{"id":"6966b153-9b6c-4cb3-ac83-5af36dee8837","arxiv_id":"1908.05501","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A spin-conserving natural orbital functional for multiplets is introduced and shown to reproduce transition-metal ionization potentials with a mean absolute error of 6.5 kcal/mol.","lead":"This paper proposes a new way to describe electrons in atoms and molecules with any total spin, using a natural orbital functional built from an equal mixture of all spin states in a multiplet. It tests the method on ionization energies of the first-row transition-metal atoms, obtaining values close to high-level coupled cluster and experiment.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The (2,2)-positivity claim for the multiplet 2RDM is asserted without proof; a concrete counterexample would invalidate the central N-representability guarantee.","rationale":"The paper proposes a natural orbital functional for spin multiplets based on a 2RDM reconstruction. The derivation of the expectation value <S^2> is explicit and correct, and the transition-metal ionization potentials provide a genuine application. The most load-bearing unproven assumption is the (2,2)-positivity of the reconstructed 2RDM, because the energy functional is only meaningful if the underlying 2RDM is N-representable at least to the level of D, Q, G >= 0. The reader correctly flags this as the weakest point. I agree: the spin calculation is not evidence of positivity, and the new inter-subspace terms in Eq. (21) introduce matrix elements that have not been checked against any positivity condition. The multiple-solution issue in Section IV is secondary; it affects the specific IP numbers but does not threaten the core construction. A concrete numerical test of positivity for small model systems would either confirm or refute the claim. Until then, conditional acceptance is appropriate. No evidence of internal inconsistency or dishonesty was found; the absence of proof is the issue.","tokens_in":10357,"tokens_out":16658,"duration_ms":150697,"concrete_test":"For the minimal cases N=2 (NI=2, NII=0, S=1) and N=4 (NI=2, NII=2, S=1), implement Eqs. (19)-(21) with a few spatial orbitals and a small basis (e.g., STO-3G). Grid over occupation numbers satisfying n_p=1/2 for p in Omega_I and sum_{p in Omega_g} n_p=1 for each Omega_g in Omega_II, plus 0<=n_p<=1. For each point, build the full 2RDM D, the two-hole matrix Q, and particle-hole matrix G using standard mappings from D, and diagonalize. If the minimum eigenvalue of any of these matrices is negative for any valid grid point, the (2,2)-positivity claim is refuted; otherwise the claim is supported for these representative cases.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract and Section I assert that the reconstructed 2RDM in Eqs. (19)-(21) fulfills the (2,2)-positivity conditions (D, Q, G >= 0). Section II verifies only the total-spin expectation value (Eqs. 22-26); no derivation, reference, or numerical check supports positivity for the new inter-subspace alpha-beta terms involving Phi_p = sqrt(n_p(1-n_p)). These terms differ from the singlet PNOF7 and are essential to the multiplet energy (Eq. 29). Because the reconstructed 2RDM is not the exact ensemble average of the pure multiplet components (e.g., the alpha-beta trace for N=2, S=1 would be 1/3 in the exact multiplet, whereas Eq. (21) gives 1/4), N-representability is not automatic. If any of D, Q, or G has a negative eigenvalue for admissible occupation numbers, the central claim fails, the energy functional may be non-variational, and the NOF-MP2 reference from Section III loses its foundation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a natural orbital functional (PNOF7) for spin multiplets of maximum spin multiplicity. The author reconstructs the ensemble two-particle reduced density matrix (2RDM) from natural occupation numbers by splitting the orbital space into singly occupied orbitals (Ω_I) and singlet-coupled pairs (Ω_II), with new inter-subspace alpha-beta contributions proportional to Φ_p = sqrt(n_p(1-n_p)). The paper shows that this reconstruction yields the ensemble expectation value <S^2> = S(S+1), asserts that it satisfies the (2,2)-positivity conditions, and extends the NOF-MP2 method to multiplets. Ionization potentials of the first-row transition-metal atoms Sc–Zn are computed with a cc-pVTZ basis and compared with CCSD(T) and experiment, reporting a mean absolute error of 6.5 kcal/mol for NOF-MP2.","tokens_in":10595,"tokens_out":11623,"duration_ms":108912,"significance":"If correct, this is a useful step toward spin-adapted natural orbital functionals: the multiplet ensemble treatment differs from earlier NOFs that target only the high-spin component or break spin symmetry, and the NOF-MP2 extension provides a practical way to add dynamic correlation. The <S^2> derivation in Eqs. (22)–(26) is explicit and appears internally consistent, and the transition-metal IP benchmark is a genuine parameter-free prediction rather than a