{"id":"d008510c-4630-40b7-a29f-75932da63042","arxiv_id":"1908.09495","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Pattern formation in a reaction-diffusion system with a step in a kinetic coefficient is predicted by the classical Turing instability conditions applied independently on each side of the step.","lead":"This paper studies how a sudden change in reaction chemistry affects pattern formation in Turing reaction-diffusion systems. It finds that pattern onset can be read off by applying the standard Turing conditions separately to each side of the step.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central 'local Turing conditions' claim (Eq. 20) rests on an unproven spectral-truncation limit, explicitly acknowledged in Appendix A.3.1; an exact Evans-function eigenvalue computation for the linearized operator would settle whether the finite-matrix spectrum actually controls the…","rationale":"The reader identified the same weakest assumption: the unproven limit of the truncated-matrix spectrum to the infinite-dimensional linearized operator. This is genuinely load-bearing for Eq. (20), because the paper presents Eq. (20) as an 'if and only if' statement about pattern emergence, and the only analytical support for the linear affine case is the spectral calculation in Appendix A.3. The authors are honest about the gap, and the numerical verification via evolution problems is systematic and consistent, so I do not see an internal inconsistency or a demonstrated counterexample. However, the central claim is stated as a property of the system ('the presence of stability or instability ... is a local property'), and a proof of that property would require either a rigorous spectral approximation theorem or an independent exact eigenvalue computation. The Evans-function test would settle whether the truncation gap actually changes any stability prediction in the tested regions. If the test passes, the paper remains a well-supported conditional result; if it fails, the central classification would be wrong. Since we do not yet have that test, the appropriate verdict is unchanged: CONDITIONAL, with high confidence in the numerical phenomenology but not in the proof.","tokens_in":18826,"tokens_out":8364,"duration_ms":98509,"concrete_test":"Compute the spectrum of the 1D linearized operator L = diag(d1,d2)d_xx + J(x) with piecewise constant J(x) and Neumann boundary conditions directly via the Evans function: for lambda in a rectangle containing the suspected rightmost eigenvalues, solve the two second-order ODE systems on (0,xi) and (xi,L) with Neumann conditions at 0 and L and continuity of u, v, u', v' at xi; eigenvalues are zeros of the resulting 4x4 matching determinant. Evaluate this determinant for representative parameter points in each of the 16 regions of Table 1, using the parameters of Fig. 3, and compare the sign of the rightmost eigenvalue with the truncated-matrix result and with Eq. (20). If the Evans-function sign disagrees with M_N in any region, the truncation limit is unreliable and the central classification fails. If it agrees in all sampled regions, the numerical content of Eq.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the affine case, the derivation of Eq. (20) is partly grounded in the eigenvalues of truncated matrices M_N, Eq. (34). The decisive step is the claim that the spectrum of the infinite system is obtained from sigma(M_N) plus a set of large negative eigenvalues. This is not proved. Appendix A.3.1 contains the exact admission: 'We will not be able to show that the spectrum of the infinite matrix is a limit of the spectrum of the truncated matrices.' The asymptotic determinant calculation only shows that M_{N+1} has two eigenvalues of order -kappa_N in addition to, at leading order, the eigenvalues of M_N; it does not establish convergence of the remaining eigenvalues to those of the second-order differential operator on (0,L), nor does it exclude spectral pollution. The later statement that the coupling matrix is a 'small perturbation' is informal. Since the stability classification in Table 1 and condition (20) depend on the sign of the rightmost eigenvalue, a failure of this limit could change the classification. The evolution-equation simulations supply independent evidence for the tested parameters, but they do not prove the universal 'local property' claim. This gap is load-bearing because it is the only analytical bridge between the infinite-dimensional linear stability problem and the local conditions (18).","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies a two-species reaction-diffusion system on an interval with Neumann boundary conditions and a step-function modulation h(x) multiplying the linear activator term. For affine kinetics, the authors construct the non-constant steady state and analyse linear stability by expanding perturbations in Neumann eigenfunctions, which leads to an infinite matrix that is truncated numerically. In parallel, a formal boundary-layer analysis for a regularised step suggests that the internal