{"id":"5bff6fbc-ffc9-42c7-b0ac-7c33ff2a8f4f","arxiv_id":"2505.10963","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":3,"one_line_summary":"The paper adds imaginary (Im-uCJ) and complex (g-uCJ) orbital rotations to the unitary cluster Jastrow ansatz, implements them exactly without Trotterization, and shows they beat the real-valued Re-uCJ on small molecular benchmarks.","lead":"Scientists introduce two new wavefunction templates, Im-uCJ and g-uCJ, for quantum computers calculating molecular energies. Both use complex or imaginary rotations, give shallow exact circuits, and often reach chemical accuracy on small molecules.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"ESI Givens-decomposition angles are inconsistent with Eq. 16: with the stated φ_k = π + φ_pj − φ_qj the (q,j) entry of r'u is not zeroed for generic phases (explicitly fails for the real Hadamard matrix), so the Trotter-free exact-implementation lemma is invalid as written.","rationale":"The reader's qualitative verdict (CONDITIONAL) is retained, but the conditions must be expanded: the ESI's Givens-angle derivation, the proof of the exact Trotter-free implementation, is internally inconsistent with Eq. 16, and a reader implementing from the text would not obtain circuits realizing e^K. My direct computation with the real 2×2 Hadamard (where the ESI's φ = π leaves the off-diagonal entry at 1 instead of 0, while φ = 0 diagonalizes) settles the inconsistency analytically; the proposed numerical test would confirm it for generic complex u = exp(K). This is distinct from the reader's hypothesized degeneracy at r_pj = 0, which is a removable singularity that atan2 handles; the actual failure is generic and stems from a wrong phase formula. The underlying mathematical fact — any unitary is a product of two-level unitaries and diagonal phases — is true, and the corrected formula φ_k = φ_qj − φ_pj restores the claim, so this is a corrigible error rather than a fatal flaw: the ansatz ideas and the benchmark strategy remain sound, and the g-uCJ two-electron exactness and CSF analysis are plausible and well-presented. The reader's other points stand (no numeric data deposit, no commit hash, noiseless simulations only, 'practical feasibility' overstatement). Acceptance should require a corrected and numerically verified decomposition (and ideally release of the simulator code), in addition to the data deposit and claims tempering already requested. With those conditions the technical contribution is verifiable; without the fix, the central claim is unproven as written.","tokens_in":19702,"tokens_out":36367,"duration_ms":293782,"concrete_test":"Implement the decomposition exactly as stated and check diagonalization. Minimal decisive case: u = (1/√2)[[1,1],[1,−1]] with p = 1, q = 2, j = 1; the ESI formulas give φ_k = π + φ_11 − φ_21 = π and θ_k = arctan(−1) = −π/4, and r'_12(θ_k, φ_k)u as defined in Eq. 16 equals [[0,1],[1,0]] — the (2,1) entry is 1, not 0 — whereas φ_k = φ_21 − φ_11 = 0 gives r'u = diag(1,−1). Generic check: for a random 4×4 anti-Hermitian K, form u = exp(K) and apply the ESI procedure column-by-column; compute the Frobenius norm of the off-diagonal part of the product ∏ r'_k u. If it is not approximately zero, the published formulas are wrong as written and must be corrected or Eq. 16's sign convention clarified, and the circuits should be re-verified to realize e^K (e.g., process fidelity of the exact uCJ circuit against exp(K)).","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim of Section 2.2 — exact, Trotter-free O(N^2)-gate implementation of Im-uCJ and g-uCJ via Eq. 18 — rests entirely on the generalized Givens decomposition whose angle formulas are 'given explicitly in the ESI'. The derivation there is internally inconsistent with Eq. 16. With Eq. 16's r'_pq(θ, φ), left-multiplication gives (r'u)_qj = e^(iφ) sin θ r_pj e^(iφ_pj) + cos θ r_qj e^(iφ_qj). The ESI instead asserts the first term is sin θ r_pj e^(i(π+φ_pj−φ)), and states φ_k = π + φ_pj − φ_qj, θ_k = arctan(−r_qj/r_pj). These two descriptions agree only when φ_qj − φ_pj ≡ π/2 (mod π). Direct check: u = (1/√2)[[1,1],[1,−1]] with p = 1, q = 2, j = 1 (all phases zero); the ESI