Exact controllability to trajectories for entropy solutions to scalar conservation laws in several space dimensions

TL;DR

This paper introduces a new method for achieving exact controllability to trajectories in multidimensional scalar conservation laws, focusing on entropy solutions under minimal assumptions on flux and geometry.

Contribution

The paper presents a novel approach for exact controllability in multidimensional conservation laws, expanding the theoretical framework for entropy solutions with minimal assumptions.

Findings

Achieved exact controllability to trajectories for entropy solutions

Developed a new method under non-degeneracy and geometric conditions

Applicable to several space dimensions

Abstract

We describe a new method which allows us to obtain a result of exact controllability to trajectories of multidimensional conservation laws in the context of entropy solutions and under a mere non-degeneracy assumption on the flux and a natural geometric condition.

Equations

$${\partial }_{t}u+{\mathrm{div}}_x\left(f(u)\right)=0,t\in {\mathbb{R}}^{+},x\in \Omega \subset {\mathbb{R}}^d,d\ge 1,$$

$$\{\begin{array}{ll}{\partial }_{t}u+{\mathrm{div}}_x\left(f(u)\right)=0,& \text{in }(0,T)\times \Omega \\ u(0,x)=u_0(x),& \text{on }\Omega ,\\ u(T,x)=u_1(x),& \text{on }\Omega ,\end{array}$$

$$\forall u\in \mathbb{R},\mathrm{sign}(u):=\{\begin{array}{ll}1& \text{if }u>0\\ 0& \text{if }u=0\\ -1& \text{if }u<0,\end{array}$$

$${\partial }_{t}u+\mathrm{div}(f(u))=0,\text{for }(t,x)\in (0,+\infty )\times \Omega ,$$

$$u(0,x)=u_0(x),\text{for }x\in \Omega .$$

$${\int }_{(0,T)\times \Omega }|u(t,x)-k|{\partial }_t\varphi (t,x)+\mathrm{sign}(u(t,x)-k)$$

$$+{\int }_{\Omega }\mathrm{sign}(u_0(x)-k)\varphi (0,x)dx\ge 0.$$

$$L:=\underset{x\in \Omega }{\sup }⟨w|x⟩-\underset{x\in \Omega }{\inf }⟨w|x⟩<+\infty ,$$

$$\forall u\in I,⟨f^{\prime }(u)|w⟩\ge c,\text{and }\mathfrak{t}=\frac{L}{c}.$$

$$\mathcal{L}\left(\{z\in \mathbb{R}:\tau +⟨\zeta |f^{\prime }(z)⟩=0\}\right)=0,$$

$$u(0,x)=u_0(x),u(T_1,x)=v(T_2,x)\text{for almost every }x\in \Omega .$$

$$\{\begin{array}{ll}{\partial }_{t}u+\mathrm{div}(f(u))=0,& (0,+\infty )\times \Omega ,\\ u(t,x)=u_b(t,x),& (t,x)\in (0,+\infty )\times \partial \Omega ,\\ u(0,x)=u_0(x),& x\in \Omega ,\end{array}$$

$$\mathrm{sign}(u(t,x)-u_b(t,x))\left(f(u(t,x))\cdot \eta (x)-f(k)\cdot \eta (x)\right)\ge 0.$$

$${\omega }^{+}(x,k,\alpha ):=\underset{k\le r,s\le \max (\alpha ,k)}{\max }|⟨f(r)-f(s)|\eta (x)⟩|,$$

$${\omega }^{-}(x,k,\alpha ):=\underset{\min (\alpha ,k)\le r,s\le k}{\max }|⟨f(r)-f(s)|\eta (x)⟩|.$$

$$\begin{gathered} {\int }_0^T{\int }_{\Omega }(u(t,x)-k{)}^{+}{\partial }_t\zeta (t,x)+{\chi }_{\{u(t,x)>k\}}⟨f(u(t,x))-f(k))|\nabla \zeta (t,x)⟩dxdt \\ +{\int }_{\partial \Omega }{\int }_0^T\zeta (t,x){\omega }^{+}(x,k,u_b(t,x))dtd\sigma (x)+{\int }_{\Omega }(u_0(x)-k{)}^{+}\zeta (0,x)dx\ge 0, \end{gathered}$$

