Z. Angew. Math. Phys.

(2021) 72:2
c 2020 Springer Nature Switzerland AG Zeitschrift für angewandte
https://doi.org/10.1007/s00033-020-01430-5 Mathematik und Physik ZAMP

Existence of solution for a class of heat equation involving the p(x) Laplacian with triple
regime

Claudianor O. Alves and Tahir Boudjeriou

Abstract. In this paper, we study the local and global existence of solution and the blow-up phenomena for a class of heat
equation involving the p(x)-Laplacian with triple regime.

Mathematics Subject Classification. 35K59, 65M60, 35B44.

Keywords. Quasilinear parabolic equations, Galerkin Methods, Blow-up.

1. Introduction

In this paper, we study the local and global existence of solution for the following class of heat equation
⎧
⎨ ut − Δp(x) u = |u|q(x)−2 u in Ω × (0, +∞),
u = 0, on ∂Ω × (0, +∞), (1.1)
⎩
u(x, 0) = u0 (x), in Ω,
1,p(x)
where Ω ⊆ RN is a smooth bounded domain and u0 ∈ W0,rad (Ω) and p, q : Ω → R are continuous
functions satisfying some conditions that will be mentioned later on.
When p and q are constant functions, the problem above becomes

ut − Δp u = |u|q−2 u in Ω × (0, +∞),
(1.2)
u(x, 0) = u0 (x), in Ω.

The methods used to solve that problem are developed in relationship with the values of q with respect
to the Sobolev critical exponent p∗ of p, which is defined by
⎧
⎨ Np
if 1 < p < N
p∗ = N −p
⎩ +∞ if p ≥ N,

and only one of the situations below can occur:

(i) q < p∗ (subcritical case).
(ii) q = p∗ , provided that 1 < p < N (critical case).
(iii) q > p∗ , provided that 1 < p < N (supercritical case).

C. O. Alves was partially supported by CNPq/Brazil 304804/2017-7.

0123456789().: V,-vol
 2 Page 2 of 18 C. O. Alves and T. Boudjeriou ZAMP

In the case of variable exponents, the problem (1.2) becomes more rich, because it can fulfill even a
“subcritical–critical–supercritical” triple regime, in the sense that we can have Ω = Ω1 ∪ Ω2 ∪ Ω3 with

q(x) < p∗ (x) if x ∈ Ω1 ;

q(x) = p∗ (x) if x ∈ Ω2 ; (1.3)
q(x) > p∗ (x) if x ∈ Ω3 .

In the last few decades, special attention has been paid to the study of partial differential equations in-
volving p(x)-growth conditions. The interest in studying such problems is motivated by their applications
in image restoration, nonlinear elasticity theory, electrorheological fluids, and so forth. In particular, par-
abolic equations involving the p(.)-Laplacian are related to the field of electrorheological fluids which are
characterized by their ability to change the mechanical properties under the influence of the exterior elec-
tromagnetic; for a more physical motivation we refer the reader to [13,22] and [23] and their references.
The rigorous study for these physical problems has been facilitated by the development of Lebesgue and
Sobolev spaces with variable exponents.
It is important to point out that many results have obtained on parabolic equations with nonlinearities
of variable exponent where the authors have studied the global existence in the subcritical case (q < p∗ ),
for example, see ([5,6,24,27–30,32]) and the references therein. In [33], by using the Galerkin method,
the authors have studied the global existence and asymptotic behavior of global weak solutions for the
following class of problem
⎧ 

