{"id":"32520e51-9570-4639-b82c-9acb48be6e94","arxiv_id":"1908.02609","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For node-centric continuous-time random walks on two-layer circulant multiplexes, mixing is fastest at a finite optimal inter-layer coupling, and fractional dynamics makes inter-layer jump probabilities decay with distance as d^{-(1+2γ)}.","lead":"This paper studies a mathematical model of random movement on two-layer networks, where a walker can switch layers and take long jumps. It finds a best strength for the bridge between layers that makes spreading fastest, and derives a power-law rule for how often long cross-layer jumps happen.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Long-range inter-layer transition derivation in Sec. IV B / App. C is not fully supported: finite-N Fourier sums are replaced by integrals, and the printed A_m definition conflicts with Eq. (22).","rationale":"The reader's weakest_assumption correctly flags the thermodynamic-limit integral approximation in Appendix C as a source of uncertainty for the long-range navigation claim. I agree that this is the main unresolved issue in the paper's secondary claim, and I add the concrete internal inconsistency in the definition of A_m in Eq. (39), which the reader's rationale also mentions as needing a fix. However, the reader's strongest_claim, Eq. (38), is not affected by either issue: the regular-multiplex eigenvalue argument is exact for finite N and the location of the optimum is independent of γ. The paper's central quantitative contribution is therefore solid, and the required changes are corrections and qualifications rather than a rejection. The verdict CONDITIONAL remains appropriate, so I leave it unchanged.","tokens_in":20915,"tokens_out":32798,"duration_ms":328114,"concrete_test":"Recompute T^(γ)_{i(j+N)} from the exact spectral sum (Eqs. (16), (25)-(27), (37)) for N = 501 and N = 20001, J = 1, γ = 0.5, D_x = 2, over d = 2..N/2; fit log T vs log d and compare exponent and prefactor to Eq. (45). Independently re-derive Eq. (41) from Eq. (64) using A_m = 2 − 2 cos(2π(m−1)/N) from Eq. (22). If the finite-N exponent deviates from 1 + 2γ or the prefactor disagrees with the corrected Eq. (45), the long-range claim needs qualification; if both match, the concern does not land.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The optimal-coupling result, Eq. (38), is well supported: for regular layers the eigenvalues of L^(γ) are µ^γ/σ, and the smallest nonzero µ is min(2D_x, A_c), giving a unique maximum at D_x = A_c/2 independent of γ. The concern targets the separate long-range navigation claim. Its derivation in Sec. IV B and Appendix C replaces the exact finite-N Fourier sum over m = 1..N by an integral over θ ∈ [0,2π) (Eqs. (64)-(66)). For a finite cycle, the discrete sum is periodic in d with period N, and the trapezoidal-rule error introduces aliasing contributions from frequencies d ± N, d ± 2N, etc. Near d ~ N/2 the alias at N − d is of the same order as the leading term, so the claimed pure power law T ~ d^{-(1+2γ)} is only a thermodynamic-limit statement, not a finite-N result. Additionally, Eq. (39) defines A_m = 2 + 2 cos(2θ_m) with θ_m = 2π(m−1)/N, whereas Eq. (22) for J = 1 gives A_m = 2 − 2 cos(2π(m−1)/N). As printed, the integral identity (66) does not apply to the A_m used in Eq. (39). These two issues leave the long-range navigation claim less secure than the optimal-coupling claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper extends the fractional continuous-time random walk (CTRW) framework of Refs. [16-18] to two-layer multiplex networks. For node-centric Poissonian CTRWs, the walker's transition matrix is derived from the normalized fractional supra-Laplacian. The authors obtain exact eigenvalues and eigenvectors for multiplexes with two circulant layers, and analyze the algebraic connectivity Lambda_2 of the normalized fractional supra-Laplacian as a function of the inter-layer diffusion coefficient D_x. For regular layers with J1=J2=J they derive a closed-form expression, Eq. (38), showing that Lambda_2 is nonmonotonic in D_x and has a global maximum at D_x=A_c/2 that is independent of the fractional exponent gamma. The paper also claims that fractional dynamics produces a new type of long-range inter-layer navigation, with transition probabilities decaying as d^{-(1+2gamma)} in the thermodynamic limit, and reports numerical simulations of the mean-square displacement showing normal diffusion (epsilon=1) even in the enhanced-diffusion regime.","tokens_in":21199,"tokens_out":11987,"duration_ms":119853,"significance":"The regular-layer optimal-coupling result is the strongest part of the paper. It is derived without any fitted parameters: for J1=J2=J the eigenvalues of the normalized fractional