{"id":"b3f5afc3-cdbb-4684-9109-fec0fe923ba8","arxiv_id":"2507.11120","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Under a parameter condition (Γ0 ≤ Γ2/3), multicritical points in a third-neighbor SSH/Kitaev chain retain one localized zero-energy edge mode per end, giving a topological invariant w_mc = 1.","lead":"This paper shows that in a one-dimensional topological chain with couplings up to third neighbors, the special points where several quantum phases meet can still host stable, topologically protected edge modes, contrary to the usual expectation that such modes delocalize. A smart generalist might read it because it corrects a common assumption about topological quantum criticality and points to a parameter regime where edge modes survive even at multicritical points.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Zero-counting invariant at degenerate multicritical points is unproven; N=25 check cannot distinguish a true localized mode from a critical power-law mode.","rationale":"The paper's central claim is that M1 and M2 host a stable localized zero-energy mode at each end, with wmc = 1 obtained by counting zeros of f(ζ) strictly inside the unit circle. The load-bearing assumption is that this zero-counting rule, established in Ref. [1] for critical points with simple zeros on the unit circle, continues to hold when f(ζ) has a degenerate zero on the unit circle. The reader's weakest_assumption identifies exactly this point, and I agree it is the most direct threat to the central assertion: if the rule fails in the degenerate case, the invariant wmc = 1 does not guarantee a physical edge mode. The numerical support is thin because N = 25 is a single small size and gapless systems can exhibit finite-size edge-like states with power-law envelopes. I am not claiming the rule actually fails; the very short localization length from ζ2 = −1/2 makes the mode plausible, but the paper does not provide the necessary analytic boundary-condition check or finite-size scaling. A secondary but concrete flaw is the discriminant diagnostic: along the blue critical line C2, the cubic discriminant factorizes as Δ = Q(−1)^2 · Disc(Q), where Q is the residual quadratic. For Γ0 = 0.2, Γ2 = 1, Disc(Q) changes sign at Γ1 ≈ 0.7798, not at the multicritical point M1 at Γ1 = 0.8. Thus the sign of Δ does not flip at M1; it is positive in a neighborhood of M1 on both sides and only crosses zero at a nearby point where two roots inside the unit circle become degenerate. This invalidates the paper's claim that the sign flip uniquely identifies topological multicritical points, but it does not by itself disprove the existence of the edge mode, so it is secondary to the zero-counting concern. Because the central phenomenon is plausible and independently checkable, the reader's CONDITIONAL verdict is appropriate; no change to the verdict is needed.","tokens_in":17164,"tokens_out":24516,"duration_ms":265465,"concrete_test":"Write the real-space zero-energy equations for the open chain at M1 (Γ0 = 0.2, Γ1 = 0.8, Γ2 = 1, Γ3 = 0.4) and at M2, using the same open boundary conditions as in the numerics. Substitute the decaying ansatz ψ_n = (−1/2)^n for a left-edge mode on the appropriate sublattice and verify analytically that all boundary equations are satisfied for arbitrary chain length N. In parallel, exact-diagonalize chains of lengths N = 25, 100, 400, and 1600 at both M1 and M2 and check: (i) the zero-energy pair remains at E = 0 with splitting that decreases exponentially with N rather than as a power law; (ii) the edge probability |ψ_1|^2 stays O(1) as N grows, not ∝ 1/N; (iii) the inverse participation ratio of the zero mode is O(1), not ∝ 1/N. If the analytic ansatz fails or the edge weight decays with N, the zero-counting invariant is not valid at the degenerate multicritical points.