{"id":"33b33e51-6921-4f49-8564-8b6491282e69","arxiv_id":"1908.06630","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":10,"one_line_summary":"In the classical square-lattice XXZ antiferromagnet, the longitudinal spin conductivity diverges exponentially at the XY-type Kosterlitz-Thouless transition, while thermal conductivity shows no anomaly.","lead":"Spin and heat currents in a model two-dimensional antiferromagnet behave very differently: heat flow looks the same for all magnetic anisotropies, but spin flow sharply peaks at the Kosterlitz-Thouless transition in XY-like magnets. The peak is traced to long-lived vortices, and could offer a transport-based way to detect this topological transition in layered magnets.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed exponential divergence of sigma^s_xx at T_KT rests on finite-time integration and finite-size extrapolation of a current autocorrelation that is explicitly long-lived and L-dependent near T_KT; this needs a direct convergence test.","rationale":"The reader's weakest assumption already identified the finite simulation time and finite-size extrapolation as the key risk, and my stress-test agrees that this is the load-bearing point. I sharpen it: the paper's own data in Fig. 6(b) show long-lived, strongly L-dependent correlations just above T_KT, so the integration cutoff and the L-xi ordering are not merely hypothetical concerns but are visible in the reported data. The exponential fit value B=2.26 is also not close to b_KT~1.57, weakening the quantitative link to the KT correlation length. However, I do not think this concern by itself overturns the paper: the qualitative scenario (spin transport enhanced near T_KT, thermal transport featureless) is supported by the direct simulation data, and the issue is one of numerical convergence and fitting robustness, which a revision with additional analysis could settle. Therefore the reader's CONDITIONAL verdict remains appropriate; no verdict change is needed. A non-finding would have been inappropriate because the central divergence claim is exactly the kind of numerical extraction that requires a convergence check, and the manuscript ships no code or raw data to allow independent verification.","tokens_in":27889,"tokens_out":5656,"duration_ms":61183,"concrete_test":"For the XY case Delta=0.95 at T/|J|=0.66 and, if feasible, T/|J|=0.64, compute the running integral I(t)=int_0^t dt' <j^s_x(0) j^s_x(t')> for L=96,192,384 and evaluate it at t_max = 200, 400, 800, 1600 (and 3200 if computationally possible). Plot I(t) versus t for each L. If I(t) has not plateaued by t=800, or if the L-extrapolated value of I(t_max) changes by more than about 20% when t_max is doubled, then the thermodynamic-limit sigma^s_xx is not converged and the exponential divergence claim is not established. Independently, refit sigma^s_xx(L->infinity) with T_KT as a free parameter and compare against a power-law form (T-T_KT)^{-nu}; report whether the essential-singularity form is statistically preferred.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that for XY-type anisotropy sigma^s_xx diverges as exp[B/sqrt(T/T_KT-1)] with B=O(1), caused by an exponentially growing spin-current relaxation time tau_s (Sec. V.B, Figs. 11-12). This claim depends on Eq. (11), where sigma is the infinite-time integral of the current autocorrelation, evaluated numerically only up to t=800 |J|^{-1} (Sec. II.C) and then extrapolated to L -> infinity. The paper itself reports that in the XY case at T/|J|=0.66, slightly above T_KT, the time correlation 'persists for a long time' and shows 'a large system size dependence' (Fig. 6(b)). That is exactly the regime where a finite-time truncation is least controlled: if tau_s grows with L (e.g., through vortex diffusion or finite-size low-energy modes), the integral truncated at t=800 underestimates the infinite-time integral, and the L -> infinity extrapolation of a finite-time quantity can produce an apparent essential singularity even when the true