⎪ n
∂u p(x)−2 ∂u
⎪
⎨ ut − ∂
∂xi + |u|σ(x) = 0, x ∈ Ω, t > 0,
∂xi ∂xi
i=1 (M)
⎪
⎪ u(x, t) = 0, x ∈ ∂Ω, t > 0,
⎩
u(x, 0) = u0 (x), x∈Ω

for p+ < σ − + 1 ≤ σ(x) + 1 ≤ σ + + 1 < p∗ (x). In [31], by using the sub-differential approach, Akagi and
Mastsuura obtained the well-posedness of solutions for problem (1.1) with f (x, t) instead of |u|q(x)−2 u.
Moreover, the large time behavior of solutions also is considered. The asymptotic stability for Kirchhoff
systems with variable exponent growth conditions has been discussed by G. Autuori and P. Pucci in [34].
We recall that in the case when p(x) and q(x) are constant functions, the global existence and blow-up
of solutions for the problem (1.1) have been studied by many authors; we refer to ([4,9–11,15,16,18])
for interested reader. Finally, for the existence theorem of solutions for elliptic problems with variable
exponents we refer to ([1–3,35,36]) and the references therein.
Motivated by the above works, the main goal this paper is to study problem (1.1) by consideration of
the triple regime mentioned in (1.3), which is a novelty for this class of problem.
The plan of the paper is as follows: In Sect. 2, we recall some facts involving the variable exponent
Lebesgue and Sobolev spaces. In Sect. 3, we assume that Ω is a ball centered at origin and show the
existence local and global of solution for (1.1) with triple regime. In Sect. 4, we work with the non-radial
case; here Ω is not a ball centered at origin. Again, it is showed the existence local and global of solution
for (1.1) with triple regime. Finally, in Sect. 5, we establish the blow-up phenomena for a general situation
that also involves a triple regime.

2. Variable exponent Lebesgue and Sobolev spaces

In this section, we recall some results on variable exponent Lebesgue and Sobolev spaces. For more details,
we refer to [8,12,17] and their references.
ZAMP Existence of solution for a class of heat equation. . . Page 3 of 18 2

Let Ω ⊂ RN be a smooth bounded domain and p ∈ L∞ (Ω) with p− := essinf x∈RN p(x) > 1. The
variable exponent Lebesgue space Lp(x) (Ω) is defined by
⎧ ⎫
⎨  ⎬

Lp(x) (Ω) = u : Ω → R u is measurable and
p(x)
|u| dx < ∞
⎩ ⎭
Ω

endowed with the norm

⎧ ⎫
⎨  ⎬
u p(x)
|u|p(x) = inf λ > 0 dx ≤ 1 .
⎩ λ ⎭
Ω

The variable exponent Sobolev space is defined by

 

W 1,p(x) (Ω) = u ∈ Lp(x) (Ω) |∇u| ∈ Lp(x) (Ω)
with the norm
u1,p(x) = |u|p(x) + |∇u|p(x)
With these norms, the spaces Lp(x) (Ω) and W 1,p(x) (Ω) are reflexive and separable Banach spaces.
(Ω) is defined as the closure of C0∞ (Ω) in W0
1,p(x) 1,p(x)
The space W0 (Ω) with respect to the above
1,p(x)
norm. Moreover, by Poincaré inequality, if p ∈ C(Ω), then u = |∇u|p(x) is a norm in W0 (Ω), and
it is equivalent to norm u1,p(x) .

Proposition 2.1. The functional ρ : W 1,p(x) (Ω) → R defined by

  
p(x) p(x)
ρ(u) = |∇u| + |u| dx, (2.1)
Ω

has the following properties:

p− p+
(i) If u1,p(x) ≥ 1, then u1,p(x) ≤ ρ(u) ≤ u1,p(x) .
p+ p−
(ii) If u1,p(x) ≤ 1, then u1,p(x) ≤ ρ(u) ≤ u1,p(x) .
In particular, ρ(u) = 1 if and only if u1,p(x) = 1 and if (un ) ⊂ W 1,p(x) (Ω), then un 1,p(x) → 0 if and
only if ρ(un ) → 0.

