Abstract
In this paper, we study existence and regularity results for the solution to a nonlinear singular parabolic problems involving Hardy potential
where \(\Omega \) is a bounded open subset on \({\mathbb {R}}^{N}, N\ge 3, 0\in \Omega \) and \(\gamma >0,\; 2\le p<N,\) \(0<T<+\infty ,\; \mu >0,\) \(0\le f\in L^{m}(Q),\; m\ge 1\) and \(u_{0}\in L^{\infty }(\Omega )\) satisfies
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In this paper, we are interested to prove existence and regularity results for a class of nonlinear singular parabolic equations involving Hardy potential, as following model
where \(\Omega \) is a bounded open subset of \({\mathbb {R}}^{N},\) \((N\ge 3),\) \(2\le p<N,\) \(\gamma ,\mu >0,\) \(Q=\Omega \times (0,T),\) \(\Gamma =\partial \Omega \times (0,T),\) with \(T>0,\) f is a nonnegative function belonging a suitable Lebesgue space, the initial datum \(u_{0}\in L^{\infty }(\Omega )\) and satisfies the following bound
Moreover, the function \(a: \Omega \times (0,T)\times {\mathbb {R}}^{N}\longrightarrow {\mathbb {R}}^{N}\) is a Caratheodory function satisfying the following conditions: there exist positive constants \(\alpha , \beta \) such that
for almost every \(x\in \Omega , t\in (0,T),\) for every \(\xi ,\xi ^{\prime }\in {\mathbb {R}}^{N},\) with \(\xi \ne \xi ^{\prime }.\)
Under assumptions (1.3), (1.4) and (1.5), the differential operator defined by
is coercive and monotone operator acting from the space \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) into its dual \(L^{p^{\prime }}(0,T; W^{-1,p^{\prime }}(\Omega )).\) The simplest example is the one in which the operator A is the p-Laplacian: \(A(u)=-{\text {div}}(|\nabla u|^{p-2}\nabla u).\)
The interest in problem (1.1) (with \(p=2,\; \mu =0\) and smooth data) started in [18] in connection with the study of thermo-conductivity (\(u^{\gamma }\) represented the resistivity of the material), and later in the study of signal transmissions and in the theory of non-Newtonian pseudoplastic fluid [20, 23, 24].
From a purely mathematical point of view the literature is wide. In the case \(\mu =0\) and \(\gamma =0,\) the existence and regularity of problem (1.1) has been studied in [9, 10, 13, 21, 22, 29] under the different assumptions on the data. If \(\gamma =0, f=0\) and \(\mu >0\) the existence and nonexistence of solution of problem (1.1) depending the value of \(\mu \) has been studied by the authors in [19, 27]. When \(\gamma =0, f\ne 0\) and \(\mu >0,\) the authors in [5] has been studied the existence and summability of elliptic problem
where \(b>0, \mu>0, r>2^{*}\) and \(f\in L^{m}(\Omega ),\, m>1,\) M(x) is a matrix satisfies \(M(x)\xi \cdot \xi \ge \alpha |\xi |^{2}; |M(x)|\le \beta \) with \( \alpha ,\beta \ge 0\) for all \(\xi \in {\mathbb {R}}^{N}\) and almost every \(x\in \Omega .\) Baras in [8] studied the existence and nonexistence of problem
where \(v(x)=c/|x|^{2}, c>0, v\in L^{\infty }(\Omega \backslash B_{\epsilon })\) (where \(B_{\epsilon }=\{x: |x|<\epsilon \}),\) the function v is singular at the origin and \(u_{0},f\ge 0\) satisfies some conditions. In the same contexts Porzio [28] showed that the problem
admits a solution for \(0<\mu <\varrho _{1}\left( \frac{N-2}{2} \right) ^{2},\) where \(\varrho _{1}\) is the coercivity constant of \(a(x,t,u,\nabla u),\) \(f\in L^{r}(0,T; L^{q}(\Omega )),\) with \(r>1, q>1,\) and the summability of solution also obtained (See also [17, 26]). When \(\mu =0\) and \(\gamma >0,\) the problem of existence, regularity and uniqueness (sometimes partial uniqueness) results of (1.1) have been investigated in different contexts by several authors (see [11, 14,15,16, 25, 30, 31] and references therein). The authors in [11] proved the existence, regularity and uniqueness of solution to singular parabolic problem
where \(\gamma >0\) and \(0\le f\in L^{m}(Q),\, m\ge 1\) and \(u_{0}\in L^{\infty }(\Omega )\) satisfies
Finally in the elliptic framework when \(\mu>0, \gamma >0\) the author in [33] proved the existence of one positive solution to singular problem
where \(0\in \Omega , \gamma >0, 0\le f\in L^{m}(\Omega ),\, 1<m<\frac{N}{2}\) and \(0<\mu <\left( \frac{N-2}{2} \right) ^{2}.\) The stationary problem associated to problem (1.1) has been studied in [3]; the authors proved the existence and regularity (and partial uniqueness ) results of solution to singular problem
where \(0\in \Omega , \gamma >0, 0<\mu \le \left( \frac{N-p}{p} \right) ^{p}\) and \(0\le f\in L^{m}(\Omega ), m\ge 1.\) If \( \mu >\left( \frac{N-p}{p} \right) ^{p},\) then the problem has non solution (see [4]), and also the authors proved that if f is a singular measure with respect to the p-Capacity associated to \(W_{0}^{1,p}(\Omega )\) the problem has a non-negative solution in suitable sense.
The aim of this work is to analyze the interaction between the Hardy potential and the singular term \(u^{-\gamma }\) in order to get a solution for largest possible class of the datum f.
1.1 Preliminary and notations
The problem (1.1) is related to the following classical Hardy inequality (see [19])
where \(C_{N,p}=\left( \frac{N-p}{p} \right) ^{p}\) is optimal and is not attained.
Also we need to the Gagliardo–Nirenberg inequality that we used later in the proof of our results.