thermodynamic conductivity is regular. Moreover, at temperatures used for the exponential fit, the KT correlation length xi_s ~ exp[(pi/2)/sqrt(T/T_KT-1)] can greatly exceed the largest simulated L=384 (e.g., at T/|J|=0.62, xi_s/a is thousands), so the finite-size extrapolation is not controlled in the critical region. The reported B=2.26 +/- 0.10 is also not within error of b_KT ~ 1.57, so the stated agreement with the KT length-scale exponent is overstated. Without a direct check that the integrated correlation has plateaued before t_max and that the extrapolated sigma is stable against t_max, the exponential divergence could be an artifact of the simulation protocol rather than a property of the Hamiltonian.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies spin and thermal transport in the classical square-lattice antiferromagnetic XXZ model using hybrid Monte Carlo and spin-dynamics simulations, supplemented by linear spin-wave theory. For Ising-type (\\Delta>1), XY-type (\\Delta<1), and Heisenberg-type (\\Delta=1) anisotropies, the authors compute the thermal conductivity \\kappa and the spin-current conductivity \\sigma^s from time-dependent current correlations. They find that \\kappa shows no clear anomaly at the magnetic or KT transitions and follows a roughly power-law low-temperature increase, while \\sigma^s distinguishes the three universality classes. The central claim is that in the XY case \\sigma^s_{xx} diverges at T_\\mathrm{KT} as \\exp[B/\\sqrt{T/T_\\mathrm{KT}-1}] with B=2.26\\pm0.10, and that this divergence is caused by an exponentially growing spin-current relaxation time associated with vortex lifetimes.","tokens_in":28310,"tokens_out":5267,"duration_ms":54950,"significance":"If the central claim is substantiated, the result is significant: it identifies the longitudinal spin conductivity as a sharp transport signature of the Kosterlitz-Thouless transition, in contrast to thermal transport, and it connects the divergence to vortex dynamics. The paper is careful in using undamped spin dynamics without phenomenological damping, and the linear spin-wave results provide useful low-temperature cross-checks for both the equal-time correlations and the conductivity. The prediction is falsifiable by further simulation and, in principle, by experiments on quasi-two-dimensional XY magnets. The main limitation is that the exponential-divergence claim rests on finite-time integration and finite-size extrapolation in the critical region, so the numerical evidence needs additional controlled convergence tests before the claim can be accepted as quantitative.","major_comments":[{"comment":"The fit in Fig. 11(b) yields B=2.26±0.10, whereas the KT correlation-length exponent quoted in the same paragraph is b_\\mathrm{KT}\\simeq\\pi/2\\simeq1.57. The difference is about 7σ, so the statement that the obtained B is 'comparable to b_\\mathrm{KT}' is not supported by the reported errors. This comparison is load-bearing because the manuscript uses it to connect the conductivity divergence to the KT universality class. The claim should either be softened or accompanied by a quantitative explanation of why the fitted B differs from b_\\mathrm{KT} by this factor, supported by additional analysis.","section":"§V.B, Fig. 11(b)"},{"comment":"The conductivity in Eq. (11) is an infinite-time integral of the spin-current autocorrelation, but the numerical integration is truncated at t=800|J|^{-1}. In the XY case at T/|J|=0.66, slightly above T_\\mathrm{KT}, the manuscript itself reports that the correlation 'persists for a long time' and shows 'a large system size dependence' (Fig. 6(b)). No convergence test is shown establishing that the integrated correlation has reached a plateau before the cutoff. If slow, size-dependent tails exist beyond t=800, the finite-time integral systematically underestimates the thermodynamic conductivity, and the L→∞ extrapolation of a truncated integral could produce an apparent essential singularity even if the true conductivity is regular. Please