Abstract
We study the existence of solutions for some nonlinear elliptic problems of the type \(-{\text {div}}(b(x, u, \nabla u)+F(x,u))=\nu\) in \(\Omega ,\) in the setting of Musielak–Orlicz spaces. The lower order term F verifies the natural growth condition, no \(\Delta _{2}\)-condition is assumed on the Musielak function, and the datum \(\nu\) is assumed to belong to \(L^{1}(\Omega )+W^{-1} E_{\psi }(\Omega )\).
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction and basic assumptions
In this note we will prove an existence of a renormalized solutions for the following nonlinear boundary value problem :
where \(\Omega\) is a bounded domain of \({\mathbb {R}}^{N}, N \ge 2,\ B(u)=-{\text {div}}(b(x, u, \nabla u))\) is a Leray-Lions operator defined from the space \(W_{0}^{1} L_{\varphi }(\Omega )\) into its dual \(W^{-1} L_{\overline{\varphi }}(\Omega ),\) with \(\varphi\) and \(\overline{\varphi }\) are two complementary Musielak-Orlicz functions and where b is a function satisfying the following conditions:
There exist two Musielak-Orlicz functions \(\varphi\) and P such that \(P \prec \prec \varphi ,\) a positive function \(d(x) \in E_{\overline{\varphi }}(\Omega ),\ \alpha >0\) and \(k_{i}>0\) for \(i=1,\cdots ,4\), such that for a.e. \(x \in \Omega\) and all \(s \in {\mathbb {R}}\) and all \(\xi ,\ \xi ' \in {\mathbb {R}}^{N},\ \xi \not =\xi '\):
The lower order term F is a Carathéodory function satisfying, for a.e. \(x \in \Omega\) and for all \(s \in {\mathbb {R}},\) the following condition:
where \(c(.) \in L^{\infty }(\Omega )\) such that
The right hand side of (1.1) is assumed to satisfy
In the usual Sobolev spaces, the concept of renormalized solutions was introduced by Diperna and Lions in [22] for the study of the Boltzmann equations, this notion of solutions was then adapted to the study of the problem (1.1) by Boccardo et al. in [21] when the right hand side is in \(W^{-1, p^{\prime }}(\Omega )\) and in the case where the nonlinearity g depends only on x and u, this work was then studied by Rakotoson in [31] when the right hand side is in \(L^{1}(\Omega ),\) and finally by DalMaso et al. in [23] for the case in which the right hand side is general measure data.
On Orlicz-Sobolev spaces and in variational case, Benkirane and Bennouna have studied in [8] the problem (1.1) where \(\Phi (x,u)\equiv \Phi (u),\) and the nonlinearity g depends only on x and u under the restriction that the N-function satisfies the \(\Delta _{2}\)-condition, this work was then extended in [4] by Aharouch, Bennouna and Touzani for N-function not satisfying necessarily the \(\Delta _{2}\)-condition and \(\Phi (x,u)\equiv \Phi (u)\). If g depends also on \(\nabla u,\) the problem (1.1) has been solved by Aissaoui Fqayeh, Benkirane, El Moumni and Youssfi in [5] where \(\Phi (x,u)\equiv \Phi (u)\), and without assuming the \(\Delta _{2}\)-condition on the N-function.
In the framework of variable exponent Sobolev spaces, Bendahmane and Wittbold have treated in [7] the nonlinear elliptic equation (1.1) where \(a(x,u,\nabla u)=|\nabla u|^{p(x)-2} \nabla u,\ \Phi \equiv 0,\ g\equiv 0\) and where \(f\in L^1(\Omega )\), they proved the existence and uniqueness of a renormalized solution in Sobolev space with variable exponents \(W_{0}^{1, p(x)}(\Omega ).\)
In the variational case of Musielak-Orlicz spaces and in the case where \(g \equiv 0\) and \(\Phi \equiv 0,\) an existence result for (1.1) has been proved by Benkirane and Sidi El Vally in [10] a when the non-linearity g depends only on x and u. If g depends also on \(\nabla u,\) the problem (1.1) has recently been solved by N. El Amarty, B. El Haji and M. El Moumni in [18] where \(\Phi (x,u)\equiv \Phi (u).\)
and several researches deals with the existence solutions of elliptic and parabolic problems under various assumptions and in different contexts (see [6, 11,12,13,14,15,16, 18,19,20] for more details).
The paper is organized as follows: In Sect. 2, we give some preliminaries and background. Section 3 is devoted to some technical lemmas which can be used to our result. In Sect. 4, we state our main result and in Sect. 5 we give the proof of an existence solution .
2 Some preliminaries and background
Here we give some definitions and properties that concern Musielak-Orlicz spaces (see [17]). Let \(\Omega\) be an open subset of \({\mathbb {R}}^{N}\), a Musielak-Orlicz function \(\varphi\) is a real-valued function defined in \(\Omega \times {\mathbb {R}}^{+}\) such that
a) \(\varphi (x, .)\) is an N-function for all \(x\in \Omega\) (i.e. convex, nondecreasing, continuous, \(\varphi (x, 0)=0,\ \varphi (x, t)>0\) for all \(t>0\) and \(\displaystyle \lim _{t \rightarrow 0} \sup _{x \in \Omega } \frac{\varphi (x, t)}{t}=0\) and \(\displaystyle \lim _{t \rightarrow \infty } \inf _{x \in \Omega } \frac{\varphi (x, t)}{t}=\infty\)).
b) \(\varphi (., t)\) is a measurable function for all \(t \ge 0\).
