{"id":"dacfaec1-76fe-4f82-8faf-d5cdc90037d0","arxiv_id":"2501.16275","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors prove existence of solutions for a model where a transported crowd rho1 and a sandpile-like, capacity-respecting crowd rho2 coexist, and demonstrate the dynamics numerically.","lead":"Two populations share a room: one moves toward an exit while the other rearranges itself like sand to avoid overcrowding. This paper gives a mathematical model and simulations for that process, including an existence proof and a numerical algorithm.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof of Lemma 3.3 invokes an unverified weak-compactness product convergence that is load-bearing for the complementarity condition p(1-ρ)=0.","rationale":"The reader's weakest_assumption flags exactly the product convergence in Lemma 3.3 via compactness references. I agree that this is the most fragile link. The concern is not a disagreement with consensus; it is an internal gap: the proof of Lemma 3.3 asserts a convergence that does not follow from the displayed estimates. The bound (3.28) is in Lip' (dual of W^{1,∞}), which is too weak for the cited Aubin-Lions-type lemmas. The citations to [5,35] may cover parabolic/elliptic PDE settings with explicit W^{1,p} bounds, but here no spatial compactness for ρ̃_n is provided. Thus the complementarity condition p(1-ρ)=0 is not rigorously established. This does not mean the theorem is false; the strict mass condition (3.15) might be sufficient, and there might be a hidden compactness argument. However, as written the proof is conditional on a missing estimate. Since the central existence claim is plausible and the numerical and modeling parts are credible, the verdict remains CONDITIONAL: the compactness gap should be closed or explicitly deferred, and ideally the numerical section's missing mesh/time-step details and code would be provided. There is no basis for REJECT or for upgrading to ACCEPT.","tokens_in":23877,"tokens_out":1957,"duration_ms":17823,"concrete_test":"Re-derive Lemma 3.3 without citing [5,35]: attempt to prove directly that ∫∫ p_n ρ̃_n φ → ∫∫ p ρ φ for all φ∈C_c(Q) using only (3.24)-(3.28) and the structure p_n=K(ρ_{n-1}+τ f_{n-1}-ρ_n). If no such proof exists, exhibit or construct a counterexample family satisfying (3.24)-(3.28), p_n∈Lip1, p_n(1-ρ_n)=0, with p_n ρ̃_n failing to converge to p ρ in weak-* L^∞. Alternatively, check the hypotheses of the cited compactness theorems (e.g., Theorem 1.1 in [35] or the main result in [5]) against the sequences defined in Section 3: verify boundedness of ρ̃_n in L^1(0,T;W^{1,1}(Ω)) or a substitute such as a uniform spatial translation estimate, which is not stated in the paper; if none holds, the application fails.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central existence result Theorem 3.2 relies on Lemma 3.3, whose key step is the assertion that p_n ρ̃_n → p ρ in L^∞(Q) weak-* by 'weak compensated compactness' citing [5,35], after having only established (3.24)-(3.26): ρ_n and ρ̃_n converge weak-*, and p_n converges weakly in L^q(0,T; W^{1,q}(Ω)) for all q∈(1,∞). No uniform space-time compactness estimate for ρ̃_n, no BV or W^{1,1}-type control on the spatial gradient of ρ̃_n, and no L^1 compactness of the pressure-velocity product is shown. The cited theorems are nonlinear compactness results requiring, typically, a uniform bound on ∂_t u_n in a dual space plus a compact embedding or a control on spatial translates; here only ∥∂_t ρ̃_n∥_{L^1(0,T; (Lip',W1))} and weak-* L^∞ bounds are proved. The estimate (3.28) is a bound in the dual of Lip, with a norm equivalent to the dual of W^{1,∞}, which does not provide the needed compactness to justify the product limit. Therefore the derivation of p(1-ρ)=0 after (3.29) is not rigorous as written. This is the most load-bearing gap because without complementarity the weak solution definition is not satisfied and the correction mechanism is not enforced in the limit. The condition (3.15) is acknowledged, but the compactness gap is not acknowledged.