Remark 1. For the functional ξ : Lp(x) (Ω) → R given by


p(x)
ξ(u) = |u| dx,
Ω

the conclusion of Proposition 2.1 also holds; for example, if (un ) ⊂ Lp(x) (Ω), then |un |p(x) → 0 if and
only if ξ(un ) → 0. Moreover, from (i) and (ii),
⎧⎛ ⎞1/p− ⎛ ⎞1/p+ ⎫
⎪
⎨   ⎪
⎬
|u|p(x) ≤ max ⎝ |u|
p(x)
dx⎠ , ⎝ |u|
p(x)
dx⎠ . (2.2)
⎪
⎩ ⎪
⎭
Ω Ω

Related to the Lebesgue space Lp(x) (Ω), we have the following generalized Hölder-type inequality.
Proposition 2.2. ([19] [p. 9]) For p ∈ L∞ (Ω) with p− > 1, let p : Ω → R be such that
1 1
+  = 1, a.e. x ∈ RN .
p(x) p (x)
 2 Page 4 of 18 C. O. Alves and T. Boudjeriou ZAMP


Then, for any u ∈ Lp(x) (Ω) and v ∈ Lp (x) (Ω),

 

uv dx ≤ 1 + 1 |u|
p(x) |v|p (x) . (2.3)
p− p−
Ω

Proposition 2.3. ([12][Theorems 1.1, 1.3]) Let p : Ω → R be a Lipschitz continuous satisfying 1 < p− ≤
p+ < N and t : Ω → R be a continuous function.
(i) If 1 ≤ t ≤ p∗ , the embedding W 1,p(x) (Ω) → Lt(x) (Ω) is continuous.
(ii) If 1 ≤ t  p∗ , the embedding W 1,p(x) (Ω) → Lt(x) (Ω) is compact.
Here t  p∗ means inf (p∗ (x) − t(x)) > 0.
x∈Ω

3. Problem 1: the radial case

In this section, Ω = BR (0) and p, q : B R (0) → R are continuous functions satisfying :

2 ≤ p− = min p(x) ≤ max p(x) = p+ < N and 1 < q− = min q(x). (H1 )
x∈B R (0) x∈B R (0) x∈B R (0)

p(x) = p(|x|) and q(x) = q(|x|), ∀x ∈ B R (0). (H2 )

Our intention is to prove the existence of local and global solution for (1.1).

1,p(x)
3.1. Existence of local solution in W0 (BR )

In this subsection, we will show the local existence of solution for problem (1.1) with triple regime. In
addition to conditions (H1 ) − (H2 ), we assume that there is r ∈ (0, R) such that
 
p∗ (x)
1 < q− ≤ q(x) ≤ min p∗ (x), 1 + , ∀x ∈ B r (0). (H3 )
2
Note that there is a control on the function q in the ball B r (0), but there are no hypotheses on the
function q in the annulus AR,r = B R (0) \ Br (0); hence, close to the boundary can exist there subsets
Ω1 , Ω2 and Ω3 such that q is subcritical in Ω1 , critical in Ω2 and supercritical in Ω3 , which characterizes
a triple regime.
Note that for any r0 ∈ (0, R) we have the continuous embedding
W 1,p(x) (BR (0)) → W 1,p− (AR,r0 )
and the compact embedding
1,p
Wrad − (AR,r0 ) → C(AR,r0 ),
which is due to Strauss [20]. Therefore, the embedding
1,p(x)
Wrad (BR (0)) → C(AR,r0 ), (3.1)
is compact, where
1,p(x)
Wrad (BR (0)) = {u ∈ W 1,p(x) (BR (0)) : u(x) = u(|x|) a.e x ∈ BR (0)}.
Hence, it follows that the embedding
1,p(x)
W0,rad (BR (0)) → Lq(x) (BR (0)), (3.2)
ZAMP Existence of solution for a class of heat equation. . . Page 5 of 18 2

is also compact, which is crucial in our approach. From this, the energy functional
1,p(x)
E : W0,rad (BR (0)) → R given by
 
1 1
E(u) = |∇u|p(x)
dx − |u|q(x) dx
p(x) q(x)
BR (0) BR (0)
1,p(x)
belongs to C 1 (W0,rad (BR (0)), R) with
 