layer is diffusive and does not itself generate patterning, so the instability should be governed by the classical Turing conditions evaluated separately on each side of the step. This yields the central conditions (19) and (20): when the homogeneous kinetics are stable on both sides, a pattern emerges iff the Turing conditions (T3,T4) hold on at least one side, with left-sided, right-sided, or both-sided patterns determined by which side satisfies them. The conditions are tested against eigenvalue computations of the truncated matrix and direct simulations of the affine, Schnakenberg, and Gierer-Meinhardt systems.","tokens_in":19095,"tokens_out":6895,"duration_ms":70447,"significance":"If rigorously established, the proposed local characterisation would be practically valuable: it reduces a non-self-adjoint, non-constant-coefficient stability problem to two classical constant-coefficient Turing checks and yields testable predictions of one-sided versus two-sided patterns, including spatial frequency changes across the step. The paper's strengths are its explicit conditions, the extensive numerical parameter sweeps, and the absence of parameter fitting; the agreement between two independent numerical methods and two nonlinear kinetics is genuinely supportive, and the predicted pattern classifications are falsifiable. However, the analytical foundation is incomplete: the spectral-truncation limit is explicitly unproved, and the boundary-layer argument is heuristic. The central claim is therefore currently a well-supported conjecture rather than a theorem.","major_comments":[{"comment":"The stability classification underlying Eq. (20) depends on the assertion that the spectrum of the infinite matrix (34) is governed by the spectra of the truncated matrices M_N. The text states in Appendix A.3.1, 'We will not be able to show that the spectrum of the infinite matrix is a limit of the spectrum of the truncated matrices.' The subsequent determinant computation only shows that M_{N+1} acquires two eigenvalues of order -kappa_N in addition to, at leading order, the eigenvalues of M_N; it does not rule out spectral pollution or establish convergence of the remaining eigenvalues to those of the differential operator. Since the sign of the rightmost eigenvalue is the criterion used in Table 1, a failure of this limit could change the classification even in the affine case. Please either supply a rigorous functional-analytic argument (for example, relative compactness of the coupling term together with a sectorial-operator convergence theorem) or perform an independent exact eigenvalue computation for the affine linearised operator, such as an Evans-function or shooting method exploiting the piecewise-constant coefficients, and compare the resulting rightmost eigenvalue with Table 1 across the parameter regions.","section":"Appendix A.3.1, Eq. (34); Table 1"},{"comment":"The boundary-layer analysis is a formal asymptotic argument: after rescaling X=(x-xi)/delta and T=t/delta^2, the leading-order inner problem is pure diffusion, and the text concludes that the inner layer 'will not drive patterning'. This conclusion is load-bearing for the extension to nonlinear kinetics and for the classification (19), but the matching of the inner constant solution to the outer problems (16)-(17) is not quantified, and no error estimates are given for finite delta or for the limit delta to 0. To support the claim, please provide a rigorous matched-asymptotics justification or, failing that, numerical evidence with the regularised step h_delta (for example, tanh profiles for several small delta) showing that the predicted pattern side and frequency are unchanged as delta decreases and that no instability is introduced in the inner region.","section":"Section 2.2, Eqs. (14)-(17)"},{"comment":"The abstract and conclusion state that 'the presence of stability or instability in Turing models is a local property for piece-wise constant kinetic parameters' as a general statement. However, the nonlinear verification in Section 3.2 covers two specific kinetics and selected parameter sets, and the text elsewhere repeatedly calls conditions (19) 'hypothesised' and lacking rigorous validation. The universal 'local property' claim therefore exceeds the evidence presented. Please either restrict the conclusion to the affine case and present the nonlinear statement as a conjecture, or add a proof, or provide a substantially broader numerical study with a quantitative measure of agreement, before claiming the local property as a general result.","section":"Section 3.2 and Section 4"}],"minor_comments":[{"comment":"The statement that the largest real part of the truncated-matrix eigenvalues is negative if and only if the supremum norm of the evolution solution decreases is qualified only by 'the possible exception of the very near vicinity of the parameter space boundaries'; please quantify the size of this vicinity and report how many sample points fell inside it.","section":"Section 3.1"},{"comment":"The compressed display of the infinite matrix