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a two-population model of congested pedestrian traffic in a bounded domain: population 1 moves by a prescribed linear transport equation, while population 2 obeys a sandpile-type granular diffusion driven by a 1-Wasserstein gradient flow, with the total density constrained by 0 ≤ ρ1 + ρ2 ≤ 1 through a pressure-type Lagrange multiplier p satisfying p(1−ρ1−ρ2)=0. The main theoretical result, Theorem 3.2, asserts existence of weak solutions to the coupled problem under the strict mass condition (3.15), sup M(t) < |Ω|, using an implicit Euler scheme in the 1-Wasserstein metric and passing to the limit. The paper also proposes a prediction-correction numerical algorithm combining finite-volume transport, minimum-flow projection, and a Chambolle-Pock primal-dual solver, and presents nine numerical experiments illustrating directed, nonlocal, and diffusion-based motion of population 1.","tokens_in":24129,"tokens_out":3948,"duration_ms":41564,"significance":"If the existence theorem is correct, this is a useful extension of single-population W1-sandpile methods to coupled crossing flows: the transport population acts as a nonlocal source, and the second population implements a decongestion mechanism while preserving the hard constraint ρ1+ρ2≤1. The distinction between transport and dispersive congestion dynamics, together with the proposed numerical scheme, is well motivated and the examples cover qualitatively different regimes. The paper is also transparent in stating the strict mass condition (3.15) as a hypothesis rather than hiding it. However, the proof of the central theorem contains an unverified compactness step that is load-bearing for the complementarity condition, so the theoretical claim is not established as written.","major_comments":[{"comment":"The assertion that p_n ρ̃_n → p ρ in L∞(Q) weak-* follows from 'weak compensated compactness' (cited [5,35]) is not supported by the estimates established in the paper. From (3.24)-(3.26) one has only weak-* L∞ convergence of ρ_n and ρ̃_n and weak convergence of p_n in Lq(0,T;W1,q(Ω)); the only time-compactness estimate, (3.28), controls ∂t ρ̃_n in L1(0,T;Lip′), which is the dual of W1,∞ and does not provide the type of strong/compact embedding needed in the Aubin-Lions-type lemmas of [5,35]. The hypotheses of those lemmas are not verified. Since p(1−ρ)=0 is subsequently obtained from this product limit in (3.29) and is part of Definition 3.1, the proof of Theorem 3.2 is incomplete at a load-bearing point.","section":"Lemma 3.3, Eq. (3.29)"},{"comment":"Passing to the limit in the discrete variational inequality (3.30) to obtain (3.10) requires a convergence statement for the piecewise-constant source f_n and for the products ⟨f_n(t), p_n(t)−ξ⟩. The paper only records L∞ bounds on W1(widehat f) and ⟨f⟩ after (3.13), and p_n converges weakly in W1,q without a strong compactness result. In particular, the boundary trace term in the definition of f in (3.7) is not shown to pass to the limit. This is a second unverified limiting step in the existence proof.","section":"Proof of Theorem 3.2, Eq. (3.30)"},{"comment":"The existence result is restricted to the strict inequality (3.15), sup M(t) < |Ω|, yet several numerical examples explicitly operate at saturation, for example Figure 1 where the text states that movement of population 2 is triggered when ρ1+ρ2=1. The paper does not discuss whether the simulations satisfy (3.15), nor does it analyze the saturation time Ta introduced in Remark 5. This leaves the relation between the proven theorem and the computed dynamics unclear and should be addressed explicitly.","section":"Theorem 3.2 and Section 4 numerical experiments"}],"minor_comments":[{"comment":"The keyword list contains 'Crowed motion' instead of 'Crowded motion'; the abstract uses 'affects' where 'affect' is intended.","section":"Keywords and abstract"},{"comment":"There are typos in the introduction, e.g., 'Lagrange multiplayer associtae' should read 'Lagrange multiplier associated'.","section":"Section 1, model description"},{"comment":"The theorem statement and proof refer to equation numbers (2.1) and (2.2) interchangeably, which makes it difficult to tell which boundary-value problem is being addressed; please renumber consistently.","section":"Theorem 2.1 and surrounding text"},{"comment":"The proof of the bound on the mean of pτ uses a chain of inequalities whose direction is not fully explained; adding the variational inequality (3.20) with ψ=0 explicitly would improve readability.","section":"Lemma 3.2"},{"comment":"The parameters α, β, θ are introduced without stating the convergence condition; the condition αβ∥Λh∥2<1 appears only in Algorithm 3 and should be stated once in Algorithm 2.","section":"Algorithm 2"}],"recommendation":"major_revision","confidential_remarks":"The main theorem is not established because of the unverified compensated-compactness step in Lemma 3.3; this is a genuine gap rather than a cosmetic issue, but it is localized and potentially fixable by adding the missing compactness estimates or by proving the product convergence directly. The numerical section is suggestive but does not compensate for the missing proof, and the disconnection between the strict mass hypothesis and the saturated simulations should be clarified. Given the paper's scope, I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague — the thing worth knowing about this paper is that the model is new in a concrete way: population 1 solves a linear transport equation with prescribed velocity, population 2 evolves by the W1-sandpile gradient-flow decongestion, and they are tied by the shared constraint rho1 + rho2 <= 1 with complementarity. Earlier papers by the same group treat one population or trivial rho1; here the nonzero rho1 genuinely changes the variational structure. That is a real step forward for macroscopic crowd and evacuation modeling.