{"id":"988d88a4-1c38-4ae0-9a90-f98380579f1e","arxiv_id":"2502.05847","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Spatial modulation of interactions in 1D two-particle systems generates topological doublon edge states via an effective Aubry-André-Harper model.","lead":"This paper shows that two-particle bound states called doublons in one-dimensional interacting boson and fermion chains can be made topological by periodically modulating the interaction strength, producing edge states and butterfly-like spectra. The result extends known Aubry-André-Harper physics into the few-body interacting regime, with potential experimental realizations in superconducting qubits and optical lattices.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Flawed perturbation theory in Appendix A treats non-eigenstates |2>_j as the unperturbed basis, invalidating Eq. (A4/A5) and the derived diagonal-AAH effective Hamiltonian for weak J, though the exact J=0 mapping and numerical results for the central claim stand.","rationale":"The central claim of the paper is the existence of topological doublon edge states induced by spatially modulated interactions. The J=0 limit is exactly mapped to a diagonal AAH model, which is rigorous and yields the Chern numbers and edge states. The robustness for J≠0 is supported by exact diagonalization. The reader's identified weakness is the only clear technical error in the analytic part: the unperturbed states in Appendix A are not eigenstates. This does not overturn the central claim because the numerics are independent, but it does invalidate the paper's analytic argument that the system remains a diagonal AAH model for weak J. A concrete comparison of the numerically derived doublon bands to the effective Hamiltonian in Eq. (6) would settle whether the analytic claim is quantitatively correct. Given that the core physics is likely correct but the analytic derivation is flawed, the verdict CONDITIONAL (unchanged from the reader) is appropriate.","tokens_in":13796,"tokens_out":12842,"duration_ms":119166,"concrete_test":"For a small hopping (e.g., J=0.05, g=1, U=0.2, α=1/3, φ=2π/3), compute the two-boson spectrum by exact diagonalization on a chain of N≥60, identify the doublon bands by large IPR, and compare their dispersion to the diagonal AAH model with parameters U_j + 2J^2/g and g + J^2/g from Eq. (6). If the doublon bands deviate beyond a few percent in bandwidth or gap structure, the analytic claim fails; if they match, the effective Hamiltonian is correct despite the flawed derivation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Appendix A, the authors take the localized two-boson states |2>_j as eigenstates of the J=0 Hamiltonian with eigenvalues E_j = U_j + 2g, and use them as the unperturbed subspace for a second-order Schrieffer-Wolff expansion in J. This is incorrect: the pair-hopping term g couples |2>_j to |2>_{j±1}, as seen from the J=0 equation (5), so the actual eigenstates are extended doublon Bloch states of the diagonal AAH model, not |2>_j. Consequently, the effective Hamiltonian Eq. (A4/A5), and the simplified form in Eq. (6) (with renormalized potential U_j + 2J^2/g and hopping g + J^2/g), are not derived from a valid perturbation theory. In fact, a two-site calculation shows the second-order correction involves the doubling energy denominator (of order the doublon dispersion, not 2g) and is momentum-dependent, so the claim that the system 'again becomes the diagonal AAH single particle model' for small J is unsupported. The numerical results for J≠0 (Figs. 3) are not invalidated, but the analytic robustness argument is weakened.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies topological two-particle states in one-dimensional Hubbard models with periodically modulated interactions. For the boson-Hubbard model with pair hopping and spatially modulated on-site interaction, the authors show that in the strongly correlated limit J=0 the doublon sector maps exactly to a diagonal Aubry-André-Harper model, producing butterfly-like doublon bands with Chern numbers (1,-2,1) and topological in-gap doublon edge states. For nonzero nearest-neighbor hopping J, exact diagonalization is used to display doublon edge states, topological bound states in the continuum, doublon collapse, and, in the mapped 2D single-particle picture, corner states in one or two corners. A parallel analysis is presented for a spinless fermion model with pair hopping. The paper also gives a second-order perturbation-theory derivation in Appendix A that is meant to support the robustness of the doublon edge states for weak J.","tokens_in":14019,"tokens_out":7749,"duration_ms":79102,"significance":"If the J≠0 robustness claim