{"id":"a22050fc-2220-4545-ae92-421f41068a44","arxiv_id":"2509.16555","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":4,"one_line_summary":"Transfer-matrix simulations claim an intermediate quasi-localized phase and new critical exponents for chiral symplectic Anderson transitions, but the paper's own abstract warns the phase may be an artifact of isotropic finite-size scaling.","lead":"This paper uses large-scale numerical simulations to study the Anderson transition in the chiral symplectic symmetry class, with and without a weak one-dimensional topology. It reports an intermediate quasi-localized phase with new critical exponents, while its own abstract warns that the 2D version of the phase may be an artifact of the scaling analysis.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Abstract concedes the QL-phase identification rests on isotropic FSS; since the true critical scaling in a weak-topology system is anisotropic, Eq. (10) cannot by itself establish a separate QL phase.","rationale":"Reading in good faith, the paper's intended contribution is a numerical extension of previous chiral-class QL studies to class CII plus a claimed universality of the QL phase. The 3D Anderson-transition exponent nu = 0.878 ± 0.005 matches an earlier non-Hermitian class-AII result, which is independent support for the nontopological transition. However, the QL claim rests on a separation of critical disorder strengths obtained under an isotropic FSS assumption in a system whose defining feature is directional anisotropy. The abstract itself flags this as unresolved for 2D; the same assumption enters the 3D analysis through Eq. (10), so the concern is not limited to 2D. The body provides no anisotropic FSS, no alternative test, and no discussion of how aspect-ratio dependence was checked. The field-theory revision announced in the abstract is also absent from the body, so the theoretical corroboration is not available to the reader. These points do not prove the claim false; they show it is supported only by an analysis whose own assumption is acknowledged to be potentially artifact-producing. The proposed anisotropic-FSS test would settle it. Since the reader's conditional verdict already requires exactly this, I keep the verdict unchanged.","tokens_in":17012,"tokens_out":4678,"duration_ms":43484,"concrete_test":"Perform anisotropic transfer-matrix FSS for the 3D h_T model: run at fixed transverse sizes L_perp = 14, 18, 22 with several aspect ratios r = L_z/L_perp = 1, 2, 4, and fit ln(xi_x/L_perp) and ln(xi_z/L_z) with independent scaling dimensions 1/nu_perp and 1/nu_parallel plus an additional scaling variable for r. Check whether W_x^c and W_z^c remain separated and whether nu_perp = 0.66 ± 0.03 is recovered; failure means the QL phase is an isotropic-FSS artifact.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that a 1D weak index universally induces a QL phase in the chiral symplectic class. The numerical evidence for this phase is the inequality W_x^c < W_z^c (3D) and W_y^c < W_z^c (2D), extracted from isotropic FSS: Eq. (10) defines Lambda_mu = xi_mu/L and expands ln Lambda_mu in powers of L^{1/nu} u_t with a single transverse size L and a single correlation-length exponent. But the very phenomenon claimed is anisotropic: in the QL phase xi_z = infinity while xi_x, xi_y are finite. At a putative metal-to-QL critical point, correlation lengths along the topological and non-topological directions need not share the same scaling form, and xi_mu/L for fixed mu may depend separately on L_perp and L_parallel (or on an aspect ratio), so a data collapse using L_perp = L_parallel = L can be spurious. The abstract explicitly states this: the numerically observed 2D QL phase 'may be an artifact of the spatially isotropic scaling assumption' and that conclusive identification 'requires a finite-size scaling approach that accommodates generic (anisotropic) spatial scaling.' The main text nevertheless concludes that the QL phase is universal without performing that analysis. A second weak link is the topological-direction boundary: W_z^c is fixed by assuming the linear form Eq. (9) with exponent 1; if the true z-direction transition has anisotropic scaling corrections, this estimate can shift relative to W_x^c, and the two could merge, removing the intermediate phase.