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that the zero-counting invariant wc = Nz from Ref. [1] remains valid at M1 and M2, where the cubic f(ζ) in Eq. (10) has a double zero on the unit circle (ζ = −1 for M1, ζ = +1 for M2) in addition to the inside zero ζ2 = −1/2. Ref. [1] established the zero-counting rule for critical points with non-degenerate zeros on the unit circle; the central-charge formula c = N′z/2 explicitly assumes simple zeros. At the multicritical points the zeros on the unit circle are degenerate, so the transfer matrix has a Jordan-block structure and the standard proof mapping zeros inside the unit circle to normalizable edge modes is not directly applicable. The paper offers no independent proof or reference for this degenerate case. The only numerical evidence is a single open-chain diagonalization at N = 25 (Figs. 3 and 4). In a gapless system, a critical power-law-decaying state can mimic an exponentially localized edge mode at small N, so the edge probability at N = 25 does not rule out delocalization in the thermodynamic limit. The short localization length ξ = 1/ln 2 ≈ 1.44 from ζ2 = −1/2 makes the claim plausible, but the open-boundary equations must still be checked explicitly: the degenerate unit-circle roots alter the space of solutions of the recurrence, and it is not automatic that the single decaying exponential satisfying the bulk recurrence also satisfies the two boundary conditions at the open edge. If the boundary conditions are not satisfied except at finite N, then wmc = 1 would not correspond to a stable localized zero mode, and the paper's central assertion of topologically nontrivial multicritical points would be unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a one-dimensional spinless fermion chain (SSH/Kitaev) with hoppings and pairings up to third nearest neighbors. It identifies two multicritical points, M1 and M2, where high-symmetry and non-high-symmetry critical lines meet with quadratic dispersion, and claims that for Γ0 ≤ Γ2/3 these points host one stable localized zero-energy edge mode at each end, characterized by the zero-counting invariant wmc = 1. The identification is made through the roots of the complex polynomial f(ζ) = -(Γ0 + Γ1 ζ + Γ2 ζ^2 + Γ3 ζ^3)/2, through localization lengths ξ = -1/ln|ζ|, and through exact diagonalization of open chains. The paper further argues that the discriminant of f(ζ) vanishes at the multicritical points and flips sign along the high-symmetry critical lines only for the topologically nontrivial case, distinguishing them from trivial multicritical points. Finally, it studies the effects of bond disorder, reporting that the multicritical points survive weak disorder and that strong disorder produces a gapless, topologically nontrivial Anderson-localized phase with two zero-energy edge modes per end.","tokens_in":17454,"tokens_out":31985,"duration_ms":307150,"significance":"If correct, the main result provides a concrete counterexample to the conventional expectation that edge modes always delocalize at multicritical points, and it establishes a parameter regime in which gapped, critical, and multicritical phases are all topologically nontrivial. The clean-limit analysis is explicit and checkable: the roots in Eqs. (11)-(13), the localization lengths in Fig. 6, and the discriminant signs in Figs. 8 and 9 are concrete and falsifiable. The disorder study adds a useful robustness analysis. The main weakness is that the zero-counting invariant is applied at points where two zeros are degenerate on the unit circle, a case not covered by the cited proof, and the only numerical check at the multicritical point is a single chain size N = 25. In addition, the stated condition Γ0 ≤ Γ2/3 includes an equality case where the claim wmc = 1 fails. I also checked the stress-test suggestion that the discriminant flips at trivial multicritical points; for Γ0 = 0.5, Γ2 = 0.5 near Γ1 = ±1, Γ3 = 0, the discriminant behaves as δ^2 times a positive factor, so it does not flip, consistent with the paper's Fig. 9.","major_comments":[{"comment":"The central claim wmc = 1 at M1 and M2 is obtained by counting only the zero ζ2 strictly inside the unit circle and ignoring the double zero at ζ = ±1. The zero-counting invariant wc = Nz - Np and the central-charge formula c = N'_z/2 in Ref. [1] are established for critical points with non-degenerate zeros on the unit circle; at M1/M2 the unit-circle zeros are degenerate, which produces a Jordan block in the transfer-matrix solution and invalidates the standard proof that the number of localized edge modes equals the number of interior zeros. The paper offers no