Abstract
This paper is concerned with the Neumann problem for a class of doubly nonlinear equations for the 1-Laplacian,
and initial data in \(L^1(\Omega )\), where \(\Omega \) is a bounded smooth domain in \({\mathbb {R}}^N\) and \(\gamma \) is a maximal monotone graph in \({\mathbb {R}}\times {\mathbb {R}}\). We prove that, under certain assumptions on the graph \(\gamma \), there is existence and uniqueness of solutions. Moreover, we proof that these solutions coincide with the ones of the Neumann problem for the total variational flow. We show that such assumptions are necessary.
Similar content being viewed by others
1 Introduction
Consider the doubly nonlinear diffusion problem:
completed with boundary conditions, being \(\Omega \) a bounded domain in \({\mathbb {R}}^N\), \(\gamma \) a maximal monotone graph (possibly multivalued) in \({\mathbb {R}}\times {\mathbb {R}}\) and \(\alpha :{\mathbb {R}}^N\rightarrow {\mathbb {R}}^N\). Typical examples are \(\alpha (\xi )=\alpha _p(\xi ):=|\xi |^{p-2}\xi \), \(p>1\), and \(\gamma (r)=|r|^{m-1}r\), \(m>0\). In these particular cases, for \(p=2\) and \(m=1\) the equation reduces to the classical heat equation, while for \(0<m < 1\) it is the porous medium equation (see, e.g., [26]) and the p -Laplacian diffusion equation for \( p>1\) and \(m=1\). In a general framework, case \(0< m<p-1\) is known as a doubly nonlinear equation with slow diffusion, while the case \(m>p-1\) is named a fast diffusion equation (see, e.g., [22]). Therefore, owing to the choice of \(\alpha \) and the graph \(\gamma \), this equation may arise a variety of different situations and it possess a wide spectrum of applications, for instance, in fluid dynamics, soil science and filtration, see [11] and [25]. Observe that, for \(p =2\), other typical examples are
for a Stefan type problem, or
for a Hele-Shaw-type problem.
From a mathematical point of view, there is an extensive literature related to problem (1.1). Existence, uniqueness, regularity and asymptotic behavior of solutions are treated under different restrictions on \(\gamma \) and \(\alpha \), and we refer some literature: [1, 2, 16, 21, 23, 24, 26] and the literature therein.
Our main aim is to deal with existence and uniqueness for the limit case \(p=1\) for the function \(\alpha _p\), that is, \(\displaystyle \alpha _1(\xi ):=\frac{\xi }{|\xi |}\), \(\gamma \) a maximal monotone graph and homogeneous Neumann boundary conditions. More precisely, by means of Crandall–Liggett’s theorem we obtain existence and uniqueness of entropy solution (see Definition 4.5) of the doubly nonlinear problem
under the condition
For this purpose, first of all we deal with the following elliptic problem
In Theorems 3.8 and 3.9, we prove the existence of solutions under the condition
and we prove uniqueness for continuous \(\gamma \) in Theorem 3.7, that is, under assumption (1.3). Note that (1.3) implies (1.4). Moreover, we see that for non-continuous maximal monotone graphs there is non-uniqueness (Example 3.6). We also show that condition \( \hbox {Rang}(\gamma )={\mathbb {R}}\) is necessary for the existence of solutions (Example 3.12).
Remark 1.1
On account of our approach to solve problem (1.2) and the above comments, condition (1.3) is natural for the study of such evolution problem. \(\square \)
In [5] (see also [6]), it was studied the well-posedness of the Neumann problem
by means of the Nonlinear Semigroup Theory. For that purpose, the following operator \({\mathcal {A}}\), defined in \(L^1(\Omega )\times L^1(\Omega )\), was introduced to give mathematical sense to the formal expression of \(\Delta _1v:=\hbox {div}\left( \frac{\nabla v}{|\nabla v|}\right) \) (jointly with the homogenous Neumann boundary conditions).
Definition 1.2
and
(see notation in Sects. 2.1 and 2.2).
Moreover, it was shown that \({\mathcal {A}}\) is the closure in \(L^1(\Omega )\times L^1(\Omega )\) of the subdifferential of the energy functional \(\Phi : L^2(\Omega ) \rightarrow (-\infty ,+\infty ]\) defined by
Since \(\Phi \) is a proper convex and lower semi-continuous function, then \(\partial \Phi \) is a maximal monotone operator with dense domain, generating a contraction semigroup in \(L^2(\Omega )\) that solves problem (1.5) for \(L^2\)-data. Entropy solutions for \(L^1\)-data \(v_0\) were introduced to characterize mild solutions of the abstract Cauchy problem
given by the Crandall–Liggett’s semigroup generation theorem ( [19]).
Remark 1.3
We show that the solutions of (1.2) are given by the solutions of (1.5) (Theorem 4.4). This is a non-trivial result; we first need to prove directly existence and uniqueness of solutions of problem (1.2). Observe that, at the level of elliptic problems, we first prove Theorem 3.9 and afterward we can prove Theorem 3.11.
The fact that solutions of (1.5) are solutions of (1.2) gives a kind of invariance property for the diffusion evolution problem via the 1-Laplacian, i.e., changing variables,
Observe that, written in this way, \(\gamma ^{-1}\) can be a non-continuous maximal monotone graph, hence not necessarily Lipschitz-continuous. When \(\gamma ^{-1}\) is an increasing and Lipschitz-continuous function, solutions of (1.5) are solutions of (1.2), see Proposition 3.10 at the level of the elliptic problems. \(\square \)
2 Preliminaries
2.1 Functions of bounded variation
We will denote by \({\mathcal {M}}(\Omega )\) the set of all Lebesgue measurable functions in \(\Omega \).
