{"id":"dac5061a-9415-4dbe-8940-d4d6070715e1","arxiv_id":"2502.10494","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A single long-range hop in a non-Hermitian chain produces roughly one tentacle-like bound state per lattice spacing of the hop, with localization length proportional to the hop distance.","lead":"This paper studies a simple one-dimensional chain with uneven hopping in the two directions, and adds one long-distance hop between two distant sites. It finds that this single long-distance hop can create many new tentacle-like energy states, with the size of the localized bound states set by the hop's range, not by the chain length.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Direction-dependent counting: Eq. (22)/(23) is derived only for p<q and E−; for p>q the exact secular equation can place all roots inside the Bloch ellipse, so the stated |p−q|+1 outside-tentacle claim is not established as written.","rationale":"The reader's weakest assumption concerned the unquantified 'p+q large' condition in Sec. IV B. My check identifies a more specific and load-bearing gap: the asymptotic simplification used for Eq. (22) is applied only to the p<q case, and for p>q the exact secular equation has a different scaling in α^{2(p−q)}. At finite, physically representative values of t_pq (|t_pq| just above 1), the roots can all lie inside the Bloch ellipse rather than in the E− region to which the localization-length formula Eq. (23) applies. This does not necessarily invalidate the existence of |p−q|+1 tentacle branches, but it does invalidate the unqualified statement that the number of tentacles outside the Bloch/non-Bloch continuum is |p−q|+1 and that their localization length is given by Eq. (23). The paper's own derivation is internally consistent for p<q and E−, and the numerical figures for those cases are supportive, so I do not see a demonstrated fatal flaw. However, the abstract and conclusion overstate the range of validity. Since the prior verdict was already CONDITIONAL with moderate confidence, my concern reinforces that conditional rather than changing it to accept or reject.","tokens_in":14494,"tokens_out":40570,"duration_ms":439328,"concrete_test":"For PBC with α²=0.7, p>q, m=p−q=10, and t_pq=10, solve the exact secular equation t_pq α^{2m}=λ2^{m+1}−α²λ2^{m−1} and count roots with |λ2|>1 (Rouché suggests zero). Then diagonalize the L=600 chain with p=L/3, q=p−10 and count eigenenergies lying outside the PBC ellipse; compare with the 11 expected from Eq. (22). Repeat with p<q (p=L/3−10, q=L/3) as a positive control. If p>q yields zero outside-E− tentacles while p<q yields 11, the direction-free statement of Sec. V and the abstract must be qualified, and Eq. (23) cannot be applied without specifying the impurity direction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The quantitative tentacle count and localization length are derived in Sec. IV B for p<q and for impurity energies in E−, where Eq. (A6)/(A11) are simplified to λ2^{q−p+1}≈t_pq and Eq. (23) follows. But the self-consistent equation is direction-dependent. For PBC with p>q and m=p−q>0, Table I/Eq. (A7) gives d_qp = −α^{2m}/[λ2^{m−1}(λ2²−α²)], so the secular condition d_qp=−1/t_pq becomes t_pq α^{2m} = λ2^{m+1} − α²λ2^{m−1}. This is not the equation λ2^{m+1}=t_pq used for Eq. (22). At moderate coupling the difference is severe: take α²=0.7, m=10, t_pq=10. Then the equation is 0.28 = λ2^{11} − 0.7λ2^9. On |λ2|=1, |λ2^{11}−0.7λ2^9| ≥ 0.3 > 0.28, so Rouché's theorem gives all 11 roots with |λ2|<1; none lie in the E− region where Eq. (23) was derived. The tentacles still exist but lie inside the Bloch ellipse (E+), with different energies and localization behavior. The same asymptotic structure appears for OBC p>q via Eq. (A11). Thus the abstract/conclusion claim that the number of wings is |p−q|+1 with localization length Eq. (23) is under-supported: it holds for one direction of the directed impurity hop, and the paper does not state or prove what happens for the opposite direction at finite t_pq.