{"id":"7a0286c2-bc4e-4e1f-bb7c-4aa4f3fcf6d6","arxiv_id":"1908.09335","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A spin-charge-separated effective model with many-body Wannier function extrapolation reproduces the optical conductivity of one-dimensional Hubbard models for U/T between 5 and 10.","lead":"Researchers built a simplified electron model that keeps spin and charge separate but still allows charge fluctuations, and showed it reproduces the light absorption of one-dimensional Hubbard models. Extrapolating this model to hundreds of sites gives spectra that match DMRG benchmarks.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The MBWF extrapolation to N≈200 is the weakest link: it starts from a 16-site cluster, truncates and constants-extrapolates matrix elements by hand, and lacks any convergence check in starting size or retained peak count.","rationale":"The reader's verdict identifies the spin-charge factorization (f(M) independent of charge configuration) as the weakest assumption. That is a reasonable foundational concern, and the paper's N=14 tests do provide some support for it. However, the more immediate risk to the central claim lies in the MBWF extrapolation: even if the charge model were exact in the thermodynamic limit, the paper's procedure for reaching large systems is a sequence of manual truncations and constant extrapolations of a 7-state effective model derived from a single 16-site cluster. The paper's own text flags the need for checks at U=5 and admits high-energy tail underestimation. Moreover, all benchmark comparisons are of normalized spectral shapes, so the 'quantitative reproduction' claimed in the Introduction is not actually evidenced. This does not warrant rejection: the charge model is a plausible effective model with some small-system support, and the MBWF idea is interesting. But the central quantitative claim for large systems is conditional on a convergence check of the extrapolation. Since the reader already issued CONDITIONAL, the appropriate recommendation is UNCHANGED.","tokens_in":21357,"tokens_out":9701,"duration_ms":100567,"concrete_test":"Repeat the MBWF construction of Sec. III B with a starting cluster of N=20 (within the paper's stated exact-diagonalization limit of ~26), keeping the same seven-peak definition, and also with nine retained peaks. Compare the resulting h_{kk'} matrix elements for k≤7 to the N=16 values. If any h_{kk'} changes by more than 5%, or if the extrapolated Nex=200 spectrum shifts its main peak or lower edge by more than 0.3T, the N=16-based extrapolation is not converged and the claimed large-system spectra are not robust predictions.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline claim of the paper is that the charge model plus MBWF extrapolation reproduces the optical conductivity of Hubbard/extended Hubbard models in large systems (N≈200). The charge model itself is validated only for N=14 (Fig. 2), a size too small to determine thermodynamic-limit spectral shapes. The route to large systems is the MBWF construction of Sec. III B, and that route has no controlled error estimate. Starting from N=16, the construction (i) retains only seven 'principal peaks' as a complete orthonormal set, (ii) sets all h_{k,k+i} with i≥4 to zero, (iii) replaces all h_{k,k+i} with k beyond the N=16 range by constants determined from averages over k=2..6 (Eqs. C1–C5), and (iv) truncates current matrix elements at r_HD=7 and rescales by sqrt(N_ex/N). None of these steps is derived from the microscopic model; each is an ad hoc assumption about locality and size-independence of the MBWF matrix elements. The paper explicitly concedes that 'checks are required of the results down to U/T=5' and that the method 'underestimates the tail structure on the high-energy side.' The comparisons shown are all normalized to unit maximum, so absolute spectral weights are never tested; in the extended-Hubbard comparison (Fig. 10) the DMRG broadening parameter is adjusted to match the MBWF peak width. Under these conditions, the agreement between the extrapolated Nex=200 spectrum and DMRG could reflect the flexibility of the extrapolation protocol rather than the validity of the charge model in the thermodynamic limit.