{"id":"56d4f4e9-423b-488b-92ec-f3a602f5e641","arxiv_id":"2507.19270","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Finite-field UCC2 and UCC3 are implemented and shown to give real, accurate energies for molecules in strong magnetic fields, including cases where standard coupled-cluster gives complex energies.","lead":"This paper implements two approximate unitary coupled-cluster methods for molecules in strong magnetic fields, where standard coupled-cluster theory can produce unphysical complex energies. The new calculations yield real energies and match standard methods in accuracy, which matters for interpreting spectra of magnetic white dwarfs.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"UCCn truncation order is assumed transferable to finite-field Hamiltonians without a convergence check; this is the main unvalidated assumption.","rationale":"The paper's central claim is that ff-UCC3 yields real energies with accuracy comparable to ff-CCSD for ground and singly excited states. The support for this is substantial: field-free validation against the original UCCn authors, finite-field benchmarks against CCSDT for CH+ and boric acid, and the Hermitian construction that guarantees real eigenvalues. The weakest link is the definition of the UCCn expansion itself. Section II.D fixes the perturbative orders of sigma1 and sigma2 based on the MP ordering of the field-free Hamiltonian. In the strong-field mixing regime, the field is not a weak perturbation; it is incorporated into the zeroth-order Fock operator, and the two-electron fluctuation potential may not be small relative to that Fock operator. If the perturbation series is not converging, the difference between UCC3 and full UCCSD could be large, and the observed closeness to CCSD would be fortuitous rather than systematic. The paper reports no UCC4 or UCCSD calculations, so this cannot be ruled out. The boric acid results (Table VI) additionally show UCC3 errors up to about 0.025 Eh for some singly excited states, larger than the CCSD errors of about 0.007 Eh, which modestly tempers the 'comparable' characterization, but this is not the primary issue. A single full-UCCSD calculation on CH+ at high field would settle whether the UCCn perturbative truncation is the limiting error. Since the reader's conditional verdict already reflects this gap, the verdict remains unchanged.","tokens_in":25857,"tokens_out":7449,"duration_ms":72788,"concrete_test":"Carry out a full UCCSD calculation (Bernoulli/BCH series iterated to convergence) for CH+ at B=0.5 B0 and B=1.0 B0 using the same unc-cc-pVDZ basis and geometry, and compare ground- and excited-state energies with UCC2, UCC3, CCSD, and CCSDT. If the UCC3-UCCSD differences exceed the CCSD-CCSDT differences, the perturbative UCCn truncation is not the dominant source of error and the comparable-accuracy claim is less informative; if UCC3 is close to UCCSD, the truncation concern is empirically resolved for this system.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing assumption is the transfer of the UCCn perturbative ordering (sigma2 first order, sigma1 second order) from the field-free Moller-Plesset partitioning to the finite-field Hamiltonian. Section II.D fixes this ordering without addressing how the magnetic field modifies the perturbation series. The field terms enter the one-electron part of the Hamiltonian (Eq. 1) and are thus included in the Fock operator; the fluctuation potential V is not necessarily small relative to F in the mixing regime where magnetic and Coulomb interactions are comparable. The truncation of the Bernoulli/BCH expansion at third order in the amplitude equations may therefore be uncontrolled. Since no UCC4 or full-UCCSD results are reported, there is no evidence that UCC3 has converged with respect to the commutator/perturbation expansion. This matters because the headline comparison 'UCC3 vs CCSD' cannot distinguish an intrinsically accurate UCC3 from one that is accurate for the wrong reasons: the CCSD comparison alone does not probe the UCCn truncation error. The reader's concern is well-founded.