{"id":"0b79ab27-7e5a-4b92-826f-4f0021f70a72","arxiv_id":"2411.09196","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A newly introduced one-dimensional flat band superconductor model appears to remain integrable when the on-site/bond singlet conversion term is added, based on level spacing statistics.","lead":"This paper introduces a one-dimensional toy model of a flat band superconductor and uses exact diagonalization plus level spacing statistics to argue that it stays integrable when one coupling is turned on. The significance is that it provides a simpler setting to study quasiparticle localization and ergodicity breaking in flat band superconductors.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The data in Fig. 10 contradict the abstract's 'does not change with λ3': at λ2=0.0001 the mean gap ratio is 0.356 for λ3=0 but 0.186 for λ3=1, so the invariance claim is unsupported and the LIOM inference is not established.","rationale":"The central claim of the paper is explicitly that the λ2 = 0 level statistics are independent of λ3 and remain close to Poisson, and that this establishes LIOMs and an integrable generalization of the Creutz ladder. The most load-bearing assumption is therefore that the statistics really are unchanged when λ3 is turned on. The paper's own Fig. 10 shows a large, systematic change at small λ2, with the mean gap ratio dropping from 0.356 to 0.186 as λ3 goes from 0 to 1. If this difference is real (and not a finite-size artifact), the abstract's 'does not change' claim is false, and the evidence base for LIOMs collapses to a less specific statement that the system remains non-ergodic. The reader's weakest_assumption focused on the SGA alternative; I partially agree with that, but I find the numerical mismatch more fundamental because it challenges the factual premise before the interpretation. The SGA concern is nevertheless relevant and should be addressed: H_{0,λ3} satisfies the same SGA as H1 because H3 commutes with B+ (Eq. 18), so the level clustering could be a spectral property of the SGA rather than evidence of local conserved quantities. A careful finite-size and error-bar analysis, as proposed, would settle whether the claimed invariance holds. The paper has independent value (the three-body scattering solution, the model construction, the dynamics), and the issues are addressable by re-analysis and by tempering the claims, so keeping the reader's CONDITIONAL verdict is appropriate rather than moving to REJECT.","tokens_in":27290,"tokens_out":8842,"duration_ms":88373,"concrete_test":"Recompute the mean gap ratio of Figs. 8-10 at λ2 = 0, 0.0001, 0.001, 0.005, 0.01, 0.02 for λ3 = 0 and λ3 = 1, for L = 12 and, if feasible, L = 14 and L = 16, with error bars estimated by bootstrapping over the symmetry sectors included in each data point. If the λ3 = 1 curve does not approach the λ3 = 0 curve as λ2 → 0, and the separation persists or grows with L, then the central claim that the λ2 = 0 statistics are unchanged by λ3 is falsified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim (abstract; Sec. V; Sec. VI) is that the level spacing distribution for λ2 = 0 is essentially independent of λ3 and remains close to Poisson, implying LIOMs for H_{0,λ3}. This is not what the paper's own numerics show. In Fig. 10, at the smallest λ2 value computed, λ2 = 0.0001, the mean gap ratio changes from ⟨r⟩ = 0.3559 for λ3 = 0 to ⟨r⟩ = 0.1863 for λ3 = 1. That shift (~0.17) is larger than the whole Poisson-to-GOE window (0.386 to 0.536) and moves the system away from, not toward, the Poisson value 0.3863. Non-negligible differences persist at λ2 = 0.02 (0.4047 vs 0.3616) and λ2 = 0.06 (0.4437 vs 0.4489). The claim of 'unchanged' statistics is therefore an overstatement of a qualitative similarity: both distributions are sub-Poissonian, but the λ3 = 1 spectrum is far more degenerate at small λ2. This matters because the LIOM conclusion is the paper's headline result. Even if the statistics were identical, the inference is not secure: H_{0,λ3} inherits the spectrum generating algebra [H1, B+] = -(A+4)B+ since [H3, B+] = 0 (Eqs. 16-18), so the anomalous level clustering could be SGA towers rather than extensive LIOMs; the paper does not rule this out. The three-body scattering solution in Sec. IV is valuable, but it constrains only the few-body sector, not the existence of an extensive set of local conserved operators.