The natural energy space to study problem (1.2) is the space of functions of bounded variation. For further information concerning functions of bounded variation, we refer to [4] and [20]. Recall that if \(\Omega \) is an open subset of \({\mathbb {R}}^N\), a function \(u \in L^1(\Omega )\) whose gradient Du in the sense of distributions is a vector valued Radon measure with finite total variation in \(\Omega \) is called a function of bounded variation. The class of such functions will be denoted by \(BV(\Omega )\). The total variation of Du in \(\Omega \) is defined by the formula
The space \(BV(\Omega )\) is endowed with norm
Recall that an \({{\mathcal {L}}}^N\)-measurable subset E of \({\mathbb {R}}^N\) has finite perimeter if \({\mathrm{\chi }}_{E} \in BV({\mathbb {R}}^N)\). The perimeter of E is defined by \(\textrm{Per}(E) = \vert D {\mathrm{\chi }}_E \vert ({\mathbb {R}}^N)\).
2.2 A generalized Green’s formula
Let \(\Omega \) be an open bounded set in \({\mathbb {R}}^N\) with Lipschitz boundary. Following [10], for \(1 \le p \le \infty \) let
If \(\textbf{z}\in X_p(\Omega )\) and \(w \in BV(\Omega ) \cap L^{p'}(\Omega )\), we define the functional \((\textbf{z},Dw): C^{\infty }_{0}(\Omega ) \rightarrow {\mathbb {R}}\) by the formula
Then, \((\textbf{z},Dw)\) is a Radon measure in \(\Omega \),
and
for any Borel set \(B \subseteq \Omega \).
Moreover, when \(\textbf{z}\in X_p(\Omega )\) and \(w \in BV(\Omega ) \cap L^{p'}(\Omega )\), we have the following integration by parts formula
where \([\textbf{z}, \nu ]\) is the weak trace on \(\partial \Omega \) of the normal component of \(\textbf{z}\) (see [10]).
By (2.1), the measures \((\textbf{z},Du)\) and \(\vert (\textbf{z},Du) \vert \) are absolutely continuous with respect to the measure \(\vert Du\vert \) in \(\Omega \).
Thus, there is a density function
satisfying
The function \(\theta (\textbf{z}, Dw, \cdot )\) is called the Radon–Nikodým derivative of \((\textbf{z},Dw)\) with respect to |Dw|. Moreover, the following results hold.
Proposition 2.1
([10], Chain rule for \((\textbf{z},D(\cdot ))\)) Let \(\Omega \) be a bounded domain with a Lipschitz-continuous boundary \(\partial \Omega \), and, for \(1\le p\le N\) and \(p^{_{\prime }}\) its conjugate exponent, let \(\textbf{z}\in X_p(\Omega )\) and \(w \in BV(\Omega ) \cap L^{p^{\prime }}(\Omega )\). Then, for every Lipschitz-continuous, monotonically increasing function \(T: {\mathbb {R}}\rightarrow {\mathbb {R}}\), one has that
We shall denote
and \(\textrm{sign}^+(r):= (\textrm{sign}(r))^+\), and \(T_{k}(r):= [k - (k- \vert r \vert )^{+}]\textrm{sign}_0(r)\), \(k \ge 0\).
Remark 2.2
Let us point out that although \(T_k\) is only non-decreasing, we also have the following result
\(\square \)
2.3 Accretive operators and nonlinear semigroups
An operator A on X is a possibly nonlinear and multivalued mapping \(A: X\rightarrow 2^{X}\). It is standard to identify an operator A on X with its graph
and so, one sees A as a subset of \(X\times X\). The set \(D(A):=\{u\in X\,\vert \,Au\ne \emptyset \}\) is called the domain of A, and \(R(A):=\mathop {\bigcup }\limits _{u\in D(A)}Au\) the range of A.
Definition 2.3
An operator A on X is called m-accretive operator on X if A is accretive, that is, for every (u, v), \(({\hat{u}},{\hat{v}})\in A\) and every \(\lambda >0\),
and if for all \(\lambda >0\) the range condition
holds.
Note that A is accretive if the resolvent \(J_\lambda := (I + \lambda A)^{-1}\) are contractions for all \(\lambda >0\). The Yosida approximation of A is defined as
We have that
Moreover,
In the case that the Banach space is \(L^1(\Omega )\), with \(\Omega \subset {\mathbb {R}}^N\) an open set and norm
it is well known (see [14]) that
Definition 2.4
We say that an operator A on \(L^1(\Omega )\) is T-accretive if for every (u, v), \(({\hat{u}},{\hat{v}})\in A\) and every \(\lambda >0\),
If A is an m-accretive operator on a Banach space X, then by the classical existence theory (see, e.g., [14, Theorem 6.5], or [12, Corollary 4.2]), the first-order Cauchy problem
is well-posed for every \(u_{0}\in \overline{D(A)}^{_{X}}\), and \(g\in L^{1}(0,T;X)\) in the following mild sense.
Definition 2.5
For given \(u_{0}\in \overline{D(A)}^{_{X}}\) and \(g\in L^{1}(0,T;X)\), a function \(u\in C([0,T];X)\) is called a mild solution of Cauchy problem (2.3) if \(u(0)=u_{0}\) and for every \(\varepsilon >0\), there is a partition \( 0=t_{0}<t_{1}<\cdots < t_{N}=T\) and a step function
satisfying
and
In the case \(g =0\), the unique mild solution is given by the Crandall–Liggett’s exponential formula
Mild solutions are limits of step functions which are not necessarily differentiable in time. This leads to the notion of strong solution of the Cauchy problem (2.3).