supra-Laplacian are mu_m^gamma/sigma^(gamma), so Eq. (38) follows directly from the spectrum, and the prediction that the optimal D_x=A_c/2 is independent of gamma is confirmed by Fig. 1. The analytic machinery in Appendices A and B is detailed and self-consistent. If the long-range-navigation derivation is corrected, the paper would provide a useful, falsifiable design rule for optimizing transport on interconnected networks and a novel characterization of fractional walkers on multiplexes. The numerical check of Eq. (38) and the absence of fitted parameters are notable strengths; the long-range claim, however, needs substantial repair before the paper can be accepted.","major_comments":[{"comment":"The definition of A_m is internally inconsistent. For J=1, Eq. (22) gives A_m = 2 - 2 cos(2 pi (m-1)/N), whereas Eq. (39) and Eq. (64) state A_m = 2 + 2 cos(2 theta_m) with theta_m = 2 pi (m-1)/N, which is 2 + 2 cos(4 pi (m-1)/N). The integral identity in Eq. (66) is the standard Fourier integral for the symbol 2 - 2 cos theta, so as printed it cannot be applied to the A_m appearing in Eq. (39). Consequently, the derivation of Eqs. (41)-(45), including the power law T ~ d^{-(1+2 gamma)}, is not supported by the equations as written. The authors should either correct the definition of A_m consistently with Eq. (22) or redo the derivation from the correct finite-sum expression and re-verify Eq. (45) against the exact numerical values of T^(gamma)_{i(j+N)}.","section":"IV B and Appendix C, Eqs. (39), (64), (66)"},{"comment":"The replacement of the finite Fourier sum over m=1,...,N by an integral over theta in [0,2 pi) is a thermodynamic-limit step whose finite-N error is not controlled. For a cycle, the exact sum is periodic in the distance d with period N, and the trapezoidal-rule/Poisson-summation error introduces aliased contributions at d ± N, d ± 2N, and so on. When d is near N/2, the alias at N-d is of the same order as the leading term, so the asserted power-law decay T ~ d^{-(1+2 gamma)} is at best an asymptotic statement for d << N, not a finite-N result. Since Fig. 2 uses N=20001 and d up to about 10^4, this is not a purely formal caveat. The paper should either restrict the long-range claim explicitly to the thermodynamic limit throughout the abstract, main text, and conclusions, or provide quantitative finite-N error estimates and numerical verification of Eq. (45) in the regime where aliasing is expected to matter.","section":"IV B and Appendix C, Eqs. (64)-(66)"}],"minor_comments":[{"comment":"The text states that the eigenvalues of the normalized supra-Laplacian are lambda_m = mu_m/sigma^(gamma), but the exponent gamma is missing: the correct relation for the fractional normalized supra-Laplacian is lambda_m = mu_m^gamma/sigma^(gamma), as is in fact used in Eq. (38). Please correct this line.","section":"Section IV A, text before Eq. (38)"},{"comment":"Figure 3 shows no error bars or statistical details for the simulated MSD curves, and the inset labels appear garbled (for example '2.0001.62 1.65 1.68' and '1.001.92 1.95 1.98'). The authors should clarify what these numbers are and report the number of realizations or an equivalent error measure.","section":"Figure 3"},{"comment":"The text cites Ref. [43] for a discussion of the integral in the thermodynamic limit, but Ref. [43] is a synchronization paper and does not appear to contain such a discussion; the relevant references are likely Refs. [16-18] and [60].","section":"Section IV B, text around Eq. (41)"},{"comment":"The manuscript refers several times to a Supplemental Material file, but no supplement is included in the arXiv version; all claims attributed to the supplement should either be verifiable in the main text or the supplement should be provided.","section":"Conclusions and Supplemental Material"},{"comment":"There are several typographical errors, including 'sucessful', 'paramenter' (twice), 'thermodinamic', 'therodynamic', 'hoping' for 'hopping', and 'aditional'; a careful proofreading pass is needed.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and the optimal-coupling result is publishable after revision. The long-range-navigation section needs a corrected derivation and a clear thermodynamic-limit qualification; with those changes I would expect the paper to be acceptable. The missing error bars and unverifiable supplemental material should also be addressed in revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the central claim I’d take away is that on two-layer multiplexes with identical circulant layers, the normalized fractional supra-Laplacian has algebraic connectivity Λ2 = (2Dx)^γ/σ for small Dx and A_c^γ/σ for large Dx, so the relaxation time is minimized at Dx = Ac/2, independent of γ. That is a clean, non-obvious result, and it is supported by the eigenvalue derivation in Appendix A and by the figures. Nothing is fitted; the only inputs are Dx, γ, J, and N. This is more than a routine parameter scan—the exact circulant spectra and the optimal-coupling rule are new for multiplexes, and the extension of Refs. [16–18] is honest.