{"id":"53415af1-d5b2-45f2-97d3-bee621353440","arxiv_id":"1908.07219","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For slowly varying heterogeneity and small diffusion, a reaction-diffusion system forms localized Turing patterns precisely where the pointwise classical Turing inequalities hold.","lead":"Mathematicians derive precise conditions for Turing pattern formation in reaction-diffusion systems whose background state varies slowly across space, using WKBJ asymptotics. The result justifies the commonly used 'local Turing conditions' and shows unstable modes are spatially localized instead of simple wavy functions.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The effective Dirichlet condition at singular points is not a derived turning-point connection rule; without the missing boundary-layer phase shift, the WKBJ modes (23b)-(23c) are not established as asymptotic eigenfunctions.","rationale":"The reader's weakest assumption identifies the same gap: the singular-point boundary-layer structure is unresolved, so the piecewise WKBJ solutions are not proven to be true asymptotic eigenfunctions of the linearized operator. I agree that this is the load-bearing issue. My stress-test sharpens it: the problem is not merely that the inner layer is unanalyzed; the paper's 'effective Dirichlet' condition is a specific connection rule that is generally false at a simple turning point. Standard WKBJ connection theory gives a phase shift (pi/4 in the scalar case, with a nontrivial Stokes constant in degenerate 2x2 systems), and the paper's sine modes with zero phase at x* and quantization constants K = 0, 1 omit that shift. This affects the quantitative mode predictions in Criterion 3 and the structural claims in Section IV.B, even if the leading-order local instability condition of Criterion 2 may survive once the phase shift is absorbed by the freedom to choose a nearby integer mode. The numerical simulations in Figures 2-4 support the qualitative picture, but they do not test the discrete eigenmode relation (27) directly, so they do not resolve the mismatch. The authors' own limitation statement in the Discussion confirms the gap. None of this makes the paper's central heuristic implausible, and the asymptotic and numerical work is substantial, so the appropriate outcome remains the reader's conditional acceptance: the headline local-Turing-condition claim is plausible but not rigorously established at the singular points, and the quantitative mode-selection results need either a boundary-layer connection analysis or explicit downgrading to a conjecture. I therefore leave the verdict unchanged.","tokens_in":26517,"tokens_out":12287,"duration_ms":141598,"concrete_test":"Re-derive the connection rule across a simple zero x* of Delta_lambda. Introduce the inner variable eta = (x - x*)/epsilon^{8/9}, expand (8) about x*, and solve the leading-order inner system for the bounded solution that matches the two outer WKBJ branches. Extract the phase gamma on the oscillatory side. If gamma is nonzero (e.g., pi/4 as in scalar WKBJ), then the effective Dirichlet condition used in (23b)-(23c) and the quantization condition (27) with K = 0,1 are incorrect, so Criterion 3's predicted modes are not asymptotic eigenfunctions. If gamma = 0, the paper's effective Dirichlet condition is supported and this concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Criterion 3, and through it Criterion 2, asserts the existence of piecewise WKBJ modes that vanish at singular points x* where the discriminant Delta_lambda = [tr(B_lambda)]^2 - 4det(B_lambda) has a simple zero. This relies on the 'effective Dirichlet' condition introduced in Section III.B and supported only by the Q0 ~ |X|^{-1/4} scaling in SI S2. But a simple zero of Delta_lambda is a turning point for the coupled 2x2 system (8), not a hard wall. In the inner layer of width epsilon^{8/9} (a width the authors themselves identify in S2), the correct connection formula generically produces an additional phase shift: in scalar WKBJ the oscillatory solution behaves as sin(epsilon^{-1} integral sqrt(mu) d xi + pi/4), not sin(epsilon^{-1} integral sqrt(mu) d xi). The paper imposes phase zero. Consequently the quantization conditions (27) with K = 0, 1 are phase-shifted, and the discrete growth rates and mode supports computed in Section IV.B and Figures 5-6 need not be asymptotic eigenvalues or eigenfunctions. No matching of the outer solution to an inner solution is supplied, so the zero extension across x* is not shown to be close to a genuine eigenfunction of (3). The Discussion concedes that 'fully resolving the boundary-layer structure across these singularities is beyond our present scope.' Since every mode in (23) with an internal singular point depends on this unproved connection, the central instability claim is not established at the claimed level.