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the Hatano-Nelson chain with an additional single long-range, directed hopping t_pq c^†_p c_q. The main formal result is Eq. (5): for L→∞ and fixed p, q, the spectrum is the impurity-free spectrum σ(H0) together with discrete values λ outside σ(H0) solving d_qp(λ) = −1/t_pq, where d_qp is the (q,p) resolvent entry. The resolvent entries are tabulated for PBC and OBC and for the E± regions (Table I). The authors interpret the discrete solutions as tentacle-like spectral branches and, in Sec. IV B, approximate them for p<q and large p+q by Eq. (22), obtaining q−p+1 impurity energies outside the Bloch ellipse with localization length Eq. (23). The claims are supported by direct diagonalization in Fig. 5(a,d).","tokens_in":14809,"tokens_out":9853,"duration_ms":100713,"significance":"If the results hold as stated, this is a valuable exact analytical treatment of a nonlocal impurity in a non-Hermitian lattice: the spectrum is obtained from a scalar secular equation with no fitting parameters, the tentacle count scales with the coupling distance rather than the system size, and the localization behavior is cleanly different from both the NHSE and local-impurity scale-free localization. The Green's-function approach and the transfer-matrix setup are transferable, and the numerical validation in Fig. 5 is a genuine cross-check. However, because the clean quantitative laws are established only in one hopping direction and under an unquantified large-p+q assumption, the advertised generality is not yet demonstrated.","major_comments":[{"comment":"The asymptotic counting and localization length are derived only for p<q and λ∈E−. For the opposite direction p>q, with m=p−q>0, the PBC E− secular condition in Eq. (A7) becomes t_pq α^{2m} = λ2^{m+1} − α² λ2^{m−1}, not λ2^{m+1} ≈ t_pq. For α²=0.7, m=10, and t_pq=10, Rouché's theorem applied on |λ2|=1 gives |λ2^{11} − 0.7 λ2^9| ≥ 0.3 > 0.282, so all 11 roots satisfy |λ2|<1 and none lie in E−. Thus Eq. (23) and the statement that the wings lie outside the Bloch ellipse are not established for p>q. The OBC case in Eq. (A11) has the same direction-dependent structure. Since Eqs. (22)-(23) are the quantitative core of the paper and the abstract/conclusion are stated for the coupling distance |p−q| without direction restrictions, this needs either a separate p>q analysis or a clear direction/strength qualification.","section":"Sec. IV B, Eqs. (22)-(23); Table I and Eq. (A7)"},{"comment":"The paper says the self-consistent equations simplify when \"p+q is large enough\" but does not state how large p+q must be or characterize the corrections. Because Eq. (22) is asymptotic, the quantitative predictions q−p+1 and Eq. (23) may fail for modest p+q or for impurity sites close to the boundaries. The numerical tests in Fig. 5 use p=L/3 and |p−q|=10, which is a favorable regime. The authors should provide an error estimate or systematic numerical tests over a range of p, q, and t_pq to delineate the validity of Eqs. (22)-(23).","section":"Sec. IV B, Eq. (22)"},{"comment":"The spectral decomposition Eq. (5) is asserted as the thermodynamic-limit result, with the impurity at finite positions. The paper should state the standard relative-compactness/limit-operator argument or cite the specific theorem that guarantees the impurity only adds discrete spectrum outside σ(H0) while leaving σ(H0) unchanged. As written, Eq. (5) is validated only in examples, and the precise sense in which the finite-L spectra converge to the union in Eq. (5) is not explained.","section":"Eq. (5), Sec. III"}],"minor_comments":[{"comment":"The typeset equation \"λp−q−1_2 ≃ 1/tpq\" is inconsistent with the displayed solution (eϱ, eθn) ≈ (^{q−p+1}√|tpq|, 2π n/(q−p+1)); it should presumably be λ2^{q−p+1} ≈ t_pq. Please correct the exponent and define the notation eϱ and eθn.","section":"Sec. IV B, Eq. (22)"},{"comment":"The caption says \"the localization length ξ of all the |p−q−1| impurity states,\" but the text and Eq. (22) give q−p+1 impurity states. This should be |p−q|+1 (or q−p+1 in the p<q case).","section":"Fig. 5 caption"},{"comment":"The entries in Table I are hard to parse because of missing parentheses and ambiguous superscripts, for example \"λp−q_1 −λp−q_2 /λ1−λ2\". Please rewrite the table with explicit parentheses and with the p<q and p>q cases clearly separated.","section":"Table I"},{"comment":"There are several typographical errors and awkward phrasings, including \"Teoplitz\", \"eigenegy\", \"cann't\", \"connectting\", and \"senser\". A careful proofreading pass is needed.