{"id":"5960f77c-2a63-40ed-b87a-3535990e11f6","arxiv_id":"2411.14724","paper_version":2,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The eta-clustering state, a three-component analogue of eta-pairing, is shown to be the ground state at the phase-separation/TLL boundary of the extended attractive SU(3) Hubbard chain.","lead":"This paper predicts that a special three-fermion clustering state, previously known only as a high-energy scar, becomes the ground state of an attractive SU(3) Hubbard chain with weak two-body hopping and nearest-neighbor attraction. The result is confirmed by DMRG and gives a realistic cold-atom route to boundary off-diagonal long-range order.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Second-order perturbation at a gapless boundary may not pin the exact ground state; higher-order terms could select a state other than the eta-clustering state.","rationale":"The paper's analytic derivation is clean, the mapping to the XXZ model is standard, and the DMRG evidence supports the large-|U| effective picture: the central-cut entanglement entropy matches the analytic eta-clustering formula, the fitted central charge is close to c=1, and the PS/TLL boundary peak at t2/t1=0.04 is consistent with Eq. (19). These are genuine independent checks, and I do not see an internal inconsistency in the perturbation calculation itself. However, the most load-bearing step is the inference from the second-order effective ground state to the original Hamiltonian: the boundary point is gapless, so degenerate perturbation theory without control of higher-order terms does not guarantee that the exact ground state of H in Eq. (1) is the eta-clustering state. The reader's weakest assumption identifies the same issue. The concern is concrete and testable, and the needed check is inexpensive: exact diagonalization on small chains with fidelity and energy differences, or a fourth-order effective Hamiltonian. For that reason I would adjust the verdict from a flat accept to a conditional accept, pending this check, rather than rejecting the paper, since the existing evidence strongly supports the large-U limit and the main physics is likely correct.","tokens_in":18936,"tokens_out":9350,"duration_ms":101249,"concrete_test":"Perform exact diagonalization at half-filling for L=6,8,10,12 with t1=1 and U=-10,-20,-40, choosing t2 and V exactly on the second-order boundary from Eq. (19) (for example, t2/t1=0.04 and VU/t1^2=0.36 for U=-10). Compute the fidelity F(L,U)=||^2 and the energy difference E_GS - E_eta, scaling both in L and |U|. If F does not approach 1 and the energy difference does not decrease with |U| at fixed t2/U and V/U, the eta-clustering state is not the exact ground state at finite U. As an analytic cross-check, derive the fourth-order Schrieffer-Wolff effective Hamiltonian and test whether the maximal-spin state remains the lowest-energy state at the renormalized boundary; a nonzero third-order H1^3 term would already signal a shift.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim rests on the 'hence' step: because the eta-clustering state is the unique Sz=0 ground state of the second-order effective XXZ chain at Jz=-|Jx| (Eqs. 18-19), it is claimed to be the ground state of the full Hamiltonian (1) in the large-|U| limit. The weak point is that Eq. (19) comes from a second-order degenerate perturbation treatment (Appendix A), while the boundary point Jz=-|Jx| makes the effective XXZ model gapless (ferromagnetic XXX). With a gapless effective Hamiltonian there is no spectral gap protecting the maximal-spin multiplet against third- and higher-order terms in t1,t2,V over U. Such terms are generically nonzero: for example, three one-body hopping processes H1^3 move a triply occupied site by one lattice site, generating an effective three-body hopping term at order t1^3/U^2 that is not included in Eq. (18). That term can split the degenerate maximal-spin multiplet and shift the exact boundary away from the condition Jz=-|Jx|. The DMRG data at U=-10t1 and L=16 are consistent with the second-order prediction, but finite-size level spacings near a gapless point are of order t1/L, comparable to the neglected t1^3/U^2 corrections, so the numerics cannot by themselves certify that the exact ground state is the eta-clustering state. Thus the paper rigorously establishes the claim for the truncated effective model, but the transfer of that claim to the original finite-U Hamiltonian is not fully controlled.