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript derives generalizations of the classical Turing instability conditions to two-component reaction-diffusion systems with smoothly varying spatial heterogeneity, using a WKBJ expansion in the small parameter ε. The authors define 'permissible' growth rates through the reality and non-negativity of the local eigenvalues of D^{-1}(J−λI), prove monotonicity of the unstable support Tλ in λ, and construct piecewise WKBJ modes localized on intervals whose endpoints are simple zeros of the discriminant. They state local Turing conditions (Criterion 2) and a discrete mode-selection condition (Criterion 3). For the Schnakenberg model with spatially varying kinetics, they compare the predicted supports with direct numerical simulations for ε down to 10^{-5}, and they construct the predicted discrete modes and their supports. The paper is clearly written, and the logical structure from local eigenvalue conditions to instability criteria is mostly self-contained and presented as a series of explicit propositions.","tokens_in":26887,"tokens_out":10314,"duration_ms":117052,"significance":"If the derivation were complete, the paper would make a genuinely useful contribution to the pattern-formation literature. It provides a principled justification of the widely used heuristic of local Turing conditions, and it identifies a structurally new feature: unstable modes with different growth rates are localized in different, nested spatial regions. Proposition 8, the monotonicity of Tλ in λ, is a clean and testable prediction, and the numerical experiments in Section IV are careful and support the predicted support and localization of patterns as ε is decreased. The instability conditions are consequences of the WKBJ formalism and stability assumptions rather than fitted parameters, and the comparison with numerics is quantitative for the support. The main weakness is the treatment of the singular points where the discriminant vanishes, which is acknowledged by the authors but is load-bearing for the claimed mode construction.","major_comments":[{"comment":"The 'effective Dirichlet' condition at a singular point x* is imposed rather than derived. At a simple zero of [tr(Bλ)]²−4det(Bλ), the two WKBJ eigenvalues collide and x* is a turning point for the coupled system (8), not a hard wall. The inner layer identified in SI S2 has width ε^{8/9}, and no matching of the outer WKBJ solution to a solution of the inner problem is supplied. For scalar WKBJ turning points, the connection formula generically adds a phase (π/4 in the standard Airy case) to the oscillatory argument; the sine forms in (23b)-(23c), which have phase zero at x*, therefore constitute an unproved choice of connection. Consequently, the quantization condition (27) and the discrete modes plotted in Figures 5-6 are not established as asymptotic eigenvalues or eigenfunctions of the linearized operator (3). Because Criterion 3 and, through it, Criterion 2 rely on the existence and form of these modes, this gap is load-bearing for the central claim. A constant phase correction might preserve the existence of some mode for sufficiently small ε, but it would change the mode numbers, supports, and growth rates at finite ε, so the manuscript's proof as written depends on the uncorrected condition.","section":"Section III.B and SI S2; Eqs. (23b)-(23c) and (27)"},{"comment":"The authors state that fully resolving the boundary-layer structure across the singularities is beyond the present scope and then assert that their results 'will remain valid even with such refinements.' The second assertion is not supported by the analysis. A connection phase changes the oscillatory factor over O(1) spatial intervals, so it is not an exponentially small