Local Asymptotic Patterns for Viscous Approximations of Conservation Laws
Alberto Bressan, Laura Caravenna, Wen Shen · 2026
Read Local Asymptotic Patterns for Viscous Approximations of Conservation Laws by Alberto Bressan, Laura Caravenna, Wen Shen (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-2605-00189.txt.
Introduction
Alberto Bressan, Laura Caravenna, and Wen Shen study what smoothed versions of conservation laws look like up close, right where the true solution breaks. Conservation laws govern quantities such as traffic density or gas flow, and their solutions can form shocks, sharp jumps in the state. Many approximation schemes exist, from added viscosity to numerical grids, and broad convergence results are known.
The open question is local: near a shock, a shock collision, or the birth of a new shock, what pattern does an approximation settle into? The authors argue that, after a careful rescaling of space and time near such a point, the approximations approach a specific eternal solution defined for all space and time. They give precise results in the vanishing viscosity case, the scheme where a small smoothing term fades away.
Catalog record
| CREATOR | Alberto Bressan, Laura Caravenna, Wen Shen |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 137 KB |
| LICENSE | http://creativecommons.org/licenses/by-sa/4.0/ |
| RIGHTS | CC BY SA |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.00189
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Local Asymptotic Patterns for Viscous Approximations of Conservation Laws Alberto Bressan, Laura Caravenna, Wen Shen Abstract Solutions to hyperbolic conservation laws can be approximated in many different ways: by vanishing viscosity, relaxations, discrete or semi-discrete numerical schemes, approximation with a nonlocal flux, etc … \ldots For some of these methods, general 𝐋 1 {\bf L}^{1} convergence results are available. Aim of this paper is to understand the local behavior of these approximations, in a neighborhood of point where the hyperbolic solution has a singularity. Specifically: a point along a shock, or where two shocks interact, or where a new shock is formed. Given a sequence of ϵ \epsilon -approximate solutions, a general expectation is that, by a suitable local rescaling of coordinates, as ϵ → 0 \epsilon\to 0 a well defined limit is obtained. This corresponds to a specific “eternal solution" (globally defined both in space and in time) to the approximating equation. Precise results this direction are here given, in the case of vanishing viscosity. Key Words: Hyperbolic conservation laws, generic singularity, local asymptotic approximation. 1 Introduction The emergence of singularities in solutions to PDEs has been a topic of extensive study. For various classes of nonlinear hyperbolic equations, solutions can be uniquely extended in time beyond singularity formation. In such cases, it is of interest to identify “structurally stable singularities", which arise from an open set of initial data (in a suitable topology), and are thus more likely to be observed in a physical context. For some particular classes of PDEs, such as scalar conservation laws, variational wave equations, or Camassa-Holm equations, a classification of generic singularities can be found in [ 22 ] , [ 6 , 10 ] and [ 28 ] , respectively. A similar approach is of interest also in connection with multidimensional Hamilton-Jacobi equations. Typically, this classification corresponds to a limit of local rescalings. Namely, if a solution u(t,x) has a singularity at