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies the extended attractive SU(3) Hubbard chain with one- and two-body hopping and nearest-neighbor attraction at half-filling. In the large-|U| limit it maps the low-energy subspace (empty and triply occupied sites) to an effective spin-1/2 XXZ chain, Eq. (18), with couplings Jx=6t1t2/U and Jz=-3(t1^2+t2^2)/U+9V, Eq. (19). From the XXZ phase diagram the authors identify PS, TLL, and CDW phases, and show that at the PS/TLL boundary Jz=-|Jx| the eta-clustering state, (η†)^{L/2}|0>, is the unique Sz=0 ground state of the effective model. They verify the phase boundaries and the eta-clustering signature with DMRG at U=-10t1 for L=16 chains, including entanglement entropy profiles, central-cut entropy, singlet correlations with boundary peaks, and level-spectroscopy detection of the BKT transition.","tokens_in":19230,"tokens_out":10399,"duration_ms":105614,"significance":"If the central claim survives scrutiny, the paper gives a clean, experimentally plausible mechanism for realizing eta-clustering states as ground states rather than high-energy scars, and it introduces a useful notion of boundary ODLRO that is distinct from localized edge modes. The analytic strong-coupling mapping is explicit and standard, the entanglement-entropy formula is checked against the exact analytic result, and the authors have made the data openly available. The main limitation is that the transfer of the exact result from the second-order effective model to the finite-U Hamiltonian is not fully controlled at the gapless boundary, and the numerical evidence at L=16 is suggestive but not conclusive on that point.","major_comments":[{"comment":"The claim that the eta-clustering state is the ground state of the full Hamiltonian (1) follows by transferring the ferromagnetic-XXX result of the second-order effective XXZ chain to the original model, but this transfer is not controlled at the boundary Jz=-|Jx|. The effective model is gapless there, so there is no spectral gap protecting the degenerate maximal-spin multiplet against third- and higher-order terms in t1, t2, V over U. Such terms are generically present (e.g., three one-body hops generate a three-site hopping process at order t1^3/U^2), and they can split the multiplet and shift the PS/TLL boundary by an amount of order t1^3/U^2. The paper should either compute the leading higher-order corrections to Eq. (19), bound them, or present an independent argument that the eta-clustering state remains a ground state of the full model. Footnote [85] only says the state is 'expected' to remain a ground state under such perturbations, which is not sufficient for the paper's central claim.","section":"III and Appendix A, Eq. (19)"},{"comment":"The numerical verification at U=-10t1 and L=16 cannot by itself control the perturbation-theory issue. Near the gapless boundary the finite-size level spacing is of order t1/L, while the neglected third-order corrections are of order t1^3/U^2; for L=16 these numbers are comparable (about 0.06 versus 0.01), and for larger L the level spacing decreases further. I ask the authors to report the squared overlap between the DMRG ground state and the eta-clustering state, or at least an estimate of the energy splitting within the Sz=0 sector, as a function of L, and to check whether the peak position t_c(L) converges to the second-order prediction as L grows. This would directly test whether the 'hence' step from Eq. (19) to the original Hamiltonian is valid at the parameters used in the DMRG runs.","section":"IV.A, Figs. 5-7"}],"minor_comments":[{"comment":"The text says that H_U and H_V represent 'on-site and next-nearest attractive interactions,' but H_V in Eq. (5) is a nearest-neighbor interaction; please correct this wording.","section":"II.A"},{"comment":"The first binomial coefficient in the displayed formula for the singlet correlation function is written with an index j that is undefined; it should presumably be the summation index l.","section":"Eq. (11)"},{"comment":"The paths C1, C2, and C3 are used throughout the numerical section but are not defined explicitly; please state the fixed parameter combinations (e.g., the value of VU/t1^2 and the relation between t2/t1 and V/t1 along each path) so the numerical choices are