For a Musielak–Orlicz function \(\varphi\), let \(\varphi _{x}(t)=\varphi (x, t)\) and let \(\varphi _{x}^{-1}\) be the nonnegative reciprocal function with respect to t, i.e. the function that satisfies
The Musielak–Orlicz function \(\varphi\) is said to satisfy the \(\Delta _{2}\) -condition if for some \(k>0,\) and a nonnegative function h, integrable in \(\Omega ,\) we have
When (2.1) holds only for \(t \ge t_{0}>0,\) then \(\varphi\) is said to satisfy the \(\Delta _{2}\)-condition near infinity. Let \(\varphi\) and \(\gamma\) be two Musielak–Orlicz functions, we say that \(\varphi\) dominate \(\gamma\) and we write \(\gamma \prec \varphi ,\) near infinity (resp. globally) if there exist two positive constants c and \(t_{0}\) such that for a.e. \(x \in \Omega :\)
We say that \(\gamma\) grows essentially less rapidly than \(\varphi\) at 0 (resp. near infinity) and we write \(\gamma \prec \prec \varphi\) if for every positive constant c we have
Remark 1
(see [33]) If \(\gamma \prec \prec \varphi\) near infinity, then \(\forall \varepsilon >0\) there exists a nonnegative integrable function h, such that
For a Musielak-Orlicz function \(\varphi\) and a measurable function \(u: \Omega \longrightarrow {\mathbb {R}},\) we define the functional
The set \(K_{\varphi }(\Omega )=\Big \{u: \Omega \longrightarrow {\mathbb {R}}\ \text{ measurable/ }\ \rho _{\varphi , \Omega }(u)<\infty \Big \}\) is called the Musielak-Orlicz class (or generalized Orlicz class). The Musielak-Orlicz space (the generalized Orlicz spaces) \(L_{\varphi }(\Omega )\) is the vector space generated by \(K_{\varphi }(\Omega ),\) that is, \(L_{\varphi }(\Omega )\) is the smallest linear space containing the set \(K_{\varphi }(\Omega ) .\) Equivalently
For a Musielak-Orlicz function \(\varphi\) we put:
Note that \(\overline{\varphi }\) is the Musielak-Orlicz function complementary to \(\varphi\) (or conjugate of \(\varphi\)) in the sense of Young with respect to the variable s. In the space \(L_{\varphi }(\Omega )\) we define the following two norms:
which is called the Luxemburg norm and the so-called Orlicz norm by:
where \(\overline{\varphi }\) is the Musielak-Orlicz function complementary to \(\varphi .\) These two norms are equivalent (see [17]). The closure in \(L_{\varphi }(\Omega )\) of the bounded measurable functions with compact support in \(\overline{\Omega }\) is denoted by \(E_{\varphi }(\Omega )\), It is a separable space (see [17], Theorem 7.10).
We say that sequence of functions \(u_{n} \in L_{\varphi }(\Omega )\) is modular convergent to \(u \in\) \(L_{\varphi }(\Omega )\) if there exists a constant \(\lambda >0\) such that
For any fixed nonnegative integer m we define
and
where \(\alpha =\left( \alpha _{1}, \ldots , \alpha _{n}\right)\) with nonnegative integers \(\alpha _{i},|\alpha |=\left| \alpha _{1}\right| +\ldots +\left| \alpha _{n}\right|\) and \(D^{\alpha } u\) denote the distributional derivatives. The space \(W^{m} L_{\varphi }(\Omega )\) is called the Musielak-Orlicz Sobolev space. Let for \(u \in W^{m} L_{\varphi }(\Omega ):\)
these functionals are a convex modular and a norm on \(W^{m} L_{\varphi }(\Omega ),\) respectively, and the pair \(\left( W^{m} L_{\varphi }(\Omega ),\Vert .\Vert _{\varphi , \Omega }^{m}\right)\) is a Banach space if \(\varphi\) satisfies the following condition (see [17]):
The space \(W^{m} L_{\varphi }(\Omega )\) will always be identified to a subspace of the product \(\displaystyle \prod _{|\alpha | \le m} L_{\varphi }(\Omega )=\Pi L_{\varphi },\) this subspace is \(\sigma \left( \Pi L_{\varphi }, \Pi E_{\overline{\varphi }}\right)\) closed.
The space \(W_{0}^{m} L_{\varphi }(\Omega )\) is defined as the \(\sigma \left( \Pi L_{\varphi }, \Pi E_{\overline{\varphi }}\right)\) closure of \({\mathcal {D}}(\Omega )\) in \(W^{m} L_{\varphi }(\Omega ),\) and the space \(W_{0}^{m} E_{\varphi }(\Omega )\) as the closure of the Schwartz space \({\mathcal {D}}(\Omega )\) in \(W^{m} L_{\varphi }(\Omega ).\)
Let \(W_{0}^{m} L_{\varphi }(\Omega )\) be the \(\sigma \left( \Pi L_{\varphi }, \Pi E_{\overline{\varphi }}\right)\) closure of \({\mathcal {D}}(\Omega )\) in \(W^{m} L_{\varphi }(\Omega ).\) The following spaces of distributions will also be used:
and
We say that a sequence of functions \(u_{n} \in W^{m} L_{\varphi }(\Omega )\) is modular convergent to \(u \in W^{m} L_{\varphi }(\Omega )\) if there exists a constant \(k>0\) such that
We recall that
For \(\varphi\) and her complementary function \(\overline{\varphi },\) the following inequality is called the Young inequality (see [17]):
This inequality implies that
In \(L_{\varphi }(\Omega )\) we have the relation between the norm and the modular
and
For two complementary Musielak-Orlicz functions \(\varphi\) and \(\overline{\varphi },\) let \(u \in L_{\varphi }(\Omega )\) and \(v \in L_{\overline{\varphi }}(\Omega ),\) then we have the Hölder inequality (see [17]):
3 Some technical lemmas
This section concern some technical lemmas that will be used in our main result.
Definition 3.1
We say that a Musielak function \(\varphi\) verifies the log-Hölder continuity hypothesis on \(\Omega\) if there exists \(A>0\) such that
\(\forall t \ge 1\) and \(\forall x, y \in \Omega\) with \(|x-y| \le \frac{1}{2}\)
Lemma 3.1
[2] Let \(\Omega\) be a bounded Lipschitz domain in \({\mathbb {R}}^N(N \ge 2)\) and let \(\varphi\) be a Musielak function verifying the log-Hölder continuity such that
Then \({\mathfrak {D}}(\Omega )\) is dense in \(L_{\varphi }(\Omega )\) and in \(W_{0}^{1} L_{\varphi }(\Omega )\) for the modular convergence.
Remark 2
Note that if \(\lim _{t \rightarrow \infty } \inf _{x \in \Omega } \frac{\varphi (x, t)}{t}=\infty ,\) then (3.1) holds (see [2]).