fit. However, the central (2,2)-positivity claim is asserted without proof, and the IP benchmark contains a multiple-solution ambiguity; both issues need to be resolved before the central claims can be accepted.","major_comments":[{"comment":"The claim that the reconstructed 2RDM in Eqs. (19)–(21) satisfies the (2,2)-positivity conditions (D, Q, G ≥ 0) is stated in the abstract and Section I, but it is nowhere proved or numerically tested. Section II verifies only the ensemble expectation value of S^2 (Eqs. (22)–(26)); no argument is given for positivity of the new inter-subspace alpha-beta terms in Eq. (21), which differ from the singlet PNOF7 reconstruction and are essential to the multiplet energy in Eq. (29). Because the reconstructed 2RDM is not the exact ensemble average of the pure multiplet components, positivity is not inherited from the pure-state 2RDMs and must be demonstrated directly. This is load-bearing: without D, Q, G ≥ 0, the functional may be non-variational and the NOF-MP2 reference built on it lacks a guaranteed N-representable 2RDM.","section":"Abstract and Section II, Eqs. (19)–(26)"},{"comment":"The NOF-MP2 benchmark is not fully determined. For the neutral atoms Co, Ni, Cu and for the Co and Ni ions, the calculation yields two solutions, and the energetically preferred solution changes between the PNOF7 and NOF-MP2 levels. The paper reports one set of PNOF7-MP2 IPs but does not specify an a priori criterion for selecting between the two stationary points. Consequently, the reported MAE of 6.5 kcal/mol is not attributable to a single well-defined method unless a selection rule is stated, such as always choosing the lowest NOF-MP2 root after solving the PNOF7 equations.","section":"Section IV, text after Fig. 2 and Table I"}],"minor_comments":[{"comment":"The text states that the MAE with respect to experiment is 7.9 and 7.6 kcal/mol for PNOF7 and CCSD(T), respectively, but Table I reports a CCSD(T) MAE of 7.3 kcal/mol; the text and table should be reconciled.","section":"Section IV, text and Table I"},{"comment":"The abstract says that the obtained values agree with CCSD(T) and experiment, but the PNOF7 IPs for Ni and Zn deviate from CCSD(T) by about 23 and 19 kcal/mol, respectively; the agreement statement should be qualified to the NOF-MP2 results or to the overall MAE.","section":"Abstract and Section IV"},{"comment":"The sentence 'Cter_p is not considered if the orbital is below NΩ' is unclear; the intended domain of Cter_p should be stated explicitly, since Eq. (34) appears to define it for p > NΩ only.","section":"Eq. (34)"},{"comment":"The caption describes the arrows as 'alpha (↓) or beta (↑)', which is the reverse of the usual spin convention; please clarify or correct the labeling.","section":"Figure 1 caption"},{"comment":"There are several typographical errors in the affiliations and abstract ('Donos tia', 'E uskadi', 'an y', 'tw o-particle') and the reference [20] author name appears as 'Paunez' rather than 'Pauncz'.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is a methods contribution from an established author and relies significantly on the author's prior PNOF machinery, which is acceptable for this type of work. The main risk is the unproven (2,2)-positivity assertion, which is central to the variational claim; the multiple-solution issue in the benchmark also needs a clear protocol before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper is a real step forward for NOF theory: it constructs a mixed-state 2RDM for a multiplet of maximum spin multiplicity, allowing PNOF7 to handle any total spin instead of just the high-spin component. That is genuinely new relative to refs 11-14. The orbital-space split into Omega_I for singly occupied orbitals and Omega_II for singlet-coupled pairs is clean, and the <S^2> algebra in Eqs. (22)-(26) checks out. The IP benchmark is also a genuine prediction: no parameters are fitted to the target IPs, and the NOF-MP2 MAE of 6.5 kcal/mol with cc-pVTZ is respectable for transition metals.\n\nThe main soft spot is exactly what the reader flagged. The abstract and Section I claim the reconstructed 2RDM satisfies the (2,2)-positivity conditions (D, Q, G >= 0), but Section II only verifies the total spin expectation value. The new inter-subspace alpha-beta terms in Eq. (21), with Phi_p = sqrt(n_p(1-n_p)), differ from the singlet PNOF7 and are essential to the multiplet energy, so N-representability is not inherited automatically. I don't have a counterexample, and the author may well be right, but a proof or at least a numerical check of D, Q, G positive semidefiniteness should be required before the central claim is accepted.\n\nA second issue, also real, is the multiple solutions. For Co, Ni, Cu neutrals and Co/Ni ions the paper reports two solutions, one lower at the PNOF7 level and the other lower for NOF-MP2. The reported IPs use one of them, but the paper doesn't give a selection criterion. That ambiguity matters for the late-TM numbers, where PNOF7 already deviates by up to 23 kcal/mol for Ni and Zn. It is an acknowledged limitation, but a referee should ask for the energies of both solutions and a rule for choosing.