reproducible.","section":"IV and Fig. 4"},{"comment":"The label 'UV/t2 1 = 0.4' is cryptic and appears to mix notation; please write the fixed ratio in a consistent form such as VU/t1^2 and ensure it matches the value used for the analytic eta-clustering curve.","section":"Fig. 5 caption"},{"comment":"The two-particle operators ar c^\\dagger_{j,\\alpha} are introduced only in a sentence and use a bar that can be easily confused with the one-particle operators; a more distinct notation (e.g., d^\\dagger_{j,\\alpha}) would improve readability.","section":"II.A, Eq. (3)"},{"comment":"The central charge fit at t2/t1=0.2 reports c≈1.052 but does not give the bond dimension, the number of sweeps, or the fit residuals; please add these details so the reader can judge convergence.","section":"IV.A, Eq. (20)"}],"recommendation":"major_revision","confidential_remarks":"The paper is well-written and presents a genuinely interesting mechanism: an eta-clustering state promoted from a scar to a ground state in a cold-atom-realistic parameter regime. The main risk is exactly what the skeptic note identifies: the un-controlled transfer from the second-order effective XXZ model to the original finite-U Hamiltonian at the gapless boundary. This is not a reason to reject, because the claim is plausible and the DMRG evidence is consistent, but it is load-bearing enough that the revision should contain either a higher-order estimate or a direct numerical overlap check. The reliance on Ref. [21] for the analytic properties of eta-clustering states is appropriate, and the data availability statement is a strength."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a clean, useful paper. The genuinely new thing is that the three-particle eta-clustering state, previously known as an exact scar or as the ground state of an artificially tuned Hamiltonian, becomes the ground state at the PS/TLL boundary of a realistic extended SU(3) Hubbard chain. The boundary condition Jz=-|Jx| from the second-order XXZ mapping is simple, and the DMRG data line up with it: the sharp peak in entanglement entropy, the analytic entropy profile, and the edge-edge singlet correlation all match the eta-clustering state at the predicted point.\n\nWhat the paper does well: the strong-coupling derivation in Appendix A is explicit and standard, the mapping to the XXZ chain is careful, and the numerical checks are appropriate (central charge fit, level crossing for the BKT transition). The notion of boundary ODLRO without localized edge modes is worth having, and the discussion is honest that the eta-clustering state is gapless and not a conventional topological phase. The data are openly available, which helps reproducibility.\n\nThe soft spot is the one the stress-test flags: at Jz=-|Jx| the effective XXZ model is gapless, so there is no spectral gap protecting the maximal-spin multiplet against third- and higher-order terms in t/U. Those terms can in principle shift the boundary or select a slightly different ground state. The paper does not control this, and the L=16 DMRG at U=-10 has level spacings comparable to the neglected corrections, so the numerics do not by themselves certify the exact ground-state identity for finite U. That said, the paper is careful to phrase the claim in the large-|U| limit, and the numerical fingerprints of the eta-clustering state at the predicted point are strong. I would treat this as a caveat, not a fatal flaw. Minor issues: no error bars on DMRG, and the BKT boundary is inferred from level crossings only.\n\nWho it is for: people working on SU(N) Hubbard models, eta-pairing/scars, and one-dimensional phase diagrams. It deserves a serious referee; I would recommend sending it to review with a request to address the higher-order perturbation caveat explicitly.","headline":"Solid strong-coupling analysis with a new ground-state identification at the PS/TLL boundary; the gapless-boundary caveat is real but minor.","tokens_in":19752,"tokens_out":8473,"would_cite":true,"duration_ms":112109,"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":"In the large-attraction limit of a half-filled SU(3) Hubbard chain with two-body hopping, the eta-clustering state is the ground state exactly at the phase-separation/TLL boundary, and on an open chain it shows boundary off-diagonal…","keywords":["SU(3) Hubbard model","eta-clustering state","boundary off-diagonal