Example 3.1
Let \(p \in P(\Omega )\) a bounded variable exponent on \(\Omega ,\) such that there exists a constant \(A>0\) such that for all points \(x, y \in \Omega\) with \(|x-y|<\frac{1}{2},\) we have the inequality
We can show that the Musielak function defined by \(\varphi (x, t)=t^{p(x)} \log (1+t)\) satisfies the hypothesis of Lemma 3.1.
Proof
(see [2]). \(\square\)
Lemma 3.2
[2] (Poincare’s inequality: Integral form) Let \(\Omega\) be a bounded Lipschitz domain of \(R ^{N}(N \ge 2)\) and let \(\varphi\) be a Musielak function satisfying the hypothesis of Lemma 3.1. Then there exists \(\beta , \eta > 0\) and \(\lambda > 0\) depending only on \(\Omega\) and \(\varphi\) such that
\(\square\)
Corollary 3.3
[2] (Poincare’s inequality) Let \(\Omega\) be a bounded Lipchitz domain of \({\mathbb {R}}^N(N \ge 2)\) and let \(\varphi\) be a Musielak function satisfying the same hypothesis of Lemma 3.2. Then there exists \(C>0\) such that
Lemma 3.4
( [30]) Let \(F: {\mathbb {R}} \longrightarrow {\mathbb {R}}\) be uniformly Lipschitzian, with \(F(0)=0 .\) Let \(\varphi\) be a Musielak-Orlicz function and let \(u \in W_{0}^{1} L_{\varphi }(\Omega ) .\) Then \(F(u) \in W_{0}^{1} L_{\varphi }(\Omega )\).
Hawever, if the set D of discontinuity points of \(F^{\prime }\) is finite, we obtain
Lemma 3.5
[1] (Poincare’s inequality). Let \(\varphi\) a Musielak-Orlicz function which satisfies the hypothesis of Lemma 3.1, let \(\varphi (x, t)\) decreases with respect of one of coordinate of x, then, that exists \(c>0\) depends only of \(\Omega\) such that
Lemma 3.6
[9] Let \(\Omega\) satisfies the segment property and suppose that \(u \in\) \(W_{0}^{1} L_{\varphi }(\Omega ) .\) Then, there exists a sequence \(\left( u_{n}\right) \subset {\mathcal {D}}(\Omega )\) such that
In addition to this, if \(u \in W_{0}^{1} L_{\varphi }(\Omega ) \cap L^{\infty }(\Omega )\) then \(\left\| u_{n}\right\| _{\infty } \le (N+1)\Vert u\Vert _{\infty }\).
Lemma 3.7
Suppose that \(\left( g_{n}\right) ,\ g \in L^{1}(\Omega )\) such that
(i) \(g_{n} \ge 0\) a.e in \(\Omega ,\)
(ii) \(g_{n} \longrightarrow g\) a.e in \(\Omega ,\)
(iii) \(\displaystyle \int _{\Omega } g_{n}(x) \,dx \longrightarrow \displaystyle \int _{\Omega } g(x) \,dx.\)
Then \(g_{n} \longrightarrow g\) strongly in \(L^{1}(\Omega ).\)
Lemma 3.8
[10] If a sequence \(h_{n} \in L_{\varphi }(\Omega )\) converges in measure to a measurable function h and if \(h_{n}\) remains bounded in \(L_{\varphi }(\Omega ),\) then \(h \in L_{\varphi }(\Omega )\) and \(h_{n} \rightharpoonup h\) for \(\sigma \left( \Pi L_{\varphi }, \Pi E_{\overline{\varphi }}\right)\).
Lemma 3.9
[10] Let \(v_{n},\ v \in L_{\varphi }(\Omega ).\) If \(v_{n} \rightarrow v\) with respect to the modular convergence, then \(v_{n} \rightarrow v\) for \(\sigma \left( L_{\varphi }(\Omega ), L_{\overline{\varphi }}(\Omega )\right) .\)
Lemma 3.10
[25] If \(\gamma \prec \varphi\) and \(u_{n} \rightarrow u\) for the modular convergence in \(L_{\varphi }(\Omega )\) then \(u_{n} \rightarrow u\) strongly in \(E_{\gamma }(\Omega )\).
Lemma 3.11
(The Nemytskii Operator). Suppose that \(\Omega\) be an open subset of \({\mathbb {R}}^{N}\) with finite measure and let \(\varphi\) and \(\psi\) be two Musielak Orlicz functions. Suppose that \(g: \Omega \times {\mathbb {R}}^{p} \longrightarrow {\mathbb {R}}^{q}\) be a Carathéodory function such that for a.e. \(x \in \Omega\) and all \(s \in {\mathbb {R}}^{p}:\)
where \(k_{1}\) and \(k_{2}\) are real positives constants and \(c(.) \in E_{\psi }(\Omega )\). Then the Nemytskii Operator \(N_{g}\) defined by \(N_{g}(u)(x)=g(x, u(x))\) is continuous from
into \(\left( L_{\psi }(\Omega )\right) ^{q}\) for the modular convergence. However if \(c(\cdot ) \in E_{\gamma }(\Omega )\) and \(\gamma \prec \prec \psi\) then \(N_{g}\) is strongly continuous from \({\mathcal {P}}\left( E_{M}(\Omega ), \frac{1}{k_{2}}\right) ^{p}\) to \(\left( E_{\gamma }(\Omega )\right) ^{q}\).
4 Main result
We now give the definition of a renormalized solution of (1.1).