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents an implementation of unitary coupled-cluster (UCC) theory at second and third order (UCC2 and UCC3) for molecules in strong magnetic fields, based on the Bernoulli-expansion UCCn formalism of Liu et al. The implementation uses complex algebra and gauge-including atomic orbitals, and solves ground- and excited-state (EOM-UCC) equations. The authors validate the field-free implementation against reference data provided by the authors of ref. 71, and test Hermiticity and symmetry in the finite-field case. Benchmarks for CH+, H2O, and B(OH)3 show that UCC3 yields real energies in all cases, with mean errors relative to CCSDT comparable to those of CCSD for singly excited states, while UCC2 fails for states with double-excitation character. The boric acid example demonstrates that UCC3 gives real degenerate excited states for a complex Abelian point group, where EOM-CC requires complex algebra. The paper concludes that UCC3 is a practical Hermitian alternative to conventional CC in strong magnetic fields.","tokens_in":26028,"tokens_out":6736,"duration_ms":66157,"significance":"If the reported accuracy holds, this work provides a valuable Hermitian alternative to standard coupled-cluster theory for molecules in strong magnetic fields, where non-Hermitian CC often produces unphysical complex energies. The implementation is validated against external reference data and against CCSDT benchmarks, and the paper provides a diagrammatic derivation scheme that is a useful methodological contribution. The results are parameter-free and the data are made available in the supplementary material. The main unvalidated assumption is the transferability of the field-free perturbative truncation ordering to the finite-field Hamiltonian, which is not probed by a convergence study with respect to the UCCn truncation order.","major_comments":[{"comment":"The UCCn truncation scheme orders the sigma1 and sigma2 amplitudes according to field-free Moller-Plesset perturbation theory (sigma2 first order, sigma1 second order), and this ordering is used without modification for the finite-field Hamiltonian in Eq. (1). In the mixing regime, the field-dependent terms enter the Fock operator, so the relative magnitudes of the 'zero-order' and 'first-order' amplitudes may change, and the Bernoulli/BCH expansion truncated at third order may become uncontrolled. Since no UCC4 or full UCCSD results are reported, the agreement between UCC3 and CCSD cannot separate the intrinsic accuracy of the method from the truncation error of the UCCn expansion. Please add a UCCSD (or at least UCC4) calculation for CH+ at a few representative field strengths and orientations, or otherwise demonstrate that the UCCn series is converged at the UCC3 level.","section":"Section II.D"},{"comment":"The UCC3 amplitude equations are not presented in the paper; the authors refer to ref. 71 but also state that eqs. 64 and 65 of that reference are missing terms. Since the published equations are evidently unreliable, the complete corrected UCC3 equations should be given in the manuscript or in the supplementary material so that the implementation is fully reproducible. The diagrammatic rules and examples in Sec. III and the appendix are helpful, but they do not substitute for the final working equations, especially because the authors had to supply corrections to the literature.","section":"Section II.D and footnote 1"}],"minor_comments":[{"comment":"The sentence 'The magnetic field strength is varied up to 0.8 B0, in steps of 0.5 B0' appears to contain a typo; the step size should likely be 0.05 B0, as is used in similar scans in the paper.","section":"Section VI.D"},{"comment":"There is a typo in the concluding section: 'Boric acid invstigated' should read 'Boric acid was investigated'.","section":"Section VII"},{"comment":"The phrase 'within a strong magnetic field of of B=0.5 B0' contains a duplicated 'of'.","section":"Section VI.C"},{"comment":"The energy expression in Eq. (17) is not explicitly labeled as the correlation energy; please clarify whether it refers to the total energy or the correlation energy, given that the water analysis plots the correlation energy for the ground state.","section":"Section II.D, Eq. (17)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and the central claim is supported by the provided benchmarks. The requested convergence study (e.g., UCCSD or UCC4 on CH+) is a modest additional calculation and should be feasible. I also recommend that the authors coordinate with the authors of ref. 71 regarding the reported corrections to eqs. 64 and 65, and consider presenting the corrected UCC3 equations in the paper or supplementary material."