long-range order","XXZ effective Hamiltonian","phase separation","Tomonaga-Luttinger liquid","charge density wave","density matrix renormalization group"],"falsifier":"A DMRG computation at $U=-10\\,t_1$ along the $C_1$ path could falsify the claim: if the half-filling ground state at $t_2/t_1=0.04$ has overlap with $(\\eta^\\dagger)^{L/2}|0\\rangle$ that tends to zero with system size, or if the end-to-end singlet correlation $|\\langle\\eta^\\dagger_1\\eta_L\\rangle|$ decays with $L$ instead of staying finite, then the eta-clustering state is not the thermodynamic ground state at that point.","tokens_in":18744,"feed_emoji":"⚛️","tokens_out":11600,"duration_ms":98309,"temperature":0.7,"pith_summary":"The paper aims to show that the eta-clustering state—a three-fermion generalization of eta-pairing—can be the ground state of a strongly attractive SU(3) Hubbard chain, not merely an exact excited eigenstate. Adding a two-body hopping term and a nearest-neighbor attraction to the usual attractive chain, and working at half-filling in the large-$|U|$ limit, the authors reduce the low-energy theory to a spin-1/2 XXZ chain and identify the line $J_z=-|J_x|$ where the ground state is $(\\eta^\\dagger)^{L/2}|0\\rangle$. This matters because the required parameters are modest ($|t_2/t_1|\\sim0.04$, $|V/t_1|\\sim0.036$ at $|U/t_1|=10$), much easier to reach in cold-atom experiments than the equal hopping amplitudes previously needed. The paper further claims that this ground state has boundary off-diagonal long-range order—a persistent end-to-end correlation on an open chain—without localized edge modes, and confirms the strong-coupling predictions with DMRG.","feed_headline":"Three-fermion clusters become the ground state of an SU(3) chain","feed_subtitle":"A small two-body hopping makes the three-fermion cluster state the ground state and imprints edge-to-edge order.","key_machinery":"The load-bearing object is the eta-clustering state $|\\Phi^L_{L/2}\\rangle=(\\eta^\\dagger)^{L/2}|0\\rangle$, where $\\eta^\\dagger=\\sum_j(-1)^j U_{j-1}\\eta^\\dagger_j$ and $\\eta^\\dagger_j=c^\\dagger_{j,1}c^\\dagger_{j,2}c^\\dagger_{j,3}$; the factor $U_{j-1}$ is a Jordan-Wigner string that lets powers of $\\eta^\\dagger$ remain nonzero. The argument is carried by second-order degenerate perturbation theory in the strong-attraction limit: projecting onto sites that are either empty or triply occupied turns the Hamiltonian into the spin-1/2 XXZ chain with $J_x=6t_1t_2/U$ and $J_z=-3(t_1^2+t_2^2)/U+9V$, whose exactly known ground states then identify the eta-clustering state at $J_z=-|J_x|$.","core_discovery":"The paper's central claim is that, at half-filling and in the large-$|U|$ limit, the extended attractive SU(3) Hubbard chain with one-body hopping $t_1$, two-body hopping $-t_2$, on-site attraction $U<0$, and nearest-neighbor attraction $V<0$ is governed by an effective spin-1/2 XXZ chain with $J_x=6t_1t_2/U$ and $J_z=-3(t_1^2+t_2^2)/U+9V$. The ground-state phase diagram of that XXZ chain has three phases: phase separation for $J_z<-|J_x|$, a Tomonaga-Luttinger liquid for $-|J_x|3$, since their bulk correlations already decay and the edge-edge correlation is the robust feature.","Because the paper notes that exact SU(3) flavor symmetry is not necessary in the large-$|U|$ limit, a cold-atom realization with slightly flavor-dependent hopping might still produce the eta-clustering ground state; a direct experimental test would be measuring the end-to-end three-fermion correlation function.","The bODLRO-without-edge-modes picture predicts that boundary correlation functions, rather than local density of states or zero modes, are the experimental signature of this order, which could be sought in noise-correlation measurements on an open chain.","At $t_1=0$ the model's fragmented sectors map to the XXC/XXZ chain, so exact integrability could supply dynamical correlation functions for the eta-clustering subspace that DMRG could then be benchmarked against."],"forward_implications":["Along the PS/TLL boundary the ground state is exactly $(\\eta^\\dagger)^{L/2}|0\\rangle$, so the eta-clustering state moves from being an excited scar-like eigenstate to being the ground state for experimentally moderate parameters ($|t_2/t_1|\\sim0.04$, $|V/t_1|\\sim0.036$ at $|U/t_1|=10$).","On an open chain, this ground state has boundary off-diagonal long-range