{"id":"b9cd74f7-0e41-4ac5-a8a8-cc8b56ef51c8","arxiv_id":"2411.13444","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For scalar conservation laws with flux switching on the sign of u_x and f>g convex, the authors construct Riemann solutions, exhibit infinite non-uniqueness for smooth data, and prove existence plus conditional uniqueness of minimal-interface solutions for piecewise monotone data.","lead":"This paper studies a traffic-inspired conservation law whose flux switches between two functions depending on the sign of the solution's spatial gradient, in the unstable regime where the accelerating flux lies below the decelerating flux. It shows that the initial-value problem admits infinitely many weak solutions, and constructs a unique global solution for piecewise monotone data when the number of switching interfaces is kept minimal.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The global uniqueness theorem rests on unverified local cases: Cases 3 and 4B of the interface ODE are dismissed as 'entirely similar', and the finite-restarting/minimal-interface argument at shock interactions is not justified.","rationale":"The reader's weakest assumption identifies the same core gap in the local interface-ODE uniqueness, which is indeed the most load-bearing assumption for the conditional-uniqueness claim. I agree that Cases 2A, 3, and especially 4B are not proved as written, and that the appeal to [9] is incomplete. My stress-test adds a second, equally load-bearing point that the reader's rationale touches only implicitly: even if every local case were proved individually, the global restarting mechanism in Section 4, item 6, is not justified. The paper asserts that each interaction reduces the number of interfaces by at least one and that the minimal-interface continuation is unique, but it does not analyze wave interactions involving non-interface shocks, nor does it prove that a collision cannot lead to a Case 4B restart that recreates two interfaces without decreasing the count. If that assertion is false, the construction may have infinitely many restart times and the claim 'defined for all t >= 0' would not follow from the local analysis. I therefore keep the reader's conditional verdict: the construction is plausible and the examples are genuinely new, but the proof as written is incomplete in a way that directly affects the main theorem. The proposed concrete test is to complete Case 4B and to verify the strict-decrease claim on a concrete shock-interface collision; this would settle whether the omitted steps are routine or hide a real obstruction.","tokens_in":17835,"tokens_out":32205,"duration_ms":364303,"concrete_test":"Write out a complete proof of the local generalized Riemann problem for Case 4B: state the two interface ODEs for y and z explicitly, and either verify the hypotheses of a suitable discontinuous-ODE uniqueness theorem for the system or exhibit two distinct minimal-interface solutions with the same initial interface set. In addition, test the Section 4, item 6, strict-decrease claim by computing the interface count before and after a representative shock-interface collision, e.g. a theta=0 g-shock hitting a theta=1 -> theta=0 interface for f(u)=u^2/2+1, g(u)=u^2/2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 3.1 claims existence for all t >= 0 and uniqueness under the minimal-interface rule. The proof rests on two pillars that are not fully established. First, the local generalized Riemann problem is solved in detail only for Case 2B. Case 2A invokes the discontinuous-ODE theorem of [9] after asserting bounded directional variation of H on the cone Gamma near (4.11), but the verification is only sketched. Cases 3 and 4B are dismissed with 'entirely similar techniques' (Section 4); this is especially problematic for Case 4B, which requires uniqueness for a system of two interface ODEs for y and z enclosing the centered f-rarefaction u_nabla, and no vector-valued version of [9] is stated. Second, Section 4, item 6, asserts without proof that at every interaction the number of interfaces decreases by at least one, so that only finitely many restarts occur. This is not automatic: a non-interface g-shock can hit an interface without reducing the interface count, and the local data at a collision can fall into Case 4B, whose minimal continuation creates two interfaces. If either pillar fails, the claimed global existence and minimal-interface uniqueness of Theorem 3.1 are not established.