{"id":"ce40d087-2037-4bc7-9204-a53e5bd2b5f1","arxiv_id":"2505.04768","paper_version":2,"verdict":"REJECT","confidence":"HIGH","novelty_score":2.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A pedagogical review of VQE for molecular ground-state chemistry, with an asymptotic complexity estimate of O(N^9/epsilon^2) for UCC/UCCG and O(k N^7/epsilon^2) for k-UpCCG.","lead":"This paper is a tutorial introduction to the Variational Quantum Eigensolver (VQE) for chemistry, written in Portuguese, that also sketches a complexity analysis of three coupled-cluster ansatz circuits. A generalist might read it to learn the standard workflow of near-term quantum chemistry simulations and to see a rough cost estimate for UCC-style VQE.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Headline complexity claims omit the optimizer iteration count K, so the quoted O(N^9/epsilon^2) and O(kN^7/epsilon^2) are per-evaluation costs, not the total cost defined in Eq. (44).","rationale":"The paper is explicitly an introduction/tutorial; the reader rejects it for lack of novelty. My role is to stress-test the central claim. The most significant quantitative contribution is the complexity analysis in Section 5, culminating in Eqs. (52)-(53). In examining that analysis, the most load-bearing weakness is not the Trotter-absorption heuristic (which is standard in VQE practice, where l is often fixed to 1 and parameters absorb the approximation) but the mismatch between the formal total complexity defined in Eq. (44) and the results later labeled as 'total'. This is an internal inconsistency, not merely a disagreement with the literature. The number of optimizer iterations K is explicitly acknowledged as uncalculated, yet the final claims omit it. Since K can be exponentially large in N, the quoted O(N^9/epsilon^2) and O(kN^7/epsilon^2) give a misleading bound on the total runtime of VQE, affecting the paper's stated implications (e.g., 'total cost' impractical even with ideal hardware). The issue is load-bearing because the headline results are the primary takeaway of the paper; if they are only per-step costs, the conclusion about total complexity is unsupported.\n\nThe reader's weakest_assumption about Trotter steps is plausible but less decisive: the variational parameters can absorb a fixed l, and the ansatz with l=1 is a well-defined (if less expressive) variational family; whether it achieves chemical accuracy is an empirical question that does not invalidate the scaling as a cost estimate for that ansatz family. Our concern, in contrast, is a direct consequence of the paper's own equations.\n\nDespite this additional correctness concern, the verdict remains REJECT: the paper is a pedagogical review, not a novel research advance, and the correctness issue only reinforces that it does not meet the bar for a research preprint. Therefore no change to the reader's verdict is needed.","tokens_in":50747,"tokens_out":9402,"duration_ms":93310,"concrete_test":"Re-derive the total temporal complexity from Eq. (44) using the values P=O(N^5) (UCC/UCCG) or P=O(kN^3) (k-UpCCG) and Q=O(N^4/epsilon^2), obtaining X + K·O(N^9/epsilon^2) and X + K·O(kN^7/epsilon^2). Then search the paper for any bound or discussion of K; the paper explicitly says K is not computed (Section 5, after Eq. (44)). If no bound on K is provided, Eqs. (52)-(53) cannot be called the total cost. A direct check is to replace 'complexidade total' in the conclusion with 'custo por avaliação da função custo' and verify whether any implication in Section 5.2 or the abstract changes; if the claims are reinterpreted as per-evaluation costs, the paper's conclusion about 'complexidade total' being an impediment would require revisiting.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 5 (around Eq. (44)), the paper correctly decomposes the total temporal complexity of a VQA as X + (I+P)·Q·K, where K is the number of cost-function evaluations required by the classical optimizer. It then explicitly states it will not compute K because it depends on unpredictable factors. However, Eqs. (52) and (53) and the concluding paragraph quote the 'total cost' as O(N^9/epsilon^2) and O(kN^7/epsilon^2), respectively, on the basis of (I+P)·Q only. This is not the total complexity defined in Eq. (44); it is the cost of a single cost-function evaluation. K is not a constant: it can scale with N and with the required accuracy (e.g., exponentially in the presence of barren plateaus, as discussed in Section 5.2). Therefore the central quantitative claims of the paper—the scaling of the total VQE cost—are not derived; the derived statements are about per-evaluation cost. The paper's own caveat that K is not calculated (after Eq. (44)) is not reflected in the later 'complexidade total' statements, so the headline results are internally inconsistent with the paper's complexity framework.