{"id":"b1c6596d-bbc0-4d07-8cda-797c048d7c72","arxiv_id":"2607.26660","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"In a disordered Weyl semimetal, the entanglement-spectrum peak at eigenvalue 1/2 that mirrors topological Fermi arcs disappears by melting into the background at the disorder-driven transition, but by losing spectral weight when the Weyl nodes annihilate.","lead":"This paper computes how quantum entanglement in a disordered Weyl semimetal — a topological metal — changes as its Weyl nodes are destroyed in two different ways. It finds that the entanglement fingerprint of the topology melts into the background at the disorder-driven transition but shrinks away when the nodes are annihilated by band-structure tuning.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-mechanism distinction (melting vs. spectral-weight loss) is drawn from visual inspection of the EDOS at a single small size L=10; no quantitative peak measure or finite-size scaling is provided, so the distinction may be a finite-size artifact.","rationale":"The reader's weakest assumption — that the L=10 EDOS peak is dominated by topological arc states and that W_c from L=100 DOS is the correct reference — is exactly the load-bearing concern I identify. My independent reading of the manuscript confirms the central claim hinges on distinguishing two disappearance mechanisms of the ξ=1/2 peak. The paper provides no finite-size scaling, no quantitative peak metric, and no check that the disordered-system peak is topologically protected rather than a generic bulk-state feature. The concern is not that the authors are wrong, but that the evidence as presented is insufficient to rule out a mundane finite-size/interpretation artifact. I agree with the CONDITIONAL verdict: the claim is plausible and the phase diagram is computed independently (reducing circularity), but the numerical evidence needs strengthening. The proposed concrete test — repeating the EDOS computation at L=12/14 with quantitative peak statistics and a surface-localization check — would directly settle whether the melting-vs-weight-loss distinction survives finite-size scrutiny. Therefore no verdict change is warranted; the manuscript should remain conditional pending these checks.","tokens_in":8559,"tokens_out":2702,"duration_ms":32897,"concrete_test":"Compute ν_S(ξ) for the same model at L=12 and L=14, for (a) m=0, W=2.0, 3.0, 3.5, 3.75, 4.0 and (b) W=2.0, m=-0.8, -1.0, -1.1, -1.2, -1.3, using the same number of disorder realizations (or scaled to keep statistical error comparable). For each run, extract a quantitative measure: integrated spectral weight of the ξ∈[0.5−δ, 0.5+δ] peak above a linear interpolation of the surrounding background, and the peak's full width at half maximum. If the W-scan peak area drops sharply near a W_c(L) that extrapolates to ≈3.75t while the peak width diverges, the 'melting' picture holds; if the area/width behavior changes substantially with L, the L=10 distinction is a finite-size artifact. Additionally, for one representative disorder realization at L=10 and W=3.0, compute the inverse participation ratio of the correlation-matrix eigenstates with ξ≈0.5 across the entanglement cut; if they are not sur","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the ξ=1/2 EDOS peak disappears by 'melting into the background' across the disorder-driven WSM→DM transition, but by 'spectral-weight loss' across the node-annihilation WSM→NI transition. This distinction is the paper's main new physics, yet it is supported only by visual inspection of ν_S(ξ) at L=10 (Figs. 2 and 4). No quantitative measure of peak weight, width, or area above background is given, and no finite-size scaling (L=12, 14, ...) is shown. At L=10 with periodic boundary conditions, the clean-limit arc-state signal is just a handful of ξ=1/2 eigenvalues along k_y=0 (Fig. 1b); once disorder breaks translational invariance and the full eigenvalue spectrum becomes dense, the broadened 'peak' could in principle be dominated by bulk states near zero energy rather than by topologically protected arc states. If that were the case, the observed difference between the W-scan and m-scan could simply reflect the known behavior of the single-particle DOS at E=0 (finite at the WSM→DM transition, zero in the NI), rather than a distinct entanglement signature of the two transitions. Additionally, the reference W_c≈3.75t is taken from an L=100 KPM DOS computation and compared directly to L=10 entanglement data without discussing finite-size mismatch; the cited rare-region/avoided-criticality scenario (refs. [34-36]) is acknowledged in the Introduction but not engaged, so the sharpness of W_c used for the comparison is an unsupported assumption. These are numerical/interpretational weaknesses rather than logical inconsistencies; the underlying claim is plausible and internally consistent, but the evidence as presented is insufficient to rule out the finite-size-artifact alternative.