\n\nThe transport part is routine and handled correctly. The numerical section shows the expected qualitative behavior: directed, dispersive, and diffusive strategies for rho1 all trigger rho2 rearrangement only where the sum hits capacity. The prediction-correction algorithm is a sensible adaptation of their prior minimum-flow approach, and the discrete proximal operators are explicit. That is honest, usable work.\n\nThe soft spot is in the proof of Theorem 3.2. Lemma 3.3 asserts p_n rho_tilde_n -> p rho in L∞ weak-* by \"weak compensated compactness,\" citing [5,35], but the hypotheses of those nonlinear compactness results are not checked. The estimates in the paper give weak-* convergence of rho_n and tilde_rho_n, weak convergence of p_n in L^q(W^{1,q}), and a time-derivative bound in the dual of Lip—no spatial compactness for tilde_rho_n, no BV-type control, no L^1 compactness of the product. The product convergence is what yields p(1-rho)=0, so the complementarity condition, the heart of the correction mechanism, is not rigorously established as written. This is a genuine gap, not a nitpick, but I do not think it sinks the paper: the scheme is natural, the discrete complementarity holds exactly, and the missing piece looks repairable with Aubin-Lions-type arguments or a different compactness route. The strict slack condition (3.15), M(t) < |Omega|, is acknowledged by the authors and is a real modeling restriction: fully saturated crossing is not covered. Minor issues: the numerics are not reproducible from the manuscript—no code, no explicit mesh or time-step values in the text, no convergence checks—and the primal-dual notation has inconsistencies. The citation pattern leans on their own prior work, but that is building on an existing formalism, not circularity; the main theorem is not a restatement.\n\nWho this is for: people working on crowd motion, W1 gradient flows, or constrained transport. It deserves a serious referee. I would send it out and ask for a repaired Lemma 3.3 and a reproducible numerical section. If the compactness step gets fixed, this is a solid paper. Reading group yes.","headline":"A genuinely novel two-population coupling—transport for one crowd, W1-sandpile decongestion for the other—but the main existence proof leans on an unverified compactness step that needs repair before the theorem is taken as established.","tokens_in":24757,"tokens_out":2462,"would_cite":false,"duration_ms":24901,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["22E46","53C35","57S20"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a two-population crossing flow, where one crowd moves and the other rearranges like a sandpile, has a weak solution as long as total mass stays below capacity.","keywords":["crowd motion","congestion","transport equation","1-Wasserstein distance","W1-gradient flow","minimum-flow problem","primal-dual numerical optimization","pedestrian flow"],"falsifier":"Test the compactness step directly: for a smooth test case, compute the discrete pressure $p^\\tau$ and density $\\tilde{\\rho}^\\tau$ produced by the scheme, and check numerically whether $p^\\tau\\tilde{\\rho}^\\tau \\to p\\rho$ in $L^\\infty(Q)$ weak-* and whether the stated uniform bounds on $\\partial_t\\tilde{\\rho}^\\tau$ in $L^1(0,T;(\\mathrm{Lip}',W_1))$ hold as $\\tau\\to 0$; a failure of either provides a concrete counterexample to the proof's Lemma 3.3.","tokens_in":23620,"feed_emoji":"🚶","tokens_out":13705,"duration_ms":112756,"temperature":0.7,"pith_summary":"The paper proposes a macroscopic model for a space where one population ($\\rho_1$) must cross territory occupied by another ($\\rho_2$), treating the first as a passive tracer carried by a given velocity field and the second as a granular medium that re-arranges so that the combined density never exceeds the maximum value $1$. The central theoretical result, Theorem 3.2, states that the coupled system has a weak solution whenever the initial data satisfy $0 \\le \\rho_1^0 + \\rho_2^0 \\le 1$ and the total mass $M(t) = \\int_\\Omega(\\rho_1+\\rho_2)\\,dx$ remains strictly below $|\\Omega|$ on the time interval. The proof rewrites the dynamics for the aggregate density $\\rho=\\rho_1+\\rho_2$ as a $1$-Wasserstein gradient flow with a forcing term coming from the transport of $\\rho_1$, and passes to the