Lemma 1.1
[12, Proposition 3.1] Let v be a function in \(W_{0}^{1,h}(\Omega )\cap L^{\rho }(\Omega ),\) with \(h\ge 1, \rho \ge 1.\) Then there exist a positive constant \(C_{G}\) depending on \(N, h, \rho \) and \(\sigma \) such that
for every \(\eta \) and \(\sigma \) satisfying
An immediate consequence of the previous Lemma is the following embedding results
which holds for every function \(v\in L^{h}(0,T; W_{0}^{1,h}(\Omega ))\cap L^{\infty }(0,T; L^{\rho }(\Omega )),\) with \(h\ge 1, \rho \ge 1\) and \(\sigma =\frac{h(N+\rho )}{N}.\)
We denote by |E| measure of Lebesgue measurable subset E of \({\mathbb {R}}^{N}.\) For any \(q>1,\) \(q^{\prime }=\frac{q}{q-1}\) is the Hölder conjugate exponent of q, while for any \(1\le p<N,\) \(p^{*}=\frac{Np}{N-p}\) is the Sobolev conjugate exponent of p. For fixed \(k>0\) we will make use of the truncation functions \(T_{k}\) and \(G_{k}\) defined as
and
for every \(s\in {\mathbb {R}}.\) For the sake of simplicity we will often use the simplified notation
when referring to integrals when no ambiguity to the variable of integration is possible. If not otherwise specified, we will denote by C several constants whose value many change from line to line and, some times in the same line. These values will only depend on the data ( for instance C can depend on \(N, p, C_{N,p},C_{G}, \alpha , \gamma , m, T, \Omega , Q)\) but they will never depend on the indexes of the sequences we will often introduce.
Definition 1.2
We will say that a function \(u\in L^{1}(0,T; W_{loc}^{1,1}(\Omega ))\) is a distributional solution of (1.1) if
i.e., some positive power of u belongs to a Sobolev space \(L^{r}(0,T; W_{0}^{1,r}(\Omega )),\) \(r>1.\) Moreover, we require that
and that
for all \( \varphi \in C^{1}_{c}(\Omega \times [0,T)).\)
2 The approximation scheme
Let \(n\in {\mathbb {N}}\) and \(f_{n}(x,t)\) be defined by \(f_{n}(x,t)=\min (f(x,t), n);\) we will consider the following approximation of (1.1)
Lemma 2.1
The problem (2.1) has a nonnegative solution belonging to \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(Q)\) for all \(\mu <\alpha C_{N,p}\) and \(2\le p<N.\)
Proof
Let \(v\in L^{p}(Q)\) and we define \(S: L^{p}(Q)\longrightarrow L^{p}(Q)\) such that \(S(v)=w,\) with \(w\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap C([0,T]; L^{2}(\Omega ))\) the unique solution of problem
The existence of solution of above problem assured by [22]. Let us take w as a test function in the above problem, from (1.3), we have
since \(u_{0}\in L^{\infty }(\Omega )\) and by Hardy inequality implies
Dropping the first non-negative term and thanks to Hölder’s inequality, we have
By application of Poincaré inequality in the right hand side, it hold that
where \(C_{p}\) is the Poincaré constant. This implies that
for some constant R independent of v. So that the ball of radius R is invariant under S. Using Sobolev embedding Theorem, it is easy to prove that S is both continuous and compact on \(L^{p}(Q),\) so that by Shauder’s fixed point Theorem there exist \(u_{n}\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(Q)\) such that \(S(u_{n})=u_{n},\) for all \(n\in {\mathbb {N}}, \, 2\le p<N,\) i.e. \(u_{n}\) solves
Moreover, since \(\frac{f_{n}}{(|u_{n}|+\frac{1}{n})^{\gamma }}\ge 0,\) taking \(u_{n}^{-}=\min (u_{n},0)\) test function in (2.1) and using (1.3), then we have
dropping the first nonnegative term and by Hardy inequality, we can get
as \(\alpha -\frac{\mu }{C_{N,p}}>0,\) then we deduce that
that implies that \(u_{n}^{-}=0\) a.e and hence \(u_{n}\ge 0.\) a.e.. \(\square \)
Lemma 2.2
Let \(u_{n}\) be a solution of (2.1). Then
Proof
Since \(u_{n}\) solution of (2.1), then
as \(\mu >0,\) then we obtain
this implies that the sequence \(u_{n}\) is a sub-solution to problem
Thanks to Proposition 2.2 in [11], \(\exists \; c_{\omega }>0\) (independent of n) such that
since \(u_{n}\ge v,\) so
\(\square \)
3 A priori estimate and main results
Now, we prove some a priori estimates on the sequence of approximated solutions \(u_{n}.\)
Lemma 3.1
Assume that (1.3)–(1.5) hold true, \(f\in L^{\frac{p(N+2)}{p(N+2)-N(1-\gamma )}}(Q).\) If \(\gamma <1\) and \(\mu <\alpha C_{N,p},\) then the sequence \(u_{n}\) is uniformly bounded in \( L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(0,T; L^{2}(\Omega )).\)
Proof
Take \(u_{n}\chi _{(0,t)}\) as a test function in (2.1) (with \(0< t\le T\)), from (1.3) and \(f_{n}\le f\) we have
since \(u_{0}\in L^{\infty }(\Omega ),\) thanks to Hölder’s and Hardy inequalities imply that
Passing to the supremum for \(t\in [0,T]\)
By Lemma 1.1, we can write
since \(0<\gamma <1\) then \(\frac{(p+N)(1-\gamma )}{p(N+2)}<1,\) this implies the sequence \(u_{n}\) is bounded in \(L^{\frac{p(N+2)}{N}}(Q),\) hence \(u_{n}\) is bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(0,T; L^{2}(\Omega ))\) with respect to n. \(\square \)
Lemma 3.2
Assume that (1.3)–(1.5) hold true, \(\gamma \ge 1,\) \(\mu <\alpha C_{N,p}\) and \(f\in L^{1}(Q),\) then
- i):
-
If \(\gamma =1,\) then \(u_{n}\) is bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(0,T; L^{2}(\Omega )).\)
- ii):
-
If \(\gamma >1,\) then \(u_{n}\) is bounded in \(L^{p}(0,T; W_{loc}^{1,p}(\Omega ))\) and \(T_{k}(u_{n})^{\frac{\gamma +p-1}{p}}\) is bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega )).\) Moreover if \( \alpha \left( \frac{p}{\gamma +p-1}\right) ^{p}-\frac{\mu }{C_{N,p}} >0,\) then \(u_{n}^{\frac{\gamma +p-1}{p}}\) is bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) and \(u_{n}\) is bounded in \(L^{\infty }(0,T; L^{\gamma +1}(\Omega )).\)
Proof
First case: \(\gamma =1\)
Choosing \(u_{n}\chi _{(0,t)}\) as a test function in (2.1) (with \(0< t\le T\) ), by (1.3) and the fact that \(0\le \frac{u_{n}}{u_{n}+\frac{1}{n}}\le 1,\, f_{n}\le f,\) we have
thanks to Hardy inequality, there result that
Passing to the supremum for \(t\in [0,T]\) and the fact that \(u_{0}\in L^{\infty }(\Omega ),\) we get
since \(\alpha -\frac{\mu }{C_{N,p}}>0,\) then the sequence \(u_{n}\) is bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) and in \(L^{\infty }(0,T; L^{2}(\Omega ))\) with respect to n. Hence the proof of item i) is achieved.