independent derivation or reference for this degenerate case. The only numerical evidence is the N = 25 open-chain diagonalization in Figs. 3 and 4, which cannot distinguish an exponentially localized mode from a critical power-law mode. Please provide a direct solution of the open-boundary recurrence at the degenerate point showing that the localized solution satisfies both boundary conditions, or a reference covering degenerate unit-circle zeros, and add finite-size scaling (e.g., energy splitting or IPR versus N) to support the thermodynamic-limit statement.","section":"Section IV, Eq. (10), Figs. 5-6"},{"comment":"The parameter condition is stated as Γ0 ≤ Γ2/3 for nontrivial multicritical points with wmc = 1. At equality, Γ0 = Γ2/3, the polynomial f(ζ) at M1 has a triple zero on the unit circle; for example, with Γ2 = 1, Γ0 = 1/3, Γ1 = 1, and Γ3 = 1/3, one obtains f(ζ) = -(1/6)(1 + ζ)^3. There is then no zero strictly inside the unit circle, so the paper's own counting gives wmc = 0, not 1. Thus the statement 'for Γ0 ≤ Γ2/3 ... nontrivial multicritical points with wmc = 1' is false at the boundary. The condition should be Γ0 < Γ2/3, or the equality case should be analyzed separately; this affects the abstract and the conclusion.","section":"Section IV, parameter condition"},{"comment":"The paper claims that the discriminant 'uniquely identifies the topological multicritical points and distinguishes them from the trivial ones.' The evidence for the topological case is a single parameter set (Γ0 = 0.2, Γ2 = 1), while the trivial cases are illustrated for three parameter sets. Since Δ = 0 also occurs at trivial quadratic multicritical points (Fig. 9), the distinguishing feature is not Δ = 0 itself but the sign flip along the high-symmetry critical lines. The paper does not prove that this sign flip occurs for all Γ0 < Γ2/3, nor that it is absent for all trivial multicritical points. Please either provide an analytic proof (it is accessible from the explicit roots) or state the criterion more cautiously as a numerically demonstrated property.","section":"Section IV, discriminant criterion"}],"minor_comments":[{"comment":"The sentence assigning wc = 2 segments to 'Γ1 > M1 and Γ1 < M2' on blue and red lines is inconsistent with the root configurations reported in Section IV and Fig. 7; for the blue line the wc = 2 segment is Γ1 < M1, and for the red line it is Γ1 > M2. Please correct the inequalities.","section":"Section II, phase description"},{"comment":"Equation (14) omits the factor 1/16 that comes from the leading coefficient -Γ3/2 in the cubic discriminant; the sign analysis is unaffected, but the formula as written is not the exact discriminant.","section":"Eq. (14)"},{"comment":"The caption of Fig. 3 lists the same parameter point for panels (a) and (b); panel (b) should be the M2 point.","section":"Fig. 3 caption"},{"comment":"In Section IV, the statement 'ζ2 < 1' should be '|ζ2| < 1', since the zero is negative and the condition for being inside the unit circle is on the modulus.","section":"Section IV, zero condition"}],"recommendation":"major_revision","confidential_remarks":"The stress-test note's specific claim that the discriminant flips at the trivial case Γ0 = 0.5, Γ2 = 0.5 is not borne out by the explicit expansion near Γ1 = ±1, Γ3 = 0 (Δ ∝ δ^2 > 0 on both sides), so I do not see a contradiction with Fig. 9. The main concern is the degenerate unit-circle zero-counting, which is a genuine gap in the central argument; it is fixable with an open-boundary solution or a reference and finite-size scaling. The equality case Γ0 = Γ2/3 is a clear error that should be corrected to a strict inequality."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core claim of this paper is likely correct: in the third-neighbor SSH/Kitaev chain with Γ0 ≤ Γ2/3, the multicritical points host one localized zero-energy mode per end, characterized by wmc = 1. That corrects the prior belief that edge modes always delocalize at multicritical points in these chains. The zero-counting argument is applied sensibly, and the numerical check at N = 25 is more convincing than the stress-test note suggests, because the localization length from the inside zero at ζ = −1/2 is about 1.44 lattice constants, so a power-law state would still have visible weight across the chain, which Figure 4 does not show. The derivation of the parameter condition Γ0 ≤ Γ2/3 is clean and the phase diagram is easy to follow.