is difficult to read; please present its block structure explicitly, since the construction of M_N is central to the numerical spectral results.","section":"Appendix A.3.1, Eq. (34)"},{"comment":"The grayscale legend for pattern types appears only in the caption; please also include an explicit legend inside the figure so that the symbols are self-explanatory.","section":"Figure 4"},{"comment":"The manuscript does not justify the choices of terminal time tau=10^3 and the norm threshold 10^7 used to classify patterns; a brief justification or a sensitivity check would strengthen the classification protocol.","section":"Section 3.2"},{"comment":"The statement that 'the discreteness of eigenmodes results in a lower bound on the size of the supporting intervals of each step' is not quantified; consider giving the expected critical-domain-size relation for the affine case.","section":"Section 4"},{"comment":"Minor grammatical issues: 'to distinguish among all the one-sided pattern and both-sided pattern' should read 'to distinguish among one-sided and both-sided patterns'.","section":"Section 2.3"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the journal's scope and the numerical evidence is substantial, but the advertised 'local property' theorem rests on an explicitly unproved spectral-truncation limit and a heuristic boundary-layer argument. The gap is fixable in principle by an exact Evans-function computation for the affine linearised operator, which would either confirm Eq. (20) or reveal its failure. I would also encourage the editors to ask for a reproducibility statement or data/code availability, since the numerical results are a central part of the evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a genuinely useful paper, and I think the main claim is probably right. The step function premultiplying a linear kinetic term is not the same as the additive step in Page et al.: it shifts the base steady state, produces one-sided patterns, and changes the local spatial frequency. The proposed local conditions (18)–(20) are new and clean, and the numerical verification is unusually thorough. Linear evolution simulations, truncated spectral calculations, and nonlinear sweeps with Schnakenberg and Gierer-Meinhardt kinetics all line up with the classification table. That is real evidence, not just a story.\n\nThe soft spot is exactly what the stress-test note identifies. Appendix A.3.1 admits they cannot show the spectrum of the infinite matrix is the limit of spectra of truncated matrices. The determinant calculation only shows that each truncation adds two large negative eigenvalues; it does not prove convergence of the rest of the spectrum, nor does it rule out spectral pollution. The boundary layer argument is leading-order and heuristic, and the nonlinear conditions come from a linearisation whose range of validity is not established. All three gaps are acknowledged in the text, which I respect, but they mean Eq. (20) is a well-evidenced hypothesis, not a theorem. I don't think this is fatal: the evolution simulations cover many parameter combinations and agree with the criteria away from region boundaries. For the affine case, an exact Evans-function computation for the piecewise-constant linearized operator would close the main gap, and the 1D setup makes that look feasible.\n\nMinor points: the notion of \"pattern\" is defined pragmatically via persistence on a large domain, so the conclusions are tied to that definition. The authors also flag the step-size limitation themselves, and the extension to nonlinear kinetics is explicitly conjectural. I do not see a circularity problem: no parameter is fitted to the outcome, and the spectral and evolution methods are independent enough to be mutually supporting.\n\nWho is this for? Anyone working on reaction-diffusion systems with spatial heterogeneity, especially Turing-type classification in heterogeneous domains. It deserves a serious referee and, with the gaps stated honestly, is citable as a strong conjecture with systematic numerical support. I would accept it for peer review and would bring it to a reading group.","headline":"Useful, honest paper whose central claim is a numerically supported conjecture rather than a proof; worth refereeing and citing, but Eq. (20) should not be treated as an established theorem.","tokens_in":19597,"tokens_out":1818,"would_cite":true,"duration_ms":19483,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35K57","35B36","92C15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that a reaction-diffusion system with a piecewise-constant kinetic coefficient patterns according to the classical Turing conditions evaluated separately on each side of the step: pattern emergence is a local property of…","keywords":["reaction-diffusion systems","Turing instability","piecewise-constant kinetics","spatial heterogeneity","linear stability analysis","boundary layer analysis","Schnakenberg kinetics","Gierer-Meinhardt kinetics"],"falsifier":"Take a parameter set in the affine system where equation (20) predicts