Second case: \(\gamma >1\)
Now taking \(G_{k}(u_{n})\) as test function in (2.1), from (1.3) we arrive to
dropping the first nonnegative term and as \(G_{k}(u_{n})=0\) if \(u_{n}\le k\) and the fact that \(G_{k}(u_{0}(x))\le u_{0}(x),\) then
Notice that for all \(a,b\ge 0\) and for all \(\epsilon >0,\) we have
For \(u_{n}>k,\) we have \(u_{n}^{p-1}G_{k}(u_{n})=(G_{k}(u_{n})+k)^{p-1}G_{k}(u_{n})\) and \(p\ge 2,\) then from the previous estimate we reach that
In view of (3.2) and (3.3), it follows that
as \(\mu <\alpha C_{N,p},\) choosing \(\epsilon \) small enough and by Hardy inequality, we get
Applying Hölder, Young and Hardy inequalities we conclude that
Testing now (2.1) by \((T_{k}(u_{n}))^{\gamma },\) so that, from (1.3) and (3.6)
There hold
this implies that the sequence \(T_{k}(u_{n})^{\frac{\gamma +p-1}{p}}\) is bounden in \(L^{p}(0,T; W_{0}^{1,p}(\Omega )).\) By Lemma 2.2 and (3.6), yields that \(T_{k}(u_{n})\) is bounded in \(L^{p}(0,T; W_{loc}^{1,p}(\Omega )).\) Collecting the last affirmation with (3.6), assume that the sequence \(u_{n}\) is bounded in \(L^{p}(0,T; W_{loc}^{1,p}(\Omega )).\) Using \(u_{n}^{\gamma }\chi _{(0,t)}\) as test function in (2.1) (with \(0< t\le T\)), from (1.3), \(u_{0}\in L^{\infty }(\Omega )\) and applying Hardy inequality, we get
since \( \alpha \left( \frac{p}{\gamma +p-1}\right) ^{p}-\frac{\mu }{C_{N,p}}>0,\) passing to the supremum for \(t\in [0,T],\) we deduce that
this implies that \(u_{n}^{\frac{\gamma +p-1}{p}}\) is bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) and \(u_{n}\) is bounded in \(L^{\infty }(0,T; L^{\gamma +1}(\Omega ))\) with respect to n. Since the proof of item ii) is achieved. \(\square \)
Theorem 3.3
Assume that (1.3)–(1.5) holds true. If \(\gamma<1, \mu <\alpha C_{N,p}\) and \(f\in L^{\frac{p(N+2)}{p(N+2)-N(1-\gamma )}}(Q).\) Then, there exists a solution u to problem (1.1) in the sense of Definition 1.2. Moreover \(u\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(0,T; L^{2}(\Omega ))\) and \(\frac{\partial u}{\partial t}\in L^{p^{\prime }}(0,T; W^{-1,p^{\prime }}(\Omega ))+ L^{1}(0,T; L_{loc}^{1}(\Omega )).\)
Remark 3.4
If \(\mu =0,\) then the result of Theorem 3.3 coincide with result of Theorem 1.3 in [11].
Theorem 3.5
Suppose that (1.3)–(1.5) holds true. If \(\gamma \ge 1, \, \mu <\alpha C_{N,p}\) and \(f\in L^{1}(Q).\) Then, there exists a solution u to problem (1.1) in the sense of Definition 1.2 with the following regularity:
-
a)
If \(\gamma =1,\) then \(u\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(0,T; L^{2}(\Omega ))\) and \(\frac{\partial u}{\partial t}\in L^{p^{\prime }}(0,T; W^{-1,p^{\prime }}(\Omega ))+ L^{1}(0,T; L_{loc}^{1}(\Omega )).\)
-
b)
If \(\gamma >1,\) then \(u\in L^{p}(0,T; W_{loc}^{1,p}(\Omega ))\) and \(T_{k}(u)^{\frac{p+\gamma -1}{p}}\in L^{p}(0,T; W_{0}^{1,p}(\Omega )).\) If \( \alpha \left( \frac{p}{p+\gamma -1}\right) ^{p}-\frac{\mu }{C_{N,p}} >0,\) then \(u^{\frac{p+\gamma -1}{p}}\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) and \(u\in L^{\infty }(0,T; L^{\gamma +1}(\Omega ))\) and \(\frac{\partial u}{\partial t}\in L^{p^{\prime }}(0,T; W^{-1,p^{\prime }}(\omega ))+ L^{1}(0,T; L^{1}(\omega ))\) for all \(\omega \subset \subset \Omega .\)
Remark 3.6
If \(\mu =0,\) then the result of Theorem 3.5 coincide with result of Theorem 1.3 in [11].