order: the end-to-end singlet correlation persists in the thermodynamic limit while bulk correlations decay exponentially.","The extended model acquires a Tomonaga-Luttinger liquid phase at half-filling, whereas the half-filled attractive SU(3) Hubbard chain without the two-body hopping and nearest-neighbor attraction has only a CDW ground state.","The TLL-to-CDW transition is BKT-type and can be located numerically by level crossings caused by an emergent SU(2) symmetry; DMRG yields central charge $c\\approx1.05$, consistent with a Tomonaga-Luttinger liquid.","The persistence of bODLRO does not imply localized edge modes, and the paper leaves open whether the eta-clustering ground state should be classified as a gapless symmetry-protected topological phase."],"supporting_citations":[{"why":"Defines eta-pairing states and off-diagonal long-range order, the concept the eta-clustering state generalizes.","marker":"[1]"},{"why":"Supplies the definition and exact correlation properties of eta-clustering states, including exponential bulk decay and persistent end-to-end correlation.","marker":"[21]"},{"why":"Provides the strong-coupling mapping of the Hubbard model to a spin chain underlying the effective Hamiltonian.","marker":"[81]"},{"why":"Gives the strong-on-site-interaction effective-Hamiltonian formalism used to derive the XXZ couplings.","marker":"[82]"},{"why":"Supplies the ground-state phase diagram of the spin-1/2 XXZ chain that the paper translates into PS, TLL, and CDW phases.","marker":"[83]"},{"why":"Gives the conformal-field-theory entanglement scaling used to identify the TLL phase and the gapless nature of the eta-clustering state.","marker":"[99]"},{"why":"Provides the previous numerical phase diagram of the SU(3) Hubbard chain with nearest-neighbor interactions that the new model extends.","marker":"[61]"}],"fun_headline_variants":["eta-clustering state becomes exact ground state in SU(3) chain","Edge-edge order from eta-clustering in SU(3) chain","eta-clustering ground state gives edge-to-edge correlations","Edge-edge order without edge states from eta-clustering"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central claim rests on the assumption that second-order degenerate perturbation theory in $1/U$ is quantitatively accurate for the finite parameters tested ($U=-10t_1$, $|t_2/t_1|\\lesssim0.25$), so the low-energy space really consists only of empty and triply occupied sites and the effective couplings $J_x,J_z$ are the ones given.","fun_headline_variants_meta":{"raw":{"variants":["eta-clustering state becomes exact ground state in SU(3) chain","Edge-edge order from eta-clustering in SU(3) chain","eta-clustering ground state gives edge-to-edge correlations","Edge-edge order without edge states from eta-clustering"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000795,"raw_usage":{"total_tokens":3548,"prompt_tokens":1042,"completion_tokens":2506,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":658,"completion_tokens_details":{"reasoning_tokens":2436}},"tokens_in":658,"tokens_out":2506,"duration_ms":18666,"temperature":1.0,"reasoning_tokens":2436,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:58:40.838918+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A DMRG computation at $U=-10\\,t_1$ along the $C_1$ path could falsify the claim: if the half-filling ground state at $t_2/t_1=0.04$ has overlap with $(\\eta^\\dagger)^{L/2}|0\\rangle$ that tends to zero with system size, or if the end-to-end singlet correlation $|\\langle\\eta^\\dagger_1\\eta_L\\rangle|$ decays with $L$ instead of staying finite, then the eta-clustering state is not the thermodynamic ground state at that point.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the strong-coupling mapping of the Hubbard model to a spin chain underlying the effective Hamiltonian."},{"cited_title":"Nonne, P","cited_arxiv_id":null,"evidence_quote":"Gives the strong-on-site-interaction effective-Hamiltonian formalism used to derive the XXZ couplings."},{"cited_title":"Capponi, P","cited_arxiv_id":null,"evidence_quote":"Supplies the ground-state phase diagram of the spin-1/2 XXZ chain that the paper translates into PS, TLL, and CDW phases."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the previous numerical phase diagram of the SU(3) Hubbard chain with nearest-neighbor interactions that the new model extends."}],"review_version":1}