb two-body terms become ZZ-type gates. The circuit is a first-order Trotterized sequence $\\prod_n \\exp(i\\alpha_n H_\\alpha)\\exp(i\\beta_n H_\\beta)\\exp(i\\gamma_n H_\\gamma^{\\mathrm{cut}})$ acting on the Hartree-Fock state, with parameters initialized along the adiabatic ramp. Keeping only $H_\\gamma^{\\mathrm{cut}}$ in the circuit is what removes the expensive nonlocal gates while preserving the dominant physics.","core_discovery":"The paper's central claim is that the non-Coulomb two-electron terms of a molecular Hamiltonian can be ordered by the size of their antisymmetrized integrals, and the long tail of small terms is redundant for the variational ground-state problem. Building the ansatz circuit from only the terms above a truncation threshold $p$, defined by the fraction of total integral magnitude retained, reduces the number of parameters and the number of CNOT ladders without a proportionate loss of accuracy. In the LiH benchmark, tVHA with $p\\approx 0.5$ and two Trotter steps lands about 3 mHa above FCI with roughly 2000 CNOT gates, a factor of 2.5 reduction compared with full VHA; the hydrogen chain drops from over 1100 to under 500 CNOTs at $p\\approx 0.5$; and H2 plus the CH2 active space reach chemical accuracy with one Trotter step. The truncated Hamiltonian is used only to build the state; the full Hamiltonian is still measured, so the truncation changes which states the ansatz can reach rather than the observable being evaluated.","pith_inferences":["The paper does not develop a per-symmetry or per-Trotter-layer truncation, but the H2 and CH2 symmetry artifacts suggest that truncating in a symmetry-adapted way could preserve Pauli-term cancellations and cut circuits further.","A natural test the paper leaves open is to apply the same magnitude cutoff after fermion-to-Pauli mapping; for mappers other than Jordan-Wigner, small fermionic integrals need not map to small Pauli terms, so the savings could differ.","The measurement truncation threshold mentioned but not implemented could be made adaptive: a circuit-level p and a separate measurement-level p would let users trade sampling cost against energy precision without touching gate count.","Because tVHA ranks operators by integral magnitude at almost no classical cost, it could seed iterative adaptive ansatze with a physically motivated operator pool instead of starting from an unranked excitation list."],"forward_implications":["At $p\\approx 0.5$, LiH requires about 2000 CNOT gates with two Trotter steps instead of more than 5000 for full VHA, while staying about 3 mHa above FCI.","For the hydrogen chain, the CNOT count falls from over 1100 to under 500 at $p\\approx 0.5$, and tVHA captures strong correlation that the hardware-efficient ansatz misses.","H2 and the CH2 active space reach chemical accuracy with a single Trotter step using at most half of the non-Coulomb two-body terms.","The low parameter count lets the classical optimizer converge within the allotted 1000 function evaluations, whereas UCCSD and HEA get stuck in local minima on the same benchmarks.","Accuracy can be improved systematically by adding Trotter steps, which increases the circuit only linearly, giving a direct knob for trading circuit size against accuracy on a given device."],"supporting_citations":[{"why":"Introduces the Hamiltonian variational ansatz whose Trotterized layers tVHA truncates.","marker":"[34]"},{"why":"Supplies the adiabatic-ramp parameter initialization that tVHA adopts.","marker":"[35]"},{"why":"Defines the variational quantum eigensolver loop used to optimize the ansatz parameters.","marker":"[17]"},{"why":"Defines the UCCSD reference ansatz whose parameter and CNOT counts are compared with tVHA.","marker":"[18]"},{"why":"Defines the hardware-efficient ansatz baseline used in the benchmarks.","marker":"[19]"},{"why":"Analyzes the entangling structure and optimization of the Hamiltonian variational ansatz, motivating the three-part decomposition.","marker":"[36]"},{"why":"Supplies the one- and two-electron integrals that define the Hamiltonian parts for the benchmark molecules.","marker":"[26]"},{"why":"Documents the Jordan-Wigner mapping used to convert the truncated fermionic terms into Pauli gates.","marker":"[37]"},{"why":"Selects the active space for the methylene benchmark.","marker":"[33]"}],"fun_headline_variants":["Truncated ansatz shrinks quantum chemistry circuits by up to 5×","Dropping weak electron terms cuts quantum circuit size 2–5×","tVHA prunes electron terms to slash quantum chemistry circuit costs","Trimming small integrals reduces quantum ansatz parameters and CNOTs","Quantum chemistry circuits shrink 2–5× with truncated variational ansatz"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the smallest non-Coulomb two-electron terms, judged by integral magnitude, are redundant for the ground-state energy and can be dropped because the retained variational parameters compensate for them; this fails if many small terms collectively contribute.","fun_headline_variants_meta":{"raw":{"variants":["Truncated ansatz shrinks quantum chemistry circuits by up to 5×","Dropping weak electron terms cuts quantum circuit size 2–5×","tVHA prunes electron terms to slash quantum chemistry circuit costs","Trimming small integrals reduces quantum ansatz parameters and CNOTs","Quantum chemistry circuits shrink 2–5× with truncated variational ansatz"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000798,"raw_usage":{"total_tokens":3536,"prompt_tokens":998,"completion_tokens":2538,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":614,"completion_tokens_details":{"reasoning_tokens":2455}},"tokens_in":614,"tokens_out":2538,"duration_ms":14540,"temperature":1.0,"reasoning_tokens":2455,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:06:21.613220+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run tVHA at $p=0.5$ on a molecule whose non-Coulomb two-electron integrals have no magnitude gap, so the omitted tail contains many terms comparable to the kept ones, and compare against FCI while increasing the number of Trotter steps. If the omitted terms remove a substantial share of correlation energy that neither more Trotter steps nor re-optimization of retained parameters recovers, the redundancy assumption is falsified; a complementary test is to build the circuit from the omitted terms alone and check whether it approaches FCI after optimization.","supporting_citations":[],"review_version":1}