Definition 2.6
For given \(u_{0}\in \overline{D(A)}^{_{X}}\) and \(g\in L^{1}(0,T;X)\), a function \(u\in C([0,T];X)\cap W^{1,1}_{\textrm{loc}}((0,T);X)\) is called a strong solution of the Cauchy problem (2.3) if \(u(0)=u_{0}\) and, for a.e. \(t\in (0,T)\), \(u(t)\in D(A)\) and \(Au(t)\ni g(t)-\frac{d u}{dt}(t)\).
We now recall a Bénilan–Crandall relation between functions \(u, v\in L^1(\Omega ,\nu )\). Denote by \(J_0\) and \(P_0\) the following sets of functions:
Assume that \(\nu (\Omega ) < +\infty \) and let \(u,v\in L^1(\Omega ,\nu )\). The following relation between u and v is defined in [15]:
Moreover, the following equivalences are proved in [15, Proposition 2.2]:
The following result is given in [15]
Proposition 2.7
Let \(\Omega \subset {\mathbb {R}}^N\) be an open bounded set.
-
(i)
For any \(u,v \in L^1(\Omega )\), if \(\int _\Omega u p(u) dx \le \int _\Omega v p(u) dx\) for all \(p \in P_0\), then \(u \ll v\).
-
(ii)
If \(u,v \in L^1(\Omega )\) and \(u \ll v\), then \(\Vert u \Vert _p \le \Vert v \Vert _p\) for all \(1 \le p \le \infty \).
-
(iii)
If \(v \in L^1(\Omega )\), then \(\{ u\in L^1(\Omega ) \, : \, u \ll v \}\) is a weakly compact subset of \(L^1(\Omega )\).
-
(iv)
If \(u_n, u \in L^1(\Omega )\) satisfy \(u_n \ll u\) and \(u_n \rightarrow u\) weakly in \(L^1(\Omega )\), then \(u_n \rightarrow u\) in \(L^1(\Omega )\).
Let \(\gamma \subset {\mathbb {R}}\times {\mathbb {R}}\) be a maximal monotone graph. We denote by \(\gamma ^0(r)\) the element of \(\gamma (r)\) of minimal absolute value. Then, for the Yosida approximations of \(\gamma \) we have that ([18, Proposition 2.6])
3 The elliptic problem
From [4, Theorem 2], given \(f \in L^1(\Omega )\) there exists a unique entropy solution v of the elliptic problem
defined as follows: \(v \in L^1(\Omega )\) with \(T_k(v) \in BV(\Omega )\) for all \(k >0\) and such that there exists \(\textbf{z}\in X_1(\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\),
and
Let \(\gamma \) be a maximal monotone graph in \({\mathbb {R}}\times {\mathbb {R}}\) with \(0\in \gamma (0)\). Following such concept, we give the following concept of entropy solution of the following elliptic problem
Definition 3.1
For \(f \in L^1(\Omega )\), we say that v is an entropy solution of problem \((S^\gamma _f)\) if \(v \in L^1(\Omega )\) and there exist \(u \in {\mathcal {M}}(\Omega )\) with \(T_k(u) \in BV(\Omega )\) for all \(k>0\) and \(\textbf{z}\in X_1(\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\) such that
For data in \(f \in L^\infty (\Omega )\), we also define the following concept of weak solution of problem \((S^\gamma _f)\).
Definition 3.2
For \(f \in L^\infty (\Omega )\), we say that v is a weak solution of problem \((S^\gamma _f)\) if \(v\in L^\infty (\Omega )\) and there exist \(u \in BV(\Omega )\cap L^\infty (\Omega )\) and \(\textbf{z}\in X_1(\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\) such that
We have that every weak solution is an entropy solution.
Working as in [6, Lemma 2.4], it is easy to see the two following results.