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a 'charge model' for the half-filled one-dimensional Hubbard and extended Hubbard models. The model is defined by projecting the full Hamiltonian onto a subspace in which spin and charge degrees of freedom are factorized: every basis state with M holon-doublon pairs carries the same spin wave function, taken to be the ground state of a Heisenberg chain with N-2M sites. Hopping matrix elements that change the number of holon-doublon pairs are renormalized by a factor cS(M)=0.82, obtained from an independent spin-chain overlap calculation. The authors validate the charge model against exact diagonalization for N=14 at U/T=10, then introduce many-body Wannier functions (MBWFs) built from the seven principal optically active eigenstates of a 16-site cluster. Matrix elements are extrapolated to larger systems, yielding optical conductivity spectra for effective sizes up to Nex=200, which are compared with t-DMRG and DDMRG results for U/T=5, 6, and 10 and for (U,V)/T=(10,2.5). The central claim is that spin-charge separation works well in the intermediate-coupling regime U=5-10T and that the MBWF extrapolation reproduces the optical spectra of the original models in large systems.","tokens_in":21747,"tokens_out":5558,"duration_ms":56196,"significance":"If the central claim is correct, the paper offers a practical route to optical spectra of strongly correlated 1D systems at sizes approaching the thermodynamic limit while retaining explicit wave-function information, which DMRG-based methods do not provide. The N=14 exact comparison is a genuine and useful benchmark, and the renormalization factor cS(M)=0.82 is not fitted to the optical spectra, which strengthens the model validation. The comparisons against t-DMRG and DDMRG are appropriate external checks. However, the extrapolation from a 16-site cluster to Nex=200 rests on several ad hoc truncation and averaging steps without a controlled error estimate, so the large-system part of the claim is not yet established at the level claimed in the abstract.","major_comments":[{"comment":"The MBWF extrapolation is the load-bearing step for the headline claim of spectra at N≈200, but it lacks any controlled convergence check. Starting from a 16-site cluster, the construction (i) retains only seven 'principal peaks' as a complete orthonormal set, (ii) sets h_{k,k+i}=0 for i≥4, (iii) replaces h_{k,k+i} for k beyond the 16-site range by constants m0-m3 averaged over k=2..6, and (iv) zero-pads the current matrix elements and rescales them by sqrt(Nex/N). None of these steps is derived from the microscopic model, and the paper does not test stability under changes in the starting cluster size (e.g., N=12 or 20) or in the number of retained peaks (e.g., 9 or 11). Without such checks, the observed agreement with DMRG at Nex=200 could reflect the flexibility of the extrapolation protocol rather than the validity of the charge model or the MBWF construction.","section":"III B, Eqs. (19)-(21) and Appendix C"},{"comment":"The abstract claims validity in the intermediate regime U=5-10T, but the direct validation of the charge model against exact diagonalization is shown only for U/T=10 and N=14 (Fig. 2). The U/T=5 and 6 comparisons in Fig. 9 are obtained by first applying the MBWF extrapolation to the charge model and then comparing with DMRG. This conflates the validity of the charge model with the validity of the extrapolation, which is precisely the uncontrolled step identified above. The text itself states that 'checks are required of the results down to U/T=5,' but no direct charge-model-versus-Hubbard exact comparison at U/T=5 or 6 is presented. A direct N=14 comparison at these values is needed before the U=5-10T claim can be supported.","section":"III A and Fig. 9"},{"comment":"All comparisons between the extrapolated charge-model spectra and the DMRG benchmarks are normalized to unit maximum, so absolute spectral weights (and hence the f-sum or total optical weight) are never compared. In addition, the t-DMRG spectrum in Fig. 10(c) is computed with an adjusted broadening γ=0.14T chosen to match the MBWF peak width, and in Fig. 8(a) the spin-derived hump around ω/T=11 is explicitly excluded from the comparison. These choices reduce the stringency of the claimed quantitative agreement. The paper should either compare absolute conductivities or clearly state that the claim concerns spectral shape only.","section":"Figs. 8-10 and Eq. (18)"}],"minor_comments":[{"comment":"The phrase 'zero center-of-gravity momentum' should be 'zero total momentum' or 'zero center-of-mass momentum'.","section":"Throughout"},{"comment":"The caption states that spectra are normalized so that <g|J^† J|g>=1, whereas later figures are normalized to unit maximum; the normalization convention should be stated consistently for every figure.","section":"Fig. 2 caption"},{"comment":"The notation cS(M) is used both for the overlap in Eq. (14) and, in the following sentence, for the weight cS^2(M); please distinguish these quantities explicitly.","section":"Sec. II around Eq. (14)"},{"comment":"For M=N/2 the spin wave function f(M) is defined on zero singly occupied sites, which is a limiting case; please specify the convention used for the Heisenberg ground state when