On approximation of local conservation laws by nonlocal conservation laws (Q2633831): Difference between revisions

This manuscript addresses scalar nonlocal conservation laws in one spatial dimension of the form \[ \partial_t q + \partial_x \left( \lambda\left( W_\eta[q] \right) q \right) = 0 \] with initial condition \(q(0,x)=q_0(x)\) for \(x \in \mathbb{R}\). Such equations occur, for instance, in modeling traffic flow, chemical ripening processes, supply chains, crowd dynamics, opinion formation and pedestrian flow. The difference from the nonlocal conservation law to the local conservation law \[ \partial_t q + \partial_x \left( \lambda\left( q \right) q \right) = 0 \] is that the dynamics (velocity) at a time-space point \((t,x)\) does not only depend on the solution value \(q(t,x)\) but rather on the solution at time \(t\) integrated over a neighborhood around \(x \in \mathbb{R}\). This is realized by the nonlocal term \(W_\eta\) which yields a nonlocal impact \[ W_\eta[q](t,x) = \frac{1}{\eta} \int_x^{x+\eta} \gamma\left( \frac{y-x}{\eta} \right) q(t,x) \, \mathrm{d}y. \] Hence, nonlocal conservation laws are often considered as (regularized) approximations to the corresponding local conservation law. Unfortunately, it has already been shown [\textit{M. Colombo} et al., Arch. Ration. Mech. Anal. 233, No. 3, 1131--1167 (2019; Zbl 1415.35188)] that in general one does not have convergence of the unique weak solutions \(q_\eta\) of the nonlocal conservation law to the unique entropy solution \(q\) of local conservation law when the nonlocal reach \(\eta >0\) converges to zero. However, in the present manuscript, the authors prove that for monotone increasing (decreasing) initial data \(q_0\) and concave (convex) velocity function \(\lambda\) convergence in \(L^1_{\mathrm{loc}}\), in fact, holds true. This result can be found in Theorem 5.5 (main convergence result) and justifies the nonlocal models as (regularized) approximations to the corresponding local conservation law in this case. The theoretical findings are illustrated by numerical tests.