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this paper delivers the first finite-field implementation of UCC2 and UCC3 in a conventional quantum chemistry code, and the central claim holds up—UCC3 gives real energies with accuracy comparable to CCSD for ground and singly excited states. It is worth a serious referee.\n\nWhat is new: the UCCn framework is Liu et al.'s, but adapting it to the finite-field regime in QCUMBRE is non-routine, with complex algebra and GIAOs. The validation is credible: field-free comparison to the original UCCn authors' data, symmetry checks, and CCSDT benchmarks for CH+ and boric acid. The water and boric acid results make a practical point—the imaginary parts of CCSD/CC3 energies can reach the mEh range in excited states, and a Hermitian ansatz removes them by construction. The boric acid case, with its complex Abelian point group, is genuinely useful: UCC3 gives the degenerate E' and E'' pairs with real algebra where EOM-CC needs a complex code.\n\nSoft spots, in proportion. The main one is exactly what the stress test flags: the UCCn perturbative ordering (sigma2 first order, sigma1 second order) is carried over from the field-free Moller-Plesset partitioning, and in the mixing regime the magnetic field is not a small perturbation. Section II.D fixes the truncation without a convergence study—no UCC4, no full UCCSD in the finite-field regime—so the accuracy claim rests on agreement with CCSD/CCSDT rather than on evidence that the UCCn expansion itself has converged. That is a missing check, not a demonstrated failure. The UCC3 equations are not derived in the paper; they are cited from ref. 71 with a note about missing terms in eqs. 64 and 65. Acceptable, but it leaves a verification gap. The code is not pinned to a version, which makes reproduction harder, though the data tables are there.\n\nThe paper does not oversell. The authors state plainly that UCC2 fails for doubly excited states, and they only claim UCC3 is comparable to CCSD, not superior. That restraint adds credibility.\n\nWho this is for: people working on molecules in strong magnetic fields, white-dwarf spectra, and anyone needing a Hermitian alternative to CC near conical intersections or for complex Abelian symmetries. My recommendation: send it to peer review, with a request that the authors either add a truncation-order convergence check in the finite-field regime or explicitly justify why the field-free ordering should transfer. That is an addressable revision, not a reason to reject.","headline":"First finite-field UCC2/UCC3 implementation, with real energies at CCSD-like accuracy; main gap is an unverified truncation-order transfer to the mixing regime.","tokens_in":26552,"tokens_out":1665,"would_cite":true,"duration_ms":15007,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["31.15.Ar","31.15.-p"],"model":"deepseek-v4-flash","headline":"This paper implements finite-field UCC2 and UCC3 and shows that UCC3 returns real-valued molecular energies in strong magnetic fields with accuracy comparable to standard CCSD, whereas standard coupled-cluster can produce unphysical…","keywords":["unitary coupled-cluster","finite magnetic fields","complex energies","excited states","white dwarf atmospheres","Bernoulli expansion","EOM-UCC","real-valued energies"],"falsifier":"Compute full untruncated UCCSD (all commutator orders) for CH+ at $B=0.5\\,B_0$ or for water at $B=0.5\\,B_0$ at orientations with large CCSD imaginary parts, and compare with ff-UCC3 and ff-UCC2. If UCC3's difference from full UCCSD is not smaller than its difference from CCSDT, or if the two truncated methods move away from full UCCSD as the field increases, the perturbative truncation is the weak link.","tokens_in":25657,"feed_emoji":"🧲","tokens_out":12338,"duration_ms":104494,"temperature":0.7,"pith_summary":"Standard coupled-cluster theory can return unphysical complex energies when a molecule is placed in a strong magnetic field, because its energy expression is not Hermitian. This paper argues that unitary coupled-cluster (UCC) removes the problem by construction: the energy is the expectation value of a Hermitian operator, so eigenvalues are real. The authors implement finite-field UCC2 and UCC3 in the QCUMBRE program package and test them on the methylidyne ion, water, and boric acid. They find that UCC3 is comparable in accuracy to standard CCSD for ground states and singly excited states, and that only UCC3 resolves the complex-energy problems, including degenerate excited states in the complex Abelian point group $C_{3h}$. UCC2, by