no pattern, push the truncation size far beyond $N=1000$ and the domain so large that many eigenmodes fit, and check numerically whether any perturbation grows; a growing mode there would disprove the locality criterion, as would a parameter set where equation (20) predicts a pattern but no growth appears at any accessible truncation.","tokens_in":18615,"feed_emoji":"🌀","tokens_out":8623,"duration_ms":78212,"temperature":0.7,"pith_summary":"This paper asks when a reaction-diffusion system with a step-function kinetic coefficient produces a Turing pattern, and on which side of the step the pattern appears. It claims the answer is local: evaluate the classical Turing conditions separately on the two constant-parameter subintervals, and a pattern emerges exactly when the homogeneous steady state is stable in both parts and at least one part satisfies the remaining Turing conditions. For nonlinear kinetics the same local conditions classify the outcome as left-sided, right-sided, or both-sided patterning. The practical upshot is that a heterogeneous domain with sharp kinetic transitions can be assessed by local Turing-space checks, with the jump itself only imposing a matching condition rather than participating in the instability.","feed_headline":"Pattern emergence at a kinetic step is a local property","feed_subtitle":"The classical Turing conditions on each side of the jump decide the pattern on their own, no coupled spectrum needed.","key_machinery":"The central object is the step function $h(x)=0$ on $(0,\\xi)$, $h(x)=s$ on $(\\xi,L)$ multiplying the activator's linear kinetic term, together with the four classical Turing conditions $T_1$--$T_4$ evaluated separately on each subinterval. The machinery has three parts: an explicit stationary solution built by solving the affine system on each subinterval and connecting the solutions at the jump; a spectral analysis in which the spatially dependent Jacobian couples all Fourier modes through an infinite matrix, then a truncated-matrix analysis showing that the two largest eigenvalues are of order $-\\kappa_N$ while the remaining spectrum follows the smaller truncated matrix; and a boundary-layer analysis of a smoothed step showing that the inner layer has trivial, purely diffusive dynamics. This combination converts the heterogeneous stability question into two independent classical Turing calculations.","core_discovery":"The central discovery, stated as equation (20), is that for the affine system a Turing pattern emerges if and only if the two stability conditions $T_1,T_2$ hold for both subintervals $(0,\\xi)$ and $(\\xi,L)$ and, in addition, at least one of the two subintervals satisfies the wavenumber conditions $T_3,T_4$. For nonlinear kinetics the paper extends this to a four-way classification: left-sided, right-sided, both-sided, or no pattern. A boundary-layer analysis of a smooth regularisation of the step shows that the inner region near the jump only homogenises by fast diffusion and does not itself drive patterning, so the outer problems on the two sides determine stability. Numerical sweeps with Schnakenberg and Gierer-Meinhardt kinetics agree with this classification. The paper therefore concludes that stability or instability in Turing models is a local property for piecewise-constant kinetic parameters.","pith_inferences":["Beyond the paper, if the locality principle extends to slowly varying kinetics, spatially resolved Turing-space maps could be built pointwise from local parameter values, making the criterion predictive at tissue scale rather than only for two constant patches.","One could push beyond the paper's finite-step remark by asking how the locality criterion degrades with step size: as $s$ grows, the localised inhomogeneity at the jump eventually swamps the Turing oscillation, so there should be a threshold curve in $(s, L)$ separating genuine patterning from passive inhomogeneity; measuring that threshold numerically would sharpen the practical reach of the clas","In developmental contexts where kinetic parameters are graded rather than stepped, the locality claim suggests that patterns can be seeded in a Turing-active zone and propagate into a non-Turing zone by diffusion, so a local Turing-space check would predict where pattern first appears; testing this with a smoothed gradient would connect the criterion to the gradient studies the paper cites."],"forward_implications":["A kinetic boundary can act as a pattern boundary: parameters can be tuned so that one side of the step patterns while the other does not, producing one-sided patterns.","The spatial frequency of the pattern is set independently on each side of the step, so sharp changes in wavelength across a domain can be read as a signature of a sharp kinetic transition.","The detailed shape of the transition region is not load-bearing: a smoothed step gives the same local classification as the sharp step, up to the small-step-size caveat.","For a parameter point to support patterning, it suffices that the homogeneous kinetics is stable on both sides and at