\n\nThe citation pattern is fine; prior work is credited, and the paper is honest about deviations. This is not a paradigm shift, but it is a useful methodological extension. The audience is quantum chemists working in RDMFT and anyone computing open-shell transition-metal properties. I would send it to review, asking for the positivity proof and a clarification of the solution-selection issue.","headline":"Genuine extension of PNOF7 to arbitrary spin multiplets with correct spin algebra and a real IP benchmark, but the (2,2)-positivity claim is asserted without proof and the multiple-solution ambiguity needs addressing.","tokens_in":11127,"tokens_out":4786,"would_cite":true,"duration_ms":38831,"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":"A natural orbital functional for any spin multiplet conserves total spin and reproduces first-row transition-metal ionization potentials to 6.5 kcal/mol mean absolute error when combined with a modified MP2 correction.","keywords":["natural orbital functional","2RDM reconstruction","spin multiplet","N-representability","PNOF7","NOF-MP2","transition-metal ionization potentials","electron correlation"],"falsifier":"Compute the eigenvalues of the $D$, $Q$, and $G$ matrices built from Eqs. (19)-(21) for a small multiplet (for example, a triplet with $N_{II}=6$ and $N_I=2$ in a modest basis); a negative eigenvalue would disprove the (2,2)-positivity claim. Alternatively, compare the reconstructed $\\langle\\hat{S}^2\\rangle$ for a mixed state against an explicitly spin-projected ensemble wavefunction for the same natural orbitals.","tokens_in":10137,"feed_emoji":"⚛️","tokens_out":9532,"duration_ms":80043,"temperature":0.7,"pith_summary":"This paper claims that a natural orbital functional can describe electronic systems with any total spin by reconstructing the two-particle reduced density matrix (2RDM) for the ensemble of all spin projections of a multiplet of maximum multiplicity. The reconstruction is built so the ensemble conserves total spin, with no opposite-spin interactions in singly occupied orbitals, and the paper states it fulfills the necessary (2,2)-positivity N-representability conditions. On top of this static functional (PNOF7 for multiplets), a modified MP2 correction (NOF-MP2) adds the missing dynamic correlation. Tested on first-row transition-metal ionization potentials with a triple-zeta basis, the combined method gives a mean absolute error of 6.5 kcal/mol against experiment, an accuracy comparable to CCSD(T) benchmark values.","feed_headline":"Spin-conserving functional reproduces transition-metal IPs","feed_subtitle":"A natural-orbital functional plus modified MP2 reaches 6.5 kcal/mol mean error on Sc-Zn ionization energies.","key_machinery":"The key machinery is the reconstructed ensemble 2RDM defined by Eqs. (19)-(21), expressed through the occupation numbers $n_p$, the pairing subspaces $\\Omega_g$, and the functions $\\Phi_p=\\sqrt{n_p(1-n_p)}$. Its intrasubspace $\\alpha\\beta$ blocks are carried by the matrix $\\Pi$, while the intersubspace $\\alpha\\beta$ blocks contain exchange integrals when one subspace lies in $\\Omega_I$. The load-bearing identity is Eq. (26): the reconstruction yields $\\langle\\hat{S}^2\\rangle = (N_I/2)(N_I/2+1)$ for a mixed state of maximum multiplicity, which is what makes the functional spin-conserving. This 2RDM serves both as the static correlation functional (PNOF7 for multiplets) and as the reference for a modified MP2 that adds dynamic correlation without double counting.","core_discovery":"On the paper's own terms, the central discovery is that a 2RDM reconstruction based on natural orbital occupation numbers can represent a maximum-multiplicity spin multiplet as a mixed state without breaking spin symmetry. The orbital space is split into a singly occupied subspace $\\Omega_I$ and an electron-paired subspace $\\Omega_{II}$; intrasubspace $\\alpha\\beta$ blocks use the pairing matrix $\\Pi$, while new intersubspace $\\alpha\\beta$ terms are proportional to $\\Phi_p\\Phi_r$ with $\\Phi_p=\\sqrt{n_p(1-n_p)}$. Inserting these blocks into the standard expression for $\\langle\\hat{S}^2\\rangle$ gives exactly $(N_I/2)(N_I/2+1)$, the value for total spin $S=N_I/2$, so the functional is spin-conserving. The resulting PNOF7 functional for multiplets retains static and intrapair dynamic correlation, and the NOF-MP2 extension recovers the remaining inter-subspace dynamic correlation. Tested on first-row transition-metal atoms, the method reproduces ionization potentials Sc-Zn with a mean absolute error of 6.5 kcal/mol relative to experiment.","pith_inferences":["If the (2,2)-positivity claim survives numerical testing, the functional would be variational, making it a practical low-cost route to geometries and dynamics on open-shell and high-spin systems.","Because the ensemble automatically averages over all spin