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces the on-site/bond singlet (OBS) model, a one-dimensional fermionic lattice model inspired by the projected dice lattice Hamiltonian, with three terms: H1 (the Creutz-ladder-type term with known local integrals of motion), H2 (bond singlet hopping), and H3 (on-site/bond singlet conversion). The authors use exact diagonalization to study the energy spectrum and time evolution, solve the two-body and three-body problems analytically for λ2=0, and analyze level spacing statistics as a function of λ2 and λ3. The central claim is that for λ2=0 the level spacing distribution is essentially independent of λ3 and remains close to Poisson, which is interpreted as evidence for an extensive number of LIOMs for H_{0,λ3}, establishing the OBS model as a nontrivial integrable generalization of the projected Creutz ladder.","tokens_in":27664,"tokens_out":4408,"duration_ms":45476,"significance":"If the LIOM claim were established, the paper would provide a useful one-dimensional platform for studying quasiparticle localization in flat-band superconductors and would demonstrate level spacing statistics as a practical probe of integrability in such systems. The analytic few-body results, particularly the three-body scattering solution and the transmission coefficient, are concrete and valuable contributions, and the presentation is generally honest about the indirect nature of level statistics. However, the headline claim is not quantitatively supported by the paper's own numerics, so the significance is currently prospective rather than established.","major_comments":[{"comment":"The claim that the level spacing distribution for λ2=0 is essentially independent of λ3 is not supported by the reported data. At λ2=0.0001, the mean gap ratio changes from ⟨r⟩=0.3559 (λ3=0) to ⟨r⟩=0.1863 (λ3=1), a shift comparable to the entire Poisson-to-GOE interval (0.386 to 0.536) and in the direction away from the Poisson value. The difference persists at λ2=0.02 (0.4047 vs 0.3616) and is also visible in the histograms of Fig. 8, where the low-s peak is much larger for λ3=1. Because the LIOM conclusion in the abstract and Sec. V rests on this invariance, the authors need to either quantify the similarity (with error bars and a stated criterion) or soften the central claim to a qualitative one.","section":"Sec. V, Fig. 10"},{"comment":"The sub-Poissonian level spacing for λ2=0 is also consistent with the spectrum generating algebra [H1,B+]=-(A+4)B+ and [H3,B+]=0, which holds for all λ3 and produces equally spaced towers of states; this mechanism is not exclusive to systems with LIOMs. The paper does not rule out the SGA as the source of the anomalous level clustering for H_{0,λ3}, so the inference of an extensive set of LIOMs from the level statistics alone is not secure. A control study (for example, a model with SGA but no LIOMs) or an explicit perturbative construction of the LIOMs is needed.","section":"Sec. V; Sec. II A"},{"comment":"The level statistics are computed only for L=10 and L=12 with no finite-size scaling; the paper itself notes, in the discussion of Fig. 10, that the finite-size effect on the degeneracies cannot be confirmed due to computational limits. For a claim of integrability, one should show that ⟨r⟩ and P(s) do not drift with system size, or at least provide a two-size scaling analysis. The histograms in Figs. 8 and 9 also lack error bars, so it is unclear whether the differences between the λ3=0 and λ3=1 rows are statistically significant.","section":"Sec. V, Figs. 8–11"}],"minor_comments":[{"comment":"In the sentence introducing the three-body scattering problem, 'interacts with a a single unpaired particle' contains a duplicated article 'a'.","section":"Sec. IV"},{"comment":"The horizontal axis label appears to be missing; please label it explicitly (e.g., k a).","section":"Fig. 7"},{"comment":"The sentence 'The level spacing distribution of a model that behaves chaotically will follow that of a Gaussian ensemble' should specify the Gaussian orthogonal ensemble for time-reversal-symmetric systems, as done later in the same paragraph.","section":"Sec. V"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is honest about the indirect nature of level-statistics evidence and about computational limitations, but the abstract and conclusion state the invariance with λ3 more strongly than the data in Fig. 10 support. The self-citation to Ref. [25] is appropriate given the direct connection between the models. If the authors revise the central claim to a qualitative similarity and add the suggested controls, the paper could become a solid contribution to the flat-band superconductivity literature."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The OBS model is worth knowing about: it is a clean one-dimensional stand-in for the projected dice lattice Hamiltonian, and the two-body and three-body analytic results in Sec. IV are the strongest part of the paper. The three-body scattering solution, including the mechanism by which an unpaired particle moves by two sites via a second-order process, is genuinely useful for people studying quasiparticle dynamics in flat band superconductors.