N-2M=0.","section":"Sec. II, Eq. (2)"},{"comment":"The averaging ranges differ among m0, m1, m2, and m3 (e.g., m3 averages only k=2..4); a sentence explaining why these ranges are chosen would help the reader assess the extrapolation.","section":"Appendix C, Eqs. (C1)-(C5)"}],"recommendation":"major_revision","confidential_remarks":"The exact-diagonalization benchmark for the charge model at N=14 is a solid piece of work, and the use of an independently determined cS(M) is a strength. The main obstacle is the MBWF extrapolation: its uncontrolled truncations and averaging steps are load-bearing for the large-system claim. I would be willing to reconsider after the authors add convergence checks in the starting cluster size and retained-peak count, and after they provide a direct exact-diagonalization comparison at U/T=5 and 6."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [colleague],\n\nWhat you should know: this paper has a real idea and a real weak spot. The real idea is a 'charge model' for half-filled 1D Hubbard and extended Hubbard models that preserves spin-charge separation while allowing unlimited H-D pairs. That is a clear step beyond the holon-doublon model, which is stuck at one pair; the N=14 exact-diagonalization comparison shows the multi-pair fluctuations matter and that a single renormalization factor c_S=0.82 (taken from an independent spin-chain calculation, not fitted) reproduces the optical conductivity. This is the strongest part of the paper.\n\nThe weak spot is the Many-Body Wannier Function extrapolation. It starts from a 16-site cluster, keeps seven 'principal peaks', sets matrix elements to zero beyond third-neighbor distance, replaces longer-range elements by constants averaged over the 16-site data, and rescales current elements by sqrt(N_ex/N). Each step is an ad hoc locality/size-independence assumption; there is no convergence test in starting size or retained peak count. Comparisons are all normalized to maximum, so absolute spectral weights are never tested. In the extended-Hubbard case the DMRG broadening is adjusted to match the MBWF peak width. The authors honestly concede the under-reproduction of the high-energy tail and the need for checks down to U/T=5, but the abstract still claims quantitative reproduction. I find that overstatement.\n\nAm I being too harsh? The extrapolation could be right: the agreement with DMRG at U=10, and even at U=5 for the overall shape, is suggestive rather than coincidental. But the manuscript does not currently provide the evidence that the extrapolation is controlled.\n\nWho is this for: people computing or interpreting optical spectra of 1D Mott insulators, and anyone developing effective models in strongly correlated systems. The charge model itself is likely to be useful; the MBWF method is an interesting but not-yet-solid technical proposal.\n\nMy recommendation: send this to referees. It deserves serious scrutiny, not a desk reject. A referee should ask for a clearer statement of what is controlled, a sum-rule or absolute-intensity check, and a modest rewriting of the claims. The ideas are genuinely new.","headline":"A genuinely new effective model with a solid small-cluster validation; the MBWF extrapolation to large systems is heuristic and the abstract overstates the agreement, but the paper deserves refereeing.","tokens_in":22247,"tokens_out":3612,"would_cite":false,"duration_ms":36788,"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":"An effective model that separates spin from charge reproduces the optical spectra of one-dimensional Mott insulators at realistic interaction strengths.","keywords":["charge model","spin-charge separation","one-dimensional Hubbard model","extended Hubbard model","optical conductivity","many-body Wannier functions","holon-doublon pairs","intermediate coupling"],"falsifier":"Compute the charge-model optical spectrum at $U/T=3$ and compare it with dynamical DMRG: the paper already notes that the high-energy tail is underestimated and that the one-pair nature of the Wannier states degrades as $U$ decreases, so a clearly visible mismatch there would set the model's lower boundary below $U/T=5$. Alternatively, compute the exact ground-state overlap $c_S$ in the original Hubbard model at intermediate $U$; a value that departs from $0.82$ as the singlet insertion point moves would falsify the single-parameter renormalization.","tokens_in":21149,"feed_emoji":"","tokens_out":13366,"duration_ms":121051,"temperature":0.7,"pith_summary":"The paper claims that the optical conductivity of half-filled one-dimensional Hubbard and extended Hubbard models can be reproduced, at the intermediate interaction strengths found in real quasi-one-dimensional Mott insulators ($U/T = 5$–$10$), by an effective model in which spin and charge are