{"id":"669df3f1-c184-4f5a-8530-7c2c6b7463fe","arxiv_id":"2602.16670","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"n-root versions of graphene and the Lieb lattice turn Dirac cones into exceptional horns with E∼|q|^{1/n} and Landau levels scaling as ϕ^{1/(2n)}.","lead":"This paper builds 2D lattices whose energy spectra are exact nth roots of graphene and the Lieb lattice, turning their Dirac points into high-order exceptional points with unusual sublinear dispersion. The result offers a systematic route to engineer high-order exceptional points and their Landau levels in photonic lattices.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unproven no-flux-through-loop-module ansatz underpins LL scaling; direct H^n block check would settle it.","rationale":"The central claim has two pillars: EP conversion at zero field and LL scaling at finite flux. The EP/horn dispersion follows from H^n block diagonality and is less at risk; the LL part explicitly depends on the no-flux-through-modules ansatz. The reader's weakest assumption matches. A direct algebraic check can settle it; no change to the conditional verdict is needed.","tokens_in":24061,"tokens_out":29926,"duration_ms":284415,"concrete_test":"For n=3 and n=4, build the full Peierls-substituted n-root Hamiltonian on a finite torus with rational flux phi=2*pi*l/q, enforcing zero phase accumulation around every loop module, and numerically compute H^n. Compare the SL1-SL1 block with the standard graphene/Lieb Hofstadter Hamiltonian at the same flux (after removing static-phase reference). If the blocks are not equal to machine precision, the parent-block identity fails and (21)/(33) need re-derivation; if equal, the ansatz is validated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Sec. II B and III A, the authors assume the magnetic field is applied only to the parent plaquettes, with zero additional flux through each loop module, so that after Peierls substitution H^n has the standard parent Hofstadter Hamiltonian as its SL1 block; the Landau levels (21) and (33) and the exceptional LL then follow by nth-rooting. This is asserted, not derived. Zero flux around a loop fixes only the total phase around that loop; it does not by itself fix the phases of the individual h_l blocks, and the effective H1 hoppings could pick up k-independent or flux-dependent corrections that change the parent LL spectrum. The numerics in Figs. 5 and 8 support the ansatz for n=3,4 at low flux, but the general-n claim and the exceptional-LL derivation rest on this unproven bracket structure.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper constructs n-th root tight-binding models (n>=3) for two 2D parent lattices, graphene and the Lieb lattice, built from loop modules of unidirectional couplings. The authors show that the Bloch spectrum consists of n rotated copies of the n-th root of the parent spectrum plus zero-energy flat bands protected by a generalized chiral symmetry. With a suitable phase configuration, the finite-energy Dirac points of the parent can be shifted to zero energy, converting the Dirac cones into 'exceptional horns': zero-energy exceptional points of order n (or higher) with dispersion E ~ |q|^(1/n). The same n-root relation is used to predict Landau-level energies scaling as phi^(1/(2n)) under a uniform magnetic field; for the n-root Lieb lattice a zero-energy exceptional Landau level is identified. A photonic ring-resonator implementation for n=3 is proposed and studied with finite-element simulations.","tokens_in":24324,"tokens_out":11767,"duration_ms":104541,"significance":"Should the results hold, the paper provides a clean, parameter-free algebraic route to high-order exceptional points in two-dimensional lattices, with distinctive sublinear dispersion and unusual Landau-level flux scaling. The spectrum-rooting construction is exact, and the low-energy perturbation theory in Appendix A is analytic and general. The explicit photonic implementation, including quantitative modeling of imperfections in Appendix D, is a strong practical contribution. The main limitations are the unproven Peierls-substitution ansatz and the unproven Jordan-form statements, which gate the Landau-level and EP-order claims.","major_comments":[{"comment":"The statement that imposing no flux through the loop modules ensures that the n-th power of the flux-inserted Hamiltonian has the parent Hofstadter Hamiltonian as its H1 block is an assertion, not a derivation. For each loop module the zero-flux condition fixes only the product of Peierls phases around that module, not the phases of the individual h_l entries. The matrix elements of H1 = h1...hn can therefore pick up