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper is a Portuguese-language pedagogical review of the Variational Quantum Eigensolver (VQE) applied to molecular electronic structure. It develops the required background in quantum chemistry, second quantization, basis sets, and the UCC/UCCG/k-UpCCG ansätze, and then presents a theoretical complexity analysis in Section 5. For a molecule described by N basis functions per spin with Jordan-Wigner encoding, the paper derives a variational circuit depth O(N^5) for UCC/UCCG and O(kN^3) for k-UpCCG, a measurement count O(N^4/epsilon^2), and per-evaluation costs O(N^9/epsilon^2) and O(kN^7/epsilon^2), stated in Eqs. (52) and (53). It closes with a discussion of NISQ limitations, especially barren plateaus.","tokens_in":50980,"tokens_out":11223,"duration_ms":121181,"significance":"If correct, these estimates would form a useful consolidation of known scaling results for practitioners and students: the paper explicitly separates preprocessing, initial state preparation, circuit depth, measurement, and optimizer iterations in Eq. (44), and it gives helpful worked examples such as Hartree-Fock state preparation and the decomposition of a single Pauli exponential. The bibliography is broad and the review is readable. However, the central quantitative claims are not new, and several load-bearing steps are either heuristic or inconsistent with the paper's own complexity framework. The paper therefore cannot currently serve as a reliable reference for the scaling of VQE, but the issues are localizable and correctable in revision.","major_comments":[{"comment":"The paper defines the total VQE complexity in Eq. (44) as X + (I+P)·Q·K and explicitly states that K will not be calculated because it depends on unpredictable factors. The text around Eqs. (52)-(53) then calls O(N^9/epsilon^2) and O(kN^7/epsilon^2) the 'cálculo da função custo total', and Section 6 refers to O(kN^7/epsilon^2) as the 'complexidade total' of the method. These statements identify the per-evaluation cost (I+P)·Q with the total algorithm cost. Since K is not a constant and can scale with N and with the required accuracy (a point the paper itself makes in the barren-plateau discussion), the headline total-cost scalings are not established. The claims should be explicitly reworded as per-cost-function-evaluation costs, or the K-dependence must be included.","section":"Section 5, Eq. (44); Section 5.2, Eqs. (52)-(53); Section 6"},{"comment":"The text states that the factor 1/l in the Trotterized exponential can be absorbed into the optimizable parameters theta_j so that the circuit becomes a sequence of Pauli exponentials. Absorbing 1/l changes the angle values, but the circuit still contains l repetitions of the entire Trotter step, so the circuit depth is multiplied by l. The Trotter error is O(1/l) and depends on the commutators of the Pauli terms; for fixed chemical accuracy, l will in general grow with N and with 1/epsilon. The reported depths O(N^5) and O(kN^3) are therefore per Trotter step, not total circuit depths, unless an explicit argument is supplied that l = O(1) suffices. This gap is load-bearing for the final scaling claims.","section":"Section 5.1, Trotterization paragraph"},{"comment":"The measurement scaling O(N^4/epsilon^2) is derived from S_i ≈ Var(P_i)/epsilon^2 per term and then summing over M terms. This is a per-term precision budget, not a total-error budget: if each of the M = O(N^4) terms is measured to precision epsilon, the variance of the summed estimator is approximately M·epsilon^2, so the total error grows as sqrt(M). To achieve total error epsilon, the per-term precision must be epsilon/sqrt(M), yielding at least O(M^2/epsilon^2) = O(N^8/epsilon^2) shots under the independent-measurement strategy used here (or O((sum |c_i|)^2/epsilon^2) with coefficient-aware allocation). This changes the Q factor in Eqs. (52) and (53) and should be corrected or explicitly justified as a different estimation strategy.","section":"Section 5.2, Eqs. (49)-(51)"},{"comment":"The paper asserts that because the average Jordan-Wigner Pauli weight is N/2, the number of sequential exponentials is of the same order as the total number of terms. This is a heuristic statement, not a derivation: the minimum depth of a scheduled circuit depends on the conflict graph of the Pauli supports, and an average-weight argument does not bound the chromatic number of that graph. Since the circuit depth P depends directly on this sequential count, the authors should either provide a counting or scheduling argument or explicitly flag this step as an assumption.","section":"Section 5.1, sequential-term count"}],"minor_comments":[{"comment":"The formula presented after 'a partir do teorema de Baker-Hausdorff' is the Zassenhaus expansion, not the Baker-Campbell-Hausdorff