 1,p(x)
E (u)v = |∇u| p(x)−2
∇u∇v dx − |u|q(x)−2 uv dx, ∀u, v ∈ W0,rad (BR (0)).
BR (0) BR (0)
1,p(x)
Hereafter, we endow W0,rad (BR (0)) with the norm
u = |∇u|p(x)
and BR denotes the ball BR (0).
The main result in this section has the following statement:
1,p(x)
Theorem 3.1. (Local weak solution) Assume (H1 ) − (H3 ), Ω = BR and u0 ∈ W0,rad (BR ). Then, there
L∞ ([0, T ], W0
1,p(x)
exist T > 0 and a function u ∈ (BR )) with ut ∈ L2 ((0, T ), L2 (BR )) such that
1,p(x)
u(0) = u0 ∈ W0 (BR ),
1,p(x)
and for each v ∈ W0 (BR )
  
ut (t)v dx + |∇u(t)|p(x)−2
∇u∇v dx = |u(t)|q(x)−2 u(t)v dx, a.e. t ∈ (0, T ).
BR BR BR

In order to prove Theorem 3.1, let us introduce some useful properties of subdifferentials of proper,
convex and lower semicontinuous functional on a Hilbert space.
Let H be a Hilbert space with the inner product (., .)H and the norm  . H . For a functional ϕ from
H to (−∞, +∞], we shall write

D(ϕ, δ) = {u ∈ H, ϕ(u) ≤ δ} for δ ∈ R and D(ϕ) = D(ϕ, δ). (3.3)
δ∈R

If ϕ : H → (−∞, +∞] is a convex functional, the subdifferential ∂ϕ of ϕ is defined by

∂ϕ(u) = {w ∈ H, ϕ(v) − ϕ(u) ≥ (w, v − u)H , ∀v ∈ H}.
It is well known that the subdifferential ∂ϕ is a maximal monotone operator and D(∂ϕ) ⊂ D(ϕ). The
following type chain rule of subdifferentials is taken from [7][Lemma 3.3, p. 73].
Lemma 3.2. Let ϕ : H → (−∞, +∞] be a proper, convex and lower semicontinuous functional. For some
T > 0, let u ∈ W 1,1 (0, T, H) and u(t) ∈ D(∂ϕ) a.e in [0, T ]. If there exists a function f ∈ L2 (0, T, H)
such that f (t) ∈ ∂ϕ(u(t)) a.e in [0, T ], then the function t → ϕ(u(t)) is absolutely continuous on [0, T ]
and
d
ϕ(u(t)) = (f (t), u (t))H , a.e. in [0, T ].
dt
In what follows, let us recall a very important result found in [15][Theorem 3.4, p. 297].
Theorem 3.3. Let ϕ, φ : H → (−∞, +∞] be a proper, convex and lower semicontinuous functionals.
Under the following assumptions :
(1) the set D(ϕ, δ) is compact in H for any δ ∈ R,
(2) D(ϕ) ⊂ D(φ),
 2 Page 6 of 18 C. O. Alves and T. Boudjeriou ZAMP

(3) the set {(∂φ)0 (u), u ∈ D(ϕ, δ)} is bounded in H for any δ ∈ R.
where (∂φ)0 is the unique element of least norm. Then, for each w ∈ D(ϕ), there exist T > 0 and a
strong solution of this initial value problem
 du(t)
dt + ∂ϕ(u(t)) − ∂φ(u(t))  0, in H, 0 < t < T,
u(0) = w,
1,p(x)
Proof of Theorem 3.1. Hereafter, we will assume that u0 ∈ W0,rad (BR ) and H = L2 (BR ); then, by
Proposition 2.3 the embedding
1,p(x)
W0,rad (BR ) → L2 (BR ) (3.4)
is compact.
In what follows, let us consider the operators ϕ : L2 (BR ) → (−∞, +∞] and φ : L2 (BR ) → (−∞, +∞]
given by
⎧ 
⎨ p(x) |∇u|
1 p(x) 1,p(x)
dx, if u ∈ W0,rad (BR )
ϕ(u) = BR
⎩ 1,p(x)
+∞, if u ∈ L2 (BR ) \ W0,rad (BR )
and
 
q(x) |u| if u ∈ Lq(x) (BR )
1 q(x)
dx,
φ(u) = BR
+∞, if u ∈ L2 (BR ) \ Lq(x) (BR ).