k-independent or flux-dependent corrections that would change the parent Landau-level spectrum. The LL energies in (21) and (33) and the exceptional LL in Sec. III A depend on this block structure. Please provide an explicit check (e.g., compute H^n for a Peierls-substituted n=3 or n=4 model and verify the H1 block) or derive the gauge constraints under which the identity holds. The numerics in Figs. 5 and 8 support the ansatz only for low flux and specific n.","section":"Sec. II B, first paragraph; Sec. III A, first paragraph"},{"comment":"The Jordan decompositions J = J_n(0) ⊕ J_{2n-1}(0) for n-root graphene and J = J_1(0) ⊕ J_{2n-1}(0) ⊕ J_{2n-1}(0) for n-root Lieb are central to the claimed EP orders, but they are stated without proof. Without a derivation or construction of generalized eigenvectors, the order of the exceptional points is an assertion rather than a demonstrated result. Please provide a proof or a direct computation of the Jordan form for general n.","section":"Sec. II A (Jordan form after Eq. (18)); Sec. III (Jordan form before Fig. 7)"},{"comment":"The phase configuration that removes the 3J offset in n-root graphene is explicitly shown only for n=3. Since the exceptional-horn phenomenon and the Landau-level analysis are claimed for general n, an explicit general-n phase assignment (analogous to Eq. (24) for the Lieb case) or a proof of its existence is required. The statement that the self-loop contributions cancel for general n is not demonstrated.","section":"Sec. II A (Fig. 2(a)) and paragraph after Eq. (18)"},{"comment":"The exceptional Landau level result rests on the uniqueness of the zero-energy solution at each LL: 'it is possible to show that there is a single exact solution at each LL'. This is not proven. If there are multiple linearly independent zero-energy states at the m-th LL, they do not coalesce and the level is not exceptional. Please provide the proof of uniqueness or state the result as a conjecture. The lower-bound statement 'at least order n' is acceptable, but the underlying uniqueness must be established.","section":"Sec. III A (around Eq. (36))"}],"minor_comments":[{"comment":"The eigenvector formula uses E_± in the denominators and is singular at E=0; this is appropriate since these eigenvectors apply to non-zero eigenvalues, but a brief remark would avoid confusion.","section":"Eq. (12)"},{"comment":"The Landau-level substitution is applied to h_n and h_1; the ordering of b and b^† for the K' valley should be checked or briefly explained.","section":"Sec. II B, Eq. (14)"},{"comment":"The name 'exceptional horn' is evocative but the connection to Gabriel's horn could be clarified: the horn refers to the divergent derivative (infinite group velocity) at the EP, not to a geometric shape of the spectrum.","section":"Sec. II A"},{"comment":"The notation H(phi) -> H^†(-phi) is clear in context, but a one-sentence derivation would improve readability.","section":"Appendix D, Eq. (D1)"},{"comment":"The paper relies heavily on the authors' own generalized index theorem [47]. A short recapitulation of the theorem's statement would make the paper more self-contained.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is a natural extension of the authors' previous work on n-root non-Hermitian models and the generalized index theorem. The missing derivations flagged in the major comments are fixable in a revision; the Peierls-substitution step is the most important. If the authors can supply the explicit block check for H^n, the Landau-level part will be on solid ground. The Jordan-form and phase-construction gaps are also local and can be repaired with direct algebraic proofs."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a credible paper from a group that knows the n-root construction well. The genuinely new items are the phase-configuration mechanism that pushes the parent Dirac point to zero energy, the resulting exceptional horns with E ~ |q|^{1/n}, and the Landau-level scalings E ~ phi^{1/(2n)}, including the exceptional zeroth Landau level for the n-root Lieb lattice. These are not in the earlier n-root SSH/index-theorem papers. The algebraic spectrum-rooting and the low-energy perturbation theory are clean and parameter-free, and the photonic ring simulations are a real attempt to connect to experiment, with honest discussion of deviations near the EPs. No circularity: the scalings follow from the algebraic n-root relation, not from fitting.