formula; the terminology should be corrected.","section":"Section 5.1, Eq. (47)"},{"comment":"The text says practical Hartree-Fock scales as O(N^3) for many problems, but the summary paragraph before Eq. (52) states that preprocessing grows as O(N^4); these two statements should be reconciled.","section":"Section 5, preprocessing discussion and Eq. (52)"},{"comment":"The assumption N_f = 2N_e = 2N is introduced for the big-O analysis but its effect on the constants and on the range of chemical systems covered is not discussed; it should be stated more prominently.","section":"Section 5.1"},{"comment":"There are typographical slips: 'Bravye-Kitaev' and 'Jorda-Wigner' should read 'Bravyi-Kitaev' and 'Jordan-Wigner'.","section":"Page 33, Section 5.1"},{"comment":"The electron-electron repulsion sums in Eqs. (3) and (5) run over all i,j without excluding i=j; the standard expression should use i<j (or include the explicit ordering), otherwise the divergent self-interaction terms are formally present.","section":"Section 2.2, Eq. (5)"},{"comment":"The commonly cited 'chemical accuracy' threshold of 1.6 mHa is attributed to the Pople Nobel lecture; a more specific reference for this convention would be helpful.","section":"Section 5.2"}],"recommendation":"major_revision","confidential_remarks":"This is a review/tutorial, so the absence of a new algorithm should not count against it if the journal publishes such pedagogical papers. My main concern is that Section 5 is likely to be used as a reference for VQE scaling, and it currently contains a per-evaluation/total-cost conflation, an unjustified treatment of the Trotter step count, and a measurement-scaling error. These are fixable in a thorough revision, but until then the headline complexity claims should not be relied upon."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a solid didactic review, not a research contribution. The complexity chapter is a careful assembly of known results: O(N^4) Hamiltonian and UCC/UCCG terms, average Jordan-Wigner Pauli weight N/2, per-term variance scaling, so P=O(N^5), Q=O(N^4/epsilon^2), and (I+P)Q=O(N^9/epsilon^2). Those scalings are standard; I don't see a new derivation or a new phenomenon here. But for a tutorial, it does its job.\n\nWhat the paper does well: the exposition from basis sets through second quantization to UCC variants is unusually complete for an introduction, and Section 5 is honest in ways many papers aren't. Eq. (44) defines total cost as X+(I+P)QK, and the text explicitly says K is not computed because it depends on the optimizer and the landscape. It also cites O'Gorman et al. for swap networks and mentions barren plateaus as a caveat. That is a fair frame.\n\nSoft spots: the stress-test note is right in one specific place. Eqs. (52)-(53) and the conclusion present O(N^9/epsilon^2) and O(kN^7/epsilon^2) as though they were the full 'complexidade total', when they are only (I+P)Q, the per-evaluation cost of the cost function. The body's wording right before Eq. (52) is actually careful—'cálculo da função custo total'—but the conclusion drops that caveat, which invites the misreading. That should be fixed. \n\nThe other assumptions—Trotter step count absorbed into the variational parameters, average weight N/2, and the per-Pauli-term measurement count—are stated but not rigorously justified. For a tutorial they are acceptable, since they are standard claims in the cited literature. If the authors wanted to claim an original derivation, more would be needed, but they don't really claim that.\n\nCitation pattern is fine. No code or data, but none is promised. The full text is in Portuguese with only an English abstract, which limits reach but not correctness.\n\nVerdict: as a research preprint it should not be accepted on novelty grounds. As a peer-reviewed teaching resource in a venue that publishes tutorials, it would deserve a fair referee. I would not cite it in my own work, but I might hand it to a student starting out in VQE.","headline":"A competent Portuguese-language tutorial on VQE for chemistry, not a research advance; the complexity scalings are standard, and the main caveat is that the quoted O(N^9/epsilon^2) and O(kN^7/epsilon^2) are per-evaluation costs, not the full VQE cost including optimizer iterations.","tokens_in":51567,"tokens_out":3465,"would_cite":false,"duration_ms":39351,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A full cost model for VQE chemistry yields scaling of N^9 over error squared.","keywords":["VQE","quantum chemistry","computational complexity","Jordan-Wigner mapping","unitary coupled cluster","Trotterization","measurement scaling","barren plateaus"],"falsifier":"Run a UCCSD VQE for a small molecule with increasing basis size (for example, water from STO-3G toward cc-pVTZ), fix chemical accuracy at 1.6 mHa, and record