a point (\tau,\xi) , one can consider the rescaled functions u^{\delta}(t,x)~=~u\bigl(\tau+\delta t,\xi+\delta x\bigr), (1.1) or, depending on the type of singularity, u^{\delta}(t,x)~=~\delta^{-\gamma}\Big[u\bigl(\tau+\delta^{\alpha}t,\,\xi+\delta^{\beta}x\bigr)-u(\tau,\xi)\Big], (1.2) for suitable exponents \alpha,\beta,\gamma . In a favorable case, taking the limit as \delta\to 0 , one obtains a self-similar function, depending on a finite number of parameters, which describes the asymptotic behavior of the solution in a neighborhood of the singular point (\tau,\xi) . Purpose of the present paper is to point out that, in many cases, this classification of generic singularities can be carried out on a further level, involving various classes of approximate solutions. As an example, consider a scalar conservation law with smooth, convex flux u_{t}+f(u)_{x}~=~0,\qquad\qquad t\in[0,T]. (1.3) For a generic initial data u(0,x)~=~\bar{u}(x), (1.4) with \bar{u} in the intersection of countably many open dense subsets of {\cal C}^{3}({\mathbb{R}}) , it is well known [ 22 , 31 ] that the unique entropy weak solution is piecewise smooth, with finitely many shock curves. Structurally stable singular points can be of three types (see Fig. 1 ). (I) points along a shock curve, (II) points where two shock curves merge, (III) points where a new shock is formed. Various approximations of ( 1.3 ) have been extensively studied in the literature. In particular: (i) Vanishing viscosity approximations [ 17 , 23 , 25 ] , such as u_{t}+f(u)_{x}~=~\varepsilon\bigl(b(u)u_{x}\bigr)_{x}\,. (1.5) (ii) Relaxation approximations [ 26 , 29 , 33 , 34 ] . For example \left\{\begin{array}[]{cl}v_{t}+v_{x}&=\,\varepsilon^{-1}\bigl(w-g(v)\bigr),\\[5.69054pt] w_{t}&=-\varepsilon^{-1}\bigl(w-g(v)\bigr),\end{array}\right. (1.6) where g^{\prime}(v)>0 and the function v=f(u) is the inverse of u=v+g(v) . (iii) Approximations with nonlocal flux. More specifically, motivated by models of traffic flow [ 12 , 16 , 24 ] , one can consider u_{t}+F(u,v)_{x}~=~0,\qquad\qquad v(t,x)\doteq\,\int_{0}^{+\infty}\varepsilon^{-1}e^{-y/\varepsilon}\,u(t,x+y)\,dy, (1.7) where F(u,u)=f(u) . (iv) Various numerical approximations [ 23 , 27 ] . Let now u=u(t,x) be a generic solution of ( 1.3 ) with {\cal C}^{3} initial data. In addition, let u_{\varepsilon} be a family of approximations of one of the types (i)–(iv), such that the initial data u_{\varepsilon}(0,\cdot) converge to u(0,\cdot) in {\cal C}^{3} , as \varepsilon\to 0 . If (\tau,\xi) is a point where u has a structurally stable singularity, as already remarked the rescaled functions in ( 1.1 ) or ( 1.2 ) converge to a well defined limit as \delta\to 0 . In this same setting we conjecture that, by suitably choosing points (\tau_{\varepsilon},\xi_{\varepsilon})\to(\tau,\xi) , for suitable exponents \alpha,\beta,\gamma (not necessarily the same as in ( 1.2 )), the rescaled functions U^{\varepsilon}(t,x)~\doteq~u_{\varepsilon}\bigl(\tau+\varepsilon^{\alpha}t,~\xi+\varepsilon^{\beta}x\bigr), (1.8) or, depending on the type of singularity, U^{\varepsilon}(t,x)~\doteq~\varepsilon^{-\gamma}\Big[u_{\varepsilon}\bigl(\tau_{\varepsilon}+\varepsilon^{\alpha}t,\,\xi_{\varepsilon}+\varepsilon^{\beta}x\bigr)-u(\tau,\xi)\Big], (1.9) converge to a unique limit U=U(t,x) , uniformly on compact sets. This limit is defined for all (t,x)\in{\mathbb{R}}^{2} . It provides an “eternal solution" (i.e, with time ranging from -\infty to +\infty ) to an equation of the same type as ( 1.5 ), ( 1.6 ) or ( 1.7 ), but with \varepsilon=1 . The