{"id":"4f5dce0d-c1a2-4d35-bfd0-5dfd5a574534","arxiv_id":"2506.21789","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A carefully tuned three-site extended Bose-Hubbard model with nearest-neighbour interactions is integrable and solved exactly by a Bethe ansatz.","lead":"This paper constructs a new exactly solvable three-site cold-atom model with nearest-neighbour interactions and a tilted potential, then derives its energy spectrum with a Bethe ansatz. The value is a clean integrable benchmark for a small system where generic three-site Bose-Hubbard models are chaotic.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 3's Bethe ansatz is invalid: identity (3.2) fails for n=1, so roots satisfying (3.6) do not yield eigenstates of the form (3.1).","rationale":"The reader's weakest assumption concerned the E=0 exceptional point where the Bethe equations are singular. That concern is valid but secondary. The more serious issue is that the Bethe ansatz solution fails for generic E>0 already in the simplest case. The identities (3.2)-(3.5) are the only link between the Hamiltonian action and the Bethe equations (3.6); since (3.2) is false, the derived equations (3.6) do not characterize the eigenstates of the form (3.1). Appendix B contains a correct polynomial method, but its roots are roots of Q, not the roots appearing in the product ansatz (3.1), so it does not rescue Section 3. The direct n=1 computation is unambiguous and does not rely on any approximation or special limit. Because the paper's principal solved result is incorrect as stated, the manuscript cannot be accepted even conditionally without a fundamental reformulation of the Bethe ansatz construction.","tokens_in":114,"tokens_out":61071,"duration_ms":1277826,"concrete_test":"Reproduce the n=1, N=1, U=1, mu=0, E/sqrt(2)=1 check in the basis |0>=|N,0,1>, |1>=|N,1,1>, |2>=|N,2,1>. Construct the 3x3 Hamiltonian using e1|m>=(2-m)|m+1>, f1|m>=m|m-1>, h1|m>=2(m-1)|m> and H1=U/4 h1^2 + (E/sqrt(2))(e1+f1), plus U N^2. Verify u1=1, u2=-1 solve (3.6). Then evaluate H on |Psi>=(e1^2-1)|0>=2|2>-|0> and compare with the eigenvector |2>-|0> at the same energy. If the residual |1> component is nonzero, the ansatz (3.1)+(3.6) is disproved.","verdict_should_be":"REJECT","load_bearing_attack":"The central claim rests on the operator identities (3.2)-(3.5), but (3.2) is algebraically false. For the smallest nontrivial module (n=1, N=1), take U=1, mu=0, and tunneling coupling E/sqrt(2)=1. The values u1=1, u2=-1 satisfy the Bethe equations (3.6) and give the energy 2 from (3.7). The ansatz state (3.1) is |Psi>=(e1^2-1)|0>=2|2>-|0>. Using the paper's su(2) actions e1|m>=(2-m)|m+1>, f1|m>=m|m-1>, h1|m>=2(m-1)|m>, a direct calculation gives e1|Psi>=-2|1>, whereas the right-hand side of (3.2) equals -4|1>. Consequently H|Psi>=4|2>-2|0>+2|1>, which is not proportional to |Psi>; the true eigenvector with energy 2 is |2>-|0>. Thus the claimed correspondence between (3.1) and (3.6) fails even at generic nonzero tunneling. Appendix B's polynomial method is not equivalent to the state ansatz (3.1): the roots v in (B.7) satisfy (3.6), but the eigenstate (B.2) is not prod(e1-v)|0>, because the coefficient map defining Q includes 1/j! factors. The paper conflates two different sets of roots.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes an integrable three-site extended Bose-Hubbard model with nearest-neighbour interactions, obtained by imposing the parameter constraints (2.2)-(2.4). The Hamiltonian is expressed through two commuting su(2) algebras inside o(4), and conserved operators are identified. A Bethe ansatz is then formulated: eigenstates are claimed to have the product form (3.1), with roots satisfying the Bethe ansatz equations (3.6) and energies given by (3.7). Section 4 asserts that the solution is complete and free of spurious solutions, with Appendix B supplying an ODE-polynomial completeness proof.","tokens_in":1536,"tokens_out":1630,"duration_ms":249242,"significance":"If correct, the paper would provide a notable example of an integrable open-chain extended Bose-Hubbard model with strictly nearest-neighbour couplings and an integrability-preserving tilt. The algebraic embedding into o(4) and the explicit parameter constraints are attractive, and the construction is genuinely parameter-free rather than fitted. However, the central Bethe ansatz is incorrect: the operator identities on which it relies fail already in the first nontrivial sector, so the claimed exact eigenstates and the correspondence with the Bethe equations are not established. The o(4) construction may contain useful ideas, but the main result as stated is not valid.","major_comments":[{"comment":"The claimed Bethe ansatz identities are false. Counterexample: take n=1, N=1, U=1, mu=0, and E/sqrt(2)=1. The roots u1=1, u2=-1 satisfy the Bethe equations (3.6): for k=1 the left side is 2*1^2/(1-(-1))=1 and the right side is 1+1-1=1, and