Lemma 3.3
For \(f \in L^\infty (\Omega )\), the following assertions are equivalent:
-
(a)
v is weak solution of problem \((S^\gamma _f)\);
-
(b)
there exist \(u \in BV(\Omega )\cap L^\infty (\Omega )\) and \({\textbf {z}}\in X_\infty (\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\) satisfying (3.1), (3.2) and
$$\begin{aligned} \begin{aligned} \int _{\Omega } (\varphi - u)(f- v) \, dx\le&{} \int _{\Omega } {\textbf {z}}\cdot \nabla \varphi \, dx - \int _{\Omega } \vert Du \vert , \\{}&{} \qquad \forall \varphi \in W^{1,1}(\Omega ) \cap L^\infty (\Omega ); \end{aligned} \end{aligned}$$(3.4) -
(c)
there exist \(u \in BV(\Omega )\cap L^\infty (\Omega )\) and \({\textbf {z}}\in X_\infty (\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\) satisfying (3.1), (3.2) and
$$\begin{aligned} \begin{aligned} \int _{\Omega } (\varphi - u)(f- v) \, dx\le&{} \int _{\Omega } ({\textbf {z}}, D\varphi ) - \int _{\Omega } \vert Du \vert , \\{}&{} \qquad \forall \varphi \in BV(\Omega )\cap L^\infty (\Omega ); \end{aligned} \end{aligned}$$(3.5) -
(d)
there exist \(u \in BV(\Omega )\cap L^\infty (\Omega )\) and \({\textbf {z}}\in X_\infty (\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\) satisfying (3.1), (3.2) and
$$\begin{aligned} \int _{\Omega } \varphi (f- v) \, dx = \int _{\Omega } (\textbf{z}, D\varphi ), \ \forall \varphi \in BV(\Omega )\cap L^\infty (\Omega ). \end{aligned}$$
Lemma 3.4
For \(f \in L^1(\Omega )\), the following assertions are equivalent:
-
(a)
v is an entropy solution of problem \((S^\gamma _f)\);
-
(b)
there exist \(u \in {\mathcal {M}}(\Omega )\) with \(T_k(u) \in BV(\Omega )\) for all \(k>0\) and \(\textbf{z}\in X_1(\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\) satisfying (3.1), (3.2) and
$$\begin{aligned} \begin{aligned} \int _{\Omega } (\varphi - T_k(u))(f- v) \, dx\le&{} \int _{\Omega } {\textbf {z}}\cdot \nabla \varphi \, dx - \int _{\Omega } \vert DT_k(u) \vert , \\{}&{} \qquad \forall \varphi \in W^{1,1}(\Omega ) \cap L^\infty (\Omega ); \end{aligned} \end{aligned}$$ -
(c)
there exist \(u \in {\mathcal {M}}(\Omega )\) with \(T_k(u) \in BV(\Omega )\) for all \(k>0\) and \(\textbf{z}\in X_1(\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\) satisfying (3.1), (3.2) and
$$\begin{aligned} \begin{aligned} \int _{\Omega } (\varphi - T_k(u))(f- v) \, dx\le&{} \int _{\Omega } ({\textbf {z}}, D\varphi ) - \int _{\Omega } \vert DT_k(u) \vert , \\{}&{} \qquad \forall \varphi \in L^\infty (\Omega ) \cap BV(\Omega ); \end{aligned} \end{aligned}$$(3.6)
As can be verified in the above lemma, the notion of entropy solution for the 1-Laplacian defined here is analogous to the concept of entropy solution for the p-Laplacian (\(1<p<N\)) introduced in the pioneering paper [13].
Remark 3.5
Let v be an entropy solution of problem \((S^\gamma _f)\). Then, there exist \(u \in {\mathcal {M}}(\Omega )\) with \(T_k(u) \in BV(\Omega )\) for all \(k>0\) and \(\textbf{z}\in X_1(\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\) such that \( v\in \gamma (u) \) and (3.6) holds true. Then, if we take \(T_k(u) \pm 1\) as test function in (3.6), it follows that
Therefore, denoting
the following condition must be satisfied
Therefore, in the case \(\hbox {Rang}(\gamma )={\mathbb {R}}\) this is always true for any \(f\in L^1(\Omega )\). \(\square \)
It is worthy to mention that if \(\gamma \) is a multivalued maximal monotone graph, the corresponding problem \((S^\gamma _f)\) has more than one weak solution, as we show in the next example.
Example 3.6
Let \(\gamma \) be a multivalued graph such that
Consider \(\Omega := ]-1,1[\) and \(f(x):= \frac{1}{2}\) for all \(x \in ]-1,1[\). We define \(\textbf{z}: ]-1,1[ \rightarrow {\mathbb {R}}\) as
Then, \(\Vert \textbf{z}\Vert _\infty \le 1\), \([\textbf{z}, \nu ]=0\) and
Clearly, \(v \in \gamma (0)\). Therefore, v is a weak solution of problem \((S^\gamma _f)\). Now, taking \(\textbf{z}=0\), it follows that f is also a weak solution of problem \((S^\gamma _f)\). In particular, for the Stefan type problem, there is not uniqueness of weak solution of problem \((S^\gamma _f)\).
Due to the above example, we need to impose some restriction to the maximal monotone graph \(\gamma \) in order to get uniqueness of entropy solution of problem \((S^\gamma _{f_i})\). In this direction, we have the following result for graphs satisfying (1.3) without the range condition.
Theorem 3.7
Assume that \(\gamma : D(\gamma ) \subset {\mathbb {R}}\rightarrow {\mathbb {R}}\) is a non-decreasing continuous function with \(\gamma (0)=0\). Given \(f_i \in L^1(\Omega )\) and \( v_i\) entropy solutions of \((S^\gamma _{f_i})\), for \(i=1,2\), then
In particular,
Proof
For \(i=1,2\), we have that there exists \(u_i \in L^1(\Omega )\) with \(T_k(u_i) \in BV(\Omega )\) for all \(k>0\) and there exists \(\textbf{z}_i \in X_1(\Omega )\) with \(\Vert \textbf{z}_i \Vert _\infty \le 1\) such that \( v_i=\gamma (u_i) \) and
Let \(p_\epsilon \) be a smooth strictly monotone approximation of the sign function. Then, applying integration by parts formula (2.2), we have
Thus,
Now, from (3.9) and (2.1), we have
This implies that
Since, according to Proposition 2.1, we have
\(\vert D(T_k(u_1) - T_k(u_2))\vert \hbox {-a.e.}\) and \(\vert Dp_\epsilon (T_k(u_1) - T_k(u_2)) \vert \hbox {-a.e.}\), we get
Therefore,
Taking limit as \(k \rightarrow \infty \), we get
Taking now limit as \(\epsilon \rightarrow 0^+\), we have that there exists \(\xi (x) \in \textrm{sign}((u_1(x) - u_2(x))\) \({\mathcal {L}}^N\)-a.e. \(x \in \Omega \) such that
Now, since \(v_i = \gamma (u_i)\), \(i=1,2\), and \(\gamma \) is non-decreasing and \(\gamma (0)=0\), we have \(\xi (x) \in \textrm{sign}((v_1(x) - v_2(x))\), if \(v_1(x) \not = v_2(x)\). Hence, since \(\gamma \) is continuous, which for a maximal monotone graph is equivalent to say that \(\gamma (r)\) is always univalued for any \(r\in D(\gamma )\),
and (3.8) holds.