\n\nThe soft spot is the headline claim. The abstract states that the level spacing distribution 'does not change with λ3' and treats this as evidence for LIOMs and an integrable generalization of the Creutz ladder. The paper's own numerics do not support that. In Fig. 10, at λ2=0.0001, the mean gap ratio drops from 0.3559 (λ3=0) to 0.1863 (λ3=1). That is a large change, not a negligible one, and it moves the system away from the Poisson value. Smaller but nonzero differences persist at λ2=0.02 (0.4047 vs 0.3616). So the 'essentially independent of λ3' claim is an overstatement. The Sec. V text is more careful ('does not alter the statistics significantly'), but even that needs a quantitative check.\n\nThere is also a deeper issue with the inference. H1 has a spectrum generating algebra: [H1,B+]=-(A+4)B+, and since [H3,B+]=0, H_{0,λ3} inherits it. SGA towers of equally spaced states can produce sub-Poissonian level clustering without an extensive set of LIOMs. The paper does not rule this out. The level spacing evidence is thin: no error bars, no finite-size scaling beyond L=10 and 12, and the histogram at λ2=0.0001 for λ3=1 is very far from the Poisson value, which is more consistent with degeneracies from the SGA than with LIOMs.\n\nWhat is good: the model definition is clear, the symmetries are carefully listed, and the few-body analysis is executed with care. The authors also correctly note in Sec. VI that near-Poisson statistics does not automatically imply Bethe ansatz integrability. The numerical methods are standard and the code is sensible.\n\nBottom line: the paper deserves a serious referee, but the central claim needs to be reframed or substantially strengthened. A revision that either constructs the putative LIOMs (or shows SGA degeneracies are not responsible) and adds finite-size scaling would make the case credible. As it stands, it is a solid model-introduction with a suggestive but unproven integrability claim. I would send it to peer review and ask for major revision.","headline":"A useful new model with solid few-body analytics, but the headline claim that level statistics do not change with λ3 is contradicted by the paper's own Fig. 10, so the LIOM inference is not established.","tokens_in":28249,"tokens_out":3517,"would_cite":true,"duration_ms":33424,"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":"This paper claims that the level spacing distribution of a one-dimensional flat-band superconductor model stays Poissonian when a singlet-conversion term is added, signaling an extensive set of local integrals of motion.","keywords":["flat band superconductivity","level spacing statistics","local integrals of motion","ergodicity breaking","exact diagonalization","Creutz ladder","projected dice lattice","quasiparticle localization"],"falsifier":"Compute the level spacing distribution for a model that has the same spectrum-generating algebra but is known to have no local integrals of motion; if it too stays sub-Poissonian, then the near-Poissonian signal at $\\lambda_2=0$ does not by itself prove local integrals of motion. A positive check would be the explicit construction of a local operator commuting with $\\hat{\\mathcal{H}}_{0,\\lambda_3}$ or a Bethe-ansatz solution for nonzero $\\lambda_3$.","tokens_in":27024,"feed_emoji":"⚛️","tokens_out":6979,"duration_ms":71859,"temperature":0.7,"pith_summary":"Flat-band superconductors are hard to analyze because projecting onto a flat band produces long-range, many-body interactions, and quasiparticle excitations may be localized even without disorder. This paper introduces the one-dimensional on-site/bond singlet (OBS) model as a tractable stand-in for the projected dice-lattice superconductor and uses exact diagonalization to study its spectrum and dynamics. Its central claim is that when the bond-singlet hopping term is absent, the level spacing statistics remain close to Poissonian as the on-site/bond singlet conversion term is varied, even though the explicitly known conserved quantities of the Creutz-ladder limit no longer commute. This is presented as evidence for an extensive number of local integrals of motion at $\\lambda_2=0$ for all $\\lambda_3$, making the OBS model a nontrivial integrable generalization of the projected Creutz ladder. The result matters because explicit local conserved quantities are rarely available for flat-band superconductors, and level spacing statistics may serve as a general diagnostic of quasiparticle localization.","feed_headline":"Level-spacing test finds hidden conservation laws in flat band model","feed_subtitle":"Poissonian level spacings persist when a singlet-conversion term is added, signaling hidden local conservation laws.","key_machinery":"The central object is the three-term OBS