similarly for k=2. Using the paper's su(2) actions e1|m>=(2-m)|m+1>, f1|m>=m|m-1>, h1|m>=2(m-1)|m>, the state (3.1) is |Psi>=2|2>-|0>. Direct computation gives e1|Psi>=-2|1>, whereas the right-hand side of (3.2) equals -4|1>, so (3.2) is false. Moreover, with H=1+(1/4)h1^2+e1+f1, one obtains H|Psi>=4|2>-2|0>+2|1>, which is not proportional to |Psi>, although formula (3.7) gives energy 2. Thus the pair (1,-1) is a spurious solution of (3.6), directly contradicting the assertion in Section 4 that spurious solutions cannot occur.","section":"Section 3, Eqs. (3.2)-(3.7)"},{"comment":"The derivation of (3.2)-(3.5) is invalid. In the x^m representation, the raising operator is e1=2nx-x^2 d/dx, not multiplication by x. The polynomial corresponding to the product state (3.1) is obtained by applying the differential operators (e1-u_j) to the constant function 1, which does not produce the polynomial product over j of (x-u_j). The manipulations in Appendix A treat e1 as multiplication by x and omit the derivative contributions; this is exactly why the n=1 counterexample in Major Comment 1 violates (3.2).","section":"Appendix A"},{"comment":"The completeness proof does not establish the claimed correspondence for the states (3.1). The polynomial Q(x) in (B.6) is defined through coefficients alpha_j/j!, whereas the product state (3.1) involves coefficients without the factorial weightings. Consequently, roots v_j of Q(x) satisfying (3.6) do not parametrize the state (3.1) in the manner claimed; the final one-to-one statement relates (B.6) to (3.6), not (3.1) to (3.6). The counterexample in Major Comment 1 shows that this distinction is not a minor technicality. In addition, the recursion (B.3)-(B.4) divides by the tunneling coupling E, leaving the E=0 case untreated, although that is a secondary issue relative to the mismatch with (3.1).","section":"Appendix B"}],"minor_comments":[{"comment":"The symbol E is used both for the tunneling coupling in (2.2)-(2.5) and for the energy eigenvalue in (3.7); this notational clash should be fixed.","section":"Notation"},{"comment":"The statement that the listed operators 'realise the o(4) Lie algebra' is under-specified: the cross-commutators such as [e1,e2], [h1,e2], and [h1,h2] are not listed. They should be stated explicitly to justify the o(3) direct-sum decomposition.","section":"Section 2"},{"comment":"The argument that spurious solutions cannot occur 'since boson creation operators do not admit a non-trivial kernel' is not applicable, because e1 is not a creation operator; the counterexample in Major Comment 1 is a spurious solution of the Bethe equations.","section":"Section 4"},{"comment":"The treatment of the U=0 case as 'diagonalisable by a canonical transformation' is asserted without proof; if U=0 is meant to be included in the parameter domain, this claim should be substantiated.","section":"Appendix B"}],"recommendation":"reject","confidential_remarks":"The stress-test counterexample is correct and decisive: the Bethe ansatz identities fail in the smallest nontrivial sector, and the paper's own parameter regime admits a spurious solution of (3.6). The o(4) algebraic construction and the ODE-polynomial method of Appendix B may be salvageable, but a valid exact solution would require replacing the product ansatz (3.1) with a different polynomial construction and re-deriving the Bethe equations from correct identities. This is a rewrite of the central result, not a local correction, so I recommend rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper constructs an integrable three-site extended Bose-Hubbard model with nearest-neighbor interactions and a tilt, and claims a Bethe ansatz solution. The integrable point itself is genuinely new: the constraints (2.2)–(2.4) reduce the Hamiltonian to a single su(2) sector, and that reduction is clean and worth knowing. The counting argument for the basis and the ODE-based completeness argument in Appendix B are also standard and mostly sound.\n\nThe problem is that the central Bethe state (3.1) is not an eigenstate. Identity (3.2) is false. For N=1, n=1, with U=1, mu=0, and E/sqrt(2)=1, the roots u1=1, u2=-1 satisfy the Bethe equations (3.6) and give energy 2 from (3.7), but the state (e1^2-1)|0> = 2|2>-|0> is not the eigenvector for energy 2. Direct calculation gives e1|Psi> = -2|1>, while the right-hand side of (3.2) equals -4|1>. Then H|Psi> is not proportional to |Psi>; the true eigenvector is |2>-|0>. So the claimed correspondence between (3.1) and (3.6) fails at generic tunneling.\n\nThe root cause is that Appendix B's polynomial method is not equivalent to the product ansatz. The polynomial Q(x) is defined from coefficients alpha_j divided by j!, so its roots v_j satisfy (3.6) but the eigenstate is the linear combination (B.2), not prod(e1-v_j)|0>. The paper conflates these two distinct sets of roots.