Before giving the proof of Theorems 3.3 and 3.5 , we need the following results:
Proposition 3.7
Under the assumptions of Lemmas 3.1 and 3.2 there exists \(u\in L^{p}(0,T; W_{loc}^{1,p}(\Omega ))\) such that, up to a subsequence, \(u_{n}\) converges to u a.e. on Q, weakly in \(L^{p}(0,T; W_{loc}^{1,p}(\Omega ))\) and strongly in \(L^{1}(0,T; L^{1}_{loc}(\Omega )).\)
Proof
From Lemmas 3.1 and 3.2 we know that \(u_{n}\) is bounded in the space \(L^{p}(0,T; W_{loc}^{1,p}(\Omega )).\) The last affirmation and Lemma 2.2 imply the sequence \(\bigg \{\mu \frac{u_{n}^{p-1}}{|x|^{p}+\frac{1}{n}}+\frac{f_{n}}{(u_{n}+\frac{1}{n})^{ \gamma }}\bigg \}\) is bounded in \(L^{1}(0,T; L_{loc}^{1}(\Omega )).\) Hence, let \(\varphi \in C_{c}^{1}(\Omega )\) then one has that \(\{\frac{\partial (u_{n}\varphi )}{\partial t}\}\) is bounded in \(L^{p^{\prime }}(0,T; W^{-1,p^{\prime }}(\Omega ))+L^{1}(Q),\) which is sufficient to apply [32, Corollary 4] in order to deduce that \(u_{n}\) converges to a function \(u\in L^{1}(0,T; L_{loc}^{1}(\Omega ))\) and \(u_{n}\) converges to u a.e. in Q. \(\square \)
In the following proposition, we are going to prove the almost everywhere convergence of the gradient of \(u_{n}.\)
Proposition 3.8
Let \(u_{n}\) be a solution of problem (2.1) and assume that \(f\in L^{\frac{p(N+2)}{p(N+2)-N(1-\gamma )}}(Q)\) if \(\gamma <1\) and \( f\in L^{1}(Q)\) if \(\gamma \ge 1\) respectively. Then the sequence \(T_{k}(u_{n})\) strongly converges to \(T_{k}(u)\) in \(L^{p}(0,T; W_{loc}^{1,p}(\Omega ))\) and so, in particular, \(\nabla u_{n}\) converges to \(\nabla u\) almost everywhere in Q.
Proof
Let \(n,m\in {\mathbb {N}}\) denote two value of the parameter describing the approximation. Since (2.1) is non-singular problem, we can take \(T_{2k}(u_{n}-u_{m})\varphi \in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(Q)\) as a test function in the difference of the approximating equations solved by \(u_{n}\) and \(u_{m},\) with \(\varphi \in C_{c}^{1}(\Omega )\) independent of \(t\in [0,T]\) and such that \(0\le \varphi \le 1\), obtaining
Observe that
where \(\theta _{2k}(t)\) is the primitive of \(T_{2k}(t)\) which vanishes for \(t=0,\) and so we can drop the parabolic term (3.8) (since it is nonnegative) obtaining
We denote by
since \(A_{k,n, m}\subset \{(x,t)\in Q: |u_{n}-u_{m}|\le 2k \},\) we have
In conclusion, we found that
The right-hand side of the previous inequality is infinitesimal for \(n,m\rightarrow +\infty \) and we denote by r(n, m) a quantity that goes to zero from \(n,m\rightarrow +\infty .\)
By using the same proof as Proposition 3.2 in [11], we have
and
Now, we prove
First of all we prove that
Notice that it results \(1<\frac{pN}{p+(p-1)N}<p^{\prime }.\) As matter of fact, for every compact \(\omega \subset \Omega \) it results (thanks to Hardy inequality and Lemmas 3.1 and 3.2 )
where the last integral in the right-hand side is finite since it results
Hence, by (3.11) and the convergence a.e. of \(u_{n}\) to u in Q we deduce that
Notice that, thanks to the Lebesgue Theorem, it results
and thus it convergences also in \(L^{{\bar{h}}^{\prime }}(Q)\) and (3.10) follows.
Then, the rest of the proof, we proceed as Proposition 3.2 in [11], we obtain up to subsequences, \(T_{k}(u_{n})\rightarrow T_{k}(u)\) in \(L^{p}(0,T; W_{loc}^{1,p}(\Omega )),\) and so \(\nabla u_{n}\rightarrow \nabla u\) a.e. in Q. \(\square \)
Proof of Theorems 3.3 and 3.5
If \(\gamma <1,\) by Lemma 3.1, we have \(u_{n}\) is bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) and in \(L^{\infty }(0,T; L^{2}(\Omega )).\) Then, by Lemma 2.2, Proposition 3.7 and Fatou’s Lemma \(u\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(0,T; L^{2}(\Omega )),\) and moreover \(\frac{u^{p-1}}{|x|^{p}}, \frac{f}{u^{\gamma }}\in L^{1}(0,T; L^{1}_{loc}(\Omega ))\) since u satisfies (1.8), in particular
If \(\gamma =1,\) thanks to Lemma 3.2, we have \(u_{n}\) is bounded in
as before, we get \(u\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(0,T; L^{2}(\Omega ))\) and u satisfies (1.8); Moreover \(\frac{\partial u}{\partial t}\in L^{p^{\prime }}(0,T; W^{-1,p^{\prime }}(\Omega ))+ L^{1}(0,T; L^{1}_{loc}(\Omega )).\)