can be properly established, the paper provides a clean mechanism by which spatially modulated interactions induce topological doublon states in few-body systems, and it connects one-dimensional two-particle topology to corner states in two-dimensional single-particle models. The exact J=0 mapping to AAH models is elegant and appears fully correct, and the numerical spectra are plausible. The paper is also commendable in that it does not fit parameters to data: the Chern numbers and edge-state locations are derived from the model and its exact mapping. However, the analytic perturbation theory used to argue robustness at small J is flawed, and this weakens the support for several claims in the main text.","major_comments":[{"comment":"The perturbation calculation is not valid as written because the unperturbed states |2>_j are not eigenstates of the J=0 Hamiltonian in Eq. (4): the pair-hopping term couples |2>_j to |2>_{j±1}, as is evident from Eq. (5), so the exact J=0 doublon eigenstates are extended Bloch states of the diagonal AAH model, not localized states. Consequently, P H_b P in Eq. (A3) is not diagonal in the |2>_j basis, the eigenvalues E_j = U_j + 2g are not the correct unperturbed doublon energies, and the effective Hamiltonian in Eqs. (A4) and (A5) does not follow from a Schrieffer-Wolff expansion. This is load-bearing because Eq. (6) in Sec. II A is used to claim that the system “again becomes the diagonal AAH single particle model” for weak J and that the topological doublon edge states are robust. The exact J=0 mapping and the numerical ED results in Figs. 2 and 3 are not invalidated, but the analytic support for the J≠0 topological phases must be rederived using the exact J=0 doublon eigenstates (which will introduce momentum-dependent corrections) or explicitly replaced by direct numerical characterization.","section":"Appendix A, Eqs. (A1)–(A5); Sec. II A, Eq. (6)"},{"comment":"The numerical evidence for the “topological insulator” and “topological bound states in the continuum” at J=0.5 is not accompanied by a direct topological invariant of the interacting two-boson spectrum. The Chern numbers (1,-2,1) are computed for the J=0 diagonal AAH model, and the identification of the in-gap states at J≠0 as topological relies on Eq. (6), whose derivation is the issue described above. I recommend either computing an invariant that is valid for the interacting problem at J≠0 (for example, a flux-averaged Chern number or Bott index of the doublon subspace) or explicitly stating that the topological classification at J≠0 is an inference from the J=0 limit rather than a direct computation.","section":"Sec. II A, Fig. 3 and the discussion of J≠0"},{"comment":"The same perturbation-theory argument is invoked for the fermion model (“the second order perturbation theory ... could also be used”). Since the derivation in Appendix A is invalid for the boson model, the analytic robustness claim for the fermionic topological doublon edge states at t≠0 is likewise unsupported. The numerical spectra in Figs. 5(b)–5(d) still provide evidence for the existence of pair edge states, but the statement that these states are topological for t≠0 requires either a corrected effective theory or a direct invariant calculation.","section":"Sec. III, fermion model, t≠0"}],"minor_comments":[{"comment":"The text reads “inverse participation patio”; this should be “inverse participation ratio.” Also, the subscript n in IPR_n is used without defining it explicitly as the eigenstate index.","section":"Sec. II A, IPR definition"},{"comment":"The caption contains the typo “respectivly”; it should be “respectively.” In the text near Fig. 2, “the topological doublon state is located in the two end regions” should read “the two topological doublon states are located in the two end regions” for the inversion-symmetric case.","section":"Fig. 2 caption and surrounding text"},{"comment":"The sentence “the energy spectra of scattering states have been eliminated” is unclear: please clarify whether the scattering states were projected out before diagonalization or simply omitted from the plotted spectra.","section":"Sec. II A, scattering states"},{"comment":"The subspace labels “U” and “V” collide with the interaction amplitude U and the complement symbol V. Consider renaming these subspaces, for example “D” and “S.” In addition, the operator S in Eq. (A2) is not a projection operator; it is the off-diagonal part of the resolvent, and the text should say so explicitly.","section":"Appendix A, notation"},{"comment":"The sentence “When the onsite interaction U becomes strong, the topological doublon edge states could become unstable (see Fig.3 [c], where the perturbation theory