Lemma 3.4. The set D(ϕ, δ) is compact in L2 (BR ) for any δ ∈ R. Moreover, D(ϕ) ⊂ D(φ).
Proof. This is an immediate consequence of (3.3) and (3.4). 
Lemma 3.5. The set {(∂φ)0 (u), u ∈ D(ϕ, δ)} is bounded in L2 (BR ) for any δ ∈ R.
Proof. It is enough to consider δ > 0. Since φ ∈ C 1 (Lq(x) (BR ), R), a simple computation gives that there
is unique hu ∈ L2 (BR (0)) such that (∂φ)(u) = {hu }, and so, (∂φ)0 (u) = {hu }. Moreover,

|u|q(x)−2 uv dx = (hu , v)L2 (BR ) , ∀v ∈ C0∞ (BR ).
BR

The above equality yields |u(x)|q(x)−2 u(x) = hu (x) a.e. in BR . From this,
|(hu , v)L2 (BR ) | ≤ |u(x)|q(x)−1 L2 (BR ) vL2 (BR ) , ∀v ∈ C0∞ (BR ).
A simple computation shows that
  
(p∗ )+ (p∗ )−
|u(x)|2(q(x)−1) dx ≤ C u + up∗ (x) + up∗ (x) ,
BR (0)

where
(p∗ )+ = max p∗ (x) and (p∗ )− = min p∗ (x).
x∈B R (0) x∈B R (0)

Using the fact that u ∈ D(ϕ, δ), there is C = C(δ) > 0 such that
(p∗ )+ (p∗ )−
u + up∗ (x) + up∗ (x) ≤ C, ∀u ∈ D(ϕ, δ).
This together with Proposition 2.1 yields
|u(x)|q(x)−1 L2 (BR ) ≤ C, ∀u ∈ D(ϕ, δ).
ZAMP Existence of solution for a class of heat equation. . . Page 7 of 18 2

Hence
hu L2 (BR ) = sup |(hu , v)L2 (BR ) | ≤ C, ∀u ∈ D(ϕ, δ),
vL2 (B ≤1
R)

showing the lemma. 

1,p(x)
From Theorem 3.3, for each u0 ∈ W0,rad (BR ) there exist T > 0 and a strong solution u ∈ D(ϕ) of
the value problem
 du(t)
dt + ∂ϕ(u(t)) − ∂φ(u(t))  0, in L (BR ), 0 < t < T,
2

u(0) = u0 .
Since
∂φ(u(t)) = {|u(t)|q(x)−2 u(t)}
and
∂ϕ(u(t)) = {ξu },
that is,

1,p(x)
−Δp(x) u(t), v = |∇u(t)|p(x) ∇u∇v dx = (ξu , v)L2 (BR )) , ∀v ∈ W0,rad (BR ),
BR
1,p(x)
the solution u must verify the equality below for all v ∈ W0,rad (BR )
  
ut (t)v dx + |∇u(t)|p(x)−2 ∇u(t)∇v dx = |u(t)|q(x)−2 u(t)v dx, (3.5)
BR BR BR

a.e. in t ∈ [0, T ]. Moreover, from Lemma 3.2,

t
ut (s)2L2 (Ω) ds + E(u(t)) = E(u0 ), a.e. in t ∈ [0, T ].
0

Here we will apply the Palais principle of symmetric criticality developed by Kobayashi and Otani
1,p(x)
[14][Theorem 2.2] to conclude that the function u is in fact a local weak solution in W0 (BR ). Hereafter,
1,p(x)
we set J ⊂ [0, T ] with med(J ) = 0 such that for all ϕ ∈ W0,rad (BR )
c

  
ut (t)ϕ dx + |∇u(t)|p(x)−2
∇u(t)∇ϕ dx = |u(t)|q(x)−2 u(t)ϕ dx, t ∈ J . (3.6)
BR BR BR