Definition 4.1
A measurable function \(u: \Omega \rightarrow {\mathbb {R}}\) is called a renormalized solution of (1.1) if:
and for every function \(h \in C_{c}^{1}({\mathbb {R}})\) such that
Remark 3
Every term in equation (4.3) is meaningful in the distributional sense. Indeed, for \(h \in C_{c}^{1}({\mathbb {R}})\) and \(u\in W_{0}^{1} L_{\varphi }(\Omega ),\) then \(h(u)\in W^{1} L_{\varphi }(\Omega )\) and for V in \({\mathcal {D}}(\Omega )\) the function \(V h(u)\in W_{0}^{1} L_{\varphi }(\Omega ).\) Since \({\text {div}} \Big (b(x, u, \nabla u)\Big ) \in W^{-1} L_{\overline{\varphi }}(\Omega ),\) we have for every \(V \in {\mathcal {D}}(\Omega )\):
Finally, \(F(x,u) h(u)\in \left( L^{\infty }(\Omega )\right) ^{N},\ F(x,u) h^{\prime }(u)\in \left( L^{\infty }(\Omega )\right) ^{N},\ {\text {div}}\Big (F(x,u) h(u)\Big ) \in W^{-1} L_{\overline{\varphi }}(\Omega )\) and \(F(x,u) h^{\prime }(u)\nabla u \in L_{\varphi }(\Omega ).\)
Our main result is the following
Theorem 4.1
Under assumptions (1.2)-(1.8) there exists at least a renormalized solution of Problem (1.1).
Remark 4
Actually the original equation (1.1) will be recovered whenever \(h(u) \equiv 1\) but unfortunately this cannot happen in general strong additional requirements on u. Therefore, (4.3) is to be viewed as a weaker form of (1.1).
Remark 5
Generalized Orlicz spaces (Musielak-Orlicz-sobolev spaces), Orlicz spaces and \(L^{p(\cdot )}\)-spaces have different nature, and neither of them is a subset of the other.
Let us list some techniques from the classical case which do not work in \(L^{p(\cdot )}{ }_{-}\) spaces and some additional ones that do not work in the generalized Orlicz case. Orlicz spaces are similar to \(L^p\)-spaces in many regards, but some differences exist.
-
Exponents cannot be moved outside the \(\Phi\)-function, i.e. \(\varphi \left( t^\gamma \right) \ne \varphi (t)^\gamma\) in general.
-
The formula \(\varphi ^{-1}\left( \int _{\Omega } \varphi (|f|) d x\right)\) does not define a norm. Techniques which do not work in \(L^{p(\cdot )}\)-spaces (from [24], pp. 9–10]):
-
The space \(L^{p(\cdot )}\) is not rearrangement invariant; the translation operator \(T_h\) : \(L^{p(\cdot )} \rightarrow L^{p(\cdot )}, T_h f(x):=f(x+h)\) is not bounded; Young’s convolution inequality \(\Vert f * g\Vert _{p(\cdot )} \leqslant c\Vert f\Vert _1\Vert g\Vert _{p(\cdot )}\) does not hold [24], Section 3.6].
-
The formula
$$\begin{aligned} \int _{\Omega }|f(x)|^p d x=p \int _0^{\infty } t^{p-1}|\{x \in \Omega :|f(x)|>t\}| d t \end{aligned}$$has no variable exponent analogue.
-
Maximal, Poincaré, Sobolev, etc., inequalities do not hold in a modular form. For instance, A. Lerner showed that the inequality
$$\begin{aligned} \int _{{\mathbb {R}}^n}|M f|^{p(x)} d x \leqslant c \int _{{\mathbb {R}}^n}|f|^{p(x)} d x \end{aligned}$$holds if and only if \(p \in (1, \infty )\) is constant [29], Theorem 1.1]. For the Poincaré inequality see [24], Example 8.2.7] and the discussion after it.
-
Interpolation is not so useful, since variable exponent spaces never result as an interpolant of constant exponent spaces (see Sect. 5.5).
-
Solutions of the \(p(\cdot )\)-Laplace equation are not scalable, i.e. \(\lambda u\) need not be a solution even if u is [24], Example 13.1.9]. New obstructions in generalized Orlicz spaces:
-
We cannot estimate \(\varphi (x, t) \lesssim \varphi (y, t)^{1+\varepsilon }+1\) even when \(|x-y|\) is small, because of lack of polynomial growth. This complicates e.g. the use of higher integrability in PDE proofs.
-
It is not always the case that \(\chi _E \in L^{\varphi }(\Omega )\) when \(|E|<\infty\).
5 Proof of Theorem 4.1
Throughout the paper, \(T_{k}\) denotes the truncation function at height \(k\ge 0:\)
5.1 Approximate problem
For \(n \in {\mathbb {N}}^*,\) let define the following approximations of \(f\) and \(\Phi .\) Let \(f_{n}\) be a sequence of \(L^{\infty }(\Omega )\) functions that converge strongly to f in \(L^{1}(\Omega ),\) and \(\left\| f_{n}\right\| _{L^{1}} \le\) \(\Vert f\Vert _{L^{1}}.\) Let \(F_{n}\left( x, s\right) =F\left( x, T_{n}\left( s\right) \right)\). Then we consider the approximate Eq. (1.1) for \(n\ge 1:\ u_{n} \in W_{0}^{1} L_{\varphi }(\Omega )\)
there exists at last one solution \(u_{n} \in\) \(W_{0}^{1} L_{\varphi }(\Omega )\) of (5.1) (see [26]).