has broken down)” is confusing because the inset of Fig. 3(c) shows the edge states still present at U=2. Please clarify what “unstable” means and how the perturbation-theory breakdown is diagnosed.","section":"Sec. II A, Fig. 3(c) discussion"}],"recommendation":"major_revision","confidential_remarks":"The main technical concern raised in the stress-test note lands exactly: Appendix A uses a non-eigenstate basis for the unperturbed problem, and this invalidates the analytic effective Hamiltonian used for the J≠0 robustness claims. The exact J=0 mapping and the numerical ED results are sound and should be preserved. The authors should be given the opportunity to fix Appendix A or to reframe the J≠0 claims as purely numerical; if the perturbation theory cannot be repaired, the analytic claim needs to be weakened, and the paper would still have value based on the exact mapping and numerics. The manuscript is within the journal's scope, and I did not find any inappropriate citation practice beyond the expected use of the authors' own prior work (Ref. 72)."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a modest but useful paper. It shows that in the two-particle sector of a 1D Bose-Hubbard model with periodically modulated on-site interaction, the J=0 problem maps exactly to a diagonal AAH model in the doublon subspace, and a similar mapping holds for modulated pair-hopping (off-diagonal AAH) and for spinless fermions with pair-hopping. The butterfly spectra, Chern numbers (1,-2,1), and the in-gap doublon edge states follow from this. That part is clean and I did not find a flaw in the mapping. The numerical exact-diagonalization results for finite J, including the bands entering the scattering continuum and the two-particle BICs, are consistent and well-presented. I also appreciate that the paper does not fit any parameters; the topological invariants are computed from the effective model, not fitted.\n\nThe real soft spot is Appendix A. The authors take the localized two-boson states |2>_j as eigenstates of the J=0 Hamiltonian with eigenvalues U_j+2g and use them as the unperturbed subspace. But the pair-hopping term g couples |2>_j to |2>_{j±1}, so those states are not eigenstates; the actual eigenstates are the extended doublon Bloch states of the AAH model. Once you see that, equations (A4) and (A5), and the claim that a weak J just renormalizes U_j and g while keeping the model diagonal AAH, are not supported by the perturbation theory as written. The numerics for weak J may still be right, but the analytic robustness argument needs to be redone, either with a proper Schrieffer-Wolff transformation on the AAH eigenstates or by dropping the analytic claim and relying on ED. A second issue: the 'topological metal' phase in the mapped 2D system is characterized only by a corner-state density profile; there is no invariant or criterion given for what makes that metal topological. That is a smaller problem, but the term 'topological metal' carries a strong claim.\n\nNone of this kills the paper. The exact J=0 mapping, the doublon band topology, and the ED evidence for edge states and BICs all stand. The appendix should be fixed, and the topological-metal language tightened. This is an incremental contribution, not a breakthrough: the underlying physics is the known two-particle-to-2D mapping plus AAH/Hofstadter topology, applied to spatially modulated interactions. For someone working on bound pairs in qubit arrays or ultracold atoms, this is a useful reference. I would send it to a serious referee, and if I worked on doublon topology I would cite it.","headline":"Exact J=0 mapping to AAH is solid and the numerics are plausible, but the Appendix A perturbation theory is wrong because the unperturbed basis isn't eigenstates; still worth a proper refereeing.","tokens_in":14571,"tokens_out":2739,"would_cite":true,"duration_ms":27747,"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":"Spatially modulated interactions can drive two-boson pairs into topologically protected edge states.","keywords":["topological doublon","spatially modulated interaction","Aubry-André-Harper model","Hofstadter butterfly","doublon bound state in the continuum","doublon collapse","corner state","Bose-Hubbard model"],"falsifier":"Run exact diagonalization of the two-boson Hamiltonian at small but nonzero $J$ (for instance $J/g=0.1$) with $\\alpha=1/3$, and compare the in-gap doublon edge-state energies and density profiles with the diagonal AAH prediction using $U_j+2J^2/g$ and $g+J^2/g$; if the edge states split, move, or disappear, or if the doublon band Chern numbers change from $(1,-2,1)$ before the bands touch the scattering continuum, the robustness claim