$$\begin{gathered} {\int }_0^T{\int }_{\Omega }(k-u(t,x){)}^{+}{\partial }_t\zeta (t,x)+{\chi }_{\{u(t,x)<k\}}⟨f(u(t,x))-f(k))|\nabla \zeta (t,x)⟩dxdt \\ +{\int }_{\partial \Omega }{\int }_0^T\zeta (t,x){\omega }^{-}(x,k,u_b(t,x))dtd\sigma (x)+{\int }_{\Omega }(k-u_0(x){)}^{+}\zeta (0,x)dx\ge 0. \end{gathered}$$

$$\begin{gathered} {\int }_0^T{\int }_{\Omega }(u(t,x)-v(t,x){)}^{+}{\partial }_t\zeta (t,x)+{\chi }_{\{u(t,x)>v(t,x)\}}⟨f(u(t,x))-f(v(t,x)))|\nabla \zeta (t,x)⟩dxdt \\ +{\int }_{\partial \Omega }{\int }_0^T{\omega }^{-}(x,u_b(t,x),v_b(t,x))\zeta (t,x)dtd\sigma (x)+{\int }_{\Omega }(u_0(x)-v_0(x){)}^{+}\zeta (0,x)dx\ge 0. \end{gathered}$$

$$u_0(x)\in J,\text{for a.e. }x\in \Omega ,$$

$$u_b(t)\in J,\text{for a.e. }t\ge 0.$$

$$u(t,x)\in J\text{for a.e. }(t,x)\text{in}(0,+\infty )\times \Omega .$$

$$u_0(x)\le B,u_b(t,y)\le B,$$

$${\int }_{\partial \Omega }{\int }_0^{\tilde{t}}{\omega }^{+}(y,B,u_b(t,y))dtd\sigma (y)+{\int }_{\Omega }(u_0(x)-B{)}^{+}-(u(\tilde{t},x)-B{)}^{+}dx\ge 0.$$

$${\omega }^{+}(y,B,u_b(t,y))=0,(u_0(x)-B{)}^{+}=0,$$

$$(u(\tilde{t},x)-B{)}^{+}=0,\text{for a.e. x in Ω}$$

$$u(\tilde{t},x)\le B,\text{for a.e. x in Ω}.$$

$$\forall t\ge \mathfrak{t},u(t,x)=v(t,x),\text{for almost every x in Ω}.$$

$$\forall t\ge 0,J_{\theta }(t):={\int }_{\Omega }|u(t,x)-v(t,x)|e^{-\theta ⟨w|x⟩}dx.$$

$$\begin{gathered} {\int }_0^{\bar{t}}{\int }_{\Omega }\left((v(t,x)-u(t,x){)}^{+}+(u(t,x)-v(t,x){)}^{+}\right){\partial }_t\zeta (t,x) \\ +⟨({\chi }_{\{v(t,x)>u(t,x)\}}f(v(t,x))-f(u(t,x)))+({\chi }_{\{u(t,x)>v(t,x)\}}f(u(t,x))-f(v(t,x)))|\nabla \zeta (t,x)⟩dxdt \\ +{\int }_{\partial \Omega }{\int }_0^{\bar{t}}2{\omega }^{-}(x,u_b(t,x),u_b(t,x))\zeta (t,x)dtd\sigma (x) \\ +{\int }_{\Omega }((v_0(x)-u_0(x){)}^{+}+(u_0(x)-v_0(x){)}^{+})\zeta (0,x)dx\ge 0. \end{gathered}$$

$$\begin{gathered} {\int }_0^{\bar{t}}{\int }_{\Omega }|u(t,x)-v(t,x)|{\partial }_t\zeta (t,x)+\mathrm{sign}(u(t,x)-v(t,x))⟨f(u(t,x))-f(v(t,x))|\nabla \zeta (t,x)⟩dxdt \\ +{\int }_{\Omega }|u_0(x)-v_0(x)|\zeta (0,x)dx\ge 0. \end{gathered}$$

$$J_{\theta }(\bar{t})\le J_{\theta }(0)+{\int }_0^{\bar{t}}{\int }_{\Omega }\mathrm{sign}(u(t,x)-v(t,x))⟨f(u(t,x))-f(v(t,x)))|-w\theta e^{-\theta ⟨w|x⟩}⟩dxdt.$$

$$\forall (a,b)\in I^2,\mathrm{sign}(a-b)⟨f(a)-f(b)|w⟩$$

Full text

11footnotetext: Université de Bourgogne Franche-Comté, Laboratoire de Mathématiques, CNRS UMR6623, 16 route de Gray, 25000 Besançon, carlotta.donadello@univ-fcomte.fr22footnotetext: Institut Denis Poisson, Université de Tours, CNRS UMR7013, Université d’Orléans, Parc de Grandmont, 37000 Tours vincent.perrollaz@lmpt.univ-tours.fr