\n\nThe main soft spot is the magnetic-field step. In Secs. II B and III A the paper assumes that after Peierls substitution, the nth power of the flux-inserted Hamiltonian still has the standard parent Hofstadter Hamiltonian as the SL1 block. The text says no flux threads the loop modules and that this 'translates into' the parent block, but it does not derive the block structure of H^n under flux. Zero total flux around each loop module fixes only the closed-loop phase; it does not fix the phases of the individual h_l blocks, and the effective parent hoppings could pick up k- or flux-dependent corrections. The numerics for n=3,4 at low flux support the ansatz, but the general-n claim depends on it. This is a missing calculation, not necessarily a wrong result.\n\nSecond, the Jordan decomposition at the high-symmetry point, J_n ⊕ J_{2n-1} (and J_1 ⊕ J_{2n-1} ⊕ J_{2n-1} for the Lieb case), is stated without proof. It is very plausible from the generalized index theorem and the module structure, but it is load-bearing for the claimed EP order and should be derived.\n\nThe photonic implementation is a plus, but the agreement near the EPs is qualitative, and the deviations are modeled with two extra parameters, α and J_cc. That is fine for a proposal; it is not a precision test.\n\nWho this is for: people working on non-Hermitian lattices, n-root topology, or EP photonics. The paper deserves a serious referee, not a desk reject, but the referee should ask for the flux-block derivation and the Jordan-block proof before publication. I'd bring it to a reading group, though I probably wouldn't cite it in my own work in the next year.","headline":"Genuinely new n-root construction with exceptional horns and root Landau-level scaling, but the magnetic-flux block structure and the Jordan decomposition are asserted rather than proven.","tokens_in":24767,"tokens_out":4402,"would_cite":false,"duration_ms":41852,"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":"Dirac cones collapse into exceptional horns in n-root lattices","keywords":["non-Hermitian lattices","n-root models","exceptional points","Dirac cones","Lieb lattice","graphene","Landau levels","photonic ring resonators"],"falsifier":"In the photonic 3-root graphene design, zoom into the p=0 branch near the K point: the exceptional-horn claim requires the two approaching bands to separate as |q|^{1/3}; observing any different power law (e.g., linear) would falsify it. Separately, compute the Hofstadter spectrum with flux also threading the loop modules: if the nth power of the butterfly no longer collapses onto the parent graphene (or Lieb) butterfly, the block-diagonal assumption that all Landau-level results rest on is broken.","tokens_in":24000,"feed_emoji":"🌀","tokens_out":7650,"duration_ms":62170,"temperature":0.7,"pith_summary":"This paper shows that a non-Hermitian lattice can be built so that its Hamiltonian is an nth root of a Hermitian parent lattice — graphene or the Lieb lattice — and that this root operation transforms the parent's Dirac cones in a controlled way. Replacing every parent hopping with a loop module of unidirectional couplings makes the nth power of the Hamiltonian block-diagonal, with the parent model as the first block; the full spectrum then consists of n rotated branches whose nth powers reproduce the parent's real bands, plus flat bands at zero energy. The central result is that a carefully chosen phase configuration on the couplings can cancel the parent's energy offset, turning each Dirac point into a zero-energy exceptional point of order n (or higher) around which energy scales as E ~ |q|^{1/n}, a sublinear regime the authors call an exceptional horn. The same algebraic recipe predicts Landau levels that scale with magnetic flux as φ^{1/(2n)}, with an exceptional zeroth Landau level for the n-root Lieb lattice. The models are matched to a photonic ring-resonator design, so the predicted horns and Landau scaling are experimentally addressable.","feed_headline":"Dirac cones become exceptional horns in n-root lattices","feed_subtitle":"Built from unidirectional loops, the lattice makes low-energy modes disperse as the nth root of momentum — a new photonic sensing regime.","key_machinery":"The loop module — a replacement of each parent hopping by n unidirectional couplings of strength J^{1/n} passing through n−1 intermediate sites — is the construction that makes H^n block-diagonal, with the Hermitian parent as the first block. The generalized chiral symmetry C_n (defined by C_n H C_n^{-1} = ω_n^{-1} H, ω_n = e^{2πi/n}) is what forces the n-branch rotation of the spectrum. The phase configuration of the loop couplings is the control knob: it either leaves the parent's energy offset (case i of the perturbation theory in Appendix A, giving Dirac points with renormalized Fermi velocity) or cancels it (case ii, giving zero-energy exceptional points and |q|^{1/n} dispersion). The