strictly separated. The charge model keeps all charge fluctuations over holon–doublon pairs while freezing the spin sector into the Heisenberg ground state of the remaining singly occupied sites. A single overlap factor $c_S(M)=0.82$ renormalizes the creation and annihilation of charge pairs, and many-body Wannier functions extrapolate the model from a 16-site cluster to roughly 200 sites. The resulting spectra match DMRG benchmarks for $U/T = 10$, $6$, and $5$, with and without a nearest-neighbor interaction $V/T = 2.5$. If the claim holds, spin-charge separation is not just an infinite-$U$ artifact, and the charge model is a compact tool for large-system optical spectra of strongly correlated one-dimensional materials.","feed_headline":"Charge model reproduces 1D Mott insulator optical spectra at U/T=5–10","feed_subtitle":"A single renormalization factor makes the model match DMRG benchmarks down to U/T = 5.","key_machinery":"The load-bearing object is the projected charge model $H^{(C)}(t)=P H(t) P$, defined on basis states in which the positions of $M$ doublons and $M$ holons are recorded while all spins are described by a single factorized wave function $f^{(M)}$, taken to be the ground state of the Heisenberg Hamiltonian with $N-2M$ sites. The key numerical input is the overlap $c_S(M)=0.82$, the thermodynamic-limit overlap between that Heisenberg ground state and the state formed by inserting a nearest-neighbor spin singlet; it renormalizes every matrix element that creates or annihilates a holon–doublon pair. The boundary twist $\\theta_M=(\\pi/2)\\,\\mathrm{mod}(N-2M,4)$ fixes the phase accumulated when a charge crosses the periodic boundary. The second mechanism is the many-body Wannier construction: the seven dominant optically active eigenstates of a 16-site cluster are unitarily rotated into spatially localized many-body states, and the resulting Hamiltonian and current matrix elements, which decay within the cluster, are extrapolated to about 200 sites.","core_discovery":"The central claim is that the effective Hamiltonian $H^{(C)}(t)=P H(t) P$, obtained by projecting the Hubbard or extended Hubbard Hamiltonian onto a subspace in which every basis state with $M$ holon–doublon pairs carries one and the same spin wave function—the ground state of the Heisenberg chain on the $N-2M$ singly occupied sites—is quantitatively faithful for linear optical response. The paper shows that all spin physics entering the projection is carried by a phase $\\theta_M=(\\pi/2)\\,\\mathrm{mod}(N-2M,4)$ and by the overlap $c_S(M)=0.82$ between a Heisenberg ground state and the state obtained by inserting a nearest-neighbor spin singlet. Within this subspace, charge fluctuations are treated exactly, and they are essential: truncating to a single H-D pair shifts the spectrum by about $0.5T$ and fails for $U/T=5$, while the full charge model matches the original models at $N=14$ and matches dynamical DMRG benchmarks for systems of about 200 sites at $U/T=10$, $6$, and $5$. The paper also claims that the many-body Wannier construction, which localizes the leading optically active eigenstates and extrapolates their Hamiltonian and current matrix elements, converges by $N\\approx200$.","pith_inferences":["One extension the paper leaves implicit is to use the same projection for pump–probe response: comparing transient spectra of the charge model with those of the full Hubbard model would isolate spin-charge coupling as the driver of the photoinduced Mott-gap collapse, a comparison the authors flag as future work.","The many-body Wannier procedure is in principle transferable to the original Hubbard and extended Hubbard models and to strongly excited states, but there the Wannier states would need to carry spin structure and the number of relevant eigenstates grows with spin-charge coupling, so the construction is not straightforward.","A testable refinement would be to let $c_S$ depend on holon–doublon distance or on the local spin environment; the paper's observation that the shortest-distance diagonal matrix element is boundary-sensitive suggests such dependence may matter just above $V/T=2.5$."],"forward_implications":["For $U/T$ between 5 and 10 and $V/T$ up to 2.5, the charge model's Hilbert space is about 20 times smaller than the original model's, so large-system optical spectra can be computed exactly in the charge sector without repeated basis transformations.","The holon–doublon two-particle model, which omits pair creation and annihilation, is not reliable in this regime: it misses the optical gap by about $T$ at $U/T=10$ and fails at smaller $U$.","The many-body Wannier extrapolation from a 16-site cluster yields spectra for about 200 sites whose low- and high-energy edges, peak positions, and asymmetry agree with t-DMRG