Hamiltonian $\\hat{\\mathcal{H}}_{\\lambda_2,\\lambda_3}=\\hat{\\mathcal{H}}_1+\\lambda_2\\hat{\\mathcal{H}}_2+\\lambda_3\\hat{\\mathcal{H}}_3$, where $\\hat{\\mathcal{H}}_1$ is the projected Creutz-ladder term (on-site singlet hopping plus spin exchange), $\\hat{\\mathcal{H}}_2$ moves bond singlets along the chain, and $\\hat{\\mathcal{H}}_3$ converts on-site singlets into bond singlets and back. The diagnostic machinery is level spacing statistics: unfolded spacings $P(s)$ and the mean gap ratio $\\langle r\\rangle$, whose Poisson value signals integrability and whose Gaussian orthogonal ensemble value signals quantum chaos. The algebra carries the argument: $\\hat{\\mathcal{H}}_1$ and $\\hat{\\mathcal{H}}_3$ both satisfy a spectrum-generating relation with the total pseudospin raising operator, while $\\hat{\\mathcal{H}}_2$ does not, so only $\\hat{\\mathcal{H}}_2$ can destroy the integrable structure. In agreement with this, the numerical level statistics depend strongly on $\\lambda_2$ but are almost independent of $\\lambda_3$ at $\\lambda_2=0$.","core_discovery":"The paper's central discovery is that the level spacing distribution of the OBS Hamiltonian $\\hat{\\mathcal{H}}_{\\lambda_2,\\lambda_3}=\\hat{\\mathcal{H}}_1+\\lambda_2\\hat{\\mathcal{H}}_2+\\lambda_3\\hat{\\mathcal{H}}_3$ is essentially independent of $\\lambda_3$ when $\\lambda_2=0$. At $\\lambda_2=\\lambda_3=0$, the Hamiltonian $\\hat{\\mathcal{H}}_1$ is the projected Creutz-ladder Hamiltonian, known to be integrable with local integrals of motion given by the site spin parities; the authors show that at $\\lambda_3=1$ the unfolded level spacing distribution and the mean gap ratio $\\langle r\\rangle$ remain sub-Poissonian and close to the $\\lambda_3=0$ values, even though the site parities are no longer conserved. They interpret this as evidence that $\\hat{\\mathcal{H}}_{0,\\lambda_3}$ has an extensive number of local integrals of motion for arbitrary $\\lambda_3$, differing from the known Creutz-ladder operators, and that the model is an integrable generalization of the projected Creutz ladder. Including $\\hat{\\mathcal{H}}_2$ at $\\lambda_2>0$ breaks this structure and drives the statistics toward the Gaussian orthogonal ensemble result, identifying $\\hat{\\mathcal{H}}_2$ as the term that restores ergodicity.","pith_inferences":["A natural next test is to construct the local integrals of motion for $\\hat{\\mathcal{H}}_{0,\\lambda_3}$ perturbatively in small $\\lambda_3$; the three-body scattering solution suggests they acquire a nonlocal component that moves an unpaired particle by two sites per collision, so candidate operators should involve bond-singlet strings rather than bare site parities.","The same level-spacing signature could also appear in models with a spectrum-generating algebra but no conserved charges, so checking the statistics against such a control model would calibrate the method before it is applied to the two-dimensional dice lattice.","If the interpretation is correct, the OBS model offers an efficient platform for studying quasiparticle transport in flat-band superconductors, since the integrable line reduces the Hilbert space and may enable a Bethe-ansatz treatment for nonzero $\\lambda_3$.","A direct finite-size check on larger chains and other fillings would clarify whether the anomalously large small-spacing bin grows with system size, as expected if it is a genuine many-body degeneracy from local integrals of motion rather than a finite-size accident."],"forward_implications":["The full line $\\lambda_2=0$ in the OBS model is integrable: an extensive number of local integrals of motion exists for every value of $\\lambda_3$, even though the explicit Creutz-ladder operators no longer commute.","The near-Poissonian level spacing at $\\lambda_2=0$ is robust to boundary conditions, appearing for both periodic and open chains, so the hidden conservation laws are not an artifact of one geometry.","Adding the bond-singlet hopping term $\\lambda_2$ destroys the integrable structure and drives the gap ratio toward the Gaussian orthogonal ensemble value, identifying $\\hat{\\mathcal{H}}_2$ as the term responsible for ergodic behavior.","Level spacing statistics can serve as an unbiased probe of quasiparticle localization in flat-band superconductors, extending the dynamics-based evidence previously obtained for the projected dice lattice."],"supporting_citations":[{"why":"Supplies the known local integrals of motion (site spin parities) and integrability of $\\hat{\\mathcal{H}}_1$ in the Creutz-ladder limit, which is the baseline for the claimed generalization.","marker":"[24]"},{"why":"Provides the projected dice-lattice Hamiltonian that motivates the OBS model and supplies the comparison for energy spectra, dynamics, and