5.2 A priori estimates
Choosing \(T_{k}(u_{n})\) as a test function in (5.1), we get
By (1.6), Lemma 3.5 and Young inequality, we obtain:
Recall that
return to (5.2) and using (5.3) and (5.4) we get
by using (1.5) we get
thus
We take \(\displaystyle \frac{1}{c_{2}}=\left[ \frac{1}{2}-\frac{\left[ \Vert c(.)\Vert _{L^{\infty }}\left( \alpha _{0} +1\right) \right] }{\alpha }\right] .\) Then we deduce that
By (1.7) we have \(c_{2}>0\) where \(C=c_{1} c_{2} .\) And by (1.5) we have
So it follows that \(\left( T_{k}\left( u_{n}\right) \right) _{n} is \text{ bounded } \text{ in } W_{0}^{1} L_{\varphi }(\Omega ),\) then there exists some \(v_{k} \in W_{0}^{1} L_{\varphi }(\Omega )\) such that
On the other hand, using (5.6), we have
Then
for all \(n\ge 1\) and for all \(k\ge 1\). Assuming that there exists a positive function \(\overline{\varphi }\) such that \(\displaystyle \lim _{t \rightarrow \infty } \frac{\overline{\varphi }(t)}{t}=+\infty\) and \(\overline{\varphi }(t) \le ess\inf _{x \in \Omega } \varphi (x, t),\ \forall t \ge 0.\) Thus, we get
Let \(\eta >0\) and \(\epsilon >0\) then
then, by using (5.7) one suppose that \(\left( T_{k}\left( u_{n}\right) \right) _{n}\) is a Cauchy sequence in measure in \(\Omega\), Let \(\varepsilon >0,\) then by (5.8) there exists some \(k=k(\varepsilon )>0\) such that
which means that \(\left( u_{n}\right) _{n}\) is a Cauchy sequence in measure in \(\Omega ,\) thus converge almost every where to u. Consequently
5.3 Boundedness of \(\left( b\left( x, u_{n}, \nabla u_{n}\right) \right) _{n}\) in \(\left( L_{\overline{\varphi }}(\Omega )\right) ^{N}\)
Let \(\vartheta \in \left( E_{\varphi }(\Omega )\right) ^{N}\) such that
\(\Vert \vartheta \Vert _{\varphi , \Omega }=1,\) we have
This implies that
By using Young’s inequality in the last two terms of the last side and (5.6) we get
Now, by using (1.3) and the convexity of \(\overline{\varphi }\) we get
Thanks to Remark 1 there exists \(h \in L^{1}(\Omega )\) such that
then by integrating over \(\Omega\) we deduce that
where \(c_{k}'\) is a constant depending on k, then \(\forall \vartheta \in \left( E_{\varphi }(\Omega )\right) ^{N}\) with \(\Vert \vartheta \Vert _{\varphi , \Omega }=1\) we have \(\displaystyle \int _{\Omega } b\left( x, T_{k}\left( u_{n}\right) , \nabla T_{k}\left( u_{n}\right) \right) \vartheta d x \le c_{k}',\) and thus \(\left\| b\left( x, T_{k}\left( u_{n}\right) , \nabla T_{k}\left( u_{n}\right) \right) \right\| _{\overline{\varphi }, \Omega } \le c_{k}',\) which implies that
5.4 Renormalization identity for the approximate solutions
Consider the function \(Z_{m}\left( u_{n}\right) =T_{1}\left( u_{n}-T_{m}\left( u_{n}\right) \right)\) and by taking \(Z_{m}\left( u_{n}\right)\) as test function in (5.1) we obtain
By the same argument used in a priori estimates, we get
where \(\displaystyle \frac{1}{C}=\left[ \frac{1}{2}-\left( \frac{\Vert c(\cdot )\Vert _{L^{\infty }(\Omega )}+\epsilon _{1}}{\alpha }\right) \right] .\) In order to pass to the limit in (5.16) as \(n \rightarrow +\infty ,\) we use the pointwise convergence of \(u_{n}\) and strongly convergence in \(L^{1}(\Omega )\) of \(f_{n},\) we get
Thanks to Lebesgue’s theorem and passing to the limit as \(m \rightarrow +\infty ,\) in every term of the right-hand side of the previous inequalities, we obtain
Using (1.6) and Young inequality, for \(n>m+1\) we have
Thanks to Lebesgue’s theorem, and by the pointwise convergence of \(u_{n}\) we can have
Passing to the limit in (5.20) as \(m \rightarrow +\infty ,\) we obtain
Finally passing to the limit in (5.16), we get
5.5 Almost everywhere convergence of the gradients
Let \(v_{j} \in {\mathcal {D}}(\Omega )\) be a sequence such that \(v_{j} \rightarrow u\) in \(W_{0}^{1} L_{\varphi }(\Omega )\) for the modular convergence. For \(m \ge k,\) we define the function \(\varrho _{m}\) by
We denote by \(\epsilon (n, \eta , j, m)\) all quantities (possibly different) such that
For fixed \(k \ge 0,\) let \(W_{\eta }^{n, j}=T_{\eta }\left( T_{k}\left( u_{n}\right) -T_{k}\left( v_{j}\right) \right)\) and \(W_{\eta }^{j}=T_{\eta }\left( T_{k}(u)-T_{k}\left( v_{j}\right) \right) .\) Multiplying the approximating equation by \(W_{\eta }^{n, j} \varrho _{m}\left( u_{n}\right)\), we obtain
Remark that if we take \(n>m+1,\) we obtain
then \(F_{n}\left( x, u_{n}\right)\) is bounded in \(L_{\overline{\varphi }}(\Omega ),\) thus, by using the pointwise convergence of \(u_{n}\) and Lebesgue’s theorem we obtain \(F_{n}\left( x, u_{n}\right)\) converges to F(x, u) with the modular convergence as \(n \rightarrow +\infty ,\) then
In the other hand for \(0 \le T_{k}\left( u_{n}\right) -T_{k}\left( v_{j}\right) \le \eta\) then \(\nabla W_{\eta }^{n, j}=\nabla \Big (T_{k}(u_{n})-T_{k}(v_{j})\Big )\) converges to \(\nabla \Big (T_{k}(u)- T_{k}(v_{j})\Big )\) weakly in \(\left( L_{\varphi }(\Omega )\right) ^{N}\) as n tends to \(+\infty\), then
By using the modular convergence of \(W_{\eta }^{j}\) as \(j \rightarrow +\infty\) and letting \(\mu\) tends to infinity, we get