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports transfer-matrix and finite-size scaling (FSS) simulations of three- and two-dimensional disordered lattice models in the chiral symplectic class CII, with and without a one-dimensional weak topological index. The numerical analysis yields a 3D Anderson-transition exponent ν=0.878±0.005 for the nontopological model, consistent with the non-Hermitian class-AII result, and reports a separate metal-to-quasi-localized transition with ν=0.66±0.03 in the topological 3D model and ν≈1.45 in the topological 2D model. The authors interpret the inequality between the critical disorder along the topological and nontopological directions as evidence of an intermediate quasi-localized phase in which the localization length diverges only along the topological direction. The abstract additionally announces a field-theory RG study that revises earlier fixed-point results and suggests that the 2D quasi-localized phase may be an artifact of the isotropic scaling assumption; this study is not presented in the main text.","tokens_in":17435,"tokens_out":7227,"duration_ms":56434,"significance":"If the central claim were established, the paper would make a strong contribution to the classification of Anderson transitions in chiral symmetry classes: it would add the chiral symplectic class to the set of classes showing weak-topology-induced quasi-localization, and would identify a new universality class for the metal-to-QL transition with a Kramers-symmetry-sensitive exponent. The 3D nontopological exponent ν=0.878±0.005 is a useful numerical benchmark and agrees with the non-Hermitian class AII result, and the fittings are presented with reasonable transparency (truncation-order scans, GOF, Monte Carlo confidence intervals) for the 3D Anderson transition. However, the decisive claim of a separate quasi-localized phase rests on an isotropic FSS assumption that the authors themselves state can produce a spurious phase, and the announced field-theory support is absent from the manuscript. The paper therefore is currently more a numerical study with an unresolved systematic caveat than a completed demonstration of universality.","major_comments":[{"comment":"The central claim of an intermediate quasi-localized phase rests on comparing W_c^μ (μ=x,y) extracted from Eq. (10) with W_c^z extracted from Eq. (9), but Eq. (10) assumes isotropic scaling with a single transverse size L for all spatial directions. In the claimed QL phase the localization length is divergent along z and finite along x,y; at a critical point separating such a phase from the metal, the scaling of ξ_μ in the topological and nontopological directions need not be governed by the same correlation length or the same L dependence. The abstract itself concedes that the 2D QL phase \"may be an artifact of the spatially isotropic scaling assumption\" and that conclusive identification \"requires a finite-size scaling approach that accommodates generic (anisotropic) spatial scaling.\" Since Sec. V concludes that the QL phase is a universal feature without performing such an anisotropic analysis, the main conclusion is not supported by the presented numerics.","section":"III D and IV B/IV C"},{"comment":"The abstract announces a field-theory study in which the authors \"revisit and extend\" a previous 2D RG analysis, include a one-loop renormalization of the weak topological term, find the previously reported quasi-localized fixed point unstable, and conclude that the 2D QL phase may be an artifact. None of this analysis appears in the main text: there is no field-theory section, no RG equations, no fixed-point analysis, and no result corresponding to this part of the abstract. The title promises \"simulational and theoretical studies,\" and the abstract's RG conclusion is used to qualify the numerical 2D phase. As written, the manuscript omits a major component of its stated results, making the abstract misleading and the theoretical claim unverifiable.","section":"Abstract vs. Sections I–V"},{"comment":"The reported 3D metal-to-QL exponent ν=0.66±0.03 is not stable under the published fitting variations. Table I(b) shows W_x^c varying from 8.287 to 8.318 and ν from 0.609 to 0.741 across the truncation orders and L ranges listed, with 95% confidence intervals that do not overlap (e.g., ν=0.609[0.542,0.650] vs. 0.741[0.623,0.808]). The spread in W_x^c is roughly five times the \"less than 2% of the distance\" between W_x^c and W_z^c quoted in Sec. IV B. Selecting the fit with best GOF (0.211) does not eliminate the systematic truncation uncertainty, so the claim that ν is distinctly different from the Anderson-transition exponent ν=0.878 is based on a statistically fragile estimate.","section":"Table I(b) and Sec. IV B"},{"comment":"The 2D Anderson-transition exponent is similarly truncation-sensitive: Table II reports ν=2.003, 2.064, 2.064 for n=2, m=2,3,4, but ν=1.760 for n=3, m=2 and ν=1.653 for n=4, m=2, with W_c shifting from 2.009 to 2.080. These systematic shifts exceed the quoted 95% confidence intervals, and no GOF values are reported for the 2D fits. The paper selects n=2, m=3 for the quoted ν=2.064±0.004, but the dependence on n makes the \"different universality class\" comparison between the 2D Anderson transition and the 2D metal-to-QL transition (ν≈1.45) less definitive than stated.","section":"Table II and Sec. IV C"}],"minor_comments":[{"comment":"The caption contains an unresolved reference \"Eq. (??)