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a minimal two-band lattice model of a Weyl semimetal with on-site disorder, focusing on how topology is destroyed in two distinct ways: by increasing disorder strength W (WSM to diffusive metal, DM) and by tuning the mass parameter m at finite disorder (WSM to normal insulator, NI). The authors compute the disorder-averaged distribution of single-particle entanglement-spectrum eigenvalues, which they call the entanglement density of states (EDOS), and Rényi entropies. In the clean limit they recover the known locus of ξ=1/2 eigenvalues mirroring Fermi arcs. With disorder, this locus broadens into a finite-width peak; the central claim is that across the WSM-DM transition the peak 'melts into the background' near W_c≈3.75t, whereas across the WSM-NI transition it disappears by gradual spectral-weight loss near m≈-1.2t at W=2.0t. They also report ΔS_i ∝ W^2 for weak disorder, breaking down near W_c, with explicit caveats about higher Rényi orders and the nonanalytic entropy kernel.","tokens_in":8968,"tokens_out":4494,"duration_ms":49610,"significance":"If the central claim holds, the EDOS provides a single entanglement-based observable that distinguishes quasiparticle destruction (disorder-driven transition) from node annihilation (band-parameter-driven transition) in a gapless topological phase. The numerical study is conscientious: 10,000 disorder realizations per parameter point, an independently computed KPM phase diagram, and explicit warnings about higher-Rényi deviations and the nonanalytic kernel. The novelty is moderate and fits the journal scope. However, the key qualitative distinction is currently supported by visual inspection of EDOS at one small system size (L=10) without quantitative measures or finite-size scaling, so the strength of the claim is not yet commensurate with the abstract's generality.","major_comments":[{"comment":"The central distinction between 'melting' (W scan at m=0) and 'spectral-weight loss' (m scan at W=2.0t) is established only by visual inspection of ν_S(ξ) at L=10. No quantitative metric is reported — e.g., integrated weight in |ξ−1/2|<δ, peak width, or peak height above a fitted background — and no finite-size scaling (L=12,14) is given. At L=10 the clean signal is a small set of ξ=1/2 arc eigenvalues; once disorder is added, the dense bulk spectrum can contribute a broad hump near 1/2. The apparent difference between the two scans might then merely track the known single-particle DOS at E=0 (finite in the DM, zero in the NI) rather than a distinct entanglement signature. Please add a quantitative peak measure and finite-size dependence, and ideally a control with the arc/trivial contributions separated.","section":"§III.B, Figs. 2 and 4"},{"comment":"W_c≈3.75t is obtained from KPM DOS on L=100 lattices, but the EDOS and entropy data are computed at L=10. The text uses W_c as a sharp reference for the L=10 data without discussing finite-size mismatch. This is load-bearing because the cited rare-region scenario (refs. [34–36]) predicts that for three-dimensional Dirac/Weyl systems the DOS at E=0 may be nonzero for all W in the thermodynamic limit, making the 'onset of metallicity' scale-dependent. Please show W-dependence of ν_A(0) or the typical DOS at several L values and state whether the L=10 'melting' is controlled by the L→∞ critical point or by a finite-size crossover.","section":"§II / §III.B, Fig. 1(a)"},{"comment":"The statement that ΔS_i is consistent with a leading-order W² correction is supported only by the local exponent α_i staying near 2 for W up to about 3t. No explicit perturbative calculation of the disorder-averaged correlation matrix or entropy is given, and a local logarithmic derivative can be close to 2 over a finite interval for unrelated reasons. Since this is a secondary claim, it need not block acceptance, but a brief derivation or a nontopological control would materially strengthen the interpretation.","section":"§III.B, Fig. 3(b)"}],"minor_comments":[{"comment":"Typo: 'qudratically vanishing DOS' should be 'quadratically vanishing DOS'.","section":"§II (text near Eq. (2))"},{"comment":"Please specify the EDOS normalization and histogram binning, and add axes/color-scale labels. Without these, the qualitative 'melting' vs 'spectral-weight loss' distinction is difficult to reproduce or compare across figures.","section":"Figs. 2 and 4"},{"comment":"The numerical differentiation used to define α_i(W)=d ln ΔS_i / d ln W is not described. Please specify the finite-difference or smoothing procedure and provide error estimates, particularly near W=0 and in the DM where the curves are noisy.","section":"§III.B, Fig. 3"},{"comment":"Ref. [42] is cited as an arXiv preprint from 2014; if it has been published, please update the reference. Also check the typesetting of author names and accents throughout the reference list.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's main new physics — the EDOS distinction between melting and spectral-weight loss — is interesting and potentially publishable, but the current numerical support is too thin: one L=10 visual comparison without finite-size scaling or a quantitative peak metric. The requested additions are within the scope of the manuscript and do not require new formalism. If the authors provide those checks, the paper would be suitable for publication; otherwise the central claim remains an unverified finite-size observation. The novelty relative to existing disordered-entanglement-spectrum studies is moderate, and the positioning against the rare-region/avoided-criticality literature should be sharpened."