limit in an implicit Euler scheme. If the result is correct, it provides a mathematically grounded way to simulate evacuation and crossing flows under a hard congestion ceiling, and the numerical method shows that different movement strategies of the traversing crowd cause qualitatively different responses in the accommodating crowd.","feed_headline":"One crowd passes; the other shifts like sand to stop overcrowding","feed_subtitle":"New PDE model lets crossing and evacuation flows be simulated with a hard density ceiling.","key_machinery":"The load-bearing object is the subdifferential of the indicator function of $\\mathrm{Lip}_1$, the set of $1$-Lipschitz functions on $\\Omega$, defined through Kantorovich potentials: a distribution $h$ belongs to $\\partial I_{\\mathrm{Lip}_1}(p)$ exactly when $p\\in\\mathrm{Lip}_1$ and $p$ maximizes $\\langle h,\\cdot\\rangle$ over $\\mathrm{Lip}_1$. Through the $1$-Wasserstein distance $W_1$ and its dual formulation as a minimum-flow problem, each time step of the implicit Euler scheme becomes a projection of the predicted density onto the admissible set $\\{0\\le u\\le 1\\}$, and the pressure $p$ emerges as the dual variable. This duality is what carries both halves of the paper: it supplies the estimates that produce the weak solution in Theorem 3.2, and it is the exact optimization problem solved numerically in the correction step of the algorithm.","core_discovery":"The paper's central claim is that congested crossing traffic can be modeled by coupling a linear transport equation for the traversing population, $\\partial_t\\rho_1 + \\nabla\\cdot(\\rho_1 V)=0$, with a sandpile-like evolution for the accommodating population, $\\frac{d\\rho_2}{dt} + \\partial I_{\\mathrm{Lip}_1}(p) \\ni 0$, where the pressure $p$ is the Lagrange multiplier enforcing $0\\le \\rho_1+\\rho_2\\le 1$ and the complementarity relation $p(1-\\rho_1-\\rho_2)=0$. Working with the total density $\\rho = \\rho_1+\\rho_2$, the authors reduce the system to the inclusion $\\partial_t\\rho + \\partial I_{\\mathrm{Lip}_1}(p) \\ni -\\nabla\\cdot(\\rho_1 V)$, in which $p$ is a Kantorovich potential for the $1$-Wasserstein projection of the discrete dynamics. Under the strict sub-saturation condition $\\sup_{t\\in[0,T)} M(t) < |\\Omega|$, they prove the existence of a weak solution by an implicit Euler scheme in the $W_1$ metric, using duality and compensated compactness to obtain the complementarity condition in the limit. The same variational structure yields a prediction-correction numerical algorithm: transport $\\rho_1$ with an upwind finite-volume step, then project $\\rho_2$ onto the admissible set by solving a minimum-flow problem with a primal-dual method. Simulations with eikonal, Gaussian-convolution, and diffusion-based velocity fields illustrate the predicted behaviors, including cases where the traversing population exits the domain and cases where reflective boundaries keep congestion inside.","pith_inferences":["The paper does not treat the limiting case $M(t)=|\\Omega|$; one could test whether allowing full saturation requires a measure-valued pressure or a different compactness argument, extending the theorem to the touching case.","The same $W_1$-projection structure suggests a direct link to discrete optimal transport solvers: comparing the primal-dual projections with entropic-regularization solvers might indicate whether the hard ceiling is best enforced by projection or by a soft penalty in practice.","Since $\\rho_1$ is assumed oblivious to congestion, a likely next test is to let $V$ depend on $\\rho_2$; the current existence theory would not automatically cover such feedback, and simulations might reveal oscillations near the boundary.","The numerical examples for reflecting boundaries suggest persistent congestion can arise purely from boundary conditions; a quantitative study of how long the congested phase lasts as a function of the reflection rule would be a concrete follow-up."],"forward_implications":["If Theorem 3.2 is right, then the model gives a rigorous macroscopic description of crossing flows with a hard density constraint, so evacuation and contraflow scenarios can be simulated with provable existence behind the numerics.","The prediction-correction scheme yields an implementable algorithm: transport the moving crowd, then solve a minimum-flow problem for the accommodating crowd, with the same duality structure in theory and code.","The strict mass condition $M(t)<|\\Omega|$ implies that contact with the maximum density acts as a barrier: solutions are guaranteed only while there remains free space, tying the theory to the geometry of the domain and the outflow through the boundary.","Spatially varying ceilings $\\kappa(x)=1-\\rho_1$ in the correction step mean the