\n\nThe main soft spot is the discriminant claim. The paper states that a sign flip in the discriminant uniquely identifies topological multicritical points and is absent for trivial ones. That is false on its own terms. For the trivial case Γ0 = 0.5, Γ2 = 0.5, the quadratic multicritical point at Γ1 = 1 has Γ3 = 0, so the cubic reduces to a perfect square (1+ζ)^2. Just below that point (Γ1 = 0.9, Γ3 = 0.1), the roots are −1 and −2 ± i, so the discriminant is negative. Just above (Γ1 = 1.1, Γ3 = −0.1), the roots are all real, so the discriminant is positive. That is a sign flip, and the point is trivial (wc goes from 0 to 1). Figure 9 in the paper shows the discriminant as positive on both sides for all three Γ0 values, which directly contradicts the equations. This is a genuine mathematical error in the flagship diagnostic, and the section needs to be corrected or removed.\n\nTwo smaller issues: the abstract promises a physical mechanism involving 'kinetic inversion in higher-order terms,' but that phrase never appears in the body. The disorder section uses only ten realizations for the zero-energy phase diagrams at N = 450, which is thin for claims about a gapless Anderson-localized phase; the IPR data with 50 realizations at N = 2000 are better.\n\nThe stress-test worry about the zero-counting invariant at a degenerate unit-circle zero is worth noting, but it is not fatal here. The paper does not prove the extension of Verresen's rule to the degenerate case, and a careful referee should ask for that proof. Still, the short localization length and the open-boundary numerics make the central phenomenon plausible.\n\nWho should read this? People working on topological quantum criticality. The central result deserves serious refereeing, but not in its current form: the discriminant overclaim must be fixed, and the mechanism should be either defined or dropped. I would send it to review, asking for major revisions on those points.","headline":"Plausible new result on nontrivial multicritical points, but the discriminant signature claimed as unique is contradicted by the paper's own equations.","tokens_in":18037,"tokens_out":9207,"would_cite":true,"duration_ms":98008,"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":"At the multicritical points of a one-dimensional topological chain with third-neighbor couplings, the paper finds one stable zero-energy edge mode at each end, characterized by the invariant $w_{\\mathrm{mc}}=1$.","keywords":["topological multicritical points","Majorana zero modes","edge mode localization","zero-counting topological invariant","third-neighbor SSH chain","third-neighbor Kitaev chain","polynomial discriminant","Anderson localization"],"falsifier":"Diagonalize the open chain exactly at M1 and M2 for sizes $N = 25, 50, 100, 200, 400$ and check that exactly one zero-energy eigenstate stays pinned at each end. If the zero-energy count changes with size, the pair of degenerate zeros on the unit circle splits, or the decay length measured numerically disagrees with $\\xi = -1/\\ln|\\zeta|$ predicted by the in-circle zero, then $w_{\\mathrm{mc}} = 1$ is not the stable characterization the paper claims.","tokens_in":16944,"feed_emoji":"🔗","tokens_out":12890,"duration_ms":125610,"temperature":0.7,"pith_summary":"The paper claims that multicritical points, the parameter values where several gapped and gapless phases of a one-dimensional topological chain meet, can host localized zero-energy edge modes instead of being topologically trivial. In a spinless-fermion chain with hopping and pairing extended to the third nearest neighbor, and with parameters restricted to $\\Gamma_0 \\le \\Gamma_2/3$, the two multicritical points M1 and M2 each carry exactly one stable zero-energy mode localized at each end, characterized by the invariant $w_{\\mathrm{mc}} = 1$. This contradicts the earlier expectation, based on localization lengths diverging as the multicritical point is approached, that edge modes always delocalize into the bulk at such points. A reader should care because it