1,p(x)
Fixing t ∈ J and defining v = u(t) ∈ and f = ut (t) ∈ L2 (BR ), we derive that
W0,rad (BR )
  
1,p(x)
|∇v|p(x)−2
∇v∇ϕ dx = |v|q(x)−2
vϕ − f ϕ dx, ∀ϕ ∈ W0,rad (BR ). (3.7)
BR BR BR
1,p(x)
Our intention is to prove that the equality above holds in whole W0 (BR ), and an important tool in
this direction is the principle of symmetric criticality developed by Kobayashi and Otani [14][Theorem
2.2]. First of all, it is very important point out that (3.7) ensures that u is a critical point of the energy
functional
  
1 1 1,p(x)
E1 (w) = |∇w|p(x) dx − |w|q(x) dx − f w dx, w ∈ W0,rad (BR ),
p(x) q(x)
BR BR BR
 2 Page 8 of 18 C. O. Alves and T. Boudjeriou ZAMP

1,p(x)
which is not well defined in whole W0 (BR ). However, we cannot use directly the principle of symmetric
1,p(x)
criticality to guarantee that u is a critical point of E1 in W0 (BR ). In order to overcome this difficulty,
we will use the following idea: Consider the function
g(x, t) = ξ(|x|)|t|q(x)−2 t + (1 − ξ(|x|))|u(x)|q(x)−2 u(x), ∀x ∈ BR ,
∞
where ξ ∈ C ([0, R], R) satisfies

1, x ∈ x ∈ B r2 (0)
ξ(x) =
0, x ∈ x ∈ B R (0) \ B 3r5 (0).

Since u ∈ C(AR, r2 ) [see (3.1)], it follows from (H3 ) that

∗
|g(x, t)| ≤ C(|t|p (x)−1
+ 1), ∀(x, t) ∈ BR × R. (3.8)
Associated with the function g, we have the problem

−Δp(x) w = g(x, w) − f in BR ,
(Pg )
w = 0 on ∂BR ,
whose the associated energy is given by
  
1
I(w) = |∇w|p(x)
dx − G(x, w) dx − f u dx,
p(x)
BR BR BR

t 1,p(x)
where G(x, t) = g(x, s) ds. From (3.8), I is well defined in the whole space W0 (BR ), I ∈
0
1,p(x)
C 1 (W0 (BR ), R) and
  
I  (w)ψ =
1,p(x)
|∇w|p(x)−2 ∇w∇ψ dx − g(x, w)ψ dx − f ψ dx, ∀w, ψ ∈ W0 (BR ).
BR BR BR

Since
g(x, u(x)) = |u|q(x)−2 u(x), ∀x ∈ BR ,
1,p(x)
we see that u is a critical point of I restricted to W0,rad (BR ). Now we can apply the Palais principle
of symmetric criticality developed by Kobayashi and Otani [14][Theorem 2.2] to conclude that u is a
1,p(x) 1,p(x)
nontrivial critical point of I in the whole W0 (BR ). This shows that (3.7) holds for all ϕ ∈ W0 (BR );
1,p(x)
then, for all ϕ ∈ W0 (BR )
  
ut (t)ϕ dx + |∇u(t)|p(x)−2 ∇u(t)∇ϕ dx = |u(t)|q(x)−2 u(t)ϕ dx, t ∈ J ,
BR BR BR

finishing the proof of Theorem 3.1.