projections, the method may extend naturally to spin-state energetics and magnetic properties, such as spin gaps in transition-metal complexes.","The pairing ansatz in $\\Omega_{II}$ is not essential to the multiplet construction, so other singlet NOF approximations could likely be lifted to multiplets in the same way.","The two coexisting solutions reported for Co, Ni, and Cu systems (one static-dominated, one dynamic-dominated) suggest a useful extension would be a criterion, based on occupation numbers or an ensemble weight, for selecting the physical state."],"forward_implications":["Systems with any total spin $S$ can be treated while conserving spin, without restricting to the high-spin component or breaking spin symmetry.","PNOF7 recovers non-dynamic and intrapair dynamic correlation, and NOF-MP2 supplies the remaining inter-subspace dynamic correlation without double-counting.","First-row transition-metal ionization potentials from Sc to Zn are reproduced with a mean absolute error of 6.5 kcal/mol using a triple-zeta basis, close to CCSD(T) benchmark values.","The ensemble formulation applies to mixed states of maximum multiplicity, so a ground state of given $S$ is described as a weighted sum of all its $2S+1$ components."],"supporting_citations":[{"why":"Supplies the two-index reconstruction of the 2RDM in terms of auxiliary matrices $\\Delta$ and $\\Pi$ on which the ansatz is built.","marker":"[22]"},{"why":"Provides the electron-pairing division of orbital space used for the singlet subspace $\\Omega_{II}$.","marker":"[21]"},{"why":"Introduces the NOF-MP2 method whose modified perturbation theory supplies the missing dynamic correlation.","marker":"[16]"},{"why":"Gives the self-consistent formulation and amplitude equations used to compute the MP2 correction.","marker":"[17]"},{"why":"Defines PNOF7 for singlets, the functional that this work extends to spin multiplets.","marker":"[25]"},{"why":"Supplies the benchmark CCSD(T) and experimental ionization potentials for the transition-metal test case.","marker":"[27]"},{"why":"Provides the expression for $\\langle\\hat{S}^2\\rangle$ in terms of 2RDM elements used to verify spin conservation.","marker":"[12]"},{"why":"States the (2,2)-positivity N-representability conditions that the reconstruction is claimed to fulfill.","marker":"[7]"}],"fun_headline_variants":["Spin-conserving NOF reproduces transition-metal IPs","Natural orbital functional for any spin multiplet","Mixed-state 2RDM functional keeps spin symmetry","NOF-MP2 mean error 6.5 kcal/mol on Sc-Zn IPs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim that the reconstructed 2RDM fulfills the (2,2)-positivity N-representability conditions is stated but not demonstrated; only the total spin expectation value is verified, and if any $D$, $Q$, or $G$ matrix had a negative eigenvalue the energy could become non-variational or unphysical.","fun_headline_variants_meta":{"raw":{"variants":["Spin-conserving NOF reproduces transition-metal IPs","Natural orbital functional for any spin multiplet","Mixed-state 2RDM functional keeps spin symmetry","NOF-MP2 mean error 6.5 kcal/mol on Sc-Zn IPs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000693,"raw_usage":{"total_tokens":3158,"prompt_tokens":992,"completion_tokens":2166,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":608,"completion_tokens_details":{"reasoning_tokens":2094}},"tokens_in":608,"tokens_out":2166,"duration_ms":15983,"temperature":1.0,"reasoning_tokens":2094,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:11:47.731115+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the eigenvalues of the $D$, $Q$, and $G$ matrices built from Eqs. (19)-(21) for a small multiplet (for example, a triplet with $N_{II}=6$ and $N_I=2$ in a modest basis); a negative eigenvalue would disprove the (2,2)-positivity claim. Alternatively, compare the reconstructed $\\langle\\hat{S}^2\\rangle$ for a mixed state against an explicitly spin-projected ensemble wavefunction for the same natural orbitals.","supporting_citations":[{"cited_title":"Piris, Adv","cited_arxiv_id":null,"evidence_quote":"Provides the electron-pairing division of orbital space used for the singlet subspace $\\Omega_{II}$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the self-consistent formulation and amplitude equations used to compute the MP2 correction."},{"cited_title":"Mitxelena, M","cited_arxiv_id":null,"evidence_quote":"Supplies the benchmark CCSD(T) and experimental ionization potentials for the transition-metal test case."},{"cited_title":"Piris, J","cited_arxiv_id":null,"evidence_quote":"Provides the expression for $\\langle\\hat{S}^2\\rangle$ in terms of 2RDM elements used to verify spin conservation."},{"cited_title":"Therefore, we are deal- ing with approximate 1RDM methods, where the 2RDM continues to play a dominant, albeit hidden, role [8]","cited_arxiv_id":null,"evidence_quote":"States the (2,2)-positivity N-representability conditions that the reconstruction is claimed to fulfill."}],"review_version":1}