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the scope of the journal and the exact Green's-function method is solid, with a genuine numerical cross-check and no fitted parameters. The main issue is that the quantitative claims in Sec. IV B are derived only for p<q and are stated in the abstract and conclusion without that restriction; the explicit p>q counterexample shows this is a load-bearing gap rather than a presentation issue. I believe it is fixable within a major revision by adding the p>q analysis or by carefully qualifying the claims."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Xiaoming and Haiping have a genuinely useful paper here. The main result—that a single long-range directed hop in the Hatano-Nelson chain produces tentacle-like spectral branches and bound states whose number and localization length are set by the hop range—is new and, as far as I can see, correctly derived. The exact secular equation d_qp = −1/t_pq follows from a standard rank-one perturbation of a Toeplitz/Laurent operator, and the Green's function entries in Table I check out. No free parameters, no circular fitting; the numerical spectra in Fig. 5 agree with the analytic points. The paper also cites the relevant mathematics (Schmidt–Spitzer, Böttcher–Embree–Sokolov) rather than rediscovering it.\n\nThe soft spots are real but concentrated. The quantitative law, q−p+1 tentacles and ξ = −(q−p+1)/ln(t_pq), is derived only for p<q and for impurity energies outside the Bloch ellipse (E−). The stress-test note is correct: for p>q the exact PBC secular equation becomes t_pq α^{2m} = λ_2^{m+1} − α^2 λ_2^{m−1}; at moderate coupling this places all roots inside the ellipse, so the outside-tentacle count and Eq. (23) do not hold for that direction. The abstract and summary state |p−q|+1 without qualification, so the paper overreaches. This is not a fatal flaw—the tentacles still exist for p>q, but their energies and localization lengths are different, and the authors need to either derive the opposite direction or state the restriction.\n\nTwo smaller issues. The 'p+q large enough' condition is never quantified; a referee should ask for a bound or a numerical check of the corrections. And the decay-direction sign convention flips between the text and the Fig. 5 captions, which is confusing but does not affect the mathematics.\n\nBottom line: this deserves a serious referee. The exact decomposition is solid and the phenomenon is worth reporting. The revision should address the asymmetry between the two hop directions and either extend or carefully scope the counting/localization claims. I'd be happy to cite the corrected version in work on non-Hermitian impurities; as written I'd cite it with a caveat.","headline":"Solid exact-solution paper with a real overgeneralization: the tentacle/loc-length law holds only for one hop direction.","tokens_in":15377,"tokens_out":4715,"would_cite":true,"duration_ms":43785,"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":"Adding a single long-range nonreciprocal hop to a Hatano-Nelson chain makes tentacle-like spectral branches emerge from the otherwise unchanged continuum, with as many branches as the hop range and bound states localized near the impurity.","keywords":["Hatano-Nelson model","non-Hermitian skin effect","long-range impurity coupling","tentacle-like spectra","bound states","localization length","transfer matrix","Toeplitz matrices"],"falsifier":"Use exact arbitrary-precision diagonalization of the finite chain in Eq. (1) with $L=100$, $p=2$, $q=4$, $\\alpha^2=0.7$, $t_{pq}=10$ and count the eigenvalues outside $\\sigma(H_0)$; if their number is not $q-p+1=3$ or their localization lengths do not follow $|q-p+1|/\\ln(t_{pq})$, the asymptotic claim in Sec. IV B is the part that fails.","tokens_in":14224,"feed_emoji":"🐙","tokens_out":10009,"duration_ms":94206,"temperature":0.7,"pith_summary":"This paper studies the Hatano-Nelson chain, a one-dimensional non-Hermitian model whose eigenstates and spectrum already depend sharply on boundary conditions, and asks what happens when one additional long-range hopping term couples two sites far from either boundary. The central claim is that this single impurity hop is not a small perturbation: in the thermodynamic limit the spectrum is exactly the original Bloch (periodic) or non-Bloch (open) continuum plus a set, possibly