\n\nThere is also a separate issue with the o(4) claim. The stated generators do not close under commutation: [e1, f2] = -sqrt(2) b1^dag b2, which is nonzero. So the o(4) framing is incorrect, although the su(2) reduction on the sector spanned by h1, e1, f1 is valid and the Hamiltonian may still have the conserved operator.\n\nBottom line: the model and the polynomial solution have merit, but the paper as written is not reliable. The product-form Bethe ansatz is invalid, and the completeness claim in Section 4 is unsupported. A referee should require a major revision: either ditch the product ansatz entirely and present the polynomial method as the solution, or fix the identities. The E=0 boundary case is a minor concern compared to these.\n\nI would not cite this in its current form, and I would not bring it to a reading group. However, it does deserve serious referee attention because the underlying integrable point, if properly stated, is a useful benchmark for the three-site Bose-Hubbard literature.","headline":"The paper's Bethe state ansatz is algebraically false, though the integrable model and the alternative polynomial method may still be salvageable.","tokens_in":11958,"tokens_out":17361,"would_cite":false,"duration_ms":162818,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["17B80","81R12"],"pacs":[],"model":"deepseek-v4-flash","headline":"A three-site Bose-Hubbard model with nearest-neighbour interactions is shown to be integrable, even with a tilting potential.","keywords":["Bose-Hubbard model","quantum integrability","Bethe ansatz","three-site model","dipolar bosons","o(4) Lie algebra","triple well","exact solution"],"falsifier":"Fix a small particle number (e.g., N = 2 or N = 4) and numerically diagonalize the Hamiltonian (2.1) under the constraints (2.2)–(2.4) for E = 0 and for E values approaching zero; compare with the energies predicted by (3.6)–(3.7). If at E = 0 the Bethe ansatz fails to reproduce the full spectrum, or if the enumeration of solutions to (3.6) becomes singular, the completeness claim fails at that point.","tokens_in":10813,"feed_emoji":"⚛️","tokens_out":2137,"duration_ms":23965,"temperature":0.7,"pith_summary":"The paper constructs an integrable 3-site extended Bose–Hubbard model whose interactions are restricted to on-site and nearest-neighbour terms, and which remains integrable when a tilting potential is added. Prior integrable 3-site models required either periodic boundary conditions or couplings that connect sites 1 and 3 directly; here the model uses only nearest-neighbour tunneling while still admitting a Bethe ansatz solution. If correct, this is the first such integrable model with a tilting term that preserves integrability, making it a useful testbed for quantum dynamics in triple-well potentials with dipolar atoms.","feed_headline":"Three-site Bose-Hubbard model is integrable even when tilted","feed_subtitle":"Exact Bethe ansatz gives the full spectrum for a dipolar triple well with nearest-neighbour tunneling.","key_machinery":"The central object is a realization of the o(4) Lie algebra in terms of the boson operators of three sites, with e1, h1, f1 forming one o(3) subalgebra and e2, h2, f2 a second. This algebra acts on the Fock space through a symmetry-adapted basis |N, m, n} where the Hamiltonian takes a simple quadratic form in h1 plus a linear coupling term. Bethe states are generated by applying products of (e1 − u_j) to lowest-weight vectors; the Bethe equations (3.6) arise from requiring the unwanted terms to cancel, and the ODE (B.8) with the distinct-roots argument provides the completeness proof.","core_discovery":"Under the parameter constraints (2.2)–(2.4), the general tilted 3-site extended Bose–Hubbard Hamiltonian (2.1) can be expressed in terms of two commuting o(3) subalgebras of an o(4) Lie algebra, yielding H = U $N^{2}$ + (U/4) $h1^{2}$ − (mu/2) h1 + (E/$\\sqrt$(2))(e1 + f1). This Hamiltonian conserves the Casimir invariants of o(4), giving it enough conserved quantities to be integrable. The paper then develops a Bethe ansatz: eigenstates take the product form (3.1) built from the o(3) raising operator e1, with rapidities satisfying the Bethe ansatz equations (3.6), and energies given by (3.7). A completeness proof in Appendix B shows that all eigenstates and all energy levels are obtained this way, with no spurious solutions, under the assumption that the tunneling strength E is nonzero.","pith_inferences":["The paper leaves the E = 0 limit unexamined; the completeness proof divides by E, and at E = 0 the Bethe equations become singular, so a separate treatment would be needed to claim complete integrability in that degenerate limit.","The existence of an integrable tilted model suggests that careful engineering of dipolar interaction strengths might realize integrable dynamics in existing triple-well cold-atom setups, since the required couplings (2.3) are in principle tunable by shaping the trapping ellipsoids.","The Bethe ansatz