gives φ = π, θ = −π/4, and r'u = [[0,1],[1,0]], i.e., the (2,1) entry is 1, not 0. The correct phase is φ_k = φ_qj − φ_pj (equivalently θ_k = arctan(r_qj/r_pj) with φ_k = π + φ_qj − φ_pj); with φ = 0, the same rotation yields r'u = diag(1,−1). Thus the algorithm as published fails generically for complex u = exp(K) — not only at the removable singularity r_pj = 0, which atan2 handles — so the proof of the central technical claim is invalid as written. The underlying theorem (any unitary factors into two-level unitaries times a diagonal phase) is true, and a corrected sign convention would restore the proof; the plausible numerical results suggest the code uses the corrected convention. But a reader implementing from the text would build circuits that do not realize e^K. This is corrigible, but it is the load-bearing step of the paper's principal technical contribution, and it must be fixed, not assumed.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript introduces two new variants of the k-fold unitary cluster Jastrow (uCJ) ansatz for molecular VQE: Im-uCJ, which uses imaginary orbital-rotation operators, and g-uCJ, with fully complex rotations. The central technical claim is that both variants, like the earlier real-rotation Re-uCJ, can be implemented exactly (Trotter-free) with O(N^2) gates by decomposing the complex unitary exp(K) into generalized Givens rotations (Section 2.2). The authors benchmark k=1 versions on H2, H3+, Be2, C2H4, C2H6, and C6H6 against FCI or CASSI references, reporting that Im-uCJ and g-uCJ recover more correlation energy than Re-uCJ, frequently reaching chemical accuracy, and that g-uCJ exactly reproduces FCI for all two-electron systems tested. All numerical results are from noiseless classical simulations with SLSQP parameter optimization, and the paper openly reports known limitations, including the failure at stretched C2H4 and spin contamination in Im-uCJ.","tokens_in":20123,"tokens_out":31084,"duration_ms":273671,"significance":"If the claims hold, the paper is a useful contribution to the shallow-ansatz literature for NISQ VQE: it extends the uCJ family at no extra asymptotic gate cost, gives evidence that imaginary/complex rotations mitigate some of the local-minimum problems documented for Re-uCJ, and provides an honest CSF-level account of the spin-symmetry breaking introduced by Im-uCJ. Strengths include the open reporting of failure modes, the use of independent FCI/CASSI reference energies, the internal consistency of the benchmark data, and the availability of the simulation code. However, the headline methodological claim—the exact, Trotter-free Givens implementation—is not correct as written, and the gate-count tables contain internal inconsistencies; both problems are corrigible, but they must be fixed before the central construction can be verified or endorsed.","major_comments":[{"comment":"The ESI angle formulas are inconsistent with the Givens rotation r'(θ,φ) defined in Eq. (16), and the claim that they zero the (q,j) entry fails as written. From Eq. (16), left-multiplication gives (r'u)_{qj} = e^{iφ} sinθ r_{pj} e^{iφ_{pj}} + cosθ r_{qj} e^{iφ_{qj}}, whereas the ESI writes u'_{qj} = sinθ r_{pj} e^{i(π+φ_{pj}−φ)} + cosθ r_{qj} e^{iφ_{qj}}; the two expressions coincide only for φ ≡ π/2 (mod π). With the ESI values φ_k = π + φ_{pj} − φ_{qj} and θ_k = arctan(−r_{qj}/r_{pj}) applied to the real Hadamard matrix u = (1/√2)[[1,1],[1,−1]], which is a valid exp(K) for the Im-uCJ class, one obtains r'(θ,φ)u = [[0,1],[1,0]] for p=1, q=2, j=1, i.e., the element meant to be zeroed is 1, not 0. The correct zeroing condition for Eq. (16) is φ_k = π + φ_{qj} − φ_{pj} with θ_k = arctan(r_{qj}/r_{pj}) (equivalently φ_k = φ_{qj} − φ_{pj} with θ_k = −arctan(r_{qj}/r_{pj})). Because Eqs. (18) and the entire Trotter-free, O(N²)-gate claim rest on this decomposition, a reader implementing from the published text cannot construct exp(K) for general complex K; the derivation of the central Section 2.2 claim is therefore invalid as written. The underlying factorization of a unitary into two-level rotations and diagonal phases is standard and true, so this is corrigible; please correct the ESI or