correction; it affects which linear combination of WKBJ exponentials satisfies the boundary conditions and hence the predicted eigenfunctions. The paper should either supply the missing connection/matching analysis or explicitly present the modes as formal approximations whose leading-order support is numerically verified, with the phase and quantization statements marked as conjectural rather than derived.","section":"Discussion, final paragraph; SI S2"},{"comment":"The reduction of Criterion 3 to Criterion 2 is only sketched. The text says that relaxing the single-interval restriction and taking 'a suitable choice of arbitrarily small ε' yields Criterion 2, but no proof is given that for every sufficiently small ε there exists a permissible pair (λ, n±) satisfying (27) with Tλ approximating T0. Since Criterion 2 is stated as the main heterogeneous instability condition, a precise statement with explicit quantifiers, and a proof (or a precise asymptotic bound) are needed. The current wording leaves the meaning of 'for sufficiently small ε' ambiguous, and the claim that the perturbation grows for all x in the interior of T0 should be justified uniformly as ε→0.","section":"Section III, paragraph after Eq. (27); Criterion 2"},{"comment":"Figures 5-6 are constructed directly from the WKBJ formulas (23b) and (27), so they do not provide an independent check of the mode-selection condition. The full simulations in Figures 2-4 confirm the spatial support and localization of the patterned state, but they do not resolve individual eigenvalues or eigenfunctions of the linearized operator. To support the discrete mode predictions, the authors should compare the WKBJ modes with numerically computed eigenfunctions of the discretized linearized problem (3) at the same parameter values; this would also help identify whether the phase correction discussed above is numerically significant.","section":"Section IV.B, Figures 5-6"}],"minor_comments":[{"comment":"The sentence 'conditions for instability which are local versions of the classical Turing conditions We find that the structure...' is missing punctuation between 'conditions' and 'We'.","section":"Abstract and Section II"},{"comment":"The expansion for p(x) reads p(x) = p0 + εp1 + ε²p1 + ...; the coefficient of ε² should presumably be p2, not p1.","section":"Section III, near Eq. (8)"},{"comment":"The sentence 'we have to determine what happens to the WKBJ solution on approaching the point where sT_* p_* ≠ 0 and beyond' should refer to points where sT_* p_* = 0, not ≠ 0, based on the surrounding discussion.","section":"Section III.B"},{"comment":"The proof references 'Proposition 9', but Proposition 9 is stated in the Supplementary Information, not in the main text; the cross-reference should be made explicit so the main text is self-contained.","section":"Criterion 3 proof"},{"comment":"The data availability statement contains the placeholder URL 'http://dx.doi.org/xx.xxxx/xxxxxxxxxxxxxxxxxx'; this should be replaced with a working link or removed before publication.","section":"Data availability statement"}],"recommendation":"major_revision","confidential_remarks":"The gap at the turning points is acknowledged by the authors in the Discussion, and the numerical evidence strongly supports the predicted support and localization. I am not recommending rejection because the existence claim is likely salvageable by a proper connection analysis, or by reformulating the mode-selection results as formal with independent numerical validation of the eigenfunctions. The revision should address the effective Dirichlet condition directly, as it is the main load-bearing assumption of the WKBJ construction."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things up front. This paper gives the standard local-Turing-condition heuristic its first real theoretical backing: it derives, not assumes, pointwise instability criteria for smoothly heterogeneous domains, and it shows that unstable modes localize to the predicted support. It also carries an acknowledged but unresolved formal gap at its singular points, and that gap is more damaging to the paper's detailed mode-counting claims than to its central existence claim.