In the other hand for \(n>m+1>k,\) we have \(\nabla u_{n}\varrho ^{'}_{m}\left( u_{n}\right) =\nabla T_{m+1}\left( u_{n}\right)\) a.e. in \(\Omega .\) By the almost every where convergence of \(u_{n}\) we have \(W_{\eta }^{n, j} \rightarrow W_{\eta }^{j}\) in \(L^{\infty }(\Omega )\) weak- \(^{*}\) and since the sequence \(\left( F_{n}\left( x, T_{m+1}\left( u_{n}\right) \right) \right) _{n}\) converge strongly \({\text {in}} E_{\overline{\varphi }}(\Omega )\) then
converge strongly in \(E_{\overline{\varphi }}(\Omega )\) as \(n \rightarrow +\infty .\) By virtue of \(\nabla T_{m+1}\left( u_{n}\right) \rightarrow \nabla T_{m+1}(u)\) weakly in \(\left( L_{\varphi }(\Omega )\right) ^{N}\) as \(n \rightarrow +\infty\) we have
with the modular convergence of \(W_{\eta }^{j}\) as \(j \rightarrow +\infty\), we get
Concerning the first term of (5.22) we have
thus
The weakly convergence of \(T_{k}\left( u_{n}\right)\) to \(T_{k}\left( v_{j}\right)\) in \(W^{0,1} L_{\varphi }(\Omega )\) as n tends to \(+\infty\), the bounded character of \(W_{\eta }^{n, j}\), we obtain
and
Appealing now (1.5), we get
In the other hand we have
Since \(b_{n}\left( x, T_{k+\eta }\left( u_{n}\right) , \nabla T_{k+\eta }\left( u_{n}\right) \right)\) is bounded in \(\left( L_{\overline{\varphi }}(\Omega )\right) ^{N},\) there exist some \(\varpi _{k+\eta } \in \left( L_{\overline{\varphi }}(\Omega )\right) ^{N}\) such that \(b_{n}\left( x, T_{k+\eta }\left( u_{n}\right) , \nabla T_{k+\eta }\left( u_{n}\right) \right) \rightharpoonup \varpi _{k+\eta }\) weakly in \(\left( L_{\overline{\varphi }}(\Omega )\right) ^{N}.\) Thus:
By letting \(j \rightarrow +\infty ,\) we get
Thanks to (5.23)–(5.33), one has
Since \(\exp \left( G\left( u_{n}\right) \right) \ge 1\) and \(\varrho _{m}\left( u_{n}\right) =1\) for \(\left| u_{n}\right| \le k\) then
Finally we show that,
For \(s>0,\) denoting by \(\Omega ^{s}=\left\{ x \in \Omega :\left| \nabla T_{k}(u)\right| \le s\right\}\) and \(\Omega _{j}^{s}=\left\{ x \in \Omega :\left| \nabla T_{k}\left( v_{j}\right) \right| \le \right.\) \(s\}\) then by \(\chi ^{s}\) and \(\chi _{j}^{s}\) the characteristic functions of \(\Omega ^{s}\) and \(\Omega _{j}^{s}\) respectively, letting \(0<\delta <1\), define
For \(s>0,\) we have
The first term of the right-side hand, with the Hölder inequality we obtain
For the second term of the right-side hand by the Hölder inequality we have
since \(a\left( x, T_{k}\left( u_{n}\right) , \nabla T_{k}\left( u_{n}\right) \right)\) is bounded in \(\left( L_{\overline{\varphi }}(\Omega )\right) ^{N},\) while \(\nabla T_{k}\left( u_{n}\right)\) is bounded in \(\left( L_{\varphi }(\Omega )\right) ^{N}\) then
We obtain
On the other hand
For each \(s,\ r\in {\mathbb {R}}^+\) with \(s>r\) one has
In the sequel we pass to the limit in \(I_i\) when \(n,\ j,\ \mu ,\) and \(s \rightarrow +\infty\). We have
Thanks to (5.35), the first term of the right hand side in \(I_1\), we get
Since \(b\left( x, T_{k}\left( u_{n}\right) , \nabla T_{k}\left( u_{n}\right) \right)\) is bounded in \(\left( L_{\overline{\varphi }}(\Omega )\right) ^{N},\) there exist some \(\varpi _{k} \in\) \(\left( L_{\overline{\varphi }}(\Omega )\right) ^{N}\) such that (for a subsequence still denoted by \(u_{n}\)):
By using in the fact \(\left( \nabla T_{k}\left( v_{j}\right) \chi _{j}^{s}-\nabla T_{k}\left( v_{j}\right) \right) \chi _{\left\{ 0 \le T_{k}\left( u_{n}\right) -T_{k}\left( v_{j}\right) \le \eta \right\} }\) strongly converges to \(\left( \nabla T_{k}\left( v_{j}\right) \chi _{j}^{s}-\nabla T_{k}\left( v_{j}\right) \right) \chi _{\left\{ 0 \le T_{k}(u)-T_{k}\left( v_{j}\right) \le \eta \right\} }\) in \(\left( E_{\varphi }(\Omega )\right) ^{N}\) as \(n \rightarrow +\infty\).
The second term of the right hand side of \(I_1\) tends to
The third term of the right-hand side tends to
Letting \(j \rightarrow +\infty\) and \(\mu \rightarrow +\infty\) of \(I_1\), it possible to conclude that
Concerning \(I_{2},\) by letting \(n \rightarrow +\infty ,\) we obtain
Since \(b\left( x, T_{k}\left( u_{n}\right) , \nabla T_{k}\left( u_{n}\right) \right) \rightarrow \varpi _{k}\) in \(\left( L_{\overline{\varphi }}(\Omega )\right) ^{N},\) for \(\sigma \left( \Pi L_{\overline{\varphi }}, \Pi E_{\varphi }\right)\) while
strongly in \(\left( E_{\varphi }(\Omega )\right) ^{N}\). Now, letting \(j \rightarrow +\infty ,\) and thanks to Lebesgue’s theorem, we obtain
and
Consequently, we obtain
Which leads to
By taking \(W_{\eta }^{n,j}=T_{\eta }\left( T_{k}\left( u_{n}\right) -T_{k}\left( v_{j}\right) \right) ^{-}\) and \(W_{\eta }^{j}=T_{\eta }\left( T_{k}(u)-T_{k}\left( v_{j}\right) \right) ^{-},\) then testing the approximating equation by \(\exp \left( G\left( u_{n}\right) \right) W_{\eta }^{n, j} \varrho _{m}\left( u_{n}\right) ,\) we obtain
Thanks to (5.43) and (5.44) we have
As a consequence, since r is arbitrary:
and for all \(k \ge 0\), we have
5.6 Renormalization identity for the solutions
We show that The limit u of the approximate solution \(u_{n}\) of (5.1) satisfies:
To this end, remark that for any \(m>0\) one has
According to (5.46), (5.47) one is at liberty to pass to the limit as n tends to infinity for fixed m and to obtain
Taking the limit as m tends to \(+\infty\) and using the estimate (5.21) show that u satisfies (5.48).