\"; it should refer to Eq. (10).","section":"Table I caption"},{"comment":"The text cites \"[Sec. III D]\" for the linear regression of γ_max^(l)(W,L), but Eq. (9) defining this quantity is presented in Sec. III C.","section":"Sec. IV B"},{"comment":"There is a stray closing bracket in \"shown in Fig. 3(a)].\"","section":"Sec. IV A"},{"comment":"Goodness-of-fit values are not reported for the 2D fits; without them the reader cannot assess the quality of the different truncation choices.","section":"Table II"},{"comment":"The expression \"t∥t′∦= 0\" is typographically ambiguous; use t_∥ t'_∥ or equivalent notation.","section":"Sec. II"}],"recommendation":"major_revision","confidential_remarks":"The abstract-body mismatch is serious: the RG part of the abstract appears nowhere in the paper. If the authors intend to submit the simulation part alone, the abstract and title must be revised. I also note that the central QL claim is already conceded in the abstract to be potentially artifactual; in revision the authors should either perform anisotropic FSS or substantially temper the universality claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know upfront: the abstract and the main text are not the same paper. The abstract tells the reader that the numerically observed 2D QL phase 'may be an artifact of the spatially isotropic scaling assumption' and that conclusive identification requires anisotropic FSS; it also says the field-theory RG analysis now finds the previously reported quasi-localized fixed point is unstable. The main text does neither. It performs only isotropic FSS, concludes the QL phase is universal across chiral classes, and contains no field-theory section at all. That mismatch is the central issue.\n\nWhat is genuinely new: this is the first transfer-matrix study of the chiral symplectic class with 1D weak topology, extending the QL phase from AIII and BDI to CII. The 3D Anderson-transition exponent ν=0.878±0.005 is stable across several FSS fits and agrees nicely with the non-Hermitian AII correspondence. The metal-to-QL exponent ν=0.66±0.03, if it holds up under anisotropic scaling, would be a distinct universality class and a real result. The DOS calculation is a useful check. The models are standard and the methods are appropriate.\n\nThe soft spots are the ones the abstract itself flags. The QL phase is defined by anisotropic divergence—ξ_z infinite, ξ_x finite—so scaling Λ_μ=ξ_μ/L with a single transverse L assumes isotropy exactly where the physics is anisotropic. The abstract concedes this, but the main text presses on to a universal conclusion. The z-direction boundary W_c^z is fixed by a linear-crossing assumption with exponent 1; anisotropic corrections could shift it relative to W_c^x and dissolve the intermediate phase. The 2D results are the weakest: the fitted exponent drifts with truncation order (ν≈2.0 to 2.06 to 1.76 to 1.65 across Table II), and the 'critical metal' side is fit only from the localized side. No code or data are provided.\n\nWho gets value: people working on chiral-class localization and topological Anderson transitions. The numerical extension to CII is worth having, and the paper would be a good referee assignment—I would send it out—but it needs major revision. Either the authors perform the anisotropic FSS and include the RG analysis, or they soften the universal claim to match what the isotropic data can actually support. As written, the conclusion overreaches the evidence in the manuscript.","headline":"A credible numerical extension of the chiral-class QL story, but the abstract itself concedes the central claim rests on an isotropic-FSS assumption the main text never checks, and the advertised field-theory RG section is missing.","tokens_in":17896,"tokens_out":3428,"would_cite":false,"duration_ms":29967,"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":"In the chiral symplectic class, a one-dimensional weak topological index induces an intermediate quasi-localized phase between metal and Anderson insulator, with critical exponent ν=0.66±0.03 in 3D, distinct from the Anderson transition's…","keywords":["Anderson transition","chiral symplectic class","weak topology","quasi-localized phase","Lyapunov exponents","finite-size scaling","disordered lattice models","universality class"],"falsifier":"A reader could test this by repeating the same transfer-matrix finite-size scaling with independent transverse sizes for each spatial direction and checking whether a single scale-invariant crossing point remains; if the apparent critical disorder strengths $W_x^c$ and $W_z^c$ merge together or the data stop collapsing on one scaling function, the quasi-localized phase is an artifact of the isotropic assumption.","tokens_in":16747,"feed_emoji":"⚛️","tokens_out":15259,"duration_ms":116958,"temperature":0.7,"pith_summary":"This paper argues that a one-dimensional weak topological index splits the metal-to-insulator transition in chiral symmetric disordered systems into two