n","core_discovery":"For n≥3, the n-root graphene and n-root Lieb lattices are formed by replacing each hopping of the parent with a closed loop of n unidirectional couplings of magnitude J^{1/n}. Because the generalized chiral symmetry C_n imposes that every finite-energy eigenvalue appears in the n-tuple {ω_n^p E}, the spectrum is a union of n branches whose nth powers coincide with the parent's real bands, plus zero-energy flat bands protected by sublattice imbalance. The paper's central discovery is that the fate of the parent's Dirac cone depends on a phase configuration on the loop couplings. If the parent block retains a finite energy offset (3J for graphene), the n-root spectrum has an n-fold set of Dira","pith_inferences":["If the photonic implementation is realized, the |q|^{1/n} dispersion near an exceptional horn should make the local density of optical states diverge at zero energy, which could be probed as an enhanced emission or absorption signal at the K point.","The same loop-module recipe should transplant the mechanism to other parent models with Dirac or nodal degeneracies, such as Weyl semimetals or the Haldane model; the authors list these targets but leave them unexplored.","Because the exceptional Landau level is pinned at zero energy by symmetry, it may offer a magnetic-field-tunable, symmetry-protected degeneracy that is robust to disorder — a testable alternative to fine-tuned EP sensors.","The predicted dependence of the zeroth LL on flux (flat at zero) could be checked in circuits or coupled-ring arrays without needing perfect unidirectionality, since the flatness only relies on the block-diagonal structure."],"forward_implications":["The n-root construction produces n rotated branches in the complex energy plane whose nth powers match the parent's real spectrum, so the parent's physics is recovered after an nth-power operation.","With the parent energy offset intact, each Dirac point survives as an n-tuple of Dirac points with a renormalized Fermi velocity; with the offset cancelled, it becomes a zero-energy exceptional point of order n or higher.","The dispersion E ~ |q|^{1/n} around the exceptional horns implies a formally infinite Fermi velocity as q→0, in contrast to the constant velocity of a Dirac cone.","Landau levels follow E ~ φ^{1/(2n)} around the exceptional region, and the n-root Lieb lattice has an exceptional zeroth Landau level.","Because the generalized chiral symmetry is unbroken by the flux, the zero-energy flat bands remain pinned and mix with the exceptional-point eigenvectors only at the high-symmetry point."],"fun_headline_variants":["Exceptional horns replace Dirac cones in n-root lattices","n-root lattices turn Dirac cones into exceptional horns","Photonic rings yield exceptional horns from Dirac cones","Low-energy modes scale as q^(1/n) in n-root lattices","Dirac cones morph into exceptional horns in n-root designs"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole magnetic-field analysis assumes the Peierls phase string places flux only on the large plaquettes of the n-root lattice, leaving the loop modules flux-free, so that the nth power of the flux-inserted Hamiltonian still has the Hermitian parent as a block; if that ansatz fails, the derived Landau-level scaling E ~ φ^{1/(2n)} and the exceptional zeroth Landau level are unsupported.","fun_headline_variants_meta":{"raw":{"variants":["Exceptional horns replace Dirac cones in n-root lattices","n-root lattices turn Dirac cones into exceptional horns","Photonic rings yield exceptional horns from Dirac cones","Low-energy modes scale as q^(1/n) in n-root lattices","Dirac cones morph into exceptional horns in n-root designs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000276,"raw_usage":{"total_tokens":1558,"prompt_tokens":892,"completion_tokens":666,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":636,"completion_tokens_details":{"reasoning_tokens":584}},"tokens_in":636,"tokens_out":666,"duration_ms":5819,"temperature":1.0,"reasoning_tokens":584,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T22:26:23.819424+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In the photonic 3-root graphene design, zoom into the p=0 branch near the K point: the exceptional-horn claim requires the two approaching bands to separate as |q|^{1/3}; observing any different power law (e.g., linear) would falsify it. Separately, compute the Hofstadter spectrum with flux also threading the loop modules: if the nth power of the butterfly no longer collapses onto the parent graphene (or Lieb) butterfly, the block-diagonal assumption that all Landau-level results rest on is broken.","supporting_citations":[],"review_version":1}