5.7 Passing to the limit
Let \(h \in {\mathcal {C}}_{c}^{1}({\mathbb {R}})\) and \(V \in {\mathcal {D}}(\Omega ).\) Using the admissible test function \(h\left( u_{n}\right) V\) in (5.1) leads to
We shall pass to the limit in each term in the previous equality, to this end, remark that since h and \(h^{\prime }\) have a compact support in h, there exists \(K>0\) such that \(supp(h)\subset [-K, K].\) For n large enough, we have:
Let us start by the third integral of the left-hand side and the right hand-side of (5.51). Since \(h \in C _{c}^{1}({\mathbb {R}})\) and \(V \in {\mathcal {D}}(\Omega ),\) then there exists two positive constants \(c_{1}\) and \(c_{1}^{\prime }\) such that \(\left\| h\left( T_{K}\left( u_{n}\right) \right) \nabla V\right\| _{\infty } \le c_{1}\) and \(\left\| h^{\prime }(t)\left( T_{K}\left( u_{n}\right) V \nabla T_{K}\left( u_{n}\right) \Vert _{\infty } \le c_{1}^{\prime }\right. \right.\) Now since \(T_{K}\left( u_{n}\right)\) is bounded in \(W_{0}^{1} L_{\varphi }(\Omega ),\) then there exists two positive constant \(\lambda _{0}\) and \(\lambda\) such that \(\displaystyle {\int _{\Omega } \varphi \left( x,\frac{\left| \nabla T_{K}\left( u_{n}\right) \right| }{\lambda }\right) d x }\le \lambda _{0} .\) Using the convexity and monotonicity of \(\varphi ,\) for \(\eta\) large enough, we can write
Then the sequence \(\left\{ \nabla \left( h\left( T_{K}\left( u_{n}\right) \right) V\right) \right\}\) is bounded in \(\left( L_{\varphi }(\Omega )\right) ^{N},\) as a consequence, we deduce
Moreover, since \(F\left( x, T_{K}\left( u_{n}\right) \right)\) is bounded in \(L_{\psi }(\Omega ),\) we have from Lemma 3.10
By (5.52), we get
Moreover we have
Concerning the first integral of (5.51), while supp \(h^{\prime } \subset [-K, K],\) we obtain
The pointwise convergence of \(u_{n}\) to u, the bounded character of \(h'V\), (5.46) and (5.47) imply that
The term \(h^{\prime }(u) V b\left( x, T_{K}(u), \nabla T_{K}(u)\right) \nabla T_{K}(u)\) is identified with \(h^{\prime }(u) V b\left( x, u, \nabla u\right) \nabla u\).
Now since \(h\left( u_{n}\right) V b\left( x, u_{n}, \nabla u_{n}\right) =h\left( u_{n}\right) V b\left( x, T_{K}\left( u_{n}\right) , \nabla T_{K}\left( u_{n}\right) \right)\) a.e. in \(\Omega\), and using the strongly convergence of \(h\left( u_{n}\right) \nabla V\) to \(h(u) \nabla V\) in \(\left( E_{\varphi }(\Omega )\right) ^{N},\) and using the weakly convergence of \(b\left( x, T_{K}\left( u_{n}\right) , \nabla T_{K}\left( u_{n}\right) \right)\) to \(b\left( x, T_{K}(u), \nabla T_{K}(u)\right)\) in \(\left( L_{\psi }(\Omega )\right) ^{N}\) for \(\sigma \left( \Pi L_{\psi }, \Pi E_{\varphi }\right) ,\) then
As a consequence of the above convergence results, we are in a position to pass to the limit as n tends to \(+\infty\) in (5.51) and to conclude that u satisfies (4.3). As a conclusion of Step 5.1 to Step 5.7, the proof of Theorem 4.1 is complete.
Remark 6
-
(1)
It is possible to extend this result to the following parabolic equation
$$\begin{aligned} {\left\{ \begin{array}{ll}\frac{\partial u}{\partial t}-{\text {div}}(a(x, t, u, \nabla u))+ F(x, t, u)=\mu &{} \text{ in } \Omega \times (0, T), \\ u=0 &{} \text{ on } \partial \Omega \times (0, T), \\ u(x, 0)=u_0(x) &{} \text{ in } \Omega .\end{array}\right. } \end{aligned}$$where \(\Omega\) is a bounded open subset of \({\mathbb {R}}^N, N \ge 1, T>0\) and \(Q_T\) is the cylinder \(\Omega \times (0, T)\). The operator \(A(u)=-{\text {div}}(a(x, t, u, \nabla u))\) is a Leray-Lions operator lefined in \(W_0^{1, x} L_\varphi \left( Q_T\right)\). The lower order term F verifies the natural growth condition, no \(\Delta _{2}\)-condition is assumed on the Musielak function, and the datum \(\mu\) is assumed to belong to \(L^{1}(Q_T)+W^{-1} E_{\psi }(Q_T)\).
-
(2)
In the case of \(F\equiv 0,\) the problem (1.1) admits a unique solution.
References
Ahmida, Y., Youssfi, A.: Poincaré-type inequalities in Musielak spaces. Ann. Acad. Sci. Fenn. Math. 44(2), 1041–1054 (2019)
Ait Khellou, M., Benkirane, A., Douiri, S.M.: Some properties of Musielak spaces with only the log-Hölder continuity condition and application Annals of Functional Analysis,Tusi Mathematical Research Group (TMRG) 2020. https://doi.org/10.1007/s43034-020-00069-7.