\n\nWhat is new and good. The WKBJ derivation of Criterion 2 proceeds from a mode-selection integral rather than from postulating that homogeneous conditions hold pointwise; Proposition 8 (monotonic containment of the supports Tλ in λ) is new; and the mode shapes in (23) genuinely differ from trigonometric modes. The numerics are the best part: two independent solvers, a range of ε down to 10^-5, and linear-plus-nonlinear runs that track the predicted boundary of T0. The comparison with Dewel-Borckmans and Kuske-Eckhaus is fair, and the self-citations are contextual rather than load-bearing.\n\nThe soft spot, in proportion. The stress-test has a real point. At a simple zero of the discriminant the paper imposes effective Dirichlet conditions, justified only by the |X|^{-1/4} growth of the amplitude factor in SI S2. For a genuine turning point of a coupled system, the inner-layer connection generically introduces a phase shift — π/4 in the scalar Airy case — and the paper does not compute it; it says so in the Discussion. The consequence splits: for Criterion 2, where ε is taken sufficiently small, an O(1) phase error is absorbed by the O(1/ε) quantization integral, so the existence of some unstable mode on T0 should survive. But the specific eigenvalues and supports in (23b)-(23c), (27), and Figures 5-6 are not yet established as asymptotic eigenvalues; they are formal predictions that may carry a constant phase correction. The paper's remark that the results 'will remain valid even with such refinements' is more confident than the evidence warrants, and a referee should push on this. Two smaller defects: the data DOI is a literal placeholder (xx.xxxx/...), and no code or data accompanies the paper, so the numerics are described rather than checkable.\n\nBottom line. This is a substantial, honest, useful paper for anyone doing heterogeneous pattern formation or developmental modeling. It deserves peer review. I'd send it out and ask for either the boundary-layer matching or an explicit demotion of Criterion 3 to a conjecture with the phase uncertainty stated, plus a working data link. If the authors fix those, this becomes the standard citation for local Turing conditions.","headline":"A genuinely useful derivation of local Turing conditions for heterogeneous domains, with an acknowledged but unresolved turning-point phase problem that a referee should push on.","tokens_in":27350,"tokens_out":7140,"would_cite":true,"duration_ms":68257,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B36","34E20","35K57","92C15"],"pacs":[],"model":"deepseek-v4-flash","headline":"A slowly varying heterogeneous background makes Turing instabilities local: patterns grow exactly where the classical conditions hold pointwise, in a nested hierarchy of mode supports.","keywords":["Turing instability","spatial heterogeneity","WKBJ asymptotics","reaction-diffusion systems","local Turing conditions","mode localization","successive pattern formation","Schnakenberg model"],"falsifier":"Compute the full numerical spectrum of the linearized operator (3) for the Schnakenberg example the paper uses ($\\alpha=1$, $d=1/40$, $\\beta(x)=3/5+(1-\\cos\\pi x)/25$) at a small value such as $\\varepsilon=10^{-3}$, and compare the eigenvalues and supports with the WKBJ predictions from (27); if the predicted localized modes are absent, or do not tend to zero at the singular point $x_*\\approx0.7774$, then Criterion 2 is not a correct asymptotic description.","tokens_in":26359,"feed_emoji":"🌀","tokens_out":11485,"duration_ms":107743,"temperature":0.7,"pith_summary":"This paper tries to establish that a reaction-diffusion system on a slowly varying heterogeneous background can undergo a Turing instability, and that the onset is governed by the classical Turing conditions applied point by point in space. Using a WKBJ approximation—an asymptotic method for rapid oscillations in a slowly changing medium—it derives the analogous instability criterion and, more surprisingly, shows that the unstable modes are not global trigonometric functions. Instead, each growth rate has its own spatial support, with faster-growing modes confined to smaller nested subregions, which explains the amplitude modulations and localized patterning seen in simulations and in heterogeneous biological tissues. If correct, this extends Turing's original mechanism from 'homogeneity to pattern' to the far more common developmental situation of 'one