extends the bulk-boundary correspondence, in zero-counting form, to the most singular points of the phase diagram and identifies where topologically protected modes survive for quantum information even while the bulk gap vanishes.","feed_headline":"Zero-energy edge modes survive where three phases meet","feed_subtitle":"A third-neighbor SSH/Kitaev chain keeps one localized zero mode at each end, even with the bulk gap closed.","key_machinery":"The central object is the complex function $f(\\zeta)$ obtained by writing the Bloch Hamiltonian in Majorana form and setting $\\zeta = e^{ik}$; for this model it is the cubic $f(\\zeta) = -\\Gamma_0/2 - \\Gamma_1\\zeta/2 - \\Gamma_2\\zeta^2/2 - \\Gamma_3\\zeta^3/2$. By Cauchy's argument principle, the number of zeros inside the unit circle (the pole count is zero) defines a topological invariant $w_c$ that counts localized edge modes even when the bulk gap closes, while non-degenerate zeros on the unit circle encode the central charge of the critical theory. At M1 and M2 the zeros arrange themselves as a degenerate pair on the unit circle ($\\zeta = -1$ for M1, $\\zeta = +1$ for M2) plus one real zero strictly inside, so $w_{\\mathrm{mc}} = 1$. The discriminant $\\Delta = \\Gamma_3^4(\\zeta_1-\\zeta_2)^2(\\zeta_1-\\zeta_3)^2(\\zeta_2-\\zeta_3)^2$ vanishes at the multicritical points and changes sign across them along the high-symmetry critical lines, which the paper uses to distinguish topological multicritical points, where the sign flips, from trivial ones, where it does not.","core_discovery":"In the third-neighbor Su–Schrieffer–Heeger or Kitaev chain, the authors show that restricting the parameter plane to $\\Gamma_0 \\le \\Gamma_2/3$ removes the trivial $w = 0$ phase, leaving only gapped phases with winding numbers $w = 1, 2, 3$ and only nontrivial critical phases. At the two multicritical points M1 and M2, located at $\\Gamma_1 = \\pm(3\\Gamma_0 + \\Gamma_2)/2$, the bulk gap closes with quadratic dispersion (dynamical exponent $z = 2$), yet the open chain retains exactly one zero-energy eigenstate localized at each end. The complex function $f(\\zeta) = -\\Gamma_0/2 - \\Gamma_1\\zeta/2 - \\Gamma_2\\zeta^2/2 - \\Gamma_3\\zeta^3/2$ has, at these points, one zero strictly inside the unit circle and two degenerate zeros on the unit circle (at $\\zeta = -1$ for M1 and $\\zeta = +1$ for M2), giving the invariant $w_{\\mathrm{mc}} = 1$. The discriminant of this cubic vanishes exactly at the multicritical points and flips sign across them along the high-symmetry critical lines, a signature the authors argue is unique to topological multicritical points and absent when a trivial $w = w_c = 0$ phase is present. Weak disorder shifts the multicritical points without destroying them, while strong disorder produces a gapless Anderson-localized phase that still carries two zero-energy modes localized at each end.","pith_inferences":["An extension the paper leaves implicit: in chains with couplings out to the $m$-th neighbor, the same zero-counting should assign $w_{\\mathrm{mc}}$ at a multicritical point by the number of zeros strictly inside the unit circle after a degenerate pair forms on it; the paper notes $w_{\\mathrm{mc}} > 1$ becomes possible with more couplings but does not give the general counting rule.","The discriminant sign flip could serve as an experimental diagnostic: scanning the local density of states across a suspected multicritical point should reveal the transition between two critical phases, with the sign flip marking where two roots coalesce through the real axis.","A testable reinterpretation follows from the paper's own comparison: the delocalization of edge modes reported at multicritical points in earlier studies of this model family may be caused by the coexisting trivial phase ($w = w_c = 0$) rather than by multicriticality itself; tuning $\\Gamma_0$ through $\\Gamma_2/3$ in the same model and watching the sign-flip property appear and disappear would set"],"forward_implications":["The conventional winding number is ill-defined exactly at a critical point, but the zero-counting invariant remains finite there and evaluates to $w_{\\mathrm{mc}} = 1$ at M1 and M2, so the bulk-boundary correspondence survives, in this generalized