provide, for the largest sizes and temperatures used in the exponential fit, plots of the running time integral of ⟨j^z_{s,x}(0)j^z_{s,x}(t)⟩ as a function of the upper cutoff, or an equivalent demonstration that the integral has converged.","section":"§II.C and Fig. 6(b)"},{"comment":"The finite-size extrapolation in Fig. 11(a) is performed at temperatures where the KT correlation length greatly exceeds the simulated sizes. For example, at T/|J|=0.62 one has T/T_\\mathrm{KT}\\simeq1.03 and ξ_s/a\\sim\\exp[(\\pi/2)/\\sqrt{T/T_\\mathrm{KT}-1}]\\sim10^3–10^4, while the largest simulated linear size is L=384. In this regime the assumption that σ^s_{xx}(L) can be extrapolated to L→∞ without explicitly accounting for ξ_s\\sim L is uncontrolled. A finite-size scaling analysis, for instance plotting σ^s_{xx} as a function of L/ξ_s or performing fits that include the ξ_s dependence, is needed to substantiate the thermodynamic-limit divergence and the fitted exponent B.","section":"§V.B, Fig. 11(a)"}],"minor_comments":[{"comment":"The text and caption label both panel (a) and panel (b) as Δ=1.05; panel (b) should correspond to the XY-type case Δ=0.95.","section":"Fig. 7 caption / §V"},{"comment":"The sentence describing the determination of τ_s refers to 'Fig. 8 (b)', but the spin-current time-correlation function for the XY case is shown in Fig. 6(b); Fig. 8 shows a different quantity.","section":"§V.B"},{"comment":"The expression '2b_XT\\simeq\\pi' contains a typographical error: it should read '2b_\\mathrm{KT}\\simeq\\pi'.","section":"§V.B"},{"comment":"The sentence 'neither a magnetic transition nor a topological one does not occur' contains a double negative; it should read 'neither a magnetic transition nor a topological one occurs'.","section":"§V.C"},{"comment":"The notation for the conductivity components would be clearer if the manuscript explicitly stated, near Eq. (11), that σ^s_{xx}=σ^s_{yy} by square-lattice symmetry and that the off-diagonal components vanish within the reported precision, as is later shown for κ and σ^s.","section":"§II.B, Eq. (11)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and the numerical study is honest and systematic. The central claim is physically interesting and defensible in principle, but the controlledness of the exponential-divergence extraction needs to be demonstrated. The requested convergence and finite-size scaling checks are feasible and should be decisive. I do not see a reason for rejection if the authors can provide those checks."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take. This paper does something genuinely useful: it runs hybrid MC + spin dynamics on the classical square-lattice XXZ model across Ising, XY, and Heisenberg anisotropies and computes both spin and thermal conductivities from current autocorrelations. The qualitative separation is convincing: thermal conductivity is featureless across all three ordering classes, while spin conductivity develops a sharp size-growing peak at T_KT in the XY case and an exponential rise toward T=0 in the Heisenberg case. The LSWT analysis is also clean and explains why the XY spin conductivity is naturally small below T_KT—the leading magnon spin current vanishes when the quantization axis is in-plane. That part is worth citing.\n\nThe soft spot is exactly where the reader put it. The claim that sigma^s_xx ~ exp[B/sqrt(T/T_KT-1)] at T_KT is extracted from finite-time autocorrelations integrated to t=800|J|^{-1} and then extrapolated in L. At T=0.66, just above T_KT, the paper itself reports long-lived, strongly L-dependent correlations. That is the regime where a finite-time truncation can masquerade as a divergence. The fit uses temperatures where the KT correlation length is far larger than the simulated L=384, so the L->infinity extrapolation is not controlled without an assumed scaling form. The reported B=2.26±0.10 is not within error of b_KT~1.57, so saying the two agree is too strong; the independent tau_s fit gives B~2.58, which is consistent with sigma but not with b_KT either. These are fixable: plateau checks, longer t_max at selected temperatures, maybe a collapse with xi_s, and an honest statement that B is O(1) but not quantitatively b_KT.