In the case \(\gamma >1,\) in view of Lemma 3.2, we have that \(u_{n}^{\frac{p+\gamma -1}{p}}\) is bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega )),\) while \(u_{n}\) is bounded in
Then
in particular, \(u=0\) on \(\partial \Omega \times (0,T)\) in weak-sense and
Using Lemma 2.2, Proposition 3.7 and Fatou’s Lemma deduce that u satisfies the condition (1.8). Now we fix \(\varphi \in C_{c}^{1}(\Omega \times [0,T)),\) by Lemma 3.1 and Lemma 3.2, we have the boundedness of the sequence \(u_{n}\) in the space \(L^{p}(0,T; W_{loc}^{1,p}(\Omega ))\) and from (1.4), implies that the sequence \(a(x,t,\nabla u_{n})\) is bounded in \(L^{p^{\prime }}(\omega \times (0,T))\) for all \(\omega \subset \subset \Omega .\) As \(supp(\varphi )\) is a compact subset of \(\Omega \times [0,T),\) then \(a(x,t,\nabla u_{n})\) is bounded in \(L^{p^{\prime }}(supp(\varphi ))\) and \(\frac{u_{n}^{p-1}}{|x|^{p}+\frac{1}{n}}\) is bounded in \(L^{1}(supp(\varphi )).\) From Propositions 3.7 and 3.8 , we have \(u_{n}\rightarrow u\) a.e. in Q and \(\nabla u_{n}\rightarrow \nabla u\) a.e. in Q and by Vitali’s Theorem we obtain
and
Concerning the passage of limit of term in the right of the approximating problem (2.1), since \(supp(\varphi )\) is a compact subset of \(\Omega \times [0,T),\) thanks to Lemma 2.2, there exists a constant \(c_{supp(\varphi )}>0\) such that \(u_{n}\ge c_{supp(\varphi )},\) then
for every \((x,t)\in supp(\varphi ),\) since it is a.e. convergent to \(\frac{f}{u^{\gamma }}\varphi \) for \(n\longrightarrow +\infty ,\) by Lebesgue Theorem, implies that
By Proposition 3.7, we have
Take now \(\varphi \in C_{c}^{1}(\Omega \times [0,T))\) as a test function in problem (2.1), by the convergences results (3.12), (3.13), (3.14), (3.15) and letting \(n\longrightarrow +\infty ,\) we get
\(\square \)
4 Regularity results
In this section we study the regularity of solutions of problem (1.1) depending on \(\mu , \gamma >0\) and the summability of f.
4.1 The case \(\gamma \ge 1\)
Theorem 4.1
Let \(\gamma \ge 1 \) and suppose that f belongs to \(L^{m}(Q)\) with \(1<m<\frac{N}{p}+1.\) If
then the solution u of (1.1) found in Theorem 3.5 satisfies the following summability \(u\in L^{\sigma }(Q),\) where \(\sigma =m\displaystyle \frac{N(p+\gamma -1)+p(\gamma +1)}{N-pm+p}.\)
Proof
Let now choosing \(u_{n}^{p\delta -p+1}\chi _{(0,t)}\) as test function in (2.1), \(\delta >\frac{p+\gamma -1}{p}\) and \(0<t<T,\) then we get
from (1.3), it follows that
Thanks to \(u_{0}\in L^{\infty }(\Omega )\) and \(u_{n}^{p\delta -p}|\nabla u_{n}|^{p}=\frac{1}{\delta ^{p}}|\nabla u_{n}^{\delta }|^{p},\) the last inequality becomes
applying Hardy and Hölder’s inequalities, yields
Passing to supremum for \(t\in (0,T)\) we have
Since \(u_{n}\in L^{\infty }(Q)\cap L^{p}(0,T; W_{0}^{1,p}(\Omega )),\) then in view to Lemma 1.1 and by (4.1), we get
hence
Choosing now \(\delta \) such that
this equivalent to
Collecting (4.2) with (4.3), we conclude that
By virtue of \(m<\frac{N}{p}+1,\) then \((\frac{p}{N}+1)\frac{1}{m^{\prime }}<1,\) since \(\delta >\frac{p+\gamma -1}{p}\) gives \(m>1\) and applying Young’s inequality implies that
this last estimate yields that the sequence \(u_{n}\) is bounded in \(L^{\sigma }(Q),\) and so \(u\in L^{\sigma }(Q).\)
\(\square \)
Theorem 4.2
Let \(\gamma \ge 1\) and \(f\in L^{m}(Q)\) with \(m\ge \frac{N}{p}+1.\) Then the solution of problem (1.1) found in Theorem (3.5) satisfies the following regularity:
If \(\lambda \ge \gamma \) and \(\frac{\alpha \lambda p^{p}}{(\lambda +p-1)^{p}}-\frac{\mu }{C_{N,p}}>0,\) then \(u^{\frac{\lambda +p-1}{p}}\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) and \(u\in L^{\infty }(0,T; L^{\lambda +1}(\Omega )).\)
Proof
Choosing \(u_{n}^{\lambda }\chi _{(0,t)}\) with \(\lambda >0\) as test function in (2.1)
From (1.3) and the fact that \(\frac{1}{(u_{n}+1)^{\gamma }}\le \frac{1}{u_{n}^{\gamma }}, u_{0}\in L^{\infty }(\Omega )\) we have
By Hardy inequality the later inequality implies
Passing to supremum for \(t\in [0,T]\) we get
applying Hölder inequality we conclude that
Using Sobolev inequality and by the above estimate, we get