1 Introduction

In this paper we consider a scalar conservation law in several space dimensions, i.e. a partial differential equation of the form

[TABLE]

where $\Omega$ is an open set with smooth boundary ($\mathcal{C}^{2}$ is sufficient), and $f$, the flux function, is in $\mathcal{C}^{1}(\mathbb{R},\mathbb{R}^{d})$.

We are interested in the following controllability problem. Given an initial datum $u_{0}\in L^{\infty}(\Omega)$, a suitable target profile $u_{1}$ and a positive time $T$, we aim at constructing an entropy weak solution $u\in L^{\infty}(\mathbb{R}^{+}\times\mathbb{R}^{d};\mathbb{R})$ of

[TABLE]

by using the boundary data on $(0,T)\times\partial\Omega$ as controls.

Given any extensive quantity $u$ defined on a domain $\Omega$, such as mass or energy, a conservation law for $u$ translates into a partial differential equation the physical observation that the total amount of $u$ in $\Omega$ changes at a rate which corresponds to the net flux of $u$, $f(u)$, through the boundary $\partial\Omega$. This kind of equations is widely used in modeling phenomena such as gas dynamics (Euler equations), electromagnetism, magneto-hydrodynamics, shallow water, combustion, road traffic, population dynamics and petroleum engineering.

It is well known that even starting from initial data in $\mathcal{C}_{c}^{\infty}(\mathbb{R}^{d})$ the classical solutions of (1) can develop singularities (jump discontinuities) in finite time, see [19] for a very complete introduction to the field.

The most general wellposedness result for classical solutions to the Cauchy problem states that for any initial datum $u_{0}$ in $H^{s}$, with $s>1+\frac{d}{2}$, there exists a solution of (1) in $\mathcal{C}^{0}([0,T],H^{s})\cap\mathcal{C}^{1}([0,T],H^{s-1})$, whose life span $T$ can be estimated depending on $f$ and $u_{0}$.

However, most of the literature devoted to conservation laws focuses on a class of weak (distributional) solutions which satisfies an additional selection criterium, necessary to ensure uniqueness, called entropy condition. In the case of a scalar conservation law in several space dimensions a complete wellposedness theory for entropy solutions to the Cauchy problem is due to Kruzhkov, [29].

The problem we aim at solving, see (1), can be adressed both in the framework of classical or of entropy solutions. In the first case controls, besides driving the state to the target, are also responsible of preventing the formation of singularities. Several results exist in this framework, see [13], [27] and [16] for a survey. Unfortunately, this approach does not allow to attain many physically relevant states involving jump discontinuities and leads to control strategies which are in general not very robust. Indeed very small perturbations of the control might lead to blow up of the derivatives of the solution before time $T$.

In the present paper we are interested in controllability of entropy solutions. The literature in this framework is less abundant also due to specific technical difficulties, even if we can notice a growing interest of researchers in this field. The classical methodology for exact controllability relies heavily on linearization which is not possible (or at least horribly technical) anymore around discontinuous solutions. Moreover, Bressan and Coclite showed in [12] that nonlinear conservation laws may fail the linear test. Indeed they provided a system for which the linearized approximation around a constant state is controllable, while the original nonlinear system cannot reach that same constant state in finite time.

A separate issue is related to the irreversibility of entropy solutions: the set of admissible target states in time $T$ is reduced and its description, often involving a number of highly technical conditions, is in itself an open problem in most cases, see [6, 7, 2, 3, 14].

In the existing literature we can distinguish essentially three approaches toward the study of exact controllability for conservation laws in one space dimension (consider equation (1) with $d=1$).

Starting from the pioneering work by Ancona and Marson, [3], several results have been obtained using the theory of generalized characteristics introduced by Dafermos in [20], as [7, 3, 14, 24, 32] or the explicit Lax-Oleinik representation formula, as [6, 1]. The latter technique is applicable only when the flux function $f$ is strictly convex/concave, while the theory of generalized characteristics covers also the (slightly) more general case of a flux function $f$ with one inflection point.