\n\nWhere I’d put the pressure: the long-range navigation claim in Sec. IV B and Appendix C. The derivation replaces finite-N Fourier sums by integrals in the N→∞ limit. That is acceptable as a thermodynamic-limit statement, but the paper also plots finite-N results up to d ≈ N/2 (Fig. 2). For a finite cycle, the discrete sum is periodic in d, and near d ~ N/2 an alias at N−d is comparable to the leading term, so the claimed pure power law T ~ d^{-(1+2γ)} is not a finite-N result without error bounds. I don’t think this sinks the paper—it is a caveat the authors should add—but it is the part a referee should probe.\n\nThere are also three concrete consistency issues. First, Eq. (39) and Appendix C define A_m = 2 + 2 cos(2θ_m) with θ_m = 2π(m−1)/N, while Eq. (22) with J=1 gives A_m = 2 − 2 cos(2π(m−1)/N). As printed, the integral identity (66) does not apply to the A_m of Eq. (39). This is almost certainly a sign/θ-convention slip, but it affects the power-law derivation and must be fixed. Second, Sec. IV A writes λ_m = μ_m/σ(γ), dropping the γ power on μ_m; Eq. (38) uses μ^γ, so the text is inconsistent. Third, Figure 3 has no error bars, which matters for MSD curves that are close at small D_x.\n\nNone of these touch the optimal-coupling result. The math there is direct: for regular layers the eigenvalues of the normalized fractional supra-Laplacian are μ^γ/σ, and the smallest nonzero μ is min(2Dx, A_c), which gives the single maximum.\n\nBottom line: this deserves a serious referee. I’d send it out with instructions to check the thermodynamic-limit approximation and to require the typos fixed. It is a paper for researchers working on fractional diffusion, multilayer transport, and random-walk search; a reading group could get a good hour out of the finite-size discussion.","headline":"The optimal-coupling result is real and well supported; the long-range interlayer power-law needs a clearer finite-size caveat and a few equation typos before this is fully publishable, but the paper deserves a serious referee.","tokens_in":21736,"tokens_out":5112,"would_cite":true,"duration_ms":54128,"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":"For node-centric continuous-time random walks on two-layer multiplexes, the normalized fractional supra-Laplacian has an algebraic connectivity that peaks at a finite inter-layer diffusion coefficient, making relaxation time minimal at…","keywords":["fractional diffusion","multiplex networks","continuous-time random walks","supra-Laplacian","algebraic connectivity","optimal coupling","long-range navigation","circulant graphs"],"falsifier":"Compute $\\Lambda_2$ exactly or numerically for a regular two-layer cycle multiplex with $J=1$ over a range of $N$, $D_x$, and $\\gamma$, and compare the location of the maximum with $D_x = A_c/2$; if the maximizing coupling shifts with $N$ or with $\\gamma$, or if the measured inter-layer transition probability exponent departs from $1+2\\gamma$ as $N$ grows, the central claim fails.","tokens_in":20707,"feed_emoji":"🌐","tokens_out":11339,"duration_ms":108200,"temperature":0.7,"pith_summary":"This paper extends fractional diffusion to two-layer multiplex networks and studies node-centric continuous-time random walks, where a walker becomes active on a Poisson clock and jumps with probability proportional to edge weight. The central claim is that the convergence rate to the stationary state, governed by the algebraic connectivity $\\Lambda_2$ of the normalized fractional supra-Laplacian, is nonmonotonic in the inter-layer diffusion coefficient $D_x$: it grows for weak coupling and shrinks for strong coupling, so relaxation time $\\tau \\sim 1/\\Lambda_2$ has a minimum at a finite optimal $D_x$. For regular multiplexes made of two identical circulant layers, the paper derives an exact two-branch formula, $\\Lambda_2 = (2D_x)^\\gamma/\\sigma^{(\\gamma)}$ for $D_x \\le A_c/2$ and $\\Lambda_2 = A_c^\\gamma/\\sigma^{(\\gamma)}$ for $D_x \\ge A_c/2$, so the optimum is $D_x = A_c/2$ for every fractional order $\\gamma$. The paper also shows that fractional dynamics creates long-range inter-layer jumps whose probability decays as $d^{-(1+2\\gamma)}$, and that on finite circulant multiplexes the mean-square displacement still grows linearly with time before saturation.","feed_headline":"Fractional walks on two-layer networks diffuse fastest at one coupling","feed_subtitle":"Relaxation time is minimized at the optimal inter-layer coupling, independent of the fractional order.","key_machinery":"The