Ait Khellou, M., Douiri, S.M., El Hadfi, Y.: Existence of solutions to parabolic equation with L1 data in Musielak spaces. J Elliptic Parabol Equ 8, 1-21 (2022). https://doi.org/10.1007/s41808-021-00139-4
Aharouch, L., Bennouna, J., Touzani, A.: Existence of Renormalized Solution of Some Elliptic Problems in Orlicz Spaces. Rev. Mat. Complut. 22(1), 91–110 (2009)
Aissaoui Fqayeh, A., Benkirane, A., El Moumni, M., Youssfi, A.: Existence of renormalized solutions for some strongly nonlinear elliptic equations in Orlicz spaces. Georgian Math. J. 22(3), 305–321 (2015)
Azraibi, O., haji, B.EL., Mekkour, M.: Nonlinear parabolic problem with lower order terms in Musielak-Sobolev spaces without sign condition and with Measure data, Palestine Journal of Mathematics, 11(3), 474-503(2022)
Bendahmane, M., Wittbold, P.: Renormalized solutions for nonlinear elliptic equations with variable exponents and \(L^{1}\) data. Nonlinear Anal. 70, 567–583 (2009)
Benkirane, A., Bennouna, J.: Existence of renormalized solutions for some elliptic problems involving derivatives of nonlinear terms in Orlicz spaces. Partial differential equations, In Lecture Notes in Pure and Appl. Math., Dekker, New York, 229,125–138(2002)
Benkirane, A., Sidi El Vally, M.: Some approximation properties in Musielak-Orlicz-Sobolev spaces. Thai. J. Math. 10, 371–381 (2012)
Benkirane, A., Sidi El Vally, M.: Variational inequalities in Musielak-Orlicz-Sobolev spaces. Bull. Belg. Math. Soc. Simon Stevin 21, (2014), 787-
Benkirane, A., El Haji, B., El Moumni, M.: On the existence of solution for degenerate parabolic equations with singular terms. Pure Appl. Math. Q. 14(3–4), 591–606 (2018)
Benkirane, A., El Haji, B., El Moumni, M.: Strongly nonlinear elliptic problem with measure data in Musielak-Orlicz spaces. Complex Var. Elliptic Equ. 1–23. https://doi.org/10.1080/17476933.2021.1882434
Benkirane, A., El Haji, B., El Moumni, M.: On the existence solutions for some Nonlinear elliptic problem. Bol. da Soc. Parana. de Mat. (3s.) 2022(40), 1–8. https://doi.org/10.5269/bspm.53111
El Haji, B., El Moumni, M.: Entropy solutions of nonlinear elliptic equations with \(L^1\)-data and without strict monotonocity conditions in weighted Orlicz-Sobolev spaces. J. Nonlinear Funct. Anal. 2021(Article ID 8), pp 1–17 (2021)
El Haji, B., El Moumni, M., Kouhaila, K.: Existence of entropy solutions for nonlinear elliptic problem having large monotonicity in weighted Orlicz-Sobolev spaces. Le Matematiche LXXVI(I), 37–61 (2021). https://doi.org/10.4418/2021.76.1.3
El Haji, B., El Moumni, M., Kouhaila, K.: On a nonlinear elliptic problems having large monotonocity with \(L^{1}\)-data in weighted Orlicz-Sobolev spaces. Moroccan J. Pure and Appl. Anal. 5(1), 104–116 (2019)
Musielak, J.: Modular spaces and Orlicz spaces, Lecture Notes in Math. (1983), 10–34
El Amarty, N., El Haji, B., El Moumni, M.: Entropy solutions for unilateral parabolic problems with \(L^{1}\)-data in Musielak-Orlicz-Sobolev spaces. Palestine J. Math. 11(1), 504–523 (2022)
El Amarty, N., El Haji, B., El Moumni, M.: Existence of renormalized solution for nonlinear Elliptic boundary value problem without \(\Delta _{2}\)-condition. SeMA. J 77(4), 389–414 (2020)
Bouzyani, R., El Haji, B., El Moumni, M.: Entropy solutions of some nonlinear elliptic problems with measure data in Musielak-Orlicz spaces. Rend. Mat. Appl. 43(7), 1–22 (2022)
Boccardo, L., Giachetti, D., Diaz, J.I., Murat, F.: Existence and regularity of renormalized solutions for some elliptic problems involving derivatives of nonlinear terms. J. Differential Equations 106(2), 215–237 (1993)
DiPerna, R. J., Lion, P.-L.: On the Cauchy problem for Boltzmann equations: Global existence and weak stability, Ann.of Math. (2) 130 (1989), no. 2, 321-366
Dal Maso, G., Murat, F., Orsina, L., Prignet, A.: Renormalized solutions of elliptic equations with general measure data, \(A\) nn. Scuola Norm. Sup. Pisa Cl. Sci. 28(4)741-808(1999)
Diening, L., Harjulehto, P., Hasto, P., Ruzicka, M.: Lebesgue and Sobolev spaces with Variable Exponents. Lecture Notes in Mathematics, vol. 2017. Springer, Heidelberg (2011)
Elarabi, R., Rhoudaf, M., Sabiki, H.: Entropy solution for a nonlinear elliptic problem with lower order term in Musielak-Orlicz spaces. Ric. Mat. (2017). https://doi.org/10.1007/s11587-017-0334
Gossez, J.P., Mustonen, V.: Variationnal inequalities in Orlicz-Sobolev spaces. Nonlinear Anal. 11, 317–492 (1987)
Gossez, J.-P.: Nonlinear elliptic boundary value problems for equations with rapidly (or slowly) increasing coeficients. Trans. Amer. Math. Soc. 190, 163–205 (1974)
Moussa, H., Rhoudaf, M.: Study of some non-linear elliptic problems with no continuous lower order terms in Orlicz spaces. Mediterr. J. Math. 13, 4867–4899 (2016)
Lerner, A.: Some remarks on the Hardy-Littlewood maximal function on variable Lp spaces. Math. Z. 251, 509–521 (2005)
Porretta, A.: Existence results for strongly nonlinear parabolic equations via strong conver- gence of truncations, Annali di matematica pura ed applicata. (IV), Vol. CL XXVII, (1999), 143–172
Rakotoson, J.M.: Uniqueness of renormalized solutions in a T-set for the \(L1\)-data problem and the link between various formulations. Indiana Univ. Math. J. 43(2), 685–702 (1994)
Rudin, W.: Real and Complex Analysis, 3rd edn. McGraw-Hill, New York (1974)
El Moumni, M.: Nonlinear elliptic equations without sign condition and \(L^1\)-data in Musielak-Orlicz-Sobolev spaces. Acta Appl. Math. 159, 95–117 (2019)
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Benkirane, A., EL Amarty, N., EL Haji, B. et al. Existence of solutions for a class of nonlinear elliptic problems with measure data in the setting of Musielak–Orlicz –Sobolev spaces. J Elliptic Parabol Equ 9, 647–672 (2023). https://doi.org/10.1007/s41808-022-00193-6
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s41808-022-00193-6