Eq. (16), state the phase convention for r_{ij} explicitly, handle r_{pj}=0, and confirm numerically that the code realizes exp(K) exactly.","section":"§2.2, Eq. (16); ESI, 'Discussion of the diagonalization of matrix u'"},{"comment":"The reported two-qubit gate counts are mutually inconsistent and cannot be reproduced from the stated counting rules. In §3.4, Eqs. (24)–(26) give N_{eK} = N_{e−K} = C(N,2)×3 and N_{eJ} = C(N,2)×2, which sum to C(N,2)×8, not the C(N,2)×5 printed in Eq. (27). For the 12-qubit H2 (6-311G) entry in Table 1 (192 CNOTs), the rules stated in §2.4 and §3.4 (spin-preserving K with 2·C(6,2) rotations per factor, and C(12,2) number-operator pairs for J) would give 180 + 132 = 312 CNOTs; the tabulated value appears to correspond to a reduced (e.g., perfect-pairing) number-operator set that is not specified in the text. Additionally, §3.1 states '192 vs. 202 two-qubit gates' for the 12-qubit H2 comparison, whereas Table 1 lists 1202 UCCSD CNOTs for that system. Please correct the arithmetic and state precisely the counting convention used for each Table 1 entry.","section":"§3.1, Table 1; §3.4, Eqs. (24)–(27)"}],"minor_comments":[{"comment":"The comparison '192 vs. 202 two-qubit gates' appears to be a typographical error for '192 vs. 1202', since Table 1 lists 1202 for UCCSD at H2 (6-311G); the claim of a roughly six-fold reduction is consistent with 1202/192 ≈ 6.3, but not with 202/192.","section":"§3.1"},{"comment":"The equation numbering skips (26), with the text jumping from (25) to (27); please renumber.","section":"§3.4"},{"comment":"Please state explicitly how Givens rotations between non-adjacent modes are ordered under the Jordan-Wigner mapping; the assertion that restricting rotations to adjacent qubits 'use[s] only local Givens rotations' is only valid together with the linear-network sweep construction from Ref. 68, which should be cited and specified so that a reader can see that an arbitrary u is still realized.","section":"§2.2 and §3.4"},{"comment":"The claim that Im-uCJ 'reliably converges to a single lower energy solution regardless of the initial guess procedure' is supported by only two initialization protocols (association and dissociation); reporting results from multiple random restarts would substantially strengthen this claim.","section":"§2.4 and §3.2"},{"comment":"The convention for r_{ij} (signed real values versus nonnegative magnitudes with phases in [0,2π)) should be stated explicitly, and the singular case r_{pj} = 0 should be addressed through atan2 or a stated pivot-selection rule.","section":"ESI"},{"comment":"The two-level decomposition of unitary matrices is standard in the quantum-optics and quantum-circuit literature (e.g., Reck et al. and Clements et al.); citing such work would help readers verify the corrected angle formulas and place the construction in context.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The Givens angle error is, in my assessment, a sign/convention slip rather than evidence of fabricated results: the benchmark energies are internally consistent, the failure modes are openly reported, and the code is available for inspection. Nevertheless, the exact-implementation proof is the paper's headline methodological claim, so I would not accept the manuscript without a corrected ESI, a consistent statement of the angle formulas, and a written confirmation that the simulations use the corrected convention. A referee-accessible code snapshot with a specific repository path and commit would aid verification. The contribution is incremental relative to the earlier Re-uCJ work of the same groups, but the expressibility comparison between the real, imaginary, and general variants is fair and useful; I view the paper as publishable after a major revision addressing the two load-bearing issues above."