machinery is the fractional supra-Laplacian $(L_M)^\\gamma$, obtained by raising the combinatorial supra-Laplacian $L_M = L_\\ell + L_x$ of the two-layer multiplex to a fractional power $\\gamma \\in (0,1)$, then normalizing node-wise to $\\mathcal{L}^{(\\gamma)} = K^{-1}(L_M)^\\gamma$. For two identical circulant layers, Fourier diagonalization yields closed-form eigenvalues and eigenvectors, and the algebraic connectivity collapses to the two-branch formula in Eq. (38). The long-range inter-layer result rests on an integral representation of the fractional Laplacian: replacing the finite Fourier sums by integrals in the thermodynamic limit and using the Gamma-function identity of Eq. (66) gives the transition probability $T^{(\\gamma)}_{i(j+N)} \\sim d^{-(1+2\\gamma)}$.","core_discovery":"The central claim is that for Poissonian node-centric continuous-time random walks on a two-layer multiplex, the spectral gap $\\Lambda_2$ of the normalized fractional supra-Laplacian $\\mathcal{L}^{(\\gamma)}$ is nonmonotonic in the inter-layer diffusion constant $D_x$: it increases as $(2D_x)^\\gamma/\\sigma^{(\\gamma)}$ when $D_x$ is small and decreases as $A_c^\\gamma/\\sigma^{(\\gamma)}$ when $D_x$ is large, attaining a unique maximum at $D_x = A_c/2$ (Eq. 38). The optimal coupling is independent of $\\gamma$, while smaller $\\gamma$ yields larger $\\Lambda_2$ and therefore faster convergence. The same nonmonotonic behavior is observed numerically in non-regular multiplexes, with a more pronounced peak for more random layer topologies. The paper further claims that fractional dynamics introduces a new inter-layer navigation mechanism: a walker can switch layers and land at a distant vertex in the same jump, and in the thermodynamic limit the inter-layer transition probability decays as $d^{-(1+2\\gamma)}$ for $d \\gg 1$. Despite this enhanced diffusion, the mean-square displacement on finite circulant multiplexes scales linearly with time before saturation, so the fractional walk remains Gaussian in that finite-size regime.","pith_inferences":["An operator tuning a multiplex transport system could treat the optimal coupling as a one-time spectral calculation: since $A_c$ is the smallest nonzero Laplacian eigenvalue of the layer, the setting $D_x = A_c/2$ is fixed by layer structure alone and remains optimal across fractional regimes.","For multiplexes with more than two layers, the same mechanism should produce a nonmonotonic spectral gap, but the exact optimum would likely depend on the full inter-layer coupling matrix rather than a single $A_c/2$; re-deriving that formula is a natural extension.","A direct test of the thermodynamic-limit derivation would be to measure inter-layer jump-length distributions on finite multiplexes: if the $d^{-(1+2\\gamma)}$ tail persists at moderate $N$, the integral approximation is robust enough for practical use, and if not, a finite-size corrected exponent should replace it."],"forward_implications":["For any fixed two-layer circulant multiplex, tuning the inter-layer diffusion to $D_x = A_c/2$ minimizes relaxation time, and this setting does not change when the fractional order $\\gamma$ is varied.","Smaller $\\gamma$ increases $\\Lambda_2$ for fixed $D_x$, so stronger fractional dynamics makes node-centric walks converge to the stationary state faster, with $\\tau \\sim 1/\\Lambda_2$.","Fractional node-centric walks on multiplexes can switch layers and land far away in the same hop, with inter-layer jump probability decaying as $d^{-(1+2\\gamma)}$; small $\\gamma$ makes the inter-layer jumps longer-tailed.","Even though fractional dynamics enhances diffusion, the mean-square displacement of these walks on finite circulant multiplexes remains linear in time before saturation, so the enhanced mixing does not imply super-diffusive transport.","The nonmonotonic dependence of $\\Lambda_2$ on $D_x$ persists in non-circular multiplexes and becomes more pronounced as the layers become more random, so choosing the inter-layer coupling optimally matters most for irregular multiplex topologies."],"supporting_citations":[{"why":"Defines fractional dynamics on networks through the fractional Laplacian, the framework this paper extends to multiplexes.","marker":"[16]"},{"why":"Provides the circulant-network spectral and integral methods used for the exact algebraic-connectivity formula and the long-range transition asymptotics.","marker":"[17]"},{"why":"Supplies the Gamma-function identities, continuum limits, and Lévy-flight interpretation behind the $d^{-(1+2\\gamma)}$ power law.","marker":"[18]"},{"why":"Introduces the normalized supra-Laplacian and the notion of physical random walkers that switch layers, which the paper generalizes to fractional jumps.","marker":"[33]"},{"why":"Establishes the supra-Laplacian