By \(m\ge \frac{N}{p}+1\) we have \(m^{\prime }\le \frac{N+p}{N}\), then for all \(\lambda \ge \gamma ,\) we get \((\lambda -\gamma )m^{\prime }\le \frac{(\lambda -\gamma )(N+p)}{N} \le \frac{N(\lambda +p-1)}{N-p}\). Thus choosing \(\lambda \) such that \(\frac{\alpha \lambda p^{p}}{(\lambda +p-1)^{p}}-\frac{\mu }{C_{N,p}}>0.\) Using Hölder’s inequality in the later estimate, we have
Since \(\frac{p}{p^{*}}=\frac{N-p}{N}> \frac{(N-p)(\lambda -\gamma )}{N(\lambda +p-1)},\) then by Young inequality we deduce that
By the fact that \((\lambda -\gamma )m^{\prime }<\frac{N(\lambda +p-1)}{N-p},\) (4.8) and using Hölder inequality in (4.6), we obtain
Since \(\lambda \ge \gamma ,\) \(\left( \frac{\alpha \lambda p^{p}}{(\lambda +p-1)^{p}}-\frac{\mu }{C_{N,p}} \right) >0\) and the later estimate we deduce that the sequence \(u_{n}^{\frac{\lambda +p-1}{p}}\) is uniformly bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) and \(u_{n}\) is bounded \(L^{\infty }(0,T; L^{\lambda +1}(\Omega ))\) with respect to n for all \(\lambda \ge \gamma ,\) so \(u^{\frac{\lambda +p-1}{p}}\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) and \(u\in L^{\infty }(0,T; L^{\lambda +1}(\Omega ))\) for all \(\lambda \ge \gamma .\) This completed the proof of Theorem 4.2. \(\square \)
4.2 The case \(\gamma <1\)
Theorem 4.3
Let \(\gamma <1,\) and suppose that \(f\in L^{m}(Q), m\ge 1\) and
Then
- (i):
-
If \(\frac{p(N+2)}{p(N+2)-N(1-\gamma )}\le m<\frac{N}{p}+1,\) then the solution u of (1.1) found in Theorem 3.3, satisfies the following regularity \(u\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\sigma }(Q),\) with \(\sigma =m\frac{N(p+\gamma -1)+p(\gamma +1)}{N-pm+p}.\)
- (ii):
-
If \(1\le m<\frac{p(N+2)}{p(N+2)-N(1-\gamma )},\) then there exists a weak solution u of problem (1.1) such that \(u\in L^{q}(0,T; W_{0}^{1,q}(\Omega ))\cap L^{\sigma }(Q),\) with
$$\begin{aligned} q=m\frac{N(p+\gamma -1)+p(\gamma +1)}{N+2-m(1-\gamma )}\; \text{ and } \; \sigma =m\frac{N(p+\gamma -1)+p(\gamma +1)}{N-pm+p}. \end{aligned}$$ - (iii):
-
If \(m\ge \frac{N}{p}+1\) and \(0< \mu <\alpha C_{N,p},\) then the solution u of (1.1) found in Theorem 3.3 satisfies the following regularity \(u\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\infty }(Q).\)
Proof
Taking \(\varphi (u_{n})=((u_{n}+a)^{\lambda }-a^{\lambda })\chi _{(0,t)}\) as a test function in (2.1), \(0<a< \frac{1}{n}, \; \lambda >0\) and using the ellipticity condition (1.3) we have
By the fact that \(\frac{1}{(u_{n}+\frac{1}{n})^{\gamma }}\le \frac{1}{(u_{n}+a)^{\gamma }}\) and \(u_{n}^{p-1}(u_{n}+a)^{\lambda }\le (u_{n}+a)^{\lambda +p-1},\) we obtain
where \(\Psi (s)=\displaystyle \int _{0}^{s}\varphi (\ell )d\ell .\) Since \((u_{n}+a)^{\lambda -1}|\nabla u_{n}|^{p}=\frac{p^{p}}{(\lambda +p-1)^{p}}|\nabla (u_{n}+a)^{\frac{\lambda +p-1}{p}}|^{p},\) then the last estimate becomes
Since \(u_{0}\in L^{\infty }(\Omega ),\) applying Hölder and Hardy inequalities, we find that
If \(\lambda \ge 1,\) by definition of \(\varphi (u_{n})\) and \(\Psi (u_{n}),\) we reach that
Therefore we obtain that
Observing that \(u_{n}^{\lambda +1}(x,t)=\left( u_{n}^{\frac{\lambda +p-1}{p}}(x,t)\right) ^{\frac{p(\lambda +1)}{\lambda +p-1}},\) then the last inequality becomes
Now passing to the supremum for \(t\in (0,T),\) we obtain
From (4.12) and applying Lemma 1.1, we have
where \(C=C(\alpha , \lambda , m, p, \mu , C_{N,p}, C_{G}, ||u_{0}||_{L^{\infty }(\Omega )}).\) Thus we get
Letting \(a\rightarrow 0,\) we reach that
choosing \(\lambda \) such that
this equivalent to
From (4.14), the estimate (4.13) becomes
The condition \(\frac{p(N+2)}{p(N+2)-N(1-\gamma )}\le m<\frac{N}{p}+1,\) ensure that \(\lambda \ge 1\) and \((\frac{p}{N}+1)\frac{1}{m^{\prime }}<1,\) and thanks to Young inequality we deduce that
From (4.9) and, by the fact that and \(|\nabla u_{n}|^{p}\le (u_{n}+a)^{\lambda -1}|\nabla u_{n}|^{p}\) \((a>0, \; \lambda \ge 1)\) going back to (4.12) and using (4.14), (4.15) yield that
Then by estimates (4.15) and (4.16) we deduce that the sequence \(u_{n}\) is bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega ))\) and in \(L^{\sigma }(Q)\) with respect to n, and so \(u\in L^{p}(0,T; W_{0}^{1,p}(\Omega )\cap L^{\sigma }(Q).\) Hence the proof of item (i) is achieved.