structure could allow closed-form expressions for observables like population imbalance or entanglement entropy in the resonant-tunnelling regime, which the paper identifies as a target for future work.","A direct numerical check for small particle numbers comparing the Bethe energies (3.7) with exact diagonalization for E close to zero would clarify the practical scope of the completeness claim."],"forward_implications":["The full spectrum and all eigenstates of the model are obtained analytically via Bethe roots, enabling exact studies of dynamics in a triple well with a tilt.","Because the model has no spurious Bethe solutions, every solution of (3.6) corresponds to a physical eigenstate, simplifying numerical and analytic work.","The formulation extends naturally to 4-site systems through o(3) ⊕ o(3) ⊕ o(3), as noted in the conclusion, which may be useful for atomtronic interferometry.","The integrability-preserving tilt offers a controlled way to tune between integrable and chaotic regimes in dipolar bosonic systems without breaking exact solvability."],"supporting_citations":[{"why":"Establishes the broader class of integrable extended Bose-Hubbard models on complete bipartite graphs, of which the present model is a nearest-neighbour specialization.","marker":"[36]"},{"why":"The previously studied 3-site model with closed boundary conditions for the quadratic interactions; the present work contrasts by restricting interactions to nearest-neighbour only.","marker":"[35]"},{"why":"The two-site Bose-Hubbard integrable case, providing the base construction that the algebraic method extends.","marker":"[20]"},{"why":"Source of the differential-operator method used to derive the Bethe ansatz equations for bosonic models.","marker":"[21]"},{"why":"Provides the Sturm oscillation results used to justify the completeness of the Bethe ansatz via the polynomial ODE.","marker":"[30]"}],"fun_headline_variants":["Exact Bethe ansatz for integrable tilted triple-well model","Bethe ansatz cracks integrable 3-site Bose-Hubbard with tilt","Integrable extended Bose-Hubbard triple well fully solved","Tilted triple-well Bose-Hubbard model solved exactly","New integrable 3-site Bose-Hubbard model solved via Bethe ansatz"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The completeness proof assumes the tunneling coupling E is nonzero and that the Bethe roots are distinct, because the recursion relations dividing by E break down at E = 0 where the spectrum becomes degenerate and the one-to-one correspondence is not established.","fun_headline_variants_meta":{"raw":{"variants":["Exact Bethe ansatz for integrable tilted triple-well model","Bethe ansatz cracks integrable 3-site Bose-Hubbard with tilt","Integrable extended Bose-Hubbard triple well fully solved","Tilted triple-well Bose-Hubbard model solved exactly","New integrable 3-site Bose-Hubbard model solved via Bethe ansatz"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000927,"raw_usage":{"total_tokens":3904,"prompt_tokens":810,"completion_tokens":3094,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":426,"completion_tokens_details":{"reasoning_tokens":2999}},"tokens_in":426,"tokens_out":3094,"duration_ms":22493,"temperature":1.0,"reasoning_tokens":2999,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:23:56.417286+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fix a small particle number (e.g., N = 2 or N = 4) and numerically diagonalize the Hamiltonian (2.1) under the constraints (2.2)–(2.4) for E = 0 and for E values approaching zero; compare with the energies predicted by (3.6)–(3.7). If at E = 0 the Bethe ansatz fails to reproduce the full spectrum, or if the enumeration of solutions to (3.6) becomes singular, the completeness claim fails at that point.","supporting_citations":[{"cited_title":"Quantum integrable multi-well tunneling models","cited_arxiv_id":"1606.00816","evidence_quote":"Establishes the broader class of integrable extended Bose-Hubbard models on complete bipartite graphs, of which the present model is a nearest-neighbour specialization."},{"cited_title":"Control of tunneling in an atomtronic switching device","cited_arxiv_id":"1710.05831","evidence_quote":"The previously studied 3-site model with closed boundary conditions for the quadratic interactions; the present work contrasts by restricting interactions to nearest-neighbour only."},{"cited_title":"The two-site Bose--Hubbard model","cited_arxiv_id":"cond-mat/0605486","evidence_quote":"The two-site Bose-Hubbard integrable case, providing the base construction that the algebraic method extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Source of the differential-operator method used to derive the Bethe ansatz equations for bosonic models."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Sturm oscillation results used to justify the completeness of the Bethe ansatz via the polynomial ODE."}],"review_version":1}