The proof of (3.7) is similar but using a smooth monotone approximation of the \(\textrm{sign}^+\). \(\square \)
Let us now prove existence for problem \((S^\gamma _f)\) for graphs satisfying condition (1.4).
Theorem 3.8
Let \(\gamma \) be a graph satisfying (1.4) and \(f \in C^\infty _c(\Omega )\). Then, there exists \(v_f\) weak solution of problem \((S^\gamma _f)\) with \(v_f\ll f.\)
Moreover, for any \({{\tilde{f}}}\in C^\infty _c(\Omega )\), it follows
for the weak solutions constructed here.
Proof
Given \(f \in C^\infty _c(\Omega )\), we must find \(v \in L^\infty (\Omega )\) and \(u \in BV(\Omega )\cap L^\infty (\Omega )\) with
and \( \textbf{z}\in X_1(\Omega ),\) with \(\Vert \textbf{z}\Vert _\infty \le 1\), satisfying (3.2) and (3.4).
By [8, Theorem 3.9 (i)], for any \(p>1\), there exist \(u_p \in W^{1,p}(\Omega )\) and \(v_p \in \gamma (u_p) \in L^1(\Omega )\) such that
for all \(\varphi \in W^{1,p}(\Omega ) \cap L^\infty (\Omega )\). Moreover,
and, since \(\hbox {Rang}(\gamma )={\mathbb {R}}\),
Taking \(\varphi =u_p\in W^{1,p}(\Omega )\cap L^\infty (\Omega )\) as a test function and taking into account that \(u_pv_p\ge 0\) it follows that
Therefore, by Hölder inequality we get
Thus, there exists \(u\in BV(\Omega )\cap L^\infty (\Omega )\) such that, up to a subsequence (no relabeled),
Moreover, inequality (3.12) allows to establish the following statements (see [5]): There exists a bounded vector field \({\textbf{z}}\in L^\infty (\Omega ;{\mathbb {R}}^N)\) with \(|\vert {\textbf{z}}|\vert _\infty \le 1\) such that
as \(p\rightarrow 1^+\). In particular,
On the other hand, by (3.11) we obtain that
being \(v\ll f\). This result, in addition to (3.13), implies that
In order to show that v is a weak solution of problem \((S^\gamma _f)\), for each \(\varphi \in W^{1,1}(\Omega )\cap L^\infty (\Omega )\) we consider the sequence \(\{ \varphi _n \}\subset {\mathcal {C}}^\infty ({\bar{\Omega }})\) such that \(\varphi _n \rightarrow \varphi \) in \(W^{1,1}(\Omega )\). Now, taking \(\varphi _n-u_p\) as a test function in (3.10) and taking limits it follows
and by (3.14)
In addition, using Young’s inequality and the weak lower semi-continuity of the total variation, we obtain
Therefore, expression (3.16) yields
Finally, taking limits as \(n\rightarrow \infty \) we obtain inequality (b) from Lemma 3.3, which means that v is a weak solution of problem \((S^\gamma _f)\).
The second part is a consequence of [8, Theorem 3.9 (ii)] and the above construction. \(\square \)
Theorem 3.9
Assume that \(\gamma \) satisfies condition (1.4). Then, for any \(f \in L^1(\Omega )\) there exists an entropy solution of problem \((S^\gamma _f)\).
Proof
Given \(f \in L^1(\Omega )\), let \(f_n\in {\mathcal {C}}_c^\infty (\Omega )\) be such that \(f_n\rightarrow f\) in \(L^1(\Omega )\). For any \(n \in {{\mathbb {N}}}\), by Theorem 3.8 there exists a weak solution \(v_n\) of problem \((S^\gamma _{f_n})\) such that \(v_n \ll f_n\). Thus, there exists \(u_n \in BV(\Omega )\cap L^\infty (\Omega )\) and there exists \(\textbf{z}_n \in X_1\) with \(\Vert \textbf{z}_n \Vert _\infty \le 1\) such that
and
Taking \(\varphi =u_n-T_k(u_n)\) in (3.18), we have
Then, by (2.1) and since \(v_nT_k(u_n)\ge 0\), we get
Then, by the compact embedding, taking subsequences and using a diagonal process, we have
with
Let us see now that (remark that this argument is not needed if \([0,+\infty [\subset D(\gamma )\), similarly for the argument with the negative part)
In fact, since \(\gamma ^{0}\) is lower semi-continuous and \(\hbox {Rang}(\gamma ^{0})={\mathbb {R}}\), by applying Fatou’s lemma it follows that
Similarly, it is shown that
By (3.19) and (3.20), if we define
we have that u is measurable and
Now, by using the second part in Theorem 3.8, we get
Therefore,
and
On the other hand, since \(\textbf{z}_n \in X_1(\Omega )\) with \(\Vert \textbf{z}_n \Vert _\infty \le 1\), we may assume that
Then, from (3.17),
Given now \(\varphi \in W^{1,1}(\Omega ) \cap L^\infty (\Omega )\) and taking \(\varphi + u_n -T_k(u_n)\) as test function in (3.18), we obtain
Thus, applying [5, Lemma 3], we arrive to
Then, taking limit as \(n \rightarrow \infty \) and having in mind (3.21), (3.22) and the lower semi-continuity of the total variation, we get
\(\square \)
Let us now prove that, under assumption (1.3), the unique solution of problem \((S^\gamma _f)\) coincides by the unique solution of \((S^{Id}_f)\). Let us first see an easy situation.