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the scalar conservation law u_t + (θ(u_x)f(u) + (1-θ(u_x))g(u))_x = 0 in the 'unstable' case f > g, with f and g strictly convex. It constructs Lax-admissible solutions to the Riemann problem in four separate cases, gives an explicit non-uniqueness example for zero initial data (Example 2.1), introduces a class of piecewise monotone solutions with interfaces, and states Theorem 3.1 claiming global existence for all t >= 0 and uniqueness under an additional minimal-interface rule. The proof is based on local generalized Riemann problems whose interface curves are determined by discontinuous ODEs, invoking uniqueness results from [9].","tokens_in":18051,"tokens_out":5047,"duration_ms":56097,"significance":"If completed, the paper would show that even in a scalar one-dimensional conservation law, Lax admissibility alone does not select a unique solution, and that a natural 'minimal interfaces' criterion can restore uniqueness in a piecewise monotone class. The explicit family of triangular spike solutions in Example 2.1 is simple, fully verified, and convincing; the local Case 2B construction through a contractive Picard operator is a substantial piece of analysis; and the paper is honest about the conditional nature of the uniqueness statement, especially in Remark 2.3. However, the global theorem is only as strong as the omitted local cases and the interaction argument, which are not proved in the manuscript as written.","major_comments":[{"comment":"The uniqueness of the pair of interface curves (y,z) is asserted by saying that 'the existence and uniqueness of the solution is proved by the same arguments as in Case 2 and Case 3', but Case 2 provides only a scalar contraction argument for one curve, and the quoted uniqueness theorem [9] is for a scalar discontinuous ODE. For the system (4.34)-(4.37), the two curves enclose a centered f-rarefaction and interact with both auxiliary solutions u♭ and u♯, so a genuine vector-valued discontinuous-ODE theorem or a two-curve contraction proof is required; none is stated or proved.","section":"Section 4, Case 4B"},{"comment":"The interface ODE (4.6), (4.33) is dismissed with 'the proof ... is achieved by the same arguments as in Case 2. We thus omit the details.' Here the flux values are g on the left and f on the right, the monotonicity of the auxiliary solutions is different from Case 2, and the tangency construction in Lemma 2.2 is not symmetric with respect to exchanging u− and u+. A detailed verification of bounded directional variation, or another complete uniqueness argument, is needed for this case; omitting it leaves the global construction incomplete.","section":"Section 4, Case 3"},{"comment":"After (4.11), the claim that H has bounded directional variation on the cone Γ and hence the theorem of [9] applies is justified only by the sentence 'This is obvious ...'. The reader needs an explicit estimate of the total variation of t ↦ u♭(t,x(t)) and t ↦ u♯(t,x(t)) for arbitrary Lipschitz curves with |ẋ(t)-λ| ≤ 2ε, including a justification that the curves do not cross the rarefaction fan boundary in a way that produces multiple oscillations. Boundedness of the BV norms of u♭ and u♯ alone does not automatically control composition with an arbitrary Lipschitz curve of positive speed.","section":"Section 4, Case 2A"},{"comment":"The statement that at the first interaction time 'the number of interfaces decreases at least by one' is asserted without proof. This is not automatic: a non-interface g-shock can meet an interface, and local Riemann data with θ− = θ+ = 0 and u− < u+ are handled by Case 4B, whose minimal solution creates two new interfaces. The proof must analyze all collision patterns and show that the minimal-interface rule still yields a strictly smaller interface count after the interaction, or else the finite-restarting argument fails.","section":"Section 4, item 6"},{"comment":"The uniqueness half of Theorem 3.1 is conditional on a 'minimum number of interfaces at each point of shock interaction' rule that is stated only informally. Since Remark 2.3 exhibits two admissible Riemann solutions with different interface counts, and Example 2.1 shows non-uniqueness for smooth data, the theorem's uniqueness claim is not a statement about the equation alone. The manuscript should formalize the selection rule and prove that it is well-defined, for instance by showing that ties cannot occur or by specifying how ties are broken.","section":"Theorem 3.1"}],"minor_comments":[{"comment":"The condition is written as u− > u∗, while the corresponding Riemann case in Section 2, Case 2A, uses u∗ ≤ u−; the boundary case u− = u∗ should be treated explicitly.","section":"Section 4, Case 2A"},{"comment":"The paragraph beginning '5.' appears to be a numbering artifact; it should be integrated into the preceding proof.","section":"Section 4, Case 2B proof"},{"comment":"The distributional formulation splits the integral over {θ = 1} and {θ = 0}; since θ is only defined a.e., the manuscript should specify the representative of θ used to define these sets.","section":"Definition 3.1(iii)"},{"comment":"In (2.4), θ = 1 is assigned for x − x0 < √2 t, which includes the region x < x0 where u is constant; this is consistent with Definition 1.1 but deserves a remark because θ is not determined by the gradient there.","section":"Example 2.1"}],"recommendation":"major_revision","confidential_remarks":"To the editor: the manuscript is part of a sequence with the companion paper [2], and the main novelty is the unstable case. The refereeing bottleneck is Section 4: the global theorem is not proven as written, although the missing pieces are plausibly fillable and the core ideas are sound. I would recommend revision rather than rejection. I have no concerns about attribution or fit with the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. First, this is a genuine advance on the unstable f>g case: Example 2.1 is explicit and correct—even zero initial data gives infinitely many Lax-admissible piecewise smooth solutions, as triangular spikes anchored at arbitrary x0. I checked the Rankine-Hugoniot speed; it works. The systematic Riemann problem in Section 2 is well organized, and the paper is honest that the minimal-interface rule is a selection choice, not a physical derivation: Remark 2.3 exhibits a second admissible solution the rule excludes, and the vanishing-viscosity route is unavailable because (1.3) is backward parabolic here. The citation pattern is normal; the companion paper [2] handles the stable case and the discontinuous-flux/hysteresis literature is cited appropriately.\n\nThe soft spots are where the reader and the stress test put them. The proof of Theorem 3.1 is detailed only for local Case 2B, and that part is real work: the Picard contraction and Lemma 4.1 are solid. Cases 2A, 3, 4A, and 4B are dismissed with 'entirely similar.' For 4B that is a genuine burden, since it requires uniqueness for a system of two coupled interface ODEs enclosing an f-rarefaction, and no vector-valued version of the discontinuous-ODE theorem [9] is stated. Case 2A's bounded-directional-variation verification is only a paragraph, though it looks right: u♭ is constant on the relevant cone and u♯ is monotone along admissible curves. The larger gap is the global restart argument in Section 4, item 6: the claim that each interaction reduces the interface count by at least one, giving finitely many restarts, is asserted without proof. It is not automatic—a non-interface g-shock can hit an interface without reducing the count, and collision data can fall into Case 4B, whose minimal continuation creates two interfaces. As written, that undercuts the global existence and uniqueness statement of Theorem 3.1.\n\nNone of this makes the paper unserious. The gaps look fixable, and the phenomena are new: infinite non-uniqueness, spike formation, minimal-interface selection. The audience is researchers in scalar conservation laws with discontinuous flux and hysteresis traffic models; I would bring it to a reading group and cite the non-uniqueness example. It deserves a serious referee: send it to review, and let the referee push for a complete treatment of Case 4B and the restart count, or a weakened claim that matches the proof.","headline":"Genuinely new treatment of the unstable f>g case with a correct infinite-non-uniqueness example, but Theorem 3.1's proof has real gaps: the 'entirely similar' local cases (especially 4B) and the finite-restart claim are not established as written.","tokens_in":18617,"tokens_out":13315,"would_cite":true,"duration_ms":126263,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35L65","35L67","34A36"],"pacs":[],"model":"deepseek-v4-flash","headline":"Infinitely many admissible solutions arise even from zero data in the unstable f>g case.","keywords":["conservation laws","discontinuous flux","gradient-dependent flux","Lax admissibility","Riemann problem","non-uniqueness","discontinuous ordinary differential equations","traffic flow"],"falsifier":"Compute or simulate the interface ODE (4.6) for the fluxes f(u)=$u^{2}$/2+1 and g(u)=$u^{2}$/2 and look for two different admissible interface curves y(t) through the origin within the cone Γ of (4.11); if two exist, uniqueness in Case 2A fails. More directly, check numerically whether H in (4.5) has bounded directional variation on that cone for a profile where u♯ contains a compression wave; a curve along which H has infinite variation would break the cited uniqueness theorem.","tokens_in":17587,"feed_emoji":"🚦","tokens_out":5715,"duration_ms":55841,"temperature":0.7,"pith_summary":"The paper treats a scalar conservation law whose flux is one of two strictly convex functions, f(u) or g(u), selected by the sign of the spatial derivative u_x: f applies where the solution increases, g where it decreases. It studies the unstable regime f(u)>g(u) for all u, where the associated viscous equation is backward parabolic and the Cauchy problem is ill-posed. The