The return method introduced by Coron, [16], is the basis of the approach developed by Horsin in [24] and, combined with the classical vanishing viscosity method, plays a key role in [23] and in the only paper to our knowledge in which the flux function $f$ is allowed to have a finite number of inflection points, [25].

The asymptotic stabilization of entropy solutions to scalar conservation laws is the topic of [10, 33, 34].

The only available tool for investigating the exact controllability of systems of conservation laws in one space dimension is wave front tracking algorithm, [11], which has been successfully applied in [2, 12, 21, 22, 28].

The asymptotic stabilization of entropy solutions to systems has been studied in [4, 9, 18].

It seems difficult to investigate the exact controllability of entropy solutions of scalar conservation laws in several space dimensions using the techniques designed for the one dimensional case. In the present paper we propose a new approach, inspired by the work on stabilization by Coron, [15], and Coron, Bastin and d’Andéa Novel, [17], see also the monography [9] for a comprehensive presentation of the method. The conditions we impose on the flux function are technical and will be detailed in the next Section, but we stress that in the special case $d=1$ they are not related to convexity (or concavity). This means that even in the one dimensional case our result is new, althought for this case stronger results are available under more restrictive hypothesis.

The first of our conditions, called later nondegeneracy condition, says that the range of $u$ does not contain any interval on which $f$ is affine. This condition is necessary to ensure the existence of traces at the boundary of $\Omega$, see [35].

The second condition, called lated replacement condition involves $f$ together with $T$ and $\Omega$. Roughly speaking, once we reduce to the one-dimensional case, it says that all generalized characteristics issued from points $(t,x)$ in $\{0\}\times\Omega$ leave the cylinder $(0,T)\times\Omega$ before time $T$, so that the dynamics in the domain only depends on the boundary data and not on the initial data for $t$ large enough.

2 Preliminary definitions and notations

In the following $u\mapsto\mathrm{sign}(u)$ is the function given by

[TABLE]

$\langle\cdot|\cdot\rangle$ denotes the scalar product between two vectors and $\chi_{E}$ is the indicator function of the set $E$.

Definition 1**.**

Given $f\in\mathcal{C}^{1}\left(\mathbb{R};\mathbb{R}^{d}\right)$ and $u_{0}\in L^{\infty}\left(\Omega\right)$, we consider the equation

[TABLE]

supplemented with the initial condition

[TABLE]

A function $u\in L^{\infty}\left([0,+\infty)\times\Omega\right)$ is an entropy solution of (2)–(3) in $[0,T)\times\Omega$ if for any real number $k$ and any non negative function $\phi$ in $\mathcal{C}_{c}^{1}([0,+\infty)\times\Omega)$ we have

[TABLE]

We will also say that a function $u$ is an entropy solution (without refering to any initial data) in $(0,+\infty)\times\Omega$ when it satisfies (4) for any non negative $\phi\in\mathcal{C}_{c}^{1}((0,+\infty)\times\Omega)$.

We now introduce a simple geometric condition which is sufficient (though clearly not necessary) to obtain our controllability result.

Definition 2**.**

Let $\Omega$ be a smooth open set of $\mathbb{R}^{d}$, $I$ be a segment of $\mathbb{R}$ and $f:\mathbb{R}\to\mathbb{R}^{d}$ a $\mathcal{C}^{1}$ flux function. We say that the triple $\left(f,\Omega,I\right)$ satisfies the replacement condition in time $\mathfrak{t}>0$ if there exists a vector $w\in\mathbb{R}^{d}$ and a positive number $c$ such that

[TABLE]

Definition 3**.**

We say that the flux $f$ is non degenerate if for any couple $(\tau,\zeta)\in\mathbb{R}\times\mathbb{R}^{d}\setminus\{(0,0)\}$ we have

[TABLE]

where $\mathcal{L}$ is the Lebesgue measure.

We can now state our main theorem on exact controllability to trajectories for a conservation law in any space dimension.

Theorem 1**.**

Let $v\in\mathcal{C}^{0}((0,+\infty);L^{1}(\Omega))\cap L^{\infty}((0,+\infty)\times\Omega)$ be an entropy solution to (2) and $u_{0}$ be a function in $L^{\infty}(\Omega)$.

We suppose that both $u_{0}$ and $v$ take values in a segment $I$ such that $\left(f,\Omega,I\right)$ satisfies the replacement condition in time $\mathfrak{t}$. We also suppose that the flux $f$ is non degenerate.