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper introduces a genuinely new observable—the disorder-averaged entanglement density of states (EDOS), i.e., the ensemble distribution of reduced-correlation-matrix eigenvalues—and uses it to compare how the ξ=1/2 Fermi-arc entanglement peak dies at the disorder-driven WSM-to-metal transition versus the band-structure-driven node-annihilation transition. That is a smart idea and the clean-limit version is already known from refs. [10,11], so the extension to strong disorder is the new content. The numerics are honest: 10,000 realizations per parameter point, the phase diagram computed independently via KPM at L=100, and a clear statement that the phase boundaries are used as reference labels rather than fit parameters. The W^2 scaling of the Rényi entropies at weak disorder is a clean confirmation of perturbation theory, with an explicit caveat about the nonanalytic kernel. The paper also advertises its own limitations, which I trust.\n\nThe main soft spot is exactly what the stress-test note flags. The central claim—melting into the background at the disorder transition versus spectral-weight loss at node annihilation—is supported only by visual inspection of ν_S(ξ) at L=10. There is no quantitative measure of peak weight, width, or area above background, and no finite-size scaling. At L=10, the clean peak is only a handful of states; once disorder broadens the spectrum, the feature could in principle be dominated by bulk states near zero energy rather than by arc-derived states. The observed difference between the W-scan and m-scan might then simply track the single-particle DOS at E=0 (finite in the diffusive metal, zero in the insulator), which would make the entanglement diagnostic redundant rather than new. The reference W_c≈3.75t comes from an L=100 DOS computation and is overlaid on L=10 entanglement data without discussing the mismatch. The rare-region/avoided-criticality scenario is acknowledged but not engaged, so the sharp W_c used for comparison is an assumption. These are numerical and interpretational gaps, not logical contradictions; the claim is plausible and internally consistent.\n\nIf I were refereeing, I would not desk-reject this. The EDOS idea is worth putting on record, and the distinction between the two transition mechanisms is a nice physics question. But the evidence as presented is not sufficient to rule out the finite-size artifact or the trivial-DOS alternative. A revised version with L=12,14 scaling, a quantitative peak measure, and an explicit discussion of the W_c sharpness would settle it. I would cite this paper for the EDOS method, with a caveat about the two-mechanism conclusion. Bring it to a reading group if you want a good discussion about what counts as evidence for an entanglement signature—just don't treat the melting-versus-weight-loss story as established yet.","headline":"Useful new EDOS diagnostic with a plausible but unproven distinction between the two transition mechanisms; needs finite-size and quantitative-peak checks before I'd lean on it.","tokens_in":9556,"tokens_out":1998,"would_cite":true,"duration_ms":22777,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["73.43.Nq","72.15.Rn","03.65.Ud","71.30.+h"],"model":"deepseek-v4-flash","headline":"Entanglement spectra tell the two ways a Weyl semimetal dies apart","keywords":["Weyl semimetal","entanglement spectrum","disorder","topological phase transition","entanglement density of states","Renyi entropy","Fermi arcs","disordered Weyl semimetal"],"falsifier":"A finite-size scaling study of the EDOS at L=10, 12, 14 for the same model would settle the issue: if the 'melting' of the ξ≈1/2 peak at W_c is mostly a small-system artifact, the peak width would decrease (or the peak would sharpen) with increasing L, rather than broaden. Alternatively, if the peak is dominated by bulk states, its center would shift away from 1/2 or its weight would scale differently with L.","tokens_in":8380,"feed_emoji":"⚛️","tokens_out":1426,"duration_ms":18396,"temperature":0.7,"pith_summary":"This paper claims that the entanglement spectrum of a disordered Weyl semimetal can distinguish the two distinct ways its topology is destroyed: by annihilating the Weyl nodes in momentum space (driving it into a normal insulator) or by strong disorder that destroys the quasiparticle pole (driving it into a diffusive metal). In the clean limit, topological surface states (Fermi arcs) appear as a set of eigenvalues exactly at ξ=1/2 in the reduced correlation matrix. The paper introduces the entanglement density of states (EDOS), the disorder-ensemble distribution of these eigenvalues, and shows that disorder broadens the δ-like peak into a finite-width distribution. Across the disorder-driven transition at W_c≈3.75t the peak melts into the background, whereas across the node-annihilation transition at m≈-1.2t the peak loses spectral weight and vanishes. It also shows that, for weak disorder, the disorder-averaged Rényi entropies scale as W², with the scaling breaking down near W_c.","feed_headline":"Entanglement detects two ways to kill a Weyl semimetal","feed_subtitle":"Disorder melts the spectrum peak; node annihilation drains it — a new probe of topological change.","key_machinery":"The entanglement density of states (EDOS), denoted ν_S(ξ), is the central object: the disorder-ensemble distribution of eigenvalues ξ of the reduced single-particle correlation matrix for a real-space bipartition. In the clean limit, topological Fermi arcs manifest as ξ=1/2 eigenvalues (maximal