fails.","tokens_in":13530,"feed_emoji":"🦋","tokens_out":12521,"duration_ms":115037,"temperature":0.7,"pith_summary":"This paper argues that periodically modulated interactions are enough to make two-particle bound states (doublons) topological in one-dimensional lattices. In the strongly correlated limit, the two-boson sector of a Bose-Hubbard chain with on-site interaction $U_j=U\\cos(2\\pi\\alpha j+\\varphi)$ and uniform pair-hopping $g$ reduces exactly to the one-dimensional Aubry-André-Harper model, so the doublon bands form a Hofstadter butterfly with Chern numbers $(1,-2,1)$. The in-gap states are doublon edge states with both bosons at the same end; weak nearest-neighbor hopping preserves them, while stronger hopping drives the doublon bands into the scattering continuum (topological bound states in the continuum) and eventually dissociates the doublons. The same construction works for modulated pair-hopping and for spinless fermions, and in the mapped two-dimensional single-particle picture it yields corner states in one or two diagonal corners.","feed_headline":"Spatially modulated interactions make boson pairs topological","feed_subtitle":"At strong interaction, two-boson bands become a Hofstadter butterfly with quantized Chern numbers and protected edge doublons.","key_machinery":"The central object is the exact reduction of the two-particle sector to a one-dimensional Aubry-André-Harper model in the strongly correlated limit. At $J=0$ the two-boson amplitudes decouple, and the pair-hopping term $g$ couples only neighboring diagonal amplitudes $\\beta_{j,j}$, yielding $g(\\beta_{j-1,j-1}+\\beta_{j+1,j+1})=(E-U_j)\\beta_{j,j}$. Treating the modulation phase $\\varphi$ as the momentum of a synthetic second dimension turns this into the Hofstadter model with flux $2\\pi\\alpha$, which supplies the butterfly spectrum, the Chern numbers, and the bulk-boundary correspondence for the doublon bands. For modulated pair-hopping $g_j=g[1+\\lambda\\cos(2\\pi\\alpha j+\\varphi)]$, the same reduction gives the off-diagonal AAH model, which becomes the SSH model at $\\alpha=1/2$ and carries chiral zero-energy doublon edge modes. For $J\\neq0$, second-order perturbation theory produces an effective single-particle Hamiltonian for the doublon subspace, with renormalized on-site and hopping terms $U_j+2J^2/g$ and $g+J^2/g$, preserving the diagonal AAH form.","core_discovery":"The paper's central claim is that the strongly interacting two-particle sector of a one-dimensional Hubbard-type model inherits the full topological structure of the Aubry-André-Harper model when interactions are spatially modulated. At $J=0$ and with $U_j=U\\cos(2\\pi\\alpha j+\\varphi)$, $g_j=g$, the Schrödinger equation for the doublon amplitudes $\\beta_{j,j}$ is $g(\\beta_{j-1,j-1}+\\beta_{j+1,j+1})=(E-U_j)\\beta_{j,j}$, which is precisely the diagonal AAH model; the phase $\\varphi$ acts as the momentum of a second dimension, mapping the system to the Hofstadter model with flux $2\\pi\\alpha$. At $\\alpha=1/3$ the three doublon bands carry Chern numbers $(1,-2,1)$, and open-boundary spectra show topological in-gap doublon edge states whose densities concentrate at the chain ends. Modulating the pair-hopping instead gives the off-diagonal AAH model (reducing to SSH at $\\alpha=1/2$) with chiral zero-energy doublon edge states. For weak $J\\neq0$ the paper derives an effective diagonal AAH model for the doublon subspace with renormalized parameters, so the edge states survive weak hopping; when the doublon bands meet the scattering continuum, topological doublon bound states in the continuum appear, and beyond that the doublons collapse. The spinless-fermion version maps to a generalized AAH model with the same topological invariants and edge states.","pith_inferences":["An immediate extension left implicit in the paper is that the exact $J=0$ reduction should hold for every rational $\\alpha$, with the number of in-gap doublon edge states following the Diophantine integers of the Hofstadter butterfly rather than only the $\\alpha=1/3$ example shown.","Because the analytical $J\\neq0$ argument uses localized two-boson states as its starting point while the pair-hopping term already delocalizes them, a direct numerical check of whether the in-gap energies track the renormalized parameters $U_j+2J^2/g$ and $g+J^2/g$ would separate the exactly solved strong-interaction limit from the weak-hopping regime.","The one- or two-corner localization in the mapped 2D system is a distinctive fingerprint: measuring the number and diagonal position of corner states as the modulation phase is swept could distinguish interaction-induced doublon topology from ordinary single-particle higher-order