\n\nThe vortex-lifetime mechanism is plausible but inferred, not directly measured—they never track vortex density or lifetime. Still, the central qualitative result, that spin transport discriminates between universality classes while thermal transport does not, survives these concerns. I do not think there is a load-bearing flaw, but the headline divergence needs hardening.\n\nWho is this for? People working on spin transport in 2D magnets and on KT physics in classical spin models. It deserves a serious referee; I would send it to review and ask for the convergence analysis. If the exponential divergence survives a direct t_max check, it becomes a much stronger paper.","headline":"A solid numerical study with a compelling qualitative claim—spin transport sees the KT transition, thermal transport doesn't—but the headline exponential divergence is less nailed down than the abstract suggests.","tokens_in":28855,"tokens_out":3182,"would_cite":true,"duration_ms":34661,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"In a classical square-lattice antiferromagnet with XY anisotropy, the spin-current conductivity diverges at the Kosterlitz-Thouless temperature as $\\exp[B/\\sqrt{T/T_{KT}-1}]$, while thermal conductivity shows no anomaly.","keywords":["classical XXZ model","square-lattice antiferromagnet","spin-current conductivity","thermal conductivity","Kosterlitz-Thouless transition","vortex unbinding","magnetic anisotropy","spin dynamics simulation"],"falsifier":"A decisive numerical test would extend the spin-dynamics integration beyond $t=800\\,|J|^{-1}$ on the largest lattices used here, or repeat the calculation with a small Gilbert damping added: if the time-integrated spin-current correlation saturates to a finite value as $T_{KT}$ is approached, or if the fitted exponent $B$ shifts beyond its quoted uncertainty, the claimed exponential divergence is a finite-window artifact rather than an intrinsic property of the XXZ model.","tokens_in":27685,"feed_emoji":"🧲","tokens_out":12275,"duration_ms":110154,"temperature":0.7,"pith_summary":"The paper asks whether a magnetic phase transition leaves a measurable fingerprint in spin and heat transport in a two-dimensional antiferromagnet. Using the classical XXZ model on the square lattice, it argues that the thermal conductivity is blind to the ordering transition, while the longitudinal spin-current conductivity $\\sigma^s_{xx}$ carries a sharp, universality-class-dependent signal. For XY-type anisotropy ($\\Delta<1$), $\\sigma^s_{xx}$ diverges at the Kosterlitz-Thouless temperature as $\\sigma^s_{xx}\\propto\\exp[B/\\sqrt{T/T_{KT}-1}]$ with $B=\\mathcal{O}(1)$; for Ising anisotropy there is no anomaly at $T_N$, and for the Heisenberg case the conductivity grows exponentially toward zero temperature. The divergence is traced to the spin-current relaxation time growing as free vortices become long-lived near $T_{KT}$. If the claim is right, spin transport—not heat transport—is the practical probe of topological vortex unbinding in these magnets.","feed_headline":"Spin current diverges at vortex-unbinding transition in 2D magnet","feed_subtitle":"Heat flow stays featureless, but spin transport peaks sharply at the vortex-unbinding transition.","key_machinery":"The load-bearing object is the time-dependent spin-current correlation function $\\langle J^z_s(0)J^z_s(t)\\rangle$, computed from the semiclassical equation of motion $d\\mathbf{S}_i/dt=\\mathbf{S}_i\\times\\mathbf{H}^{\\rm eff}_i$, whose zero-frequency integral defines $\\sigma^s_{xx}$ through the Green-Kubo formula. The paper factorizes the conductivity as $\\sigma^s_{xx}\\simeq T^{-1}\\langle|j^z_{s,x}(0)|^2\\rangle\\tau_s$, where the equal-time fluctuation term is mild but the spin-current relaxation time $\\tau_s$ diverges toward $T_{KT}$ in the XY case, fitted as $\\tau_s\\propto\\exp[\\tilde{B}/\\sqrt{T/T_{KT}-1}]$ with $\\tilde{B}\\simeq 2.6$. This divergence is attributed to the growing lifetime of diffusing free vortices, which must travel farther to find an antivortex partner as the inter-vortex distance grows. The same machinery makes the effect specific to the XY case: in Ising and Heisenberg magnets, $\\tau_s$ is controlled by ordinary magnon damping or by the spin correlation length, not by vortex unbinding.","core_discovery":"The central claim is that the conserved magnetization current in the classical antiferromagnetic XXZ model on the square lattice distinguishes the three universality classes, while the energy current does not. In the XY case ($\\Delta<1$), the longitudinal spin-current conductivity $\\sigma^s_{xx}=\\sigma^s_{yy}$ diverges as $T$ approaches $T_{KT}$ from above, following $\\sigma^s_{xx}\\propto\\exp[B/\\sqrt{T/T_{KT}-1}]$ with fitted $B\\simeq 2.3$, comparable to the KT correlation-length coefficient $b_{KT}\\simeq\\pi/2$. Below $T_{KT}$, $\\sigma^s_{xx}$ drops to a vanishingly small value, in line with the absence of a leading-order magnon spin current when the easy-plane order is perpendicular to the conserved spin component. In the Ising case ($\\Delta>1$), $\\sigma^s_{xx}$ rises monotonically toward low temperature with no clear anomaly at $T_N$, and in the Heisenberg case ($\\Delta=1$) it grows approximately as $\\exp[b_H|J|/T]$, following the exponentially growing spin correlation length. The thermal conductivity, by contrast, shows a common power-law increase toward $T=0$ in all three cases.","pith_inferences":["Since the divergence lives in $\\tau_s$, the frequency-dependent (ac) spin conductivity should develop a narrow low-frequency peak that sharpens as $T_{KT}$ is approached; measuring it would give a direct experimental window on vortex lifetimes.","In real quasi-two-dimensional XY magnets with weak interlayer coupling, the true divergence at $T_{KT}$ will be rounded by three-dimensional ordering, so the paper's mechanism predicts a crossover enhancement of $\\sigma^s_{xx}$ just above the ordering temperature rather than a true singularity.","The same vortex-lifetime logic applies to frustrated Heisenberg magnets with $Z_2$ vortices, where a spin-conductivity enhancement at the vortex-unbinding temperature $T_v$ should appear even though the correlation length stays finite—an extension the paper leaves open.","If the factorization $\\sigma^s_{xx}\\simeq T^{-1}\\langle|j^z_s|^2\\rangle\\tau_s$ is exact enough, comparing the peak height with the equal-time current fluctuation would extract $\\tau_s$ directly, turning a transport measurement into a measurement of topological-defect lifetime."],"forward_implications":["In the XY-anisotropic square-lattice antiferromagnet, $\\sigma^s_{xx}$ should show a sharp, exponentially divergent peak just above $T_{KT}$ while the thermal conductivity remains featureless, so spin transport is the discriminating probe of the KT transition.","Below $T_{KT}$, $\\sigma^s_{xx}$ drops to a near-zero value because the leading magnon contribution to the conserved spin current vanishes when the easy-plane order is perpendicular to the spin-current polarization; the same suppression holds for antiferromagnetic and ferromagnetic XY models.","In the Ising-anisotropy case, no transport anomaly marks $T_N$; $\\sigma^s_{xx}$ grows roughly as $T^{-1}$ toward low temperature, controlled by magnon damping.","In the Heisenberg case, $\\sigma^s_{xx}$ grows as $\\exp(b_H|J|/T)$ toward $T=0$, mirroring the exponential spin-correlation length and the absence of finite-temperature order.","The contrast between spin and thermal currents is a general lesson: conserved magnetization currents can be sensitive to topological excitations that carry no obvious energy anomaly, so spin-conductivity measurements can detect vortex unbinding where heat transport cannot."],"supporting_citations":[{"why":"Introduces the Kosterlitz-Thouless vortex binding-unbinding transition that the XY case