Then for any times $T_{1}$ and $T_{2}$ larger than $\mathfrak{t}$, there exists an entropy solution $u$ of (2) satisfying

[TABLE]

Remark 1**.**

For the sake of simplicity, we omit to write here the exact form of the controls we use. In the next Section we precise in which sense the boundary condition on $\partial\Omega$ are taken into account by entropy solutions and in the last Section, in the proof of Theorem 1, we write our controls in a fully explicit way.

Remark 2**.**

A characterization of the set of admissible target profiles at fixed time time $T\geq 0$ for a scalar conservation law in several space dimensions is not available in the literature. We stress however that in the statement of Theorem 1 we really need to assume that $v$ is an entropy solution on the cylinder $(0,+\infty)\times\Omega$ because the complete knowledge of $v$ is necessary in our proof.

3 Boundary conditions and entropy solutions

We have so far avoided the precise formulation of boundary conditions for (2). In general, given the initial boundary value problem

[TABLE]

its entropy solution $u$ does not satisfy the boundary condition in the usual sense, as the trace of $u$ on $\partial\Omega$ does not coincide with the prescribed Dirichlet datum. The situation is easier to visualize in the one dimensional case as we can see in the following exemple.

Example 1**.**

Assume $d=1$, $\Omega=(0,+\infty)$, $f(u)=\dfrac{u^{2}}{2}$ and impose in (7) constant initial and boundary data, $u_{0}=-1$ and $u_{b}=-\dfrac{1}{2}$. Then the initial condition is transported along characteristic curves with negative slope up to the boundary, while no characteristic can spring out of the boundary itself. The trace of the solution at $x=0$ can only take the value $u(t,0^{+})=u_{0}$, and $u_{b}$ can not be attained.

For this reason the boundary conditions should be interpreted in a broader sense, made precise by Leroux, [26], and Bardos, Leroux and Nédélec, [8]. In the setting of the exemple above we say that the boundary condition is fulfilled in the sense of Bardos-Leroux-Nédélec as soon as the solution to the Riemann problem with data $u_{L}=u_{b}$ and $u_{R}=u(t,0^{+})$ only contains waves of non positive speed (i.e. waves which do not enter the domain). In the general multidimensional case this takes the following form.

Definition 4**.**

Let $I(a,b)$ denote the interval of extrema $a$ and $b$, and let $\eta(x)$ be the outer unit normal of $\partial\Omega$ at $(t,x)\in(0,T)\times\partial\Omega$. Then we say that the boundary condition $u_{b}$ in the IBVP (7) is fulfilled at $(t,x)\in(0,T)\times\partial\Omega$ if for any $k\in I(u_{b}(t,x),u(t,x))$

[TABLE]

We have to precise, however, that the above definition is not exactly the one we adopt in the present work as existence of traces is not guaranted for solution of (7) in the $L^{\infty}$ setting. The first results dealing with this problem were by Otto, [31], see also [30]. We use more recent results by Ammar, Carillo and Wittbold, [5], which build upon those ideas. We also recall a (simplified version of a) result by Vasseur, [35], showing that if the flux satisfies the non degeneracy condition then any entropy solution $u\in L^{\infty}$ admits a trace at the boundary.

Let us recall some definitions and results in [5]. We use the following notations. For any real numbers $\alpha$ and $k$, and any point $x\in\partial\Omega$, we call $\eta(x)$ the outer unit normal at $x$ and introduce

[TABLE]

[TABLE]

For integrals on the boundary we denote the surface measure at $x\in\partial\Omega$ by $d\sigma(x)$.

Definition 5**.**

Given a boundary condition $u_{b}\in L^{\infty}((0,+\infty)\times\partial\Omega)$ and an initial data $u_{0}\in L^{\infty}(\Omega)$ we say that $u$ is an entropy solution to (7) when the following hold for any $k\in\mathbb{R}$ and any non negative function $\zeta\in\mathcal{C}^{\infty}_{c}([0,+\infty)\times\mathbb{R}^{d})$

[TABLE]

[TABLE]

The following two theorems were proven in [5] (see [5], Theorem 2.3 and 2.4).

Theorem 2**.**

Given initial and boundary data $u_{0}\in L^{\infty}(\Omega)$ and $u_{b}\in L^{\infty}((0,+\infty)\times\partial\Omega)$, there exists a unique entropy solution of (7).