empty, of discrete tentacle-like branches sticking out of that continuum. The tentacles are bound states pinned near the impurity, and their number is set by the coupling range $|p-q|+1$ rather than by the system size, while their localization length grows with that range. The authors derive this from an exact self-consistent condition involving the unperturbed Green's function and confirm the localization scaling with a transfer-matrix computation. If correct, the result gives a non-perturbative mechanism for creating and controlling localized states in non-Hermitian lattices.","feed_headline":"Single long-range hop sprouts tentacle spectra in non-Hermitian chain","feed_subtitle":"A hop between distant sites pins bound states whose count and localization length grow with the hop's range.","key_machinery":"The load-bearing object is the $(q,p)$ entry of the unperturbed resolvent, $d_{qp}(\\lambda)=\\left[(H_0-\\lambda I)^{-1}\\right]_{qp}$, the Green's function from the impurity target to the impurity source. Because the impurity is a rank-one hopping term, an eigenvalue of $H$ outside the continuum of $H_0$ appears exactly when the rank-one perturbation closes the resolvent, which is the condition $d_{qp}(\\lambda)=-1/t_{pq}$. The paper evaluates $d_{qp}$ explicitly for periodic and open boundaries by expanding the Laurent and Toeplitz resolvent in its Wiener-Hopf factors; the resulting formula depends on $p-q$, on the boundary condition, and on whether $\\lambda$ lies inside or outside the Bloch ellipse. For the localization length, the companion machinery is a one-step modified transfer matrix, iterated with QR decomposition to get the Lyapunov exponent $\\gamma=\\xi^{-1}$.","core_discovery":"The paper's main analytical result is Eq. (5): for a single long-range coupling from site $q$ to site $p$ placed far from the boundaries, the thermodynamic-limit spectrum is $$\\lim_{L\\to\\infty}\\$\\sigma$(H) = \\$\\sigma$(H_0) \\cup \\{\\$\\lambda$\\in\\mathbb{C}\\setminus\\$\\sigma$(H_0) : d_{qp}(\\$\\lambda$) = -1/t_{pq}\\},$$ where $d_{qp}$ is the $(q,p)$ entry of the resolvent (Green's function) of the unperturbed tight-binding chain, evaluated with the appropriate periodic or open boundary condition. The first piece is the unchanged Bloch ellipse or non-Bloch segment; the second piece is the impurity-induced part and is empty unless the coupling is strong enough. When the impurity sites are far from the boundaries and $p+q$ is large, the self-consistent equation simplifies to $\\lambda_2^{q-p+1}\\simeq t_{pq}$, giving approximately $q-p+1$ impurity eigenvalues, all lying on the same curve just outside the original continuum. These are the tentacle branches. The corresponding eigenstates are right- or left-decaying bound states centered at the impurity, with asymptotic localization length $\\xi = -(q-p+1)/\\ln(t_{pq})$, a scale set by the coupling range and independent of the chain length.","pith_inferences":["Beyond the paper's setup, the same resolvent condition should apply to several impurities or to higher-dimensional non-Hermitian lattices: each rank-one hop would add a finite set of tentacle branches indexed by the range of that hop, provided the branches do not overlap.","Because the localization length is controlled by the distance $|p-q|$ rather than by the system size, the result offers a design rule for experiments in photonic lattices or topolectrical circuits: choose the hop range to set the spatial extent of a bound state.","The count $q-p+1$ has the flavor of a topological index of the rank-one perturbation, and verifying that interpretation would give a bulk invariant for impurity-induced bound states.","A natural stress test is the near-boundary or small-$p+q$ regime: the paper's quantitative count is derived for asymptotically large $p+q$, so deviations there would demarcate where the simple formula breaks down."],"forward_implications":["For both periodic and open boundary conditions, the exact spectrum is the original continuum plus impurity branches determined by $d_{qp}=-1/t_{pq}$, so the long-range hop cannot be treated as a perturbative correction.","In the strong-coupling, large-$p+q$ limit, the number of tentacle eigenvalues is $q-p+1$, set by the coupling range and not by the system size.","The tentacle states are exponentially localized around the impurity, with