first result is a strong non-uniqueness statement: for the fluxes f=$u^{2}$/2+1 and g=$u^{2}$/2, even identically zero initial data admits infinitely many piecewise smooth, Lax-admissible weak solutions, each a triangular spike anchored at an arbitrary point. The main theorem then shows that for piecewise monotone initial data with a prescribed finite set of interfaces, an admissible solution exists globally in time, and it is unique under the additional rule that at every shock interaction the number of interfaces is kept as small as possible. The motivation is a hysteresis model of traffic flow in which drivers use different flux functions while accelerating and decelerating; the unstable case produces persistent stop-and-go-like spikes.","feed_headline":"Even zero data yields infinitely many solutions","feed_subtitle":"A gradient-switching conservation law with f>g is ill-posed; a minimum-interface rule selects a unique global solution.","key_machinery":"The load-bearing object is the interface curve x=y(t) separating the region where theta=1 (flux f) from the region where theta=0 (flux g), together with the Rankine-Hugoniot speed function H(t,x)=(f(u♭(t,x−))−g(u♯(t,x+)))/(u♭−u♯) that must equal dot y(t). Existence and uniqueness of the interface is reduced to a discontinuous ordinary differential equation dot y=H(t,y), solved by gluing two auxiliary solutions u♭ and u♯ of the single-flux conservation laws (2.7) and (2.18). Uniqueness of the interface uses the theory of discontinuous ODEs: in Case 2A the coefficient H is shown to have bounded directional variation on a cone Γ, so Bressan's uniqueness theorem applies, while in Case 2B a Picard contraction on a wedge Wε2 is proved with the aid of a lemma comparing values of a decreasing function along two curves. The global solution is then assembled by restarting the construction at each time when two interfaces meet, which reduces the interface count and can happen only finitely often.","core_discovery":"On the paper's own terms, the discovery is that the unstable gradient-dependent flux law (1.1)-(1.2) has a definite solution theory if the right selection criterion is imposed. For any Riemann data, an admissible solution is constructed case by case from shocks and centered rarefactions, with the switching set theta determined by the sign of u_x. Because any point of a constant profile can be read as a local maximum, zero initial data yields a continuum of admissible triangular spike solutions; new spikes can be nucleated at arbitrary times at any point where a decreasing solution is smooth. Restricting to piecewise monotone initial data with a prescribed interface set, the paper constructs a global piecewise monotone admissible solution for all t>=0, with interfaces governed by the discontinuous Rankine-Hugoniot ODE (4.6). The uniqueness statement is conditional: among all admissible solutions, exactly one has the minimal possible number of interfaces at every time, and this is the solution constructed.","pith_inferences":["If f and g cross, the model should switch from well-posed (stable, fg) as the density u crosses the intersection; the minimal-interface rule gives a plausible selection principle in the unstable regime but is not derived from any physical or entropic limit.","The spike-nucleation mechanism suggests that any numerical or viscous regularization that keeps the transition layer narrow will effectively choose one of the infinitely many admissible solutions; measuring which spike locations are selected under a given regularization would test the modeling value of the minimal-interface criterion.","The same interface-ODE machinery could be applied to systems with two convex fluxes selected by the gradient of a second variable, e.g. two-phase or hysteretic models, whenever the unstable sign condition holds."],"forward_implications":["The Riemann problem for (1.1)-(1.2) is solvable in all four combinations of θ−, θ+; solvability is explicit, with shock speeds given by tangent-line constructions on the graph of f.","Any smooth decreasing region of a solution is a nucleation site: two new spikes can appear at any time and then persist, so the set of admissible solutions is at least one-dimensional even from simple data.","For piecewise monotone initial data, a global admissible solution exists for all t≥0 and its total variation does not increase in time, because characteristics impinge on every interface from both sides.","Uniqueness is restored, within the class considered, by the minimal-interface rule: at each shock interaction the solution with the fewest interfaces is the one the construction selects."],"supporting_citations":[{"why":"Supplies the uniqueness theorem for discontinuous ODEs used to single out the interface curve in Case 2A.","marker":"[9]"},{"why":"Companion paper defining the stable case f