Proposition 3.10
Let \(\gamma ^{-1}\) be an increasing and Lipschitz-continuous function with \(\gamma (0)=0\) and \(\hbox {Rang}(\gamma )={\mathbb {R}}\). Let \(v\in BV(\Omega )\cap L^\infty (\Omega )\) be the unique weak solution of \((S^{Id}_f)\) for \(f\in L^\infty (\Omega )\). Then, v is also a weak solution of problem \((S^{\gamma }_f)\).
Proof
By setting \(u:= \gamma ^{-1}(v)\) (which is well defined since \(\hbox {Rang}(\gamma )={\mathbb {R}}\)), we have
and
(see [4, Theorems 3.101 and 3.99]). Now, by Proposition 2.1 and (3.3) it follows
consequently,
Therefore, v is a weak solution of problem \((S^\gamma _f)\). \(\square \)
Theorem 3.11
Under condition (1.3), the entropy solution of problem \((S^\gamma _f)\) is given by the entropy solution of problem \((S^{Id}_f)\), i.e., of problem
Proof
By Theorem 3.7, it is enough to prove it for data \(f\in C_c^\infty (\Omega )\). So, our aim is to see that the (weak) solution v to
is the (weak) solution of
Let \({\tilde{\gamma }}_n(r)=\gamma _{1/n}(r)+\frac{1}{n}r\), where \(\gamma _{1/n}\) is the Yosida approximation of \(\gamma \) (and where \(\frac{1}{n}r\) can be deleted if the graph of \(\gamma \) has not flat zones). Then, \({\tilde{\gamma }}_n(r)\) is a Lipschitz-continuous and increasing function also satisfying \(\hbox {Rang}({\tilde{\gamma }}_n)={\mathbb {R}}\). Therefore, from Theorem 3.8, there exist \(v_n \in L^\infty (\Omega )\), \(u_n \in BV(\Omega )\cap L^\infty (\Omega )\) and \(\textbf{z}_n \in X_1(\Omega )\) with \(\Vert \textbf{z}_n \Vert _\infty \le 1\) such that
Moreover by (3.5), we have
Since \(v_n:= {\tilde{\gamma }}_n(u_n)\), we have
and
By (3.26) and Proposition 2.1, we have
consequently,
and, therefore, we get that \(v_n\) is a (weak) solution of problem (3.23) with vector field \(\textbf{z}_n\). Therefore, by uniqueness of problem (3.23), we have
And, by (3.27), we have
Now, since \({\tilde{\gamma }}_n(u_n)=v \ll f\), we get
for \(q\in [1,\infty ]\). In particular, \(\Vert u_n\Vert _\infty \le C_1:= \max \{-\gamma _1^{-1}(-\Vert f\Vert _\infty ), \gamma _1^{-1}(\Vert f\Vert _\infty ) \}\), for all \(n\in {\mathbb {N}}\).
Finally, taking \(\varphi =0\) as a test function in (3.28) we obtain
so that \(\{u_n\}\) is bounded in \(BV(\Omega )\). It follows that there exists \(u\in BV(\Omega )\) such that up to a subsequence (no relabeled)
and
This implies that \(v=\gamma (u)\).
On the other hand, since \(\textbf{z}_n \in X_1(\Omega )\) with \(\Vert \textbf{z}_n \Vert _\infty \le 1\), we may assume that
In particular, from (3.25)
Then, by [6, Proposition C.12] and having in mind the lower semi-continuity of the total variation, taking limits in (3.28) as \(n \rightarrow \infty \), we get
Therefore, by Lemma 3.3, we have that v is a solution of (3.24). \(\square \)
In the next example will be see that the condition \(\text{ Rang }(\gamma ) = {\mathbb {R}}\) is necessary in the above theorem.
Example 3.12
Let, for \(n\in {\mathbb {N}}\), \({\beta _n}(r)=\frac{1}{n}\arctan (r)\), and let \(f\in L^1(\Omega )\), \(\int _\Omega f=0\). If \((S^{{\beta _n}}_f)\) has a solution for all \(n\in {\mathbb {N}}\), then there exists \(u_n \in {\mathcal {M}}(\Omega )\), \(T_k(u_n) \in BV(\Omega )\) for all \(k>0\), \(\beta _n(u_n)\in L^1(\Omega )\), and there exists \(\textbf{z}_n \in X_1(\Omega )\) with \(\Vert \textbf{z}_n \Vert _\infty \le 1\), such that
Now, since
taking limits in (3.29) we find \(\textbf{z}\in X_1(\Omega ),\) \(\Vert \textbf{z}\Vert _{\infty }\le 1,\) such that
and we get a contradiction with the well-known fact that there exist \(f\in L^1(\Omega )\), \(\int _\Omega f=0\), such that the above equation has not solution in \( X_1(\Omega )\) (see, for instance, [17]). Nevertheless, let us see, with an easy example, that there are \(L^\infty \)-functions for which (3.30) has no solution in \( X_1(\Omega )\) with \(\Vert \textbf{z}\Vert _{\infty }\le 1\). In fact, (3.30) implies that
Take \(\Omega =B_1(0)\) the ball in \({\mathbb {R}}^N\) centered at 0 of radius 1, and, for \(k>0\),
which satisfies \(\int _{\Omega } f=0\). Take now \(\varphi (x)=1-|x|\), which belongs to \(W^{1,1}_0(\Omega )\). Then, on the one hand,
and, on the other hand, since \(\int _{\Omega } f=0\),
which, for \(k>2(N+1)(2^N-1)\), contradicts (3.31). \(\square \)
4 The evolution problem
In this section, we study the evolution problem (1.2).