form, at multicriticalities.","The condition $\\Gamma_0 \\le \\Gamma_2/3$ makes the whole parameter plane nontrivial, so transitions between critical phases along a critical line never force the edge modes to delocalize at the crossing point.","The discriminant $\\Delta = 0$ with a sign flip across the point identifies a topological multicritical point; trivial multicritical points show $\\Delta = 0$ without a sign flip, so the sign behavior of $\\Delta$ gives a direct criterion to locate topologically nontrivial multicriticalities in other one-dimensional chains.","If the central claim is right, at weak disorder the two multicritical zero modes survive (the points merely shift in parameter space), and at strong disorder the system enters a gapless Anderson-localized phase that still carries two zero-energy modes at each end."],"supporting_citations":[{"why":"Supplies the zero-counting topological invariant $w_c = N_z - N_p$ that the paper applies to characterize critical and multicritical phases.","marker":"[1]"},{"why":"Earlier study of this model family that reported edge modes delocalizing at multicritical points; the claim this paper revises.","marker":"[15]"},{"why":"Earlier analysis of the same chain including the trivial $w = 0$ phase; provides the comparison case of trivial multicritical points.","marker":"[16]"},{"why":"Defines the Su–Schrieffer–Heeger model that the chain with extended couplings generalizes.","marker":"[41]"},{"why":"Defines the Kitaev chain that the extended-pairing Hamiltonian is based on.","marker":"[42]"}],"fun_headline_variants":["Three phases meet, edge modes persist at multicritical point","Quadratic gap closing with zero modes at multicritical points","Cubic discriminant uniquely flags topological multicritical points","Topological multicritical points keep zero modes when gap closes","Edge modes survive at topologically nontrivial multicritical points"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the number of edge-localized modes is still counted by the zeros of $f(\\zeta)$ strictly inside the unit circle at the multicritical point, where the other two zeros coincide exactly on the unit circle; this count is used without an independent proof that the on-circle degeneracy leaves it unchanged, and the numerical check runs at a single chain size, $N = 25$.","fun_headline_variants_meta":{"raw":{"variants":["Three phases meet, edge modes persist at multicritical point","Quadratic gap closing with zero modes at multicritical points","Cubic discriminant uniquely flags topological multicritical points","Topological multicritical points keep zero modes when gap closes","Edge modes survive at topologically nontrivial multicritical points"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000218,"raw_usage":{"total_tokens":1520,"prompt_tokens":1103,"completion_tokens":417,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":719,"completion_tokens_details":{"reasoning_tokens":337}},"tokens_in":719,"tokens_out":417,"duration_ms":5508,"temperature":1.0,"reasoning_tokens":337,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:17:46.543288+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Diagonalize the open chain exactly at M1 and M2 for sizes $N = 25, 50, 100, 200, 400$ and check that exactly one zero-energy eigenstate stays pinned at each end. If the zero-energy count changes with size, the pair of degenerate zeros on the unit circle splits, or the decay length measured numerically disagrees with $\\xi = -1/\\ln|\\zeta|$ predicted by the in-circle zero, then $w_{\\mathrm{mc}} = 1$ is not the stable characterization the paper claims.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the zero-counting topological invariant $w_c = N_z - N_p$ that the paper applies to characterize critical and multicritical phases."},{"cited_title":"Kumar, Nilanjan Roy, Y","cited_arxiv_id":null,"evidence_quote":"Earlier study of this model family that reported edge modes delocalizing at multicritical points; the claim this paper revises."},{"cited_title":"Kumar, Y","cited_arxiv_id":null,"evidence_quote":"Earlier analysis of the same chain including the trivial $w = 0$ phase; provides the comparison case of trivial multicritical points."}],"review_version":1}