set of all these limits depends on a finite number of parameters. In the remainder of this paper we shall give a proof of the above conjecture in the basic case of viscous approximations to a scalar conservation law with strictly convex flux: u_{t}+f(u)_{x}~=~\varepsilon\,u_{xx}\,. (1.10) The following assumptions will be used. (A1) The flux function f is smooth and uniformly convex: 0~<~c_{1}~\leq~f^{\prime\prime}(u)~\leq~c_{2}\qquad\qquad\hbox{for all}~u\in{\mathbb{R}}. (1.11) (A2) The initial data satisfies \bar{u}\in\mathbf{L}^{1}({\mathbb{R}}) . We recall that, for a generic initial data \bar{u}\in{\cal U}\subset{\cal C}^{3}({\mathbb{R}}) , the solution of ( 1.3 )-( 1.4 ) is piecewise smooth with a locally finite number of shocks [ 21 , 22 , 31 ] . Moreover, shocks interact at most two at a time. Here {\cal U} is a dense {\cal G}_{\delta} -set, i.e., the intersection of countably many open dense subsets of {\cal C}^{3}({\mathbb{R}}) . Figure 1: The three types of generic singularities for a solution to a conservation law. A: a point along a shock. B: a point where a new shock forms. C: a point where two shocks merge. Here the dashed lines denote characteristics, while the solid curves denote shocks. For such generic solutions, singularities can only be of the above types (I)–(III). For each type, there is a natural asymptotic rescaling that characterizes the singularity up to leading order. CASE 1. If (\tau,\xi) is a point along a shock, we consider the rescalings u^{\delta}(t,x)~\doteq~u\bigl(\tau+\delta t,\,\xi+\delta x\bigr). (1.12) As \delta\to 0+ , the solutions u^{\delta} converge in \mathbf{L}^{1}_{loc}({\mathbb{R}}^{2}) to a piecewise constant solution U=U(t,x) of ( 1.3 ) defined for all (t,x)\in{\mathbb{R}}^{2} and containing a single shock: U(t,x)~=~\left\{\begin{array}[]{rl}u^{-}&\hbox{if}~~x<\lambda t,\\[2.84526pt] u^{+}&\hbox{if}~~x>\lambda t.\end{array}\right. (1.13) Here u^{\pm}\doteq u(\tau,\xi\pm),\qquad\qquad\lambda={f(u^{+})-f(u^{-})\over u^{+}-u^{-}}\,. CASE 2. If (\tau,\xi) is a point where two shocks interact, the rescaled functions u^{\delta} defined at ( 1.12 ) now converge to a piecewise constant solution with two interacting shocks U(t,x)~=~\left\{\begin{array}[]{cl}u^{-}&\hbox{if}~~x<\lambda_{1}t,\\[2.84526pt] u^{*}&\hbox{if}~~\lambda_{1}t<x<\lambda_{2}t,\\[2.84526pt] u^{+}&\hbox{if}~~\lambda_{2}t<x,\end{array}\right.\qquad\qquad\hbox{for}~t<0, (1.14) U(t,x)~=~\left\{\begin{array}[]{rl}u^{-}&\hbox{if}~~x<\lambda t,\\[2.84526pt] u^{+}&\hbox{if}~~x>\lambda t,\end{array}\right.\qquad\qquad\hbox{for}~t\geq 0. (1.15) Here u^{-},u^{*},u^{+} are the limits of left, middle and right states before the interaction, while \lambda_{1}>\lambda_{2} and \lambda are the Rankine-Hugoniot speeds of the shocks, before and after the interaction. CASE 3. At the point \overline{P}=(\tau,\xi) a new shock is formed. Assuming that the initial data \bar{u} is {\cal C}^{3} , call x_{0}\,\doteq\,\xi-f^{\prime}\bigl(u(\tau,\xi)\bigr)\,\tau, the initial point of the characteristic through \overline{P} , so that \bar{u}(x_{0})~=~u(\tau,\xi),\qquad\qquad\tau\,=\,{1\over f^{\prime\prime}\bigl(\bar{u}(x_{0})\bigr)\bar{u}_{x}(x_{0})}\,. Following [ 22 ] , we say that this singularity is structurally stable if {\partial\over\partial x}f^{\prime}\bigl(\bar{u}(x)\bigr)\bigg|_{x=x_{0}}=~0,\qquad\qquad{\partial^{2}\over\partial x^{2}}f^{\prime}\bigl(\bar{u}(x)\bigr)\bigg|_{x=x_{0}}<~0. In this case, the rescaled functions u^{\delta}(t,x)~=~\delta^{-1}\Big[u\bigl(\tau+\delta^{2}t,\,\xi+\delta^{3}x\bigr)-u(\tau,\xi)\Big],