steps, with an intermediate quasi-localized phase whose localization length diverges only along the topological direction. The numerical part, transfer-matrix simulations with finite-size scaling of 3D and 2D chiral-symplectic lattice models, yields a 3D metal-to-quasi-localized exponent $\\nu=0.66\\pm0.03$ and a 3D Anderson-transition exponent $\\nu=0.878\\pm0.005$, so the two transitions differ in their critical exponents and hence belong to different universality classes. The paper's abstract adds a renormalization-group analysis in which the previously reported quasi-localized fixed point is unstable once the weak topological term is renormalized at one loop, leaving only an anisotropic scaling signature and suggesting the 2D quasi-localized phase may be an artifact of the isotropic scaling assumption. This matters because if the two-step picture is right, band topology joins symmetry and dimensionality as a factor that decides the universality class of a disorder-driven transition.","feed_headline":"Weak 1D topology turns metal-insulator transition into two steps","feed_subtitle":"Simulations find a quasi-localized phase (ν=0.66) distinct from the Anderson transition (ν=0.878).","key_machinery":"The mechanism that carries the argument is the imbalance of Lyapunov exponents—the inverse decay lengths of transmission through a disordered wire—produced by the 1D weak topological index. For a chiral Hamiltonian in block off-diagonal form, the weak index $\\nu_\\mu$ is the winding number of $\\det h(\\varphi_\\mu)$ as the twisted boundary phase $\\varphi_\\mu$ runs from $0$ to $2\\pi$, and in a quasi-one-dimensional wire it equals the normalized imbalance $(N_{+,\\mu}-N_{-,\\mu})/(N_{+,\\mu}+N_{-,\\mu})$ between positive and negative Lyapunov exponents along $\\mu$. This imbalance lets the lower Lyapunov band touch zero along the topological direction, making $\\xi_z$ diverge there while the exponent spectrum stays symmetric along the other directions. The finite-size scaling of the normalized length $\\Lambda_\\mu=\\xi_\\mu/L$, with one transverse size $L$ assumed for all directions, then extracts the critical disorder strengths and exponents that define the phase diagram.","core_discovery":"On its own terms, the paper's central claim is that a 1D weak topological index forces a two-stage delocalization–localization transition in the chiral symplectic class. In the transfer-matrix simulations of the 3D topological model, the localization length $\\xi_z$ diverges along the topological $z$ direction across an intermediate quasi-localized phase while $\\xi_x$ and $\\xi_y$ stay finite; the boundary of that phase is located at $W_x^c=8.313\\pm0.005$ with $\\nu=0.66\\pm0.03$ along the non-topological directions, and at $W_z^c=8.6447\\pm0.0006$ with $\\nu=1$ along the topological direction. Without the weak topology, the 3D Anderson transition in the same class sits at $W_c=10.248$ with $\\nu=0.878\\pm0.005$. The 2D models reproduce the same two-step structure, with $\\nu\\simeq2.06$ for the Anderson transition and $\\nu\\simeq1.45$ for the metal-to-quasi-localized transition. The manuscript's abstract also reports a renormalization-group analysis in which the quasi-localized strong-coupling fixed point is unstable after the one-loop renormalization of the weak topological term; the RG still produces an anisotropic scaling signature and suggests that the 2D quasi-localized phase seen with isotropic FSS may be an artifact.","pith_inferences":["If the anisotropic scaling test the paper calls for confirms the quasi-localized phase, then the topological direction should show a scale-invariant, metallic conductance even when the transverse directions are localized, a direction-dependent transport signature that could be observed in artificially disordered lattices.","The abstract's RG conclusion, taken seriously, implies that the robust footprint of 1D weak topology is not a distinct phase but an anisotropic critical exponent: one should look for different correlation-length exponents along the topological and non-topological directions at a single Anderson transition.","Because the 3D Hermitian chiral symplectic transition matches the non-Hermitian class AII transition, a non-Hermitian lattice with a 1D winding should show the same two-step or anisotropic-criticality scenario; this is testable in existing transfer-matrix codes and in open-system experiments.","One could probe the universality of the claimed quasi-localized phase by repeating the calculation at different values of the weak-topology parameter $t'_\\parallel/t_\\parallel$; if the intermediate phase is truly induced by the topology, its critical exponents should remain unchanged while the critical disorder strengths shift."],"forward_implications":["A 1D weak topological index changes the Anderson transition in the chiral symplectic class from a single metal-to-insulator transition into a two-step process, with an