localization length $|q-p+1|/\\ln(t_{pq})$, so the hop's range controls the spatial extent of the bound state.","The analytical self-consistent equation reproduces the numerically diagonalized spectrum, giving a cheap way to compute single-impurity spectra for large chains where direct diagonalization is error-prone.","Under open boundary conditions the spectrum is a segment plus wings, and under periodic boundary conditions an ellipse plus wings, so boundary-condition sensitivity persists in the impurity branches."],"supporting_citations":[{"why":"Defines the Hatano-Nelson model whose non-Hermitian spectra and boundary sensitivity are being probed.","marker":"[55]"},{"why":"Supplies the Toeplitz and Laurent matrix formalism used to write down the resolvent entries $d_{qp}$.","marker":"[56]"},{"why":"Provides the spectral theory of infinite Toeplitz and Laurent matrices with localized impurities that underpins the self-consistent equation $d_{qp}=-1/t_{pq}$.","marker":"[57]"},{"why":"Extends that spectral theory to band matrices with a perturbed entry, supporting the decomposition into an unperturbed continuum plus impurity branches.","marker":"[58]"},{"why":"Gives the non-Bloch band theory used to identify the open-boundary continuum $\\sigma(H_0)=2\\alpha\\cos k$.","marker":"[9]"},{"why":"Justifies the transfer-matrix Lyapunov exponent computation that yields the localization length formula.","marker":"[59]"}],"fun_headline_variants":["Tentacle spectra from long-range impurity in Hatano-Nelson chain","Distant hop yields tentacle bound states in non-Hermitian chain","Long-range coupling creates tentacle spectrum in non-Hermitian model","Impurity with long reach: tentacle eigenvalues in Hatano-Nelson","Tentacle branches in non-Hermitian chain from one distant hop"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The quantitative predictions (the number of tentacles and the localization length) hold only when the impurity is far from the boundaries and the two coupled sites are far enough apart that the self-consistent equation reduces to a simple power law, and the paper does not state the threshold.","fun_headline_variants_meta":{"raw":{"variants":["Tentacle spectra from long-range impurity in Hatano-Nelson chain","Distant hop yields tentacle bound states in non-Hermitian chain","Long-range coupling creates tentacle spectrum in non-Hermitian model","Impurity with long reach: tentacle eigenvalues in Hatano-Nelson","Tentacle branches in non-Hermitian chain from one distant hop"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000305,"raw_usage":{"total_tokens":1763,"prompt_tokens":970,"completion_tokens":793,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":586,"completion_tokens_details":{"reasoning_tokens":698}},"tokens_in":586,"tokens_out":793,"duration_ms":8036,"temperature":1.0,"reasoning_tokens":698,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T19:01:03.299137+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Use exact arbitrary-precision diagonalization of the finite chain in Eq. (1) with $L=100$, $p=2$, $q=4$, $\\alpha^2=0.7$, $t_{pq}=10$ and count the eigenvalues outside $\\sigma(H_0)$; if their number is not $q-p+1=3$ or their localization lengths do not follow $|q-p+1|/\\ln(t_{pq})$, the asymptotic claim in Sec. IV B is the part that fails.","supporting_citations":[{"cited_title":"Stegmaier, et al","cited_arxiv_id":null,"evidence_quote":"Defines the Hatano-Nelson model whose non-Hermitian spectra and boundary sensitivity are being probed."},{"cited_title":"Schmidt and F","cited_arxiv_id":null,"evidence_quote":"Provides the spectral theory of infinite Toeplitz and Laurent matrices with localized impurities that underpins the self-consistent equation $d_{qp}=-1/t_{pq}$."},{"cited_title":"B¨ ottcher, M","cited_arxiv_id":null,"evidence_quote":"Extends that spectral theory to band matrices with a perturbed entry, supporting the decomposition into an unperturbed continuum plus impurity branches."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the non-Bloch band theory used to identify the open-boundary continuum $\\sigma(H_0)=2\\alpha\\cos k$."},{"cited_title":"B¨ ottcher, M","cited_arxiv_id":null,"evidence_quote":"Justifies the transfer-matrix Lyapunov exponent computation that yields the localization length formula."}],"review_version":1}