same algorithm can handle obstacles, walls, and zones of different carrying capacity without changing the method.","Different choices of the velocity field $V$ (eikonal, Gaussian, diffusion) change only the prediction step, so the framework covers directed, dispersive, and density-responding traversal strategies within one model."],"supporting_citations":[{"why":"supplies the weak compensated compactness results used to pass $p_n\\tilde{\\rho}_n$ to the limit and to derive the complementarity condition $p(1-\\rho)=0$.","marker":"[5, 35]"},{"why":"provides the prediction-correction crowd-motion framework and the minimum-flow projection that the algorithm and the discretization are adapted from.","marker":"[21]"},{"why":"are the $W_1$/sandpile evolution works that justify the Kantorovich-potential formulation of $\\partial I_{\\mathrm{Lip}_1}$ used in the model.","marker":"[1, 20, 29, 30]"},{"why":"is the macroscopic crowd-motion projection model that the correction step adapts.","marker":"[33]"},{"why":"is the minimax theorem used in Lemma 3.1 to interchange the minimum over densities and the maximum over admissible potentials.","marker":"[10]"},{"why":"establish the well-posedness of the linear transport equation with BV coefficients, which supplies the regularity of $\\rho_1$ used throughout.","marker":"[17, 19]"},{"why":"is the variational Wasserstein time discretization that the implicit Euler scheme is patterned on.","marker":"[32]"},{"why":"is the primal-dual algorithm used to solve the discrete minimum-flow problem in the correction step.","marker":"[14]"},{"why":"is the companion primal-dual convergence result used for the same discrete minimum-flow solver.","marker":"[15]"}],"fun_headline_variants":["Sandpile model keeps crossing crowds from clumping","Crossing crowds: transport meets sandpile to avoid pileups","New model couples traffic flow and sandpile for crowded spaces","Crowd crossing: one moves, the other shifts to keep space","When two crowds cross, one flows and one piles like sand"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof relies on the assumption that the approximate pressure and density produced by the scheme have enough compactness for their product to converge correctly in the limit; if that fails, the congestion constraint may not be preserved and the proof breaks.","fun_headline_variants_meta":{"raw":{"variants":["Sandpile model keeps crossing crowds from clumping","Crossing crowds: transport meets sandpile to avoid pileups","New model couples traffic flow and sandpile for crowded spaces","Crowd crossing: one moves, the other shifts to keep space","When two crowds cross, one flows and one piles like sand"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00102,"raw_usage":{"total_tokens":4344,"prompt_tokens":1026,"completion_tokens":3318,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":642,"completion_tokens_details":{"reasoning_tokens":3234}},"tokens_in":642,"tokens_out":3318,"duration_ms":24472,"temperature":1.0,"reasoning_tokens":3234,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T13:35:50.096395+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Test the compactness step directly: for a smooth test case, compute the discrete pressure $p^\\tau$ and density $\\tilde{\\rho}^\\tau$ produced by the scheme, and check numerically whether $p^\\tau\\tilde{\\rho}^\\tau \\to p\\rho$ in $L^\\infty(Q)$ weak-* and whether the stated uniform bounds on $\\partial_t\\tilde{\\rho}^\\tau$ in $L^1(0,T;(\\mathrm{Lip}',W_1))$ hold as $\\tau\\to 0$; a failure of either provides a concrete counterexample to the proof's Lemma 3.3.","supporting_citations":[{"cited_title":"Ennaji, N","cited_arxiv_id":null,"evidence_quote":"provides the prediction-correction crowd-motion framework and the minimum-flow projection that the algorithm and the discretization are adapted from."},{"cited_title":"Maury, A","cited_arxiv_id":null,"evidence_quote":"is the macroscopic crowd-motion projection model that the correction step adapts."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"is the minimax theorem used in Lemma 3.1 to interchange the minimum over densities and the maximum over admissible potentials."},{"cited_title":"Jordan, D","cited_arxiv_id":null,"evidence_quote":"is the variational Wasserstein time discretization that the implicit Euler scheme is patterned on."},{"cited_title":"Chambolle, An algorithm for total variation minimization and applications,J","cited_arxiv_id":null,"evidence_quote":"is the primal-dual algorithm used to solve the discrete minimum-flow problem in the correction step."},{"cited_title":"Chambolle and T","cited_arxiv_id":null,"evidence_quote":"is the companion primal-dual convergence result used for the same discrete minimum-flow solver."}],"review_version":1}