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take on arXiv:2505.10963. The numerical claims look plausible and the benchmarking is genuinely useful, but the paper's central technical lemma — the exact Trotter-free decomposition for complex orbital rotations — is wrong as written. The ESI angle formulas don't zero the target matrix element for generic complex u. Left-multiplying the generalized Givens rotation of Eq. 16 into u gives (r'u)_qj = e^{i phi} sin(theta) r_pj e^{i phi_pj} + cos(theta) r_qj e^{i phi_qj}. The ESI instead writes sin(theta) r_pj e^{i(pi+phi_pj-phi)} + cos(theta) r_qj e^{i phi_qj}, which matches only when phi_pj - phi_qj = pi/2 mod pi. With the stated phi_k = pi + phi_pj - phi_qj, the (q,j) entry does not vanish; the real Hadamard matrix with p=1,q=2,j=1 gives 1 instead of 0. So a reader implementing from the text will build circuits that don't realize e^K. This is the load-bearing step for the gate-count and Trotter-free claims.\n\nWhat the paper gets right: the Im-uCJ and g-uCJ variants are a sensible extension, the benchmarks on H2, H3+, Be2, C2H4, C2H6, and C6H6 are carefully done, and the CSF analysis is illuminating. The authors are honest about spin contamination and about the C2H4 failure at stretched geometry. The g-uCJ exactness for two-electron systems is a nice empirical result. The O(N^2) scaling is not new in spirit — that comes from the earlier uCJ literature — but the exact circuit for the complex case is genuinely new, provided the decomposition is fixed.\n\nSoft spots, in proportion: the math bug is the big one, and it is not minor because it invalidates the central proof as written. The underlying theorem is true and the fix is small (a sign convention error), but the paper must be corrected, not assumed. Minor issues: no numeric data deposited, only figures; all simulations are noiseless, so the claim that the results confirm 'practical feasibility' on near-term hardware overstates what was demonstrated. The energy data and code should be deposited for reproducibility anyway.\n\nRecommendation: this deserves peer review — the paper is important enough and the flaw is corrigible — but the referee must require a corrected ESI derivation, actual numerical data, and a tempered hardware claim. As is, I would not accept; with the decomposition fixed, it becomes a solid practical paper.","headline":"The numerical benchmarks are useful, but the paper's central exact-decomposition lemma is wrong as written and must be fixed before this can be accepted.","tokens_in":20814,"tokens_out":3578,"would_cite":false,"duration_ms":29637,"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":"Imaginary or complex orbital rotations make the unitary cluster Jastrow ansatz Trotter-free at $O(N^2)$ cost and more accurate than the real-only variant, often reaching chemical accuracy.","keywords":["unitary cluster Jastrow ansatz","variational quantum eigensolver","Trotter-free implementation","generalized Givens rotation","complex orbital rotation","correlation energy","near-term quantum hardware","quantum chemistry"],"falsifier":"Apply the paper's generalized-Givens sweep to a random complex unitary $u = \\exp(K)$ chosen with a zero pivot entry ($r_{pj} = 0$, where the angle formula $\\theta_k = \\arctan(-r_{qj}/r_{pj})$ is undefined): if the sweep does not diagonalize $u$, or if the reconstructed circuit deviates from the exact exponential beyond machine precision, the Trotter-free implementation claim is refuted. A complementary chemical test is to run g-uCJ with $k=1$ on a four-electron system past the point where the paper already reports failure—e.g., C$_2$H$_4$ beyond 1.7 Å—and check whether the error returns to chemical accuracy when $k$ is increased to 2.","tokens_in":19438,"feed_emoji":"⚛️","tokens_out":22210,"duration_ms":178858,"temperature":0.7,"pith_summary":"The paper extends the $k$-fold unitary cluster Jastrow (uCJ) family of wavefunction ansätze for quantum chemistry, replacing the real orbital-rotation operator of the earlier Re-uCJ variant with an imaginary one (Im-uCJ) or a fully complex one (g-uCJ). Its central claim is that already at the simplest level, $k=1$, both new variants can be implemented exactly—without any Trotter decomposition—using generalized Givens rotations at a cost of $O(N^2)$ two-qubit gates. In variational eigensolver benchmarks on H$_2$, H$_3^+$, Be$_2$, C$_2$H$_4$, C$_2$H$_6$, and C$_6$H$_6$, the new variants recover more correlation energy than Re-uCJ and frequently keep the error within chemical accuracy (1 kcal/mol), with