is identified with.","marker":"[7]"},{"why":"Supplies the standard definitions of the spin and thermal currents used in Eqs. (5)–(6).","marker":"[13–21]"},{"why":"Gives the linear-response (Green-Kubo) formula from which both conductivities are computed.","marker":"[23]"},{"why":"Provides the magnon-damping calculation ($\\alpha\\propto T^2$) that anchors the low-temperature transport expectations.","marker":"[27]"},{"why":"Gives the exponential spin-correlation length of the two-dimensional Heisenberg magnet used to fit $\\sigma^s_{xx}\\propto\\exp(b_H|J|/T)$.","marker":"[28]"},{"why":"Locates the KT transition temperature $T_{KT}\\simeq 0.6|J|$ for the XY case, fixing the temperature scale for the divergence.","marker":"[34–36]"},{"why":"Supplies the KT correlation-length form with $b_{KT}\\simeq\\pi/2$ that the divergent conductivity and $\\tau_s$ are compared with.","marker":"[37]"},{"why":"Describes diffusive vortex motion and the vortex lifetime estimate $\\tau_{vtx}\\propto \\xi_s^2$ used to explain the growing spin-current relaxation time.","marker":"[58–62]"}],"fun_headline_variants":["Easy-plane spin current diverges at vortex unbinding","Heat transport blind, spin current sees KT transition","Spin conductivity diverges at KT, heat unaffected","Anisotropy gates spin-current divergence in 2D antiferromagnet"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the undamped spin-dynamics equation, started from Monte Carlo thermalized configurations, relaxes the spin current completely within integration windows of 100–800 inverse exchange couplings, and that finite-size extrapolation then yields the thermodynamic-limit conductivity; if longer-lived current tails or lattice-induced damping are needed, the exponential divergence at $T_{KT}$ could be an artifact of the simulation protocol rather than a property of the spin Hamiltonian.","fun_headline_variants_meta":{"raw":{"variants":["Easy-plane spin current diverges at vortex unbinding","Heat transport blind, spin current sees KT transition","Spin conductivity diverges at KT, heat unaffected","Anisotropy gates spin-current divergence in 2D antiferromagnet"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0003,"raw_usage":{"total_tokens":1861,"prompt_tokens":1206,"completion_tokens":655,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":822,"completion_tokens_details":{"reasoning_tokens":589}},"tokens_in":822,"tokens_out":655,"duration_ms":6714,"temperature":1.0,"reasoning_tokens":589,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:38:44.470447+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive numerical test would extend the spin-dynamics integration beyond $t=800\\,|J|^{-1}$ on the largest lattices used here, or repeat the calculation with a small Gilbert damping added: if the time-integrated spin-current correlation saturates to a finite value as $T_{KT}$ is approached, or if the fitted exponent $B$ shifts beyond its quoted uncertainty, the claimed exponential divergence is a finite-window artifact rather than an intrinsic property of the XXZ model.","supporting_citations":[{"cited_title":"Frangou, G","cited_arxiv_id":null,"evidence_quote":"Introduces the Kosterlitz-Thouless vortex binding-unbinding transition that the XY case is identified with."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the linear-response (Green-Kubo) formula from which both conductivities are computed."},{"cited_title":"Krech, A","cited_arxiv_id":null,"evidence_quote":"Provides the magnon-damping calculation ($\\alpha\\propto T^2$) that anchors the low-temperature transport expectations."},{"cited_title":"Okubo and H","cited_arxiv_id":null,"evidence_quote":"Gives the exponential spin-correlation length of the two-dimensional Heisenberg magnet used to fit $\\sigma^s_{xx}\\propto\\exp(b_H|J|/T)$."},{"cited_title":"Cuccoli, V","cited_arxiv_id":null,"evidence_quote":"Supplies the KT correlation-length form with $b_{KT}\\simeq\\pi/2$ that the divergent conductivity and $\\tau_s$ are compared with."}],"review_version":1}