entanglement). Disorder replaces this delta-function with a broadened peak, and the shape of the EDOS—whether it melts into the background (disorder-driven transition) or loses spectral weight (node-annihilation transition)—is the diagnostic that distinguishes the two topological phase transitions.","core_discovery":"The central claim is that in a three-dimensional Weyl semimetal with a single pair of Weyl nodes, the topological Fermi arcs leave a sharp imprint on the entanglement spectrum: a locus of eigenvalues pinned exactly at ξ=1/2 in the reduced correlation matrix. This paper shows that for any finite disorder these eigenvalues are no longer pinned, but are replaced by a finite-width distribution centered at 1/2. The evolution of this distribution—tracked via the entanglement density of states—is qualitatively different for the two topological transitions: increasing disorder strength past W_c≈3.75t causes the peak to broaden and dissolve into the background, while tuning the mass parameter across","pith_inferences":["A quantitative measure of the EDOS peak (e.g., its width and weight) would allow a finite-size scaling analysis that could confirm whether the melting at W_c is sharp or smeared by rare-region (avoided-criticality) effects. The paper notes the rare-region scenario but does not test it with its L=10 data.","The paper's clean-limit identification of the ξ=1/2 locus with Fermi arcs rests on the known correspondence between the entanglement spectrum and surface states; an extension would be to check whether the broadened EDOS peak can be directly related to the dissolution of arc states seen in local-density calculations.","The claim that the EDOS can distinguish the two transitions suggests a testable extension: study a Weyl semimetal with multiple pairs of nodes and track whether the EDOS shows multiple peaks or a single broadened one, which would reveal whether each pair contributes independently.","If the entropies' W² scaling breakdown is tied to the emergence of rare regions, then higher-order Rényi entropies (i>10, which the paper also computes) might show a sharper anomaly at W_c than the low-order ones. This could be checked by the authors' own data.","The paper uses a single disorder realization in Fig. 1(c); an ensemble-resolved view of individual spectra (not just the averaged EDOS) could clarify whether the peak broadening is due to a gradual spread of each realization's eigenvalues or to a mixture of pinned and unpinned states."],"forward_implications":["If the EDOS behavior is confirmed, entanglement measurements in cold-atom or photonic simulators of Weyl semimetals could be used to identify which mechanism destroys the topology.","The result implies that even when the Weyl nodes remain stable against weak disorder, their entanglement signature already becomes non-universal, with a disorder-dependent width.","The breakdown of the naive W² scaling of Rényi entropies near W_c provides a scalar, disorder-averaged marker for the onset of the diffusive-metal phase.","The distinction between 'melting' and 'spectral-weight loss' suggests that the entanglement spectrum can serve as a probe of whether a transition is driven by quasiparticle destruction or by gap closing.","The qualitative difference between the two EDOS evolutions adds a new observable to the study of dirty Weyl semimetals, complementing the more common density-of-states and transport diagnostics."],"fun_headline_variants":["Entanglement spectrum reveals two paths to kill a Weyl node","How entanglement fingerprints the death of a Weyl semimetal","A dirty Weyl semimetal shows its topological death in entanglement","Two ways a Weyl semimetal dies, seen in its entanglement","Disorder and annihilation leave different entanglement scars"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The main load-bearing premise is that at the small system size L=10, the ξ≈1/2 peak in the entanglement spectrum is dominated by topological Fermi-arc states and not by bulk states that happen to have eigenvalues near 1/2; without a finite-size scaling check, this identification is unverified.","fun_headline_variants_meta":{"raw":{"variants":["Entanglement spectrum reveals two paths to kill a Weyl node","How entanglement fingerprints the death of a Weyl semimetal","A dirty Weyl semimetal shows its topological death in entanglement","Two ways a Weyl semimetal dies, seen in its entanglement","Disorder and annihilation leave different entanglement scars"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00018,"raw_usage":{"total_tokens":1125,"prompt_tokens":710,"completion_tokens":415,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":454,"completion_tokens_details":{"reasoning_tokens":330}},"tokens_in":454,"tokens_out":415,"duration_ms":5343,"temperature":1.0,"reasoning_tokens":330,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T11:20:20.254681+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A finite-size scaling study of the EDOS at L=10, 12, 14 for the same model would settle the issue: if the 'melting' of the ξ≈1/2 peak at W_c is mostly a small-system artifact, the peak width would decrease (or the peak would sharpen) with increasing L, rather than broaden. Alternatively, if the peak is dominated by bulk states, its center would shift away from 1/2 or its weight would scale differently with L.","supporting_citations":[],"review_version":1}