Theorem 3**.**

Given initial data $u_{0},\,v_{0}$ in $L^{\infty}(\Omega)$ and boundary data $u_{b},\,v_{b}$ in $L^{\infty}((0,+\infty)\times\partial\Omega)$, the corresponding entropy solutions $u$ and $v$ satisfy

[TABLE]

for any non negative function $\zeta\in\mathcal{C}^{\infty}_{c}([0,+\infty)\times\mathbb{R}^{d})$.

Let us finally recall a simplified version of the result obtained by Vasseur in [35], which is sufficient for our use.

Theorem 4**.**

Assume that the flux $f$ is non degenerate and that the domain $\Omega$ is $\mathcal{C}^{2}$. Then if $u\in L^{\infty}((0,+\infty)\times\Omega)$ is an entropy solution of (2) in the sense of Definition 1, i.e. (4) is satisfied for any $k$ and any non negative function $\phi$ in $\mathcal{C}_{c}^{1}((0,+\infty)\times\Omega)$, then there exists boundary data $u_{b}\in L^{\infty}((0,T)\times\partial\Omega)$ and initial data $u_{0}\in L^{\infty}(\Omega)$ such that $u$ is the unique entropy solution of the mixed problem (7) in the sense of Definition 5.

4 Proof of the main result

Lemma 1**.**

Consider $J:=[A,B]\subset\mathbb{R}$ and suppose that

[TABLE]

Then the unique entropy solution to the IBVP (7) with initial and boundary data $u_{0}$ and $u_{b}$, $u$ satisfies

[TABLE]

Démonstration.

We prove here in full details that $u(t,x)\leq B$ for a.e. $(t,x)$ in $(0,+\infty)\times\Omega$. The inequality $A\leq u(t,x)$ can be obtained analogously.

By hypothesis we have for almost every $x$ in $\Omega$ and $(t,y)$ in $(0,+\infty)\times\partial\Omega$

[TABLE]

then for any fixed time $\tilde{t}\geq 0$, taking a sequence $\zeta_{n}\in\mathcal{C}_{c}^{\infty}(\mathbb{R})\to\chi_{(-\infty,\tilde{t}]}$ and using $k=B$, from (8) we obtain

[TABLE]

It is clear that for a.e. $x$ in $\Omega$ and $(t,y)$ in $(0,\tilde{t})\times\partial\Omega$ we have

[TABLE]

then (11) implies

[TABLE]

which is indeed

[TABLE]

∎

Proposition 1**.**

Let $u$ and $v$ be entropy solutions of (7) with respective initial data $u_{0}$ and $v_{0}$ and the same boundary datum $u_{b}$. Let us also suppose that all data take value in an interval $I$ which satisfies the replacement condition in time $\mathfrak{t}=\frac{L}{c}$ (with a direction $w$). Then we can conclude that

[TABLE]

Démonstration.

Let us define for $\theta>0$ the functional $J_{\theta}$ by

[TABLE]

Given $\bar{t}\geq 0$, we apply (10) to the ordered couples $(u,v)$ and $(v,u)$, then adding the inequalities we get

[TABLE]

which is actually

[TABLE]

We consider a sequence $\left(\zeta_{n}\right)_{n}\subset\mathcal{C}_{c}^{\infty}(\mathbb{R})$ converging in $L^{1}$ to $\chi_{(-\infty,\bar{t}]}e^{-\theta\langle w|x\rangle}$, so that in the limit $n\to\infty$ we get

[TABLE]

But since

[TABLE]

from (12), Lemma 1 and the replacement condition, we obtain

[TABLE]

Thanks to the classical Gronwall’s lemma we end up with

[TABLE]

As $\bar{t}$ was arbitrarily chosen, if $M:=\underset{x\in\Omega}{\sup}\ \langle w|x\rangle$ and $m=\underset{x\in\Omega}{\inf}\ \langle w|x\rangle$, we can write that for all $t\geq 0$

[TABLE]

So we can compute

[TABLE]

So for any $t\geq\mathfrak{t}=\dfrac{L}{c}$ letting $\theta\to+\infty$ we obtain

[TABLE]

∎

We are ready to prove Theorem 1.

Démonstration.

We aim at proving that there exists an entropy solution to the problem

[TABLE]

In view of the well-posedness result stated in Theorem 2 our goal is achieved once we construct suitable boundary conditions, which can be interpreted as controls in our setting.

