Systems of Nonlocal Conservation Laws with Memory and Their Zero Retention Limit
Aekta Aggarwal, Ganesh Vaidya · 2026
Read Systems of Nonlocal Conservation Laws with Memory and Their Zero Retention Limit by Aekta Aggarwal, Ganesh Vaidya (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-26877.txt.
Introduction
Aggarwal and Vaidya study entropy solutions for nonlocal conservation laws whose flux depends on nearby values in space and on past states in time, giving the system memory. They build a convergent finite-volume approximation, prove existence and uniqueness of the entropy solution, and track the limit as the time-memory window shrinks. That passage shows a memory-to-memoryless effect, converging to the memoryless nonlocal law, with rate estimates derived. The approximations stay asymptotically compatible with that limit, in a general setting without convexity restrictions on the kernels.
Catalog record
| CREATOR | Aekta Aggarwal, Ganesh Vaidya |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 89494 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.26877
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Systems of Nonlocal Conservation Laws with Memory and Their Zero Retention Limit Aekta Aggarwal, Ganesh Vaidya Abstract We study the entropy solution for a class of systems of nonlocal conservation laws in which the convective flux is convoluted with a kernel in both spatial and temporal variables. This formulation models the flux dependence on the solution within its spatial neighborhood (nonlocal in space) as well as on prior states in time (nonlocal in time), thereby incorporating memory effects. In addition, employing a convergent finite volume approximation, the existence of the entropy solution is discussed. The uniqueness of such entropy solutions is also established. In addition, we analyze the asymptotic behavior of the solutions as the support of the temporal convolution kernel shrinks, demonstrating the “memory- to-memoryless” effect and convergence to the entropy solution of the corresponding nonlocal conservation law without memory (i.e., nonlocal only in space). Convergence rate estimates are derived. In addition, the proposed numerical approximations are shown to be asymptotically compatible with this passage to the memoryless limit by deriving the corresponding asymptotic convergence rate estimates. The analysis is carried out in a very general set- ting, without imposing any geometric restrictions such as the convexity of the spatial and temporal convolution kernels, unlike the existing literature on the asymptotic analysis of nonlocal-in-space only conservation laws. To the best of our knowledge, this provides the first convergence and asymptotic analysis for finite volume schemes applied to nonlocal conservation laws with memory. Numerical experiments are included to illustrate the theory. MSC Classification]35L65,65M25, 35D30, 65M12, 65M15 pacs [ keywords nonlocal conservation laws, traffic flow,convergence rate, hyperbolic systems, adapted entropy 1 Introduction Many real-world phenomena exhibit dynamics in which the state at a given time depends not only on the present configuration but also on the past history of the system. In such settings, the notion of “memory” captures the influence of previous states on the current dynamics. Such hereditary structures appear in industrial applications such as damping phenomena in elastic media 22 ; 23 , viscoelasticity 32 , models involving fractional-in-time behavior 54 , gas transport models in coal seams involving adsorption and diffusion in meso and micropores 24 ; 55 , and transport models in subsurface and hyporheic zones 43 ; 41 ; 44 . A prototypical scalar conservation law modeling such events, reads as follows: \partial_{t}u+\partial_{x}\Big(\,f(u,\int_{0}^{t}u(\tau,x)\Gamma(t-\tau))\,d\tau\Big)=0,\quad(t,x)\in Q_{T}:=(0,T)\times\mathbb{R}, (1.1) where \Gamma is a temporal memory kernel. Despite their applications in modeling a wide range of phenomena, the well-posedness theory for such models remains largely incomplete. One key difficulty is that the temporal convolution term often lacks spatial smoothness, rendering the flux discontinuous as a function of the spatial variable. The presence of temporal convolution destroys the semigroup property in the usual sense and gives rise to a new mathematical object, unlike those arising in local or purely spatially nonlocal conservation laws. Consequently, classical methods for standard conservation laws do not directly apply. Some studies 23 ; 52 ; 36 ; 53 ; 51 ; 32 ; 34 ; 18 ; 33 ; 22 deal with some specific cases exploiting structural properties of particular PDEs to study the wellposedness, under some specialized assumptions on the flux. However, the case of “nonlocal space only” conservation laws, \displaystyle\partial_{t}u+\partial_{x}\Big(f(u,\int_{\mathbb{R}}u(t,\xi)\mu(x-\xi))\,d\xi\Big) \displaystyle=0,\quad(t,x)\in Q_{T}, (1.2) with \mu being a spatial horizon kernel, has gained significant interest in the recent decade, due to two reasons: their widespread use in modeling of applications