Now we prove item (ii). Let now taking \(\gamma<\lambda <1\) and by definition of \(\varphi (u_{n})\) and \(\Psi (u_{n}),\) we can get
From (4.17) and going back to (4.10), we have
We proceed as before, we obtain that
Thanks to Lemma 1.1 and repeat the above process, it hold that
Let now choosing \(\lambda \) such that
this yields that
Since \(\lambda <1,\) then \(m<\frac{p(N+2)}{p(N+2)-N(1-\gamma )}<\frac{N}{p}+1,\) and \((\frac{p}{N}+1)\frac{1}{m^{\prime }}<1,\) from (4.19), (4.20) and thanks to Young inequality it hold that
By (4.9), then we have \(\frac{\lambda \alpha p^{p}}{(\lambda +p-1)^{p}}-\frac{\mu }{C_{N,p}}>0.\) Let \(1<q<p,\) applying Hölder’s inequality and from (4.18), we get
we take q such that
this equivalent to \(q=m\frac{N(p+\gamma -1)+p(\gamma +1)}{N+2- m(1-\gamma )}.\) Using (4.23) in (4.22) and letting \(a\rightarrow 0,\) we hold that
From (4.21) it follows that
Therefore estimates (4.21) and (4.24) imply that the sequence \(u_{n}\) is bounded in \(L^{q}(0,T; W_{0}^{1,q}(\Omega ))\) and in \(L^{\sigma }(Q)\) with respect to n, and so \(u\in L^{q}(0,T; W_{0}^{1,q}\) \((\Omega ))\cap L^{\sigma }(Q).\)
Now we give the proof of item (iii). Taking \(G_{k}(u_{n})\chi _{(0,t)}\) as a test function in (2.1) for \(t\in (0,T),\) we have
We observe that the function \(G_{k}(u_{n})\) is different from zero only on the set \(A_{k,n}=\{(x,t)\in Q: u_{n}(x,t)>k\},\) and that, on this set, we have \(u_{n}+\frac{1}{n}\ge k\ge 1.\) Note that
and
applying Hardy inequality and using the fact that \(G_{k}(u_{n})\le u_{n}\) in the set \(A_{k,n},\) we can write
Inequality (4.25) becomes
Passing to the supremum in \(t\in (0,T),\) we get
From now on, we can follow the standard technique used for the non-singular case in [7], we deduce there exist a constant \(C_{\infty }\) independent of n such that
Now taking \(u_{n}\) as a test function in (2.1), by (1.3) and Hardy inequality, we have
Since \(u_{0}\in L^{\infty }(\Omega )\) and by (4.26) and Hölder’s inequality,we have
As \(0\le \mu <\alpha C_{N,p},\) and by the last estimate, we obtain
where C is a positive constant independent of n. Hence the proof of Theorem 4.3 is completed. \(\square \)
In the following Theorem we are interesting to prove regularity of u solution of (1.1) when the datum f belong to \(L^{r}(0,T; L^{q}(\Omega )),\) with \(r,q>1.\)
Theorem 4.4
Under the hypothesis (1.3)–(1.5), if \(0<\gamma <1\) and \(0\le \mu <\alpha M C_{N,p},\) with
\(f\in L^{r}(0,T; L^{q}(\Omega ))\) with q and r be real numbers such that
where
Then there exists a weak solution \(u\in L^{p}(0,T; W_{0}^{1,p}(\Omega ))\cap L^{\delta }(Q)\) to problem (1.1) with
Remark 4.5
If \(\gamma , \mu \rightarrow 0,\) then the result of Theorem 4.4 coincides with classical regularity results for parabolic problems with coercivity (see [10, Theorem 1.1]).
Proof
Let now testing (1.3) by \(\varphi (u_{n})=((u_{n}+a)^{\lambda }-a^{\lambda })\chi _{(0,t)},\) \(0<a<\frac{1}{n},\;\; \lambda >0\) and repeating the same passage of proof of item (i) of Theorem 4.3 in order to arrive to the following inequality
Passing to supremum for \(t\in [0,T],\) we obtain
Setting \(v_{n}=u_{n}^{\frac{\lambda +p-1}{p}}\) and \(I= \displaystyle \iint _{Q}f(u_{n}+a)^{\lambda -\gamma }\), formula (4.28) can be rewritten as
Using Hölder’s inequality twice, for all \(q>1\) and \(r>1\) we get
Let us define \(\eta \in (0,1)\) such that
Thus, by the Lemma 1.1, applied to
we have
Integrating on time we obtain
If \(\eta <1,\) applying the Young inequality with exponents
we deduce
Letting \(\epsilon =\frac{c_{0}}{2C_{f}}\) and collecting (4.29), (4.30) and (4.35), we have
Now we choose \(\lambda \) satisfying
such that \(\lambda>\gamma>0,\; r>1\) and \(0\le \mu <\frac{\alpha \lambda p^{p}C_{N,p}}{(\lambda +p-1)^{p}}.\) From (4.37), it hold that
Since, from (4.37)
we have
and so
with \(\beta <1.\) If \(\eta =1,\) choosing \(\lambda \) as in (4.37), (4.35) becomes (4.39) with \(\beta =\frac{r-1}{r}<1.\) Thus from (4.39) immediately follows
Thanks to Lemma 1.1, we obtain
where \(\sigma = p\frac{N+\frac{p(\lambda +1)}{\lambda +p-1}}{N}.\) Recalling the definition of \(v_{n}\) we thus have proved that
where \(c_{3}\) is a positive constant independent of n, and
From (4.31) and (4.37), we deduce that
which implies, by (4.43)
we now have to check that \(\lambda \ge 1\) and that \(\eta ,\) defined in (4.31), belong to (0, 1). After easy calculations, we obtain that \(\lambda \ge 1\) if and only if
while the condition \(\eta \le 1\) hold is satisfied and only if
The condition \(\eta \ge 0\) is automatically satisfied if \(\lambda \ge 1.\)
It remains to prove the estimate in \(L^{p}(0,T; W_{0}^{1,p}(\Omega )).\) By (4.28), (4.30), (4.40) and \(\lambda \ge 1,\) we obtain
then the sequence \(u_{n}\) bounded in \(L^{p}(0,T; W_{0}^{1,p}(\Omega )),\) and so \(u\in L^{p}(0,T; W_{0}^{1,p}\)
\((\Omega )).\) The estimates (4.41) and (4.44) completed the proof of Theorem 4.4. \(\square \)
References
Abdellaoui, B., Miri, E.H.S., Peral, I., Touaoula, M.T.: Some remarks on quasilinear parabolic problems with singular potential and a reaction term. Nonlinear Differ. Equ. Appl. 21, 453–490 (2014)