quasiparticle localization.","marker":"[25]"},{"why":"Introduces the gap ratio $r_n$ used throughout Section V to distinguish Poissonian from chaotic level statistics.","marker":"[44]"},{"why":"Origin of the Wigner level-spacing approach connecting spectral statistics to the symmetry class and integrability of a Hamiltonian.","marker":"[62]"},{"why":"Establishes the standard interpretation of unfolded level spacing distributions for integrable versus chaotic systems and the unfolding procedure.","marker":"[64]"},{"why":"Provides the specific unfolding method adopted by the paper, including discarding edge levels and binning average spacings.","marker":"[65]"},{"why":"Establishes integrability of the isotropic Heisenberg chains into which $\\hat{\\mathcal{H}}_1$ decomposes, supporting the claim that $\\hat{\\mathcal{H}}_{0,0}$ is integrable.","marker":"[73]"},{"why":"Supplies the numerical mean gap ratio values for the Gaussian orthogonal ensemble and Poisson statistics used as references in Figures 10 and 11.","marker":"[77]"}],"fun_headline_variants":["Hidden integrals of motion revealed by level spacing in flat band model","Level spacings expose local conservation laws in flat band integrable model","Flat band model's level statistics hint at hidden conserved quantities","New integrable phase in flat band superconductor from level spacing data","Poissonian level spacings betray hidden local integrals in flat band model"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the near-Poissonian level spacing seen when the conversion term is switched on is caused by hidden local conservation laws, rather than by the special algebraic symmetry that both the integrable and non-integrable settings share.","fun_headline_variants_meta":{"raw":{"variants":["Hidden integrals of motion revealed by level spacing in flat band model","Level spacings expose local conservation laws in flat band integrable model","Flat band model's level statistics hint at hidden conserved quantities","New integrable phase in flat band superconductor from level spacing data","Poissonian level spacings betray hidden local integrals in flat band model"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000411,"raw_usage":{"total_tokens":2189,"prompt_tokens":1063,"completion_tokens":1126,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":679,"completion_tokens_details":{"reasoning_tokens":1036}},"tokens_in":679,"tokens_out":1126,"duration_ms":100823,"temperature":1.0,"reasoning_tokens":1036,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T20:55:37.209150+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the level spacing distribution for a model that has the same spectrum-generating algebra but is known to have no local integrals of motion; if it too stays sub-Poissonian, then the near-Poissonian signal at $\\lambda_2=0$ does not by itself prove local integrals of motion. A positive check would be the explicit construction of a local operator commuting with $\\hat{\\mathcal{H}}_{0,\\lambda_3}$ or a Bethe-ansatz solution for nonzero $\\lambda_3$.","supporting_citations":[{"cited_title":"Tovmasyan, S","cited_arxiv_id":null,"evidence_quote":"Supplies the known local integrals of motion (site spin parities) and integrability of $\\hat{\\mathcal{H}}_1$ in the Creutz-ladder limit, which is the baseline for the claimed generalization."},{"cited_title":"Swaminathan, P","cited_arxiv_id":null,"evidence_quote":"Provides the projected dice-lattice Hamiltonian that motivates the OBS model and supplies the comparison for energy spectra, dynamics, and quasiparticle localization."},{"cited_title":"P.Wigner, On the statistical distribution of thewidths and spacings of nuclear resonance levels, Mathematical Proceedings of the Cambridge Philosophical Society47, 790 (1951)","cited_arxiv_id":null,"evidence_quote":"Origin of the Wigner level-spacing approach connecting spectral statistics to the symmetry class and integrability of a Hamiltonian."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the standard interpretation of unfolded level spacing distributions for integrable versus chaotic systems and the unfolding procedure."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the specific unfolding method adopted by the paper, including discarding edge levels and binning average spacings."},{"cited_title":"Cao, W.-L","cited_arxiv_id":null,"evidence_quote":"Establishes integrability of the isotropic Heisenberg chains into which $\\hat{\\mathcal{H}}_1$ decomposes, supporting the claim that $\\hat{\\mathcal{H}}_{0,0}$ is integrable."},{"cited_title":"Giraud, N","cited_arxiv_id":null,"evidence_quote":"Supplies the numerical mean gap ratio values for the Gaussian orthogonal ensemble and Poisson statistics used as references in Figures 10 and 11."}],"review_version":1}