contrast, fails for states with double-excitation character. This matters for interpreting spectra of magnetic white dwarfs, where fields too strong for the laboratory require accurate theoretical predictions.","feed_headline":"UCC3 delivers real energies for molecules in strong magnetic fields","feed_subtitle":"Finite-field UCC3 matches CCSD accuracy for ground and singly excited states, without complex energies.","key_machinery":"The load-bearing object is the Bernoulli-expansion truncation scheme called UCCn. In UCC the transformed Hamiltonian $\\bar H = e^{-\\tilde\\sigma}\\hat H e^{\\tilde\\sigma}$ does not terminate, so the paper adopts an ordering in which $\\sigma_2$ amplitudes enter at first order and $\\sigma_1$ at second order, keeping terms through order $n$ in the amplitude equations (UCC2 uses $H_{SS}^{(2)}, H_{SD}^{(1)}, H_{DS}^{(1)}, H_{DD}^{(0)}$; UCC3 uses $H_{SS}^{(3)}, H_{SD}^{(2)}, H_{DS}^{(2)}, H_{DD}^{(1)}$). This fixes the cost at $\\sim N^5$ for UCC2 and $\\sim N^6$ for UCC3, like CC2 and CCSD respectively. For excited states, the paper uses the ansatz $|\\Psi_k\\rangle = e^{\\tilde\\sigma}\\hat R|0\\rangle$, which satisfies the killer condition and turns the excited-state problem into the Hermitian eigenvalue problem $\\bar H \\hat R|0\\rangle = E_k \\hat R|0\\rangle$. The finite-field Hamiltonian includes the orbital and spin Zeeman terms plus the diamagnetic term; gauge-including atomic orbitals are used for gauge-origin independence, and the implementation uses complex algebra except where Hermiticity restores a real representation.","core_discovery":"The central claim is that the unitary parameterization $|\\Psi\\rangle = e^{\\tilde\\sigma}|0\\rangle$, with $\\tilde\\sigma = \\hat\\sigma - \\hat\\sigma^\\dagger$, turns the finite-field electronic-structure problem into a Hermitian one, so every computed energy is real even when the field breaks the symmetries that keep standard CC energies real. Working at the UCC2 and UCC3 truncation levels of the Bernoulli-expansion scheme, the paper reports three concrete results. For CH+ in fields up to $1\\,B_0$ at several orientations, UCC3 reproduces the CCSDT curves with mean deviations near 0.5-3 milli-Hartree for the ground and singly excited states, comparable to CCSD, while UCC2 deviates by 18-100 milli-Hartree for states with double-excitation character. For water at $B=0.5\\,B_0$ with the field pointing in arbitrary directions, UCC3 energies match the real parts of CCSD to within about 15 milli-Hartree even where the CCSD imaginary parts reach hundreds of $\\mu E_h$ for excited states; the imaginary parts carry no diagnostic signal. For boric acid, whose $C_{3h}$ group has complex-conjugate irreducible representations, UCC3 returns the $E'$ and $E''$ excited states as real degenerate pairs, while EOM-CC only finds complex-conjugate pairs, and the real parts of the two methods agree with CCSDT to similar order.","pith_inferences":["A natural follow-up the paper leaves open: use ff-UCC3 as the reference to benchmark the real parts of cheaper CC methods in the mixing regime; the water results suggest the real part of CCSD can be trusted even when the imaginary part is large, but only a Hermitian calculation can certify that.","The UCCn perturbative ordering could break down for molecules with stronger static correlation; trying commutator-rank truncation schemes in a finite field would reveal whether the field-free ordering is the limiting assumption.","The boric acid result extends by symmetry: any molecule whose field-free point group has complex irreducible representations should be treatable by real-algebra UCC3 for degenerate excited states, a class that includes many cyclic and helical molecules.","If ff-UCC3 is combined with larger uncontracted basis sets and gauge-including atomic orbitals, it may produce reference-quality transition energies for magnetic white-dwarf molecules beyond the reach of CCSDT, but the paper does not demonstrate that scaling."],"forward_implications":["ff-UCC3 provides real-valued electronic energies for molecules in strong magnetic fields with accuracy comparable to ff-CCSD, so spectral predictions for white-dwarf atmospheres no longer require interpreting or discarding imaginary CC energies.","EOM-UCC2 should be restricted to singly excited states; for states with double-excitation character it can miss the state entirely or produce errors an order of magnitude larger than UCC3.","For molecules with