g-uCJ reproducing the full configuration interaction energy for every two-electron system tested. A reader should care because these circuits are at once more accurate and far shallower than the standard UCCSD ansatz, which is exactly what noisy near-term quantum hardware needs.","feed_headline":"Complex orbital rotations reach chemical accuracy with far fewer gates","feed_subtitle":"Imaginary and complex rotations yield exact shallow circuits that beat the real-only uCJ and UCCSD on small molecules.","key_machinery":"The load-bearing object is the generalized complex Givens rotation $r'_{pq}(\\theta,\\varphi)$, a two-orbital unitary that acts as $\\cos\\theta$ on the diagonal pair and $\\pm e^{\\pm i\\varphi}\\sin\\theta$ on the off-diagonal pair, together with its fermionic counterpart $\\hat{R}'_{pq}(\\theta_k,\\varphi_k) = \\exp[A(\\hat a^\\dagger_p \\hat a_q - \\hat a^\\dagger_q \\hat a_p) + iB(\\hat a^\\dagger_p \\hat a_q + \\hat a^\\dagger_q \\hat a_p)]$. The paper's central technical claim is that any complex unitary matrix $u = \\exp(K)$ can be diagonalized by $\\binom{N}{2}$ such rotations with pivot angles $\\varphi_k = \\pi + \\varphi_{pj} - \\varphi_{qj}$ and $\\theta_k = \\arctan(-r_{qj}/r_{pj})$, exactly as real unitaries are diagonalized by ordinary Givens rotations. This machinery converts the orbital-rotation exponentials in Im-uCJ and g-uCJ from approximate Trotter products into exact sequences of local two-qubit gates, which is what pins the gate count at $O(N^2)$ and removes Trotter error from every subsequent accuracy benchmark.","core_discovery":"Stated in the paper's own terms, the discovery is that the exact, Trotter-free exponentiation previously available for real orbital rotations survives when the orbital-rotation matrix $K$ is allowed to be imaginary or complex. The mechanism is a generalized Givens rotation $R'_{pq}(\\theta_k,\\varphi_k)$ that can zero out entries of a complex unitary matrix $u = \\exp(K)$; with the angle choices $\\varphi_k = \\pi + \\varphi_{pj} - \\varphi_{qj}$ and $\\theta_k = \\arctan(-r_{qj}/r_{pj})$, any such $u$ is brought to diagonal form by $\\binom{N}{2}$ such rotations followed by single-qubit phase gates, so $e^{K}$ becomes an exact product of local two-qubit operations. With that machinery in place, the paper's numerical results for $k=1$ show that Im-uCJ and g-uCJ recover more correlation energy than Re-uCJ across a range of small molecules, that Im-uCJ avoids the initialization-dependent local minima that trap Re-uCJ, and that g-uCJ reaches the exact full-configuration-interaction energy for every two-electron system tested, all at substantially lower CNOT counts than one Trotter step of UCCSD. A wavefunction analysis shows why: Im-uCJ lowers the energy by mixing a triplet $M_s = 0$ configuration into the singlet, while g-uCJ keeps the correct spin and spatial symmetry and is exact for two-electron systems.","pith_inferences":["Because the pivot formula $\\theta_k = \\arctan(-r_{qj}/r_{pj})$ becomes singular at $r_{pj}=0$, a natural stress test the paper leaves unrun is to apply the decomposition to random complex unitaries with degenerate entries; a pivot-selection variant that stays regular there would make the exact circuit usable without assumptions.","The paper stops at $k=1$; its own C$_2$H$_4$ results suggest that $k=2$ would plausibly restore chemical accuracy across the full dissociation curve, which is a direct and untested follow-up.","Im-uCJ's persistent spin contamination suggests a cheap experimental monitor on real hardware: measure $\\langle S^2 \\rangle$ alongside the energy during VQE runs, and if spin purity matters more than energy, apply spin projection as post-processing—a route the paper does not explore.","The gate estimates assume all-to-all qubit connectivity, so on chips with limited connectivity the SWAP-routing overhead is untested and could partially erode the reported factor-of-six advantage over UCCSD."],"forward_implications":["$k=1$ Im-uCJ and g-uCJ circuits compile exactly from $O(N^2)$ generalized Givens rotations plus phase gates, eliminating Trotter error; for the 12-qubit H$_2$/6-311G case this is 192 CNOTs versus 1202 