**Case $T_{2}>T_{1}$. **

Thanks to Theorem 4 it makes sense to consider

[TABLE]

We call $w$ the unique entropy solution to the IBVP (7) with data $w_{0}$, $w_{b}$ on $(0,T_{1})\times\Omega$. The form of the equation implies that for almost every $(s,x)$ in $(0,T_{1})\times\Omega$, $w(s,x)=v(T_{2}-T_{1}+s,x)$.

By hypothesis $T_{1}\geq\mathfrak{t}$ so, as a direct application of Proposition 1, we can conclude that the entropy solutions to the mixed problems of the form (7) with initial data $u_{0}$ and $w_{0}$ respectively, and common boundary datum $w_{b}$ satisfy

[TABLE]

which means

[TABLE]

**Case $T_{1}>T_{2}$. **

We define

[TABLE]

where $\mathfrak{t}$ is the time given by the replacement condition.

We call $w$ the unique entropy solution to the IBVP (7) with data $w_{0}$, $w_{b}$ on $(0,\mathfrak{t})\times\Omega$. The form of the equation implies that for almost every $(s,x)$ in $(0,\mathfrak{t})\times\Omega$, $w(s,x)=v(T_{2}-\mathfrak{t}+s,x)$.

We consider now a boundary condition of the following form

[TABLE]

where $\mathfrak{b}$ is any constant state in the interval $I$.

The IBVP (7) with data $u_{0}$, $u_{b}$ admits a unique entropy solution $u$ in $(0,+\infty)\times\Omega$.

We call $\tilde{u}_{0}$ the profile of $u$ at time $t=T_{1}-\mathfrak{t}$.

Now it is clear that if we apply Proposition 1 to the entropy solutions $\tilde{u}$ et $w$ of (7) with respective initial data $\tilde{u}_{0}$ et $w_{0}$ and common boundary data $w_{b}$ we obtain

[TABLE]

which means

[TABLE]

∎

Acknowledgements

The first author acknowledges the financial support of the Région Franche-Comté, projet Analyse mathématique et simulation numérique d’EDP issus de problèmes de contrôle et du trafic routier. The second author was supported by ANR Finite4SOS (15-CE23-0007).