We do this through the Nonlinear Semigroup Theory, and therefore, we introduce an operator \({\mathcal {B}}\) in \(L^1(\Omega )\) that allows to rewrite problem (1.2) as the abstract Cauchy problem
Definition 4.1
\((v,w) \in {\mathcal {B}}\) if and only if \(v,w \in L^1(\Omega )\) and there exist \(u \in {\mathcal {M}}(\Omega )\) such that \(T_k(u) \in BV(\Omega )\) for all \(k >0\) and \(v \in \gamma (u)\) a.e. in \(\Omega \), and there exists \(\textbf{z}\in X_1(\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\), satisfying:
and
Note that for \(f \in L^1(\Omega )\), we have that
By Theorems 3.7 and 3.9, we have the following result.
Theorem 4.2
Under condition (1.3), \({\mathcal {B}}\) is a T-accretive and m-accretive operator on \(L^1(\Omega )\).
As a consequence of the above result, by Crandall–Liggett’s theorem, it follows that, for every initial data \(v_0 \in \overline{D({\mathcal {B}})}^{L^1(\Omega )}\), the abstract Cauchy problem (4.1) has a unique mild solution v(t) given by the exponential formula
In [5] (see also [6]), it is shown that the operator \({\mathcal {A}}\) given in Definition 1.2 is an m-completely accretive operator in \(L^1(\Omega )\) and that for every initial data \(v_0 \in L^1(\Omega )\) the mild solution
is a strong solution.
Theorem 4.3
The following equality holds
Proof
It is enough to prove that \(C_c^\infty (\Omega )\subset \overline{D({\mathcal {B}})}^{L^1(\Omega )}\). So, take \(f\in C_c^\infty (\Omega )\) and take \(f_n=\left( I+\frac{1}{n}{\mathcal {B}}\right) ^{-1}f\). Observe that \(f_n\in D({\mathcal {B}})\) and it is a (weak) solution of
Therefore, by Theorem 3.8 and Lemma 3.3 (d), \(f_n\in L^\infty (\Omega )\), \(f_n\ll f\), and there exist \(u_n \in BV(\Omega )\cap L^\infty (\Omega )\) and \(\textbf{z}_n \in X_1(\Omega )\) with \(\Vert \textbf{z}\Vert _\infty \le 1\) satisfying (3.1), (3.2) and
Now, for \(\varphi \in BV(\Omega )\cap L^\infty (\Omega )\), since
we have that
Now, since \(f_n\ll f\), by (iv) in Proposition 2.7 we get
\(\square \)
Then, as a consequence of Theorem 3.11, we obtain the following result:
Theorem 4.4
Under condition (1.3) and for every initial data \(v_0 \in L^1(\Omega )\), the abstract Cauchy problem (4.1) has a unique strong solution v(t). Moreover, this solution coincides with the unique strong solution of problem (1.6).
We introduce the following concept of solution of problem (1.2).
Definition 4.5
A measurable function \(v: (0,T)\times \Omega \rightarrow {\mathbb {R}}\) is an entropy solution of (1.2) in \((0,T)\times \Omega \) if \(v \in C([0,T],L^1(\Omega )) \cap W^{1,1}_{loc}(0,T; L^1(\Omega ))\), \(v(0)=v_0\), and, for almost all \(t \in (0, T)\), there exists \(u(t)\in {\mathcal {M}}(\Omega )\) with \(T_{k}(u(t)) \in BV(\Omega )\) for all \(k > 0\), and there exists \(\textbf{z}(t) \in L^{\infty }(\Omega )\) with \(\Vert \textbf{z}(t) \Vert _{\infty } \le 1\), such that
and
for every \(w \in W^{1,1}(\Omega ) \cap L^{\infty }(\Omega )\).
As a consequence of the above result, we have the following existence and uniqueness result.
Theorem 4.6
Under condition (1.3) and for every initial data \(v_0 \in L^1(\Omega )\), there exists a unique entropy solution of (1.2) in \((0,T)\times \Omega \) for every \(T >0\) such that \(v(0) = v_0\). Moreover, if v(t) and \({\hat{v}}(t)\) are entropy solutions corresponding to initial data \(v_0\) and \({\hat{v}}_0\), respectively, then
In particular,
Data availability statement
This manuscript has no associated data.
References
G. Akagi and U. Stefanelli, A variational principle for doubly nonlinear evolution, Applied Mathematics Letters 23 (2010), 1120–1124.
G. Akagi and G. Schimperna, A Subdifferential calculus and doubly nonlinear evolutions equations in Lp-spaces with variable exponents, Journal of Functional Analysis 267 (2014), 173–213.
L. Alvarez, P.L. Lions and J.M. Morel, Image selective smoothing and edge detection by nonlinear diffusion, SIAM J. Numer. Anal. 29 (1992), 845–866.
L. Ambrosio, N. Fusco and D. Pallara, Functions of Bounded Variation and Free Discontinuity Problems. Oxford Mathematical Monographs, 2000.