for one Trotter step of UCCSD.","Im-uCJ recovers more correlation energy than Re-uCJ at the same gate count and is far less sensitive to initialization, making it the recommended restricted variant when the number of variational parameters is a hard constraint.","g-uCJ is exact for every two-electron system tested—H$_2$ in several basis sets, Be$_2$ (2e,2o), C$_2$H$_6$ (2e,2o), and H$_3^+$—and holds chemical accuracy for four-electron C$_2$H$_4$ out to a C–C distance of 1.7 Å, beyond which $k>1$ is required because breaking the double bond into two triplet fragments needs quadruple excitations.","Because exact exponentiation lets one parameter set be reused across permuted broken-symmetry references, biradicaloid systems in non-orthogonal eigensolver calculations avoid re-optimizing $K$ and $J$ for each reference state.","The improved energy of Im-uCJ comes with a measurable cost: it mixes triplet $M_s=0$ character into the singlet wavefunction (nonzero $\\langle S^2 \\rangle$) all along the curve, unlike g-uCJ which preserves the pure singlet symmetry."],"supporting_citations":[{"why":"Introduces the k-fold uCJ ansatz that this work extends to imaginary and complex orbital rotations.","marker":"[63]"},{"why":"Introduced the real Re-uCJ variant and the NOQE framework; supplies the accuracy and gate-count baseline the new ansätze must beat.","marker":"[55]"},{"why":"Supplies the Givens-rotation decomposition of real orbital rotations that the paper generalizes to complex matrices u = exp(K).","marker":"[68]"},{"why":"Provides the O(N^4) gate-scaling analysis for UCCSD used as the comparison baseline for circuit depth.","marker":"[67]"},{"why":"Generates the RHF references and integrals for all molecular benchmarks.","marker":"[72]"},{"why":"The optimizer whose converged parameters underlie the reported correlation-energy results.","marker":"[73]"}],"fun_headline_variants":["Complex rotations make uCJ exact, Trotter-free, and more accurate","Im-uCJ and g-uCJ beat real uCJ with exact shallow circuits","Complex orbital rotations: fewer gates, chemical accuracy","Trotter-free complex rotations: better than UCCSD on small molecules"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the lemma that every complex unitary orbital-rotation matrix can be decomposed exactly into $\\binom{N}{2}$ pairwise Givens rotations using the paper's angle recipe; if that decomposition fails or degenerates for some matrix, the Trotter-free gate-count claim collapses, though a Trotterized version of the ansatz would still work. A softer reliance is that the classical optimizer reliably finds the global minimum for the new ansätze, something the paper's own Re-uCJ results show can go wrong for the real-valued variant.","fun_headline_variants_meta":{"raw":{"variants":["Complex rotations make uCJ exact, Trotter-free, and more accurate","Im-uCJ and g-uCJ beat real uCJ with exact shallow circuits","Complex orbital rotations: fewer gates, chemical accuracy","Trotter-free complex rotations: better than UCCSD on small molecules"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000265,"raw_usage":{"total_tokens":1742,"prompt_tokens":1214,"completion_tokens":528,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":830,"completion_tokens_details":{"reasoning_tokens":448}},"tokens_in":830,"tokens_out":528,"duration_ms":4953,"temperature":1.0,"reasoning_tokens":448,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:01:59.516910+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Apply the paper's generalized-Givens sweep to a random complex unitary $u = \\exp(K)$ chosen with a zero pivot entry ($r_{pj} = 0$, where the angle formula $\\theta_k = \\arctan(-r_{qj}/r_{pj})$ is undefined): if the sweep does not diagonalize $u$, or if the reconstructed circuit deviates from the exact exponential beyond machine precision, the Trotter-free implementation claim is refuted. A complementary chemical test is to run g-uCJ with $k=1$ on a four-electron system past the point where the paper already reports failure—e.g., C$_2$H$_4$ beyond 1.7 Å—and check whether the error returns to chemical accuracy when $k$ is increased to 2.","supporting_citations":[],"review_version":1}