Références

• [1] Adimurthi, Ghoshal S. S. and Gowda V., Exact controllability of scalar conservation laws with strictly convex flux, Math. Control Relat. Fields 4 (2014) 401-449.
• [2] Ancona F. and Coclite G. M., On the attainable set for Temple class systems with boundary controls, SIAM J. Control Optim. 43 (2005), no. 6, pp. 2166-2190.
• [3] Ancona F. and Marson A., On the attainable set for scalar nonlinear conservation laws with boundary control, SIAM J. Control Optim. 36 (1998) no. 1, pp. 290-312.
• [4] Ancona F. and Marson A., Asymptotic stabilization of systems of conservation laws by controls acting at a single boundary point. Control methods in PDE-dynamical systems, pp. 1–43, Contemp. Math., 426, Amer. Math. Soc., Providence, RI, 2007.
• [5] Ammar K., Wittbold P. and Carillo J., Scalar conservation laws with general boundary conditions and continuous flux function, J. Differential Equations 228 (2006) pp. 111-139.
• [6] Andreianov B. , Donadello C., Ghoshal S. S. and Razafison U., On the attainable set for a class of triangular systems of conservation laws, J. Evol. Equ., 15 (2015), no. 3, pp 503-532.
• [7] Andreianov B., Donadello C. and Marson A., On the attainable set for a scalar nonconvex conservation law, SIAM J. Control Optim. 55 (2017), no. 4, pp. 2235-2270.
• [8] Bardos C., Leroux A.-Y. and Nédélec J.-C., First order quasilinear equations with boundary conditions, Comm. in Partial Differential Equations, 9 (1979); pp. 1017-1034.
• [9] Bastin G. and Coron J.-M., Stability and boundary stabilization of 1-D hyperbolic systems, Springer PLNDE Subseries in Control, 88 (2016).
• [10] Blandin S., Litrico X., Dalle Monache M. L., Piccoli B. and Bayen T., Regularity and Lyapunov stabilization of weak entropy solutions to scalar conservation laws, IEEE Transactions on Automatic Control 62 no. 4 (2017) pp. 1620-1635.
• [11] Bressan A., Hyperbolic Systems of Conservation Laws. The One-Dimensional Cauchy Problem. Oxford University Press. Oxford (2000).
• [12] Bressan A. and Coclite G. M., On the boundary control of systems of conservation laws, SIAM J. Control Optim. 41 (2002), no. 2, pp. 607-622.
• [13] Chapouly M., Global controllability of non-viscous and viscous Burgers type equations, SIAM J. Control Optim., Volume 48, Issue 3 (2009), pp. 1567-1599.
• [14] Corghi M. and Marson A., On the attainable set for scalar balance laws with distributed control, ESAIM COCV 22 (2016) pp. 236-266.
• [15] Coron, J.-M., On the null asymptotic stabilization of the two-dimensional incompressible Euler equations in a simply connected domain. SIAM J. Control Optim., Volume 37, Issue 6 (1999), pp. 1874-1896.
• [16] Coron J.-M., Control and nonlinearity. Mathematical Surveys and Monographs, 136. American Mathematical Society, Providence, RI, 2007.
• [17] Coron, J.-M., Bastin, G., and d’Andréa-Novel, B., Dissipative boundary conditions for one dimensional nonlinear hyperbolic systems. SIAM J. Control Optim., Volume 47, Issue 3 (2008), pp. 1460-1498.
• [18] Coron J.-M., Ervedoza S., Ghoshal S. S., Glass O. and Perrollaz V., Dissipative boundary conditions for 2x2 hyperbolic systems of conservation laws for entropy solutions in BV, Journal of Differential Equations 262 (2017), pp. 1-30.
• [19] Dafermos C. M., Hyperbolic conservation laws in continuum physics. Second edition. Grundlehren der Mathematischen Wissenschaften, 325. Springer-Verlag, Berlin, 2005.
• [20] Dafermos C. M., Generalized characteristics and the structure of solutions of hyperbolic conservation laws. Indiana Math. J. 26 (1977), pp. 1097-1119.
• [21] Glass O., On the controllability of the 1-D isentropic Euler equation, J. European Math. Soc. 9, no. 3 (2007) pp. 427-486.
• [22] Glass O., On the controllability of the non-isentropic 1-D Euler equation, Journal of Differential Equations 257 (2014), pp. 638-719.
• [23] Glass O. and Guerrero S., On the uniform controllability of the Burgers equation, SIAM J. Control Optim., Volume 46, no. 4, pp. 1211-1238.
• [24] Horsin T., On the controllability of the Burger equation, ESAIM COCV 3 (1998) pp. 83-95.
• [25] Léautaud M., Uniform controllability of scalar conservation laws in the vanishing viscosity limit, SIAM J. Control Optim. 50 (2012), pp. 1661-1699.
• [26] Leroux A. Y., Etude du problème mixte pour une équation quasi-linéaire du premier ordre, C. R. Acad. Sci. Paris, série A, T. 285 (1977), pp. 351-354.
• [27] Li T., Controllability and observability for quasilinear hyperbolic systems. AIMS Series on Applied Mathematics, 3. American Institute of Mathematical Sciences (AIMS),
• [28] Li T. and Yu L., One sided exact null controllability of entropy solutions to a class of hyperbolic systems of conservation laws, Journal de Mathématiques Pures et Appliquées, 107 no. 1 (2017) pp. 1-10.
• [29] Kruzkov S. N., First order quasilinear equations with several independent variables. Mat. Sb. (N.S.) 81 (123) 1970, pp. 228-255.
• [30] Malek J., Necas J., Rokyta M., and Ruzicka M., Weak and measure-valued solutions to evolutionary PDEs, Chapman & Hall, London, 1996.
• [31] Otto F., Initial-boundary value problem for a scalar conservation law. C. R. Acad. Sci. Paris Sér. I Math., 322(8) pp. 729-734, 1996.
• [32] Perrollaz V., Exact controllability of scalar conservation laws with an additional control in the context of entropy solutions. SIAM J. Control Optim., 50(4), pp. 2025-2045.
• [33] Perrollaz V., Asymptotic stabilization of entropy solutions to scalar conservation laws through a stationary feedback law, Annales de l’Institut Henri Poincaré (C), Non Linear Analysis, 30 no. 5(2013), pp. 879-915.
• [34] Perrollaz V., Asymptotic stabilization of stationnary shock waves using boundary feedback law, Arxiv preprint 1801.06335.
• [35] Vasseur A., Strong traces for solutions to multidimensional scalar conservation laws, Arch. Ration. Mech. Anal. 170 (2001) no. 3, pp. 181-193.