pattern into another,' and it justifies practices already used in experimental papers that check local Turing conditions by eye.","feed_headline":"Patterns appear exactly where classic Turing conditions hold","feed_subtitle":"Faster-growing modes occupy smaller nested regions, so spatial modulation is built into the instability itself.","key_machinery":"The workhorse is the WKBJ ansatz $w=e^{\\lambda t}e^{i\\phi(x)/\\varepsilon}p(x)$, with $\\phi'$ determined by the eigenvalues $\\mu^\\pm_\\lambda(x)$ of $B_\\lambda(x)=D^{-1}(J(x)-\\lambda I)$. Expanding in powers of $\\varepsilon$ and applying a Fredholm solvability condition gives the amplitude prefactor $Q_0(x)\\propto[\\mu^\\pm_\\lambda]^{-1/4}\\exp(-\\int s_*^T p_*'/s_*^T p_*\\,dx)$. The zeros of $[\\mathrm{tr}(B_\\lambda)]^2-4\\det(B_\\lambda)$ are the turning points: there the two eigenvalues coalesce, the left and right eigenvectors become orthogonal, and $Q_0$ develops a $|x-x_*|^{-1/4}$ singularity. Imposing effective Dirichlet conditions at these points selects sine modes and yields the quantization condition $\\int_a^b\\sqrt{\\mu^\\pm_\\lambda(x)}\\,dx=n^\\pm\\pi\\varepsilon$, which determines the discrete growth rates and their nested supports.","core_discovery":"The central claim is that the classical Turing conditions can be applied pointwise in a heterogeneous medium. For a two-component reaction-diffusion system $\\partial_t u = \\varepsilon^2 D u_{xx} + F(u,x)$ with a slowly varying steady state $u_*(x)$ that is stable in the absence of diffusion ($\\mathrm{tr}(J)<0$ and $\\det(J)>0$ everywhere), the paper's Criterion 2 states that if $\\mathrm{tr}(D^{-1}J(x))>0$ and $[\\mathrm{tr}(D^{-1}J(x))]^2-4\\det(D^{-1}J(x))>0$ for all $x$ in the largest such region $T_0$, then for sufficiently small $\\varepsilon$ the steady state is unstable to a non-homogeneous perturbation that grows exponentially on the interior of $T_0$. The unstable modes are constructed by WKBJ theory: they are localized to nested regions $T_\\lambda\\subseteq T_0$, with support strictly shrinking as the growth rate $\\lambda$ increases, and they vanish at the singular points that bound $T_\\lambda$. This is what gives the paper its title: one pattern destabilizes into another, with the new pattern confined to the region where the local conditions hold.","pith_inferences":["A genuine solution of the unsolved boundary-layer problem at the singular points would replace the effective Dirichlet condition by a connection formula; until then, the scaling $\\varepsilon^{8/9}$ for the inner layer gives a concrete numerical target.","If the envelope of a developed pattern tracks the locally fastest-growing mode, then a single converged simulation (or experiment) could be used to read off the spatial variation of growth rates, although the paper only conjectures this envelope relation.","The same WKBJ reduction should carry to higher-dimensional slowly varying domains, where the predicted patterning regions would follow level sets of the local dispersion relation; the authors state this as an anticipation rather than a derived result.","The paper explicitly excludes rapidly varying heterogeneity; mapping the failure threshold as heterogeneity's length scale approaches the diffusion length would connect this smooth-heterogeneity theory to step-function treatments in the literature."],"forward_implications":["Slowly varying spatial heterogeneity does not destroy Turing's mechanism; it localizes it, so the patterning region is predicted in advance as the set $T_0$ where the local conditions hold.","There is no single dispersion relation or global wavenumber for the heterogeneous problem: each unstable mode has a distinct support, so heuristics based on the fastest-growing Fourier mode need to be replaced by local information.","The monotonicity $T_{\\lambda_2}\\subseteq T_{\\lambda_1}$ for $\\lambda_1<\\lambda_2$ means the pattern's spatial extent is encoded in the growth-rate spectrum: slow modes paint wide regions, fast modes sit inside them.","For fixed small $\\varepsilon$, only finitely many discrete growth rates are allowed, selected by the phase integral $\\int_a^b\\sqrt{\\mu^\\pm_\\lambda}\\,dx=n^\\pm\\pi\\varepsilon$, so the theory gives explicit countable mode predictions.","Sequential or hierarchical patterning—an established pattern