description of diffusion on multiplexes and the role of inter-layer coupling in relaxation dynamics.","marker":"[35]"},{"why":"Shows optimal inter-layer coupling in directed multiplexes, the nonmonotonic behavior this paper proves for fractional node-centric walks.","marker":"[38]"},{"why":"Distinguishes node-centric from edge-centric continuous-time random walks and relates relaxation time to the spectral gap.","marker":"[11]"},{"why":"Supplies the eigenvalues and eigenvectors of circulant matrices used in the exact diagonalization of the two-layer multiplex.","marker":"[58]"},{"why":"Gives fractional-Laplacian integrals used to evaluate the Fourier-sum-to-integral approximation in Appendix C.","marker":"[60]"},{"why":"Connects Lévy jump distributions and waiting-time statistics to normal and anomalous diffusion classifications used to interpret the results.","marker":"[57]"}],"fun_headline_variants":["Fractional multiplex walks: optimal coupling independent of order","Nonmonotonic relaxation in fractional two-layer walks","Fractional walks on multiplexes: one optimal coupling","Fastest relaxation in fractional two-layer networks at one coupling","Fractional order doesn't shift the sweet spot in multiplex walks"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the thermodynamic limit $N \\to \\infty$, in which the discrete Fourier sums are replaced by integrals with the short-distance term $K_d$ included only at $d = 0$ and $d = 1$, accurately describes finite multiplexes; if finite-size corrections alter that integral approximation, the predicted inter-layer power law $d^{-(1+2\\gamma)}$ and the claimed long-range navigation would not hold as stated.","fun_headline_variants_meta":{"raw":{"variants":["Fractional multiplex walks: optimal coupling independent of order","Nonmonotonic relaxation in fractional two-layer walks","Fractional walks on multiplexes: one optimal coupling","Fastest relaxation in fractional two-layer networks at one coupling","Fractional order doesn't shift the sweet spot in multiplex walks"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001438,"raw_usage":{"total_tokens":5803,"prompt_tokens":960,"completion_tokens":4843,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":576,"completion_tokens_details":{"reasoning_tokens":4764}},"tokens_in":576,"tokens_out":4843,"duration_ms":39726,"temperature":1.0,"reasoning_tokens":4764,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:39:54.975871+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $\\Lambda_2$ exactly or numerically for a regular two-layer cycle multiplex with $J=1$ over a range of $N$, $D_x$, and $\\gamma$, and compare the location of the maximum with $D_x = A_c/2$; if the maximizing coupling shifts with $N$ or with $\\gamma$, or if the measured inter-layer transition probability exponent departs from $1+2\\gamma$ as $N$ grows, the central claim fails.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines fractional dynamics on networks through the fractional Laplacian, the framework this paper extends to multiplexes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the circulant-network spectral and integral methods used for the exact algebraic-connectivity formula and the long-range transition asymptotics."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Gamma-function identities, continuum limits, and Lévy-flight interpretation behind the $d^{-(1+2\\gamma)}$ power law."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the normalized supra-Laplacian and the notion of physical random walkers that switch layers, which the paper generalizes to fractional jumps."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the supra-Laplacian description of diffusion on multiplexes and the role of inter-layer coupling in relaxation dynamics."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows optimal inter-layer coupling in directed multiplexes, the nonmonotonic behavior this paper proves for fractional node-centric walks."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Distinguishes node-centric from edge-centric continuous-time random walks and relates relaxation time to the spectral gap."},{"cited_title":"Van Mieghem, Graph Spectra for Complex Networks 2011 (Cambridge University Press, Cambridge)","cited_arxiv_id":null,"evidence_quote":"Supplies the eigenvalues and eigenvectors of circulant matrices used in the exact diagonalization of the two-layer multiplex."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives fractional-Laplacian integrals used to evaluate the Fourier-sum-to-integral approximation in Appendix C."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Connects Lévy jump distributions and waiting-time statistics to normal and anomalous diffusion classifications used to interpret the results."}],"review_version":1}