where the flux at a given point may depend not only on the local state but also on averaged quantities over a finite interaction horizon, for example, crowds 29 ; 1 , traffic 12 ; 17 ; 38 ; 4 ; 5 , sedimentation models 11 , laser technology 30 , granular material dynamics 8 , conveyor belt dynamics 42 , opinion formation 7 , sedimentation models 11 and granular material dynamics 8 etc, and their analytical/mathematical vicinity to local nonlinear conservation laws. Its well-posedness has been well studied in literature, a non-exhaustive list being 19 ; 39 ; 10 ; 6 ; 37 ; 50 ; 20 ; 48 ; 7 ; 8 ; 17 ; 9 ; 2 ; 12 ; 38 ; 29 ; 4 ; 5 ; 3 ; 1 ; 11 ; 42 ; 21 ; 30 etc. In contrast, the present article combines these two frameworks by considering convolution kernels that depend on both spatial and temporal variables. In particular, we study the initial-value problem (IVP) for the coupled system of N nonlocal hyperbolic conservation laws of the above type, where for every k\in\mathcal{N}:=\{1,\ldots,N\} , the k^{\rm th} equation is given by \displaystyle\partial_{t}U^{k}+\partial_{x}\Big(f^{k}(U^{k})\nu^{k}((\boldsymbol{\Theta}\circledast\boldsymbol{U})^{k})\Big) \displaystyle=0,\quad(t,x)\in Q_{T}, (1.3) \displaystyle U^{k}(0,x) \displaystyle=U_{0}^{k}(x),\quad x\in\mathbb{R}. (1.4) Here, T is the final time and the unknown is \boldsymbol{U}=(U^{k})_{k\in\mathcal{N}}:[0,\infty)\times\mathbb{R}\to\mathbb{R}^{N}. Further, for every j,k\in\mathcal{N} , {U}^{k}_{0}\in(L^{1}\cap\bv)(\mathbb{R};[0,1]),\Theta^{j,k}(t,x):=\mu^{j,k}(x)\Gamma^{j,k}(t) , with \boldsymbol{\mu} and \boldsymbol{\Gamma} being smooth N\times N matrices, and (\boldsymbol{\Theta}\circledast\boldsymbol{U})^{k}:=(\Theta^{i,k}*U^{i})_{i\in\mathcal{N}}, where for every (t,x)\in\overline{Q}_{T} , \displaystyle\begin{split}(\Theta^{i,k}*U^{i})(t,x)&:=\displaystyle\int_{0}^{t}\int_{\mathbb{R}}U^{i}(\tau,\xi)\mu^{j,k}(x-\xi)\Gamma^{i,k}(t-\tau)\mathinner{\mathrm{d}{\tau}}\mathinner{\mathrm{d}{\xi}},i\in\mathcal{N}.\end{split} (1.5) Additionally, 1. f^{k}\in\lip(\mathbb{R}) with f^{k}(0)=0=f^{k}(1) . 2. \nu^{k}\in(C^{2}\cap\bv\cap\,W^{2,\infty})(\mathbb{R}^{N},\mathbb{R}) ; 3. \Theta^{j,k}(t,x):=\mu^{j,k}(x)\Gamma^{j,k}(t) , with \boldsymbol{\mu} and \boldsymbol{\Gamma} being smooth N\times N matrices. The space kernel {\mu}^{j,k}\in C^{2}(\mathbb{R})\cap W^{2,\infty}(\mathbb{R}) . The time kernel \Gamma^{j,k}\in C^{2}([0,\infty);\mathbb{R}^{+})\cap W^{2,\infty}([0,\infty);\mathbb{R}^{+}). Modeling-wise, it is to be noted that the system ( 1.3 )–( 1.4 ) is coupled through the nonlocal operator \boldsymbol{\Theta}\circledast\boldsymbol{U} , which introduces both spatial averaging and temporal memory effects into the convective flux of ( 1.3 ). More precisely, for every k\in\mathcal{N} , the flux at a spacetime point (t_{0},x_{0}) depends not only on the instantaneous density U^{k}(t_{0},x_{0}) , but also on 1. a weighted average of the states U^{j}(t_{0},\cdot) in a neighborhood of x_{0} , determined by the spatial kernel \mu^{j,k},j,k,\in\mathcal{N} ; 2. a cumulative influence of past states U^{j}(\tau,x_{0}) , \tau<t_{0} , modulated by the time kernel \Gamma^{j,k},j,k,\in\mathcal{N} . Consequently, the characteristic speed of the k^{\mathrm{th}} equation at (t_{0},x_{0}) reflects a history-dependent interaction between all components of the system, rather than being determined solely by local instantaneous data. In practical applications, \Gamma^{j,k} is typically taken as a decreasing function of time, depicting fading memory and assigning smaller weights to older information. However, the results of this article remain valid for a broader class of \Gamma^{j,k} satisfying 3 . We will show that although a general well-posedness theory for time-nonlocal conservation laws remains largely undeveloped, the present framework enables us to systematically adapt techniques from the theory of purely spatial nonlocal conservation laws to establish well-posedness for a substantially broader class of flux functions incorporating both spatial and temporal nonlocal effects, invoking the regularity of the convective flux with respect to the space variable x . Using relaxation schemes, ( 1.3 )-( 1.4 ) has been recently studied for N=1 in 36 for a specific choice of \Theta and linear f . That said, the question of existence as well as uniqueness of the entropy solutions for the system ( 1.3 )-( 1.4 ) with a general \boldsymbol{\Theta} having temporal dependence and(or) for N>1 , remains unexplored and unsettled as of now, which is dealt with in the first part of this article. Several further interesting questions regarding the entropy solution of ( 1.3 )-( 1.4 ) are: 1. Does it converge to the entropy solution of its “nonlocal-space” counterpart \displaystyle\partial_{t}U^{k}+\partial_{x}\Big(f^{k}(U^{k})\nu^{k}((\boldsymbol{\mu}\circledast\boldsymbol{U})^{k})\Big) \displaystyle=0,\quad(t,x)\in Q_{T}, (1.6) \displaystyle(\mu^{j,k}*U^{j})(t,x) \displaystyle:=\displaystyle\int_{\mathbb{R}}U^{j}(t,\xi)\mu^{j,k}(x-\xi)\mathinner{\mathrm{d}{\xi}}, (1.7) as the radius of the temporal convolution kernel