Abdellaoui, B., Primo, A.: Influence of the Hardy potential in a semilinear heat equation. Proc. R. Soc. Edinb. 139A, 897–926 (2009)
Abdellaoui, B., Attar, A.: Quasilinear elliptic problem with Hardy potential and singular term. Commun. Pure Appl. Anal. 12(3), 1363–1380 (2013)
Abdellaoui, B., Peral, I.: Existence and nonexistence results for quasilinear elliptic equations involving the \(p\)-Laplacian a critical potential. Ann. Math. 182, 247–270 (2003)
Adimurth, A., Boccardo, L., Cirmi, G.R., Orsina, L.: The regularizing effect of lower order terms in elliptic problems involving Hardy potential. Adv. Nonlinear Stud. 17(2), 311–317 (2017)
Aguilar, J.A., Peral, I.: Global behavior of the Cauchy problem for some critical nonlinear parabolic equations. SIAM J. Math. Anal. 31(6), 1270–1294 (2000)
Aronson, D.G., Serin, J.: Local behavior of solutions to a quasilinear parabolic equations. Arch. Ration. Mech. Anal. 25, 81–122 (1967)
Baras, P., Goldstein, J.A.: The heat equations with singular potential. Trans. Am. Math. Soc. 284(1), 121–139 (1984)
Boccardo, L., Gallouet, T.: Nonlinear elliptic and parabolic equations involving measure data. J. Funct. Anal. 87, 149–169 (1989)
Boccardo, L., Dall’Aglio, A., Galloët, T., Orsina, L.: Existence and regularity results for some nonlinear parabolic equations. Adv. Math. Sci. Appl. 9, 1017–1031 (1999)
De Bonis, I., De Cave, L.M.: Degenerate parabolic equations with singular lower order terms. Differ. Integr. Equ. 27(9/10), 949–976 (2014)
DiBenedetto, E.: Degenerate Parabolic Equations. Springer, New York (1993)
El Hadfi, Y., Benkirane, A., Yousfi, A.: Existence and regularity results for parabolic equations with degenerate coercivity. Complex Var. Ellipt. Equ. 63(5), 517–529 (2017)
El Hadfi, Y., El Ouardy, M., Ifzarne, A., Sbai, A.: On nonlinear parabolic equations with singular lower order term. J. Ellip. Parabol. Equ. (2021). https://doi.org/10.1007/s41808-021-00138-5
El Ouardy, M., El Hadfi, Y., Ifzarne, A.: Existence and regularity results for a singular parabolic equations with degenerate coercivity. Discrete Contin. Dyn. Syst. S 15(1), 117–141 (2022)
El Ouardy, M., El Hadfi, Y.: Some nonlinear parabolic problems with singular natural growth term. Results Math. 77(3), 1–25 (2022)
Farroni, F., Greco, L., Moscariello, G., Zecca, G.: Nonlinear evolution problems with singular coefficients in the lower order terms. Nonlinear Differ. Equ. Appl. 28, 38 (2021)
Fulks, W., Maybee, J.S.: A singular non-linear equations. Osaka J. Math. 12, 1–19 (1960)
García Azorero, J.P., Peral, I.P.: Hardy inequalities and some critical elliptic and parabolic problems. J. Differ. Equ. 144, 441–476 (1998)
Gatica, J.A., Oliker, V., Waltman, P.: Singular nonlinear boundary-value problems for second order ordinary differential equations. J. Differ. Equ. 79, 62–78 (1989)
Ladyženskaja, O., Solonnikov, V.A., Ural’ceva, N.N.: Linear and Quasilinear Equations of Parabolic Type. Translations of the American Mathematical Society, American Mathematical Society, Providence (1968)
Lions, J.L.: Quelques méthodes de resolution des probl‘emes aux limites non linéaires. Dunod, Paris (1969)
Nachman, A., Callegari, A.: A nonlinear singular boundary value problem in the theory of pseudoplastic fluids. SIAM J. Appl. Math. 28, 271–281 (1980)
Nowosad, P.: On the integral equations \(kf=1/f\) arising in a problem in communication. J. Math. Appl. 14, 484–492 (1966)
Oliva, F., Petitta, F.: A nonlinear parabolic problem with singular terms and nonregular data. Nonlinear Anal. 194, 111472 (2020)
Porzio, M.M.: Quasilinear Parabolic and Elliptic Equations with Singular Potentials. In: Wood, D., de Gier, J., Praeger, C.E., Tao, T. (eds.) 2017 MATRIX Annals. MATRIX Book Series, vol. 2. Springer, Cham. (2019)
Porzio, M.M.: On uniform and decay estimates for unbounded solutions of partial differential equations. J. Differ. Equ. 259, 6960–7011 (2015)
Porzio, M.M., Primo, A.: Summmability and existence results for quasilinear parabolic equations with Hardy potential term. Nonlinear Differ. Equ. Appl. 20, 65–100 (2013)
Porzio, M.M.: Local regularity results for some parabolic equations. Houst. J. Math. 25(4), 769–792 (1999)
Sbai, A., El Hadfi, Y.: Degenerate elliptic problem with a singular nonlinearity. Complex Var. Elliptic Equ. (2021). https://doi.org/10.1080/17476933.2021.2014458
Sbai, A., El Hadfi, Y.: Regularizing effect of absorption terms in singular and degenerate elliptic problems. arXiv preprint arXiv:2008.03597 (2020)
Simon, J.: Compact sets in the space \(L^{p}(0, T; \, B)\). Ann. Mat. Pura Appl. 146, 65–96 (1987)
Tyagi, J.: An existence of positive solution to singular elliptic equations. Boll. Union Mat. Ital. 7, 45–53 (2014)
Youssfi, A., El Benkirane, A., Hadfi, Y.: On bounded solutions for nonlinear parabolic equations with degenerate coercivity. Mediterr. J. Math. 13, 3029–3040 (2016)
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
About this article
Cite this article
Ouardy, M.E., Hadfi, Y.E. & Sbai, A. Existence of positive solutions to nonlinear singular parabolic equations with Hardy potential. J. Pseudo-Differ. Oper. Appl. 13, 28 (2022). https://doi.org/10.1007/s11868-022-00457-8
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11868-022-00457-8