least one side enters the Turing wavenumber region; no coupled spectrum of the heterogeneous problem is needed.","A subdomain smaller than roughly one local pattern period can hide a predicted one-sided pattern, so the criterion must be read together with a minimum subdomain size."],"supporting_citations":[{"why":"Defines the diffusion-driven instability that the paper extends to piecewise-constant kinetics.","marker":"[32]"},{"why":"Supplies the step-function stationary-solution construction via Jordan normal form and connecting conditions that is reused here.","marker":"[23]"},{"why":"Gives the standard statement of the Turing conditions T1–T4 that the paper evaluates on each subinterval.","marker":"[21]"},{"why":"Provides the domain-size and discrete-mode reasoning used to define when subintervals are large enough for a pattern.","marker":"[14]"},{"why":"Supplies the Schnakenberg kinetics used in one set of nonlinear numerical verifications.","marker":"[29]"},{"why":"Supplies the Gierer-Meinhardt kinetics used in the second set of nonlinear numerical verifications.","marker":"[5]"},{"why":"Documents numerically complex pattern formation with spatially varying parameters, the phenomenon the paper's local classification addresses.","marker":"[24]"}],"fun_headline_variants":["Turing patterns at kinetic jumps are set by local stability","No coupled spectrum: local Turing conditions decide pattern at a step","Piecewise constant kinetics: Turing instability is a local property","Local stability analysis suffices for Turing patterns with kinetic jumps","Turing pattern at a kinetic step is governed by side-wise stability"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification rests on the unproven step, stated in Appendix A.3.1, that stability of the infinite linearised system is captured by the finite truncated matrix; the authors explicitly say they cannot prove that the infinite-matrix spectrum is the limit of truncated spectra.","fun_headline_variants_meta":{"raw":{"variants":["Turing patterns at kinetic jumps are set by local stability","No coupled spectrum: local Turing conditions decide pattern at a step","Piecewise constant kinetics: Turing instability is a local property","Local stability analysis suffices for Turing patterns with kinetic jumps","Turing pattern at a kinetic step is governed by side-wise stability"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000666,"raw_usage":{"total_tokens":2988,"prompt_tokens":840,"completion_tokens":2148,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":456,"completion_tokens_details":{"reasoning_tokens":2064}},"tokens_in":456,"tokens_out":2148,"duration_ms":17285,"temperature":1.0,"reasoning_tokens":2064,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:09:39.014768+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a parameter set in the affine system where equation (20) predicts no pattern, push the truncation size far beyond $N=1000$ and the domain so large that many eigenmodes fit, and check numerically whether any perturbation grows; a growing mode there would disprove the locality criterion, as would a parameter set where equation (20) predicts a pattern but no growth appears at any accessible truncation.","supporting_citations":[{"cited_title":"The chemical basis of morphogenesis.Philosophical Transactions of the Royal Society of London B: Biological Sciences, 237(641):37–72, 1952","cited_arxiv_id":null,"evidence_quote":"Defines the diffusion-driven instability that the paper extends to piecewise-constant kinetics."},{"cited_title":"Pattern formation in spatially hetero- geneous turing reaction–diffusion models.Physica D: Nonlinear Phenomena, 181(1):80–101, 2003","cited_arxiv_id":null,"evidence_quote":"Supplies the step-function stationary-solution construction via Jordan normal form and connecting conditions that is reused here."},{"cited_title":"Mathematical Biology","cited_arxiv_id":null,"evidence_quote":"Gives the standard statement of the Turing conditions T1–T4 that the paper evaluates on each subinterval."},{"cited_title":"Domain size driven instability: Self- organization in systems with advection.SIAM Journal on Applied Mathematics, 78(5):2298– 2322, 2018","cited_arxiv_id":null,"evidence_quote":"Provides the domain-size and discrete-mode reasoning used to define when subintervals are large enough for a pattern."},{"cited_title":"Simple chemical reaction systems with limit cycle behaviour","cited_arxiv_id":null,"evidence_quote":"Supplies the Schnakenberg kinetics used in one set of nonlinear numerical verifications."},{"cited_title":"A theory of biological pattern formation","cited_arxiv_id":null,"evidence_quote":"Supplies the Gierer-Meinhardt kinetics used in the second set of nonlinear numerical verifications."},{"cited_title":"Complex pattern formation in reaction–diffusion systems with spatially varying parameters","cited_arxiv_id":null,"evidence_quote":"Documents numerically complex pattern formation with spatially varying parameters, the phenomenon the paper's local classification addresses."}],"review_version":1}