complex Abelian point-group symmetry, a single UCC3 calculation gives real degenerate excited states where EOM-CC needs complex algebra and returns complex-conjugate pairs.","The imaginary part of ff-CCSD and ff-CC3 energies is not a reliable accuracy diagnostic; its size does not correlate with the difference between CC and UCC3 real energies.","Because the Hermitian construction is what removes complex eigenvalues, UCC formulations are a natural fix for other sources of CC complex energies, such as conical intersections, not just magnetic fields."],"supporting_citations":[{"why":"Defines the UCCn Bernoulli-expansion formalism and the UCC2/UCC3 equations that this paper adapts to finite magnetic fields.","marker":"[71]"},{"why":"Documents complex ground- and excited-state energies in CC theory for water in a magnetic field and supplies the benchmark problem.","marker":"[54]"},{"why":"Provides the finite-field CCSD and CCSDT reference energies for CH+ that the UCC methods are compared against.","marker":"[7]"},{"why":"Establishes the finite-field CC2 and CC3 implementations whose behavior in double-excitation regions frames the UCC2/UCC3 comparison.","marker":"[52]"},{"why":"Supplies finite-field CC methodology and the boric acid CCSD, CC3, and CISD reference data used for the complex-Abelian-point-group tests.","marker":"[78]"},{"why":"Motivates the problem by analyzing why coupled-cluster theory in strong magnetic fields produces complex energies.","marker":"[4]"},{"why":"Earlier theoretical extension of UCC to strong magnetic fields in a quantum-computing context, which the present conventional implementation builds on.","marker":"[65]"},{"why":"Identifies the program package in which the finite-field UCC2/UCC3 implementation lives.","marker":"[82]"}],"fun_headline_variants":["UCC3 keeps CC energies real in strong magnetic fields","Hermitian UCC3 fixes complex energies in strong fields","Strong-field molecules get real energies from UCC3","UCC3 matches CCSD accuracy without complex energies","Real molecular energies under strong fields via UCC3"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The ordering of single- and double-excitation amplitudes is imported from field-free perturbation theory; the paper assumes a strong magnetic field does not change that ordering and provides no next-order convergence check.","fun_headline_variants_meta":{"raw":{"variants":["UCC3 keeps CC energies real in strong magnetic fields","Hermitian UCC3 fixes complex energies in strong fields","Strong-field molecules get real energies from UCC3","UCC3 matches CCSD accuracy without complex energies","Real molecular energies under strong fields via UCC3"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001112,"raw_usage":{"total_tokens":4648,"prompt_tokens":980,"completion_tokens":3668,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":596,"completion_tokens_details":{"reasoning_tokens":3590}},"tokens_in":596,"tokens_out":3668,"duration_ms":22902,"temperature":1.0,"reasoning_tokens":3590,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:55:26.238823+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute full untruncated UCCSD (all commutator orders) for CH+ at $B=0.5\\,B_0$ or for water at $B=0.5\\,B_0$ at orientations with large CCSD imaginary parts, and compare with ff-UCC3 and ff-UCC2. If UCC3's difference from full UCCSD is not smaller than its difference from CCSDT, or if the two truncated methods move away from full UCCSD as the field increases, the perturbative truncation is the weak link.","supporting_citations":[{"cited_title":"\\ Kitsaras ,\\ title Finite magnetic-field coupled-cluster methods: efficiency and utilities ,\\ 10.25358/OPENSCIENCE-9599 Ph.D","cited_arxiv_id":null,"evidence_quote":"Supplies finite-field CC methodology and the boric acid CCSD, CC3, and CISD reference data used for the complex-Abelian-point-group tests."},{"cited_title":"Stopkowicz ,\\ title title Perspective: Coupled cluster theory for atoms and molecules in strong magnetic fields , \\ 10.1002/qua.25391 journal journal Int","cited_arxiv_id":null,"evidence_quote":"Motivates the problem by analyzing why coupled-cluster theory in strong magnetic fields produces complex energies."},{"cited_title":"Culpitt , author E","cited_arxiv_id":null,"evidence_quote":"Earlier theoretical extension of UCC to strong magnetic fields in a quantum-computing context, which the present conventional implementation builds on."}],"review_version":1}