intermediate quasi-localized phase in between.","The 3D metal-to-quasi-localized transition ($\\nu=0.66\\pm0.03$) and the Anderson transition without topology ($\\nu=0.878\\pm0.005$) belong to different universality classes, so band topology is an additional factor beyond symmetry and dimension in setting the critical behavior.","The 3D Anderson transition exponent in the chiral symplectic class differs from the chiral unitary ($\\nu=1.06\\pm0.02$) and chiral orthogonal ($\\nu=1.12\\pm0.06$) values, showing time-reversal symmetry changes the universality class.","The same two-step structure and distinct exponents appear in 2D ($\\nu\\simeq2.06$ for Anderson, $\\nu\\simeq1.45$ for metal-to-quasi-localized), so the effect is not a 3D artifact—subject to the isotropic-scaling caveat.","The matching of the 3D chiral symplectic exponent to the non-Hermitian class AII exponent supports the correspondence between Anderson transitions in Hermitian chiral classes and non-Hermitian symmetry classes."],"supporting_citations":[{"why":"Establishes the one-parameter scaling hypothesis for the localization length in disordered wires, the basis of the finite-size scaling analysis.","marker":"[12]"},{"why":"Provides the transfer-matrix finite-size scaling method used to extract localization lengths and critical points.","marker":"[13]"},{"why":"Supplies the corrections-to-scaling formalism used in the polynomial finite-size scaling fit of the normalized localization length.","marker":"[14]"},{"why":"Provides the standard finite-size scaling methodology and prior critical-exponent determination against which the new fits are calibrated.","marker":"[16]"},{"why":"Supplies the winding-number characterization of the 1D weak topological index used in Eq. (7).","marker":"[29]"},{"why":"Gives the chiral unitary and chiral orthogonal critical exponents that the chiral symplectic values are compared with.","marker":"[32]"},{"why":"Reports the quasi-localized phases in the chiral unitary and orthogonal classes that this paper extends to the chiral symplectic class.","marker":"[36]"},{"why":"Proposes the vortex and Berry-phase mechanism by which 1D weak topology generates the quasi-localized phase.","marker":"[37]"},{"why":"Provides the field-theoretic vortex picture of the Anderson transition in chiral classes used in the theory discussion.","marker":"[39]"},{"why":"Supplies the non-Hermitian class AII Anderson transition result whose exponent matches the 3D chiral symplectic value.","marker":"[41]"}],"fun_headline_variants":["Weak 1D topology forces two-step Anderson transition","Quasi-localized phase emerges from weak topological order","Two-stage delocalization in chiral symmetry classes","Scaling artifact suspected in quasi-localized phase"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the localization length obeys the same finite-size scaling in every spatial direction, so one transverse size $L$ can stand for all directions; the paper's abstract acknowledges that if scaling is anisotropic, the claimed quasi-localized phase and its exponents could be artifacts and that an anisotropic finite-size scaling test is needed.","fun_headline_variants_meta":{"raw":{"variants":["Weak 1D topology forces two-step Anderson transition","Quasi-localized phase emerges from weak topological order","Two-stage delocalization in chiral symmetry classes","Scaling artifact suspected in quasi-localized phase"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000289,"raw_usage":{"total_tokens":1800,"prompt_tokens":1162,"completion_tokens":638,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":778,"completion_tokens_details":{"reasoning_tokens":577}},"tokens_in":778,"tokens_out":638,"duration_ms":5881,"temperature":1.0,"reasoning_tokens":577,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:49:28.995303+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A reader could test this by repeating the same transfer-matrix finite-size scaling with independent transverse sizes for each spatial direction and checking whether a single scale-invariant crossing point remains; if the apparent critical disorder strengths $W_x^c$ and $W_z^c$ merge together or the data stop collapsing on one scaling function, the quasi-localized phase is an artifact of the isotropic assumption.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the winding-number characterization of the 1D weak topological index used in Eq. (7)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the chiral unitary and chiral orthogonal critical exponents that the chiral symplectic values are compared with."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proposes the vortex and Berry-phase mechanism by which 1D weak topology generates the quasi-localized phase."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the field-theoretic vortex picture of the Anderson transition in chiral classes used in the theory discussion."}],"review_version":2}