topology.","The same mapping suggests that for incommensurate $\\alpha$ the doublon sector would inherit the Aubry-André localization transition, giving a two-particle analog of Anderson localization that the paper does not discuss."],"forward_implications":["Spatially modulated interactions are a sufficient ingredient for two-particle topology: at $J=0$ the doublon bands form a Hofstadter butterfly with quantized Chern numbers, with no modulated hopping required.","At $\\alpha=1/3$ the three doublon bands carry Chern numbers $(1,-2,1)$, which implies exactly two topologically protected in-gap doublon edge states under open boundary conditions.","Weak nearest-neighbor hopping renormalizes the effective doublon model but leaves it in the Aubry-André-Harper class, so the topological edge states persist until the doublon bands touch the scattering continuum; the resulting topological bound states in the continuum are edge-localized doublon modes.","At sufficiently large hopping the doublons collapse: both bulk and edge doublon states dissociate into two weakly interacting bosons, setting an upper bound on the interaction-induced topological phase.","In the mapped 2D single-particle picture, the same phases appear as topological insulators and topological metals whose corner states occupy only one or two diagonal corners."],"supporting_citations":[{"why":"Supplies the two-particle-to-2D single-particle mapping used to turn the two-boson problem into a 2D lattice model.","marker":"[48–50]"},{"why":"Defines the Aubry-André and Harper models whose diagonal form the $J=0$ two-boson sector reduces to.","marker":"[77, 78]"},{"why":"Establishes the topological properties of the diagonal AAH model, including edge states and equivalence to the Hofstadter model.","marker":"[79, 80]"},{"why":"The Hofstadter model whose butterfly spectrum and flux-picture phases underpin the Chern-number analysis.","marker":"[81]"},{"why":"Provides the two-body bound-state and topological edge-state framework for doublons in 1D lattices, including doublon collapse.","marker":"[14, 15]"},{"why":"Supports the analysis of interaction-induced topological doublon states and their instability when hopping becomes strong.","marker":"[88, 89]"},{"why":"Defines the off-diagonal AAH model with topological zero-energy modes used for the modulated pair-hopping case.","marker":"[91]"},{"why":"Supplies the second-order perturbation method for deriving the effective doublon Hamiltonian at weak hopping.","marker":"[46, 47]"}],"fun_headline_variants":["Spatially modulated pairing makes doublons topological","Doublon edge states from periodic interaction strength","Topological doublons from modulated interactions in Hubbard chains","Butterfly doublon bands with Chern numbers via spatial modulation","Modulated interactions produce topological two-boson states"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analytic claim that weak nearest-neighbor hopping leaves the topological doublon states intact rests on a perturbation calculation whose unperturbed states are the localized two-boson states, even though the pair-hopping term already moves doublons between neighboring sites and so those states are not actual eigenstates.","fun_headline_variants_meta":{"raw":{"variants":["Spatially modulated pairing makes doublons topological","Doublon edge states from periodic interaction strength","Topological doublons from modulated interactions in Hubbard chains","Butterfly doublon bands with Chern numbers via spatial modulation","Modulated interactions produce topological two-boson states"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000243,"raw_usage":{"total_tokens":1586,"prompt_tokens":1062,"completion_tokens":524,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":678,"completion_tokens_details":{"reasoning_tokens":445}},"tokens_in":678,"tokens_out":524,"duration_ms":5469,"temperature":1.0,"reasoning_tokens":445,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T17:41:35.466484+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run exact diagonalization of the two-boson Hamiltonian at small but nonzero $J$ (for instance $J/g=0.1$) with $\\alpha=1/3$, and compare the in-gap doublon edge-state energies and density profiles with the diagonal AAH prediction using $U_j+2J^2/g$ and $g+J^2/g$; if the edge states split, move, or disappear, or if the doublon band Chern numbers change from $(1,-2,1)$ before the bands touch the scattering continuum, the robustness claim fails.","supporting_citations":[{"cited_title":"Ganeshan, K","cited_arxiv_id":null,"evidence_quote":"Defines the off-diagonal AAH model with topological zero-energy modes used for the modulated pair-hopping case."}],"review_version":1}