serving as the heterogeneous background for the next—can be modelled as an instability of a heterogeneous steady state rather than as a fresh symmetry breaking from homogeneity."],"supporting_citations":[{"why":"Introduces the diffusion-driven instability mechanism that the paper generalizes to heterogeneous kinetics.","marker":"[1]"},{"why":"Supplies the standard homogeneous Turing conditions and dispersion-relation framework used as baseline.","marker":"[5]"},{"why":"Articulates the 'from one pattern into another' view of development that motivates the heterogeneous problem.","marker":"[6]"},{"why":"Derives local Turing conditions for a step-function heterogeneity, the previous best result this work extends to smooth heterogeneity.","marker":"[36]"},{"why":"Uses an earlier WKBJ-like ansatz for slowly varying heterogeneity, the direct methodological predecessor.","marker":"[45]"},{"why":"Shows why standard Sturm-Liouville theory does not extend to multi-component systems, motivating the WKBJ route.","marker":"[49]"},{"why":"Provides the general WKBJ asymptotic framework used to build the mode solutions.","marker":"[58]"},{"why":"Supplies measured diffusion timescales used to argue the small-epsilon regime is biologically relevant.","marker":"[61]"}],"fun_headline_variants":["Local Turing conditions pick pattern hotspots","Heterogeneous media localize Turing modes","Nested eigenmodes carve out pattern regions","WKBJ shows patterns emerge where conditions hold","Turing instability shrinks to nested cores"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes the unresolved inner region at each singular point can be replaced by an effective zero boundary condition, so that the piecewise WKBJ solutions are genuine asymptotic eigenfunctions; the paper does not construct that inner solution.","fun_headline_variants_meta":{"raw":{"variants":["Local Turing conditions pick pattern hotspots","Heterogeneous media localize Turing modes","Nested eigenmodes carve out pattern regions","WKBJ shows patterns emerge where conditions hold","Turing instability shrinks to nested cores"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000221,"raw_usage":{"total_tokens":1488,"prompt_tokens":1018,"completion_tokens":470,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":634,"completion_tokens_details":{"reasoning_tokens":404}},"tokens_in":634,"tokens_out":470,"duration_ms":5079,"temperature":1.0,"reasoning_tokens":404,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:22:07.555963+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the full numerical spectrum of the linearized operator (3) for the Schnakenberg example the paper uses ($\\alpha=1$, $d=1/40$, $\\beta(x)=3/5+(1-\\cos\\pi x)/25$) at a small value such as $\\varepsilon=10^{-3}$, and compare the eigenvalues and supports with the WKBJ predictions from (27); if the predicted localized modes are absent, or do not tend to zero at the singular point $x_*\\approx0.7774$, then Criterion 2 is not a correct asymptotic description.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the standard homogeneous Turing conditions and dispersion-relation framework used as baseline."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Articulates the 'from one pattern into another' view of development that motivates the heterogeneous problem."},{"cited_title":"Pattern formation in reaction-diffusion systems with piece-wise kinetic modulation: an example study of heterogeneous kinetics","cited_arxiv_id":"1908.09495","evidence_quote":"Derives local Turing conditions for a step-function heterogeneity, the previous best result this work extends to smooth heterogeneity."},{"cited_title":"Dewel and P","cited_arxiv_id":null,"evidence_quote":"Uses an earlier WKBJ-like ansatz for slowly varying heterogeneity, the direct methodological predecessor."},{"cited_title":"Klika, M","cited_arxiv_id":null,"evidence_quote":"Shows why standard Sturm-Liouville theory does not extend to multi-component systems, motivating the WKBJ route."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the general WKBJ asymptotic framework used to build the mode solutions."},{"cited_title":"Sekine, T","cited_arxiv_id":null,"evidence_quote":"Supplies measured diffusion timescales used to argue the small-epsilon regime is biologically relevant."}],"review_version":1}