both the Trotter step count and the total circuit depth as functions of N; if depth grows faster than $N^{5}$ because the Trotter step count increases with N, the paper's central scaling estimate fails.","tokens_in":50549,"feed_emoji":"🧪","tokens_out":6254,"duration_ms":59858,"temperature":0.7,"pith_summary":"The paper is a self-contained introduction to using the Variational Quantum Eigensolver (VQE) to find molecular ground-state energies, and it ends by deriving the computational cost of evaluating the VQE cost function rather than benchmarking a new method. It claims that for a molecule described by N electrons (with 2N spin-orbitals) under the Jordan-Wigner mapping, a UCC or UCCG ansatz needs circuit depth O($N^{5}$), about O($N^{4}$/$epsilon^{2}$) measurements, and hence total time O($N^{9}$/$epsilon^{2}$) per cost-function evaluation. The k-UpCCG ansatz lowers this to O($kN^{3}$) depth and O($kN^{7}$/$epsilon^{2}$) total, where k is the number of repeated layers. A careful reader should care because these scalings say where VQE becomes impractical: not only on noisy devices but even with idealized hardware, unless the Pauli-term count or measurement number can be reduced.","feed_headline":"VQE chemistry cost hits N^9 over error squared","feed_subtitle":"A teaching paper derives how circuit depth and measurement shots grow with molecule size, and why near-term devices struggle.","key_machinery":"The machinery is the decomposition of the variational unitary U(theta) = exp(sum_j -i theta_j P_j / 2) into Trotter steps, combined with the replacement of each single Pauli exponential by a circuit whose depth is linear in the Pauli weight. The named objects are the Jordan-Wigner mapping, which assigns one qubit to each spin-orbital and enforces fermionic antisymmetry with chains of Z operators, and the ansatze UCC, UCCG, and k-UpCCG. The sequential-term count is treated as the number of Pauli terms because Jordan-Wigner strings have average weight N/2, so parallelization across disjoint qubits is rare; the Trotter step count l is absorbed into the optimizable angles theta_j, which removes l from the quoted depth.","core_discovery":"On the paper's own terms, the central claim is that the practical cost of VQE for chemistry can be computed from three pieces: the number of cluster amplitudes, the average Pauli weight of Jordan-Wigner strings, and the number of measurements needed to reach precision epsilon. Counting cluster amplitudes gives O($N^{4}$) Pauli terms for UCC and UCCG and O($kN^{2}$) for k-UpCCG; because Jordan-Wigner Pauli strings have average weight N/2 and different strings can rarely be parallelized, the circuit depth becomes O($N^{5}$) and O($kN^{3}$). Since the molecular Hamiltonian decomposes into O($N^{4}$) Pauli strings, the measurement count is O($N^{4}$/$epsilon^{2}$), and the per-step cost-function evaluation becomes O($N^{9}$/$epsilon^{2}$) for UCC/UCCG and O($kN^{7}$/$epsilon^{2}$) for k-UpCCG, with the optimizer's own step count left outside the estimate.","pith_inferences":["If the Trotter step count l must grow with system size or with 1/epsilon to maintain chemical accuracy, the depth scalings O(N^5) and O(kN^3) are optimistic; a direct count of l as a function of N for small molecules would test this.","The same counting logic, applied to low-Paulie-weight encodings, suggests that the depth factor could be reduced while the O(N^4/epsilon^2) measurement term remains, changing the quoted total cost.","The O(N^4/epsilon^2) measurement bound is a conservative per-string estimate; grouping Pauli terms into commuting families could replace the number of terms by the number of groups and lower the total cost in practice.","The paper's analysis covers one cost-function evaluation; the full VQE runtime is X + (I+P)QK, so any end-to-end complexity claim must also account for the classical optimizer cost, which the paper deliberately leaves empirical."],"forward_implications":["At fixed chemical accuracy, adding basis functions makes VQE dramatically more expensive: UCC and UCCG depth grows as N^5 and measurements as N^4, so the per-step cost grows as N^9.","The k-UpCCG ansatz is the only one of the three whose quoted depth, O(kN^3), might fit inside near-term coherence times for modest N, but its total cost O(kN^7/epsilon^2) still grows quickly enough to question a practical quantum advantage.","Measurement count, not just gate depth, is a bottleneck: O(N^4/epsilon^2) shots are needed before optimization even begins, so reducing the number of Pauli terms or grouping them into commuting families would improve the total scaling directly.","Because the optimizer's number of cost-function evaluations is left undetermined, the quoted costs are lower bounds on actual wall-clock time; barren-plateau landscapes can make that number large and also force higher measurement