F. Andreu, C. Ballester, Caselles and J.M. Mazón, Minimizing total variation flow, Differential Integral Equations 14, no 3 (2001), 321–360.
F. Andreu, V. Caselles, and J.M. Mazón, Parabolic Quasilinear Equations Minimizing Linear Growth Functionals. Progress in Mathematics, vol. 223, 2004. Birkhauser.
F. Andreu, N. Igbida, J.M. Mazón and J. Toledo A degenerate elliptic-parabolic problem with nonlinear dynamical boundary conditions. Interfaces and Free Boundaries 8 (2006), 447–479.
F. Andreu, N. Igbida, J.M. Mazón and J. Toledo \(L^1\)existence and uniqueness results for quasi-linear elliptic equations with nonlinear boundary conditions. Ann. Inst. Poincaré. Analyse Non Linéaire 24 (2007), 61–89.
F. Andreu, N. Igbida, J.M. Mazón and J. Toledo Degenerate elliptic equations with nonlinear boundary conditions and measures datas. Ann. Scuola Norm. Sup. Pisa Cl. Sci. Vol. VIII (2009), 767–803.
G. Anzellotti, Pairings between measures and bounded functions and compensated compactness, Ann. Mat. Pura Appl. (4) 135 (1983), 293–318.
D.G. Aronson, Regularity properties of flows through porous media: The interface, Arch. Rational Mech. Anal. 37 (1970), 1–10.
V. Barbu, Nonlinear differential equations of monotone types in Banach spaces, Springer Monographs in Mathematics, Springer, New York, 2010.
Ph. Bénilan, L. Boccardo, Th. Gallouët, R. Gariepy, M. Pierre and J. L. Vázquez, An\(L^1\)-theory of existence and uniqueness of solutions of nonlinear elliptic equations, Ann. Sc. Norm. Super. Pisa, Cl. Sci., 22 (1995), 241–273.
P. Bénilan, M. G. Crandall and A. Pazy, Evolution problems governed by accretive operators, book in preparation, 1994.
P. Bénilan and M. G. Crandall, Completely accretive operators, in Semigroup theory and evolution equations (Delft, 1989), vol. 135 of Lecture Notes in Pure and Appl. Math., Dekker, New York, 1991, pp. 41–75.
V. Bögelein, F. Duzaar, P. Marcellini and Ch. Schev, Doubly Nonlinear Equations of Porous Medium Type. Arch. Rational Mech. Anal. 229 (2018), 503–545.
J. Bourgain and H. Brezis. On the equation div Y = f and application to control of phases. J. Amer. Math. Soc. 16 (2002), 393–426.
H. Brézis, Opérateurs maximaux monotones et semi-groupes de contractions dans les espaces de Hilbert, North-Holland Publishing Co., Amsterdam-London; American Elsevier Publishing Co., Inc., New York, 1973. North-Holland Mathematics Studies, No. 5. Notas de Matemática (50).
M. G. Crandall and T. M. Liggett, Generation of Semigroups of Nonlinear Transformations on General Banach Spaces, Amer. J. Math. 93 (1971), 265–298.
L.C. Evans and R.F. Gariepy, Measure Theory and Fine Properties of Functions, Studies in Advanced Math., CRC Press, 1992.
O. Grange and F. Mignot, Sur la Résolution d’une Équation et d’une Inequation Paraboliques non Linéaires, Journal of Functional Analysis 11 (1972), 77–92.
A.V. Ivanov, Regularity for doubly nonlinear parabolic equations, J. Math. Sci. 83(1) (1997), 22–37.
A. Mielke, R. Rossi and G. Savaré Nonsmooth analysis of doubly nonlinear evolutions equations, Calculus of Variations 46 (2013), 253–310.
Z. Peng Existence and regularity results for doubly nonlinear inclusions with nonmonotone perturbation, Nonlinear Analysis 115 (2015), 71–88.
R. Showalter and N.J. Walkington, Diffusion of fluid in a fissured medium with microstructure, SIAM J. Math. Anal. 22(6) (1991), 1702–1722.
J. L. Vázquez, The porous medium equation. Mathematical theory. Oxford Mathematical Monographs. The Clarendon Press, Oxford University Press, Oxford, 2007. xxii+624 pp.
W. P. Ziemer, Weakly Differentiable Functions, GTM 120, Springer–Verlag, 1989.
Acknowledgements
The first and third authors have been partially supported by Conselleria d’Innovació, Universitats, Ciència i Societat Digital, project AICO/2021/223, and by Ministerio de Ciencia e Innovación (Spain), project PID2022-136589NB-I00. The second author is supported by Grant PID2021-122122NB-I00 funded by MCIN/AEI/ 10.13039/501100011033 and by “ERDF A way of making Europe,” by the European Union-Next Generation EU (ref. RR-B-2021-03), by Junta de Andalucía FQM-116, by UAL2020-FQM-B2046 (UAL/CTEICU/FEDER), by Junta de Andalucía, Consejería de Transformación Económica, Industria, Conocimiento y Universidades-Unión Europea grant P18-FR-667, and by Ministerio de Ciencia, Innovación y Universidades (MCIU), Agencia Estatal de Investigación (AEI), (FEDER) Fondo Europeo de Desarrollo Regional under Research Project PGC2018-096422-B-I00.
Funding
Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Mazón, J.M., Molino, A. & Toledo, J. Doubly nonlinear equations for the 1-Laplacian. J. Evol. Equ. 23, 67 (2023). https://doi.org/10.1007/s00028-023-00917-8
Accepted:
Published:
DOI: https://doi.org/10.1007/s00028-023-00917-8