and DDMRG benchmarks, meaning finite-size effects can be removed while keeping explicit excited-state wave functions.","Because the charge model retains wave functions, spectral features can be interpreted in terms of renormalized holon–doublon distances, a type of diagnosis that is harder with repeated-basis-transformation methods."],"supporting_citations":[{"why":"Supplies the rigorous spin-charge-separated ground state of the one-dimensional Hubbard model in the infinite-U limit, the starting point for the factorized subspace.","marker":"8"},{"why":"Defines the holon-doublon two-particle model that the charge model is compared against and shown to improve upon.","marker":"13"},{"why":"Gives the strong-coupling perturbation analysis showing that processes changing the number of H-D pairs are first order in T/(U - V), supporting the validity of the projection.","marker":"20"},{"why":"Provides the earlier one-H-D-pair optical analysis in the infinite-U limit, including the twist phase that motivates the choice of theta_M.","marker":"22"},{"why":"Supplies the dynamical DMRG optical conductivity benchmark for the Hubbard model that the many-body Wannier charge-model spectra are matched against at U/T = 5-10.","marker":"28"},{"why":"Supplies the fitted overlap c_S(M) = 0.820 + 0.740 (N - 2M)^(-2.36) whose thermodynamic-limit value renormalizes pair creation and annihilation amplitudes.","marker":"36"}],"fun_headline_variants":["Spin-charge separation validated at intermediate U in 1D","Charge model reproduces optical spectra for large 1D Hubbard","New effective model captures 1D Hubbard optical response","Many-body Wannier enables large-scale Hubbard optical calc","Charge model beats truncation for 1D Hubbard spectra"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction assumes that one and the same spin wave function, the Heisenberg ground state, is correct for every placement of holons and doublons; if the spin background rearranges around charge carriers at intermediate $U$, the projection drops the physics that controls the optical response.","fun_headline_variants_meta":{"raw":{"variants":["Spin-charge separation validated at intermediate U in 1D","Charge model reproduces optical spectra for large 1D Hubbard","New effective model captures 1D Hubbard optical response","Many-body Wannier enables large-scale Hubbard optical calc","Charge model beats truncation for 1D Hubbard spectra"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000195,"raw_usage":{"total_tokens":1377,"prompt_tokens":986,"completion_tokens":391,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":602,"completion_tokens_details":{"reasoning_tokens":309}},"tokens_in":602,"tokens_out":391,"duration_ms":4234,"temperature":1.0,"reasoning_tokens":309,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:14:57.104771+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the charge-model optical spectrum at $U/T=3$ and compare it with dynamical DMRG: the paper already notes that the high-energy tail is underestimated and that the one-pair nature of the Wannier states degrades as $U$ decreases, so a clearly visible mismatch there would set the model's lower boundary below $U/T=5$. Alternatively, compute the exact ground-state overlap $c_S$ in the original Hubbard model at intermediate $U$; a value that departs from $0.82$ as the singlet insertion point moves would falsify the single-parameter renormalization.","supporting_citations":[{"cited_title":"Ogata and H","cited_arxiv_id":null,"evidence_quote":"Supplies the rigorous spin-charge-separated ground state of the one-dimensional Hubbard model in the infinite-U limit, the starting point for the factorized subspace."},{"cited_title":"Mizuno, K","cited_arxiv_id":null,"evidence_quote":"Defines the holon-doublon two-particle model that the charge model is compared against and shown to improve upon."},{"cited_title":"Eskes, A","cited_arxiv_id":null,"evidence_quote":"Gives the strong-coupling perturbation analysis showing that processes changing the number of H-D pairs are first order in T/(U - V), supporting the validity of the projection."},{"cited_title":"Stephan and K","cited_arxiv_id":null,"evidence_quote":"Provides the earlier one-H-D-pair optical analysis in the infinite-U limit, including the twist phase that motivates the choice of theta_M."},{"cited_title":"Jeckelmann, F","cited_arxiv_id":null,"evidence_quote":"Supplies the dynamical DMRG optical conductivity benchmark for the Hubbard model that the many-body Wannier charge-model spectra are matched against at U/T = 5-10."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the fitted overlap c_S(M) = 0.820 + 0.740 (N - 2M)^(-2.36) whose thermodynamic-limit value renormalizes pair creation and annihilation amplitudes."}],"review_version":1}