precision."],"supporting_citations":[{"why":"Supplies the original VQE hybrid algorithm and the premise that a short-depth parameterized circuit can estimate ground-state energy.","marker":"[14]"},{"why":"Defines the Jordan-Wigner mapping used to convert fermionic operators into Pauli strings, whose average weight drives the depth count.","marker":"[73]"},{"why":"Supplies the circuit decomposition showing that a Pauli exponential has circuit depth linear in its Pauli weight.","marker":"[84]"},{"why":"Supplies the Trotter product formula used to split the multi-term exponential into sequential single-term exponentials.","marker":"[85]"},{"why":"Supplies the Trotter-Suzuki approximants and the O(1/l) error that justifies absorbing l into the variational parameters.","marker":"[86]"},{"why":"Supplies the barren-plateau phenomenon, which the paper uses to argue that measurement precision and optimization difficulty grow with system size.","marker":"[75]"},{"why":"Introduced the k-UpCCG ansatz and its lower-depth O(kN^3) claim, which the paper compares against UCC and UCCG.","marker":"[81]"},{"why":"Provides the Hartree-Fock preprocessing complexity that the paper uses as the classical baseline for the overall VQE cost.","marker":"[82]"}],"fun_headline_variants":["VQE chemistry cost per step: O(N^9/eps^2)","Measurement budget for VQE chemistry: O(N^4/eps^2)","VQE circuit depth N^5, measurement shots N^4 for molecules","N^9 per-step cost challenges VQE for near-term chemistry","VQE chemistry scaling: depth N^5, measurements N^4/eps^2"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The quoted scalings assume that the number of Trotter steps can be absorbed into the variational parameters, so it does not grow with system size or required accuracy; if real chemical accuracy forces that number to grow, the depth and total-cost estimates are too low.","fun_headline_variants_meta":{"raw":{"variants":["VQE chemistry cost per step: O(N^9/eps^2)","Measurement budget for VQE chemistry: O(N^4/eps^2)","VQE circuit depth N^5, measurement shots N^4 for molecules","N^9 per-step cost challenges VQE for near-term chemistry","VQE chemistry scaling: depth N^5, measurements N^4/eps^2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001761,"raw_usage":{"total_tokens":6967,"prompt_tokens":981,"completion_tokens":5986,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":597,"completion_tokens_details":{"reasoning_tokens":5884}},"tokens_in":597,"tokens_out":5986,"duration_ms":39714,"temperature":1.0,"reasoning_tokens":5884,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:20:21.970601+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a UCCSD VQE for a small molecule with increasing basis size (for example, water from STO-3G toward cc-pVTZ), fix chemical accuracy at 1.6 mHa, and record both the Trotter step count and the total circuit depth as functions of N; if depth grows faster than $N^{5}$ because the Trotter step count increases with N, the paper's central scaling estimate fails.","supporting_citations":[{"cited_title":"Trapped- ion quantum computing","cited_arxiv_id":null,"evidence_quote":"Defines the Jordan-Wigner mapping used to convert fermionic operators into Pauli strings, whose average weight drives the depth count."},{"cited_title":"A mathematical and computational review of hartree–fock scf methods in quantum chemis- try.Molecular Physics, 105(23-24):3057–3098, 2007","cited_arxiv_id":null,"evidence_quote":"Supplies the circuit decomposition showing that a Pauli exponential has circuit depth linear in its Pauli weight."},{"cited_title":"A quantum computing view on unitary coupled cluster theory.Che- mical Society Reviews, 51(5):1659–1684, 2022","cited_arxiv_id":null,"evidence_quote":"Supplies the Trotter product formula used to split the multi-term exponential into sequential single-term exponentials."},{"cited_title":"The bravyi-kitaev transformation for quantum computation of electronic struc- ture.The Journal of chemical physics, 137(22), 2012","cited_arxiv_id":null,"evidence_quote":"Supplies the barren-plateau phenomenon, which the paper uses to argue that measurement precision and optimization difficulty grow with system size."},{"cited_title":"Some comments on the coupled cluster with generalized singles and doubles (ccgsd) an- satz.Chemical physics letters, 397(1-3):174– 179, 2004","cited_arxiv_id":null,"evidence_quote":"Introduced the k-UpCCG ansatz and its lower-depth O(kN^3) claim, which the paper compares against UCC and UCCG."},{"cited_title":"Generali- zed unitary coupled cluster wave functions for quantum computation.Journal of chemical theory and computation, 15(1):311–324, 2018","cited_arxiv_id":null,"evidence_quote":"Provides the Hartree-Fock preprocessing complexity that the paper uses as the classical baseline for the overall VQE cost."}],"review_version":1}