1,p(x)
3.2. Existence of global weak solution in W0 (BR )

In order to prove the existence of a global solution, we replaced (H3 ) by the following condition:
There exists 0 < r < R such that
p+ < q − = min q(x) ≤ q−
r
= min q(x) ≤ max q(x) = q+
r
< min p∗ (x). (H4 )
x∈B R (0) x∈B r (0) x∈B r (0) x∈B R (0)
ZAMP Existence of solution for a class of heat equation. . . Page 9 of 18 2

Arguing as in the last section, it is possible to prove that the hypotheses (H1 ) − (H2 ) and (H4 ) are
enough to conclude that the compact embedding (3.2) still holds with these assumptions. Associated with
the functional E, we have the Nehari manifold given by
⎧ ⎫
⎨   ⎬
1,p(x)
N = u ∈ W0,rad (BR ) \ {0} : |∇u|p(x) dx = |u|q(x) dx
⎩ ⎭
BR BR

and the real number

d = inf {E(u) : u ∈ N } .
Hereafter, let us assume that u0 ∈ W \{0} where
1,p(x)
W := {u ∈ W0,rad (BR ), E(u) < d, J(u) > 0} ∪ {0}, (3.9)
1,p(x)
with J : W0,rad (BR ) → R given by
 
J(u) = |∇u|p(x) dx − |u|q(x) dx. (3.10)
BR BR

One can show easily that the functional E restricted to N is bounded from below, and the assumption
u0 ∈ W \{0} leads to


1 1
d > E(u0 ) > − |∇u0 |p(x) dx > 0.
p+ q−
Ω

Theorem 3.6. (Global weak solution) Assume (H1 ) − (H2 ), (H4 ), Ω = BR and u0 ∈ W \ {0}. Then, there
is a function u ∈ L∞ ((0, +∞), W0
1,p(x)
(BR )) with ut ∈ L2 ((0, +∞), L2 (BR )) such that
1,p(x)
u(0) = u0 ∈ W0 (BR ),
1,p(x)
and for each v ∈ W0 (BR )
  
ut (t)v dx + |∇u(t)|p(x)−2 ∇u(t)∇v dx = |u(t)|q(x)−2 u(t)v dx a.e. t ∈ (0, +∞).
BR BR BR

Proof. Here we are going to use the Galerkin method to prove firstly the existence of global weak solutions
1,p(x)
for problem (1.1) in W0,rad (BR ). After that, we will show that the solution obtained is in fact a global
1,p(x) 1,p(x) 1,p(x)
weak solutions for problem (1.1) in W0 (BR ). Since W0,rad (BR ) is separable, and W0,rad (BR ) is
1,p(x)
dense in L2rad (BR ), we have a base V = {wj , j ∈ N} in W0,rad (BR ) and also in L2rad (BR ) such that
. 1,p(x)
V = W0,rad (Ω) and (wi , wj )L2rad (BR ) = δi,j , i, j = 1, 2, . . . . (3.11)
1,p(x)
Thereby, as u0 ∈ W0,rad (BR ), there exists {aim , i = 1, . . . , m} such that

m
1,p(x)
um (0) = aim wi → u0 in W0,rad (BR ). (3.12)
j=1


m
For each m, we look for the approximate solutions um (x, t) = gim (t)wi (x) satisfying the following
i=1
identities :
  
(um )t wj dx + |∇um |p(x)−2 ∇um ∇wj dx = |um |q(x)−2 um wj dx, j ∈ {1, . . . , m}, (3.13)
BR BR BR
 2 Page 10 of 18 C. O. Alves and T. Boudjeriou ZAMP

with the initial condition

um (0) = u0m . (3.14)
Then, (3.13) and (3.14) is equivalent to the following initial value problem for a system of nonlinear
ordinary differential equations on gjm :
 
gjm (t) = Hj (g(t)), j = 1, 2, . . . , m, t ∈ [0, t0 ],
(3.15)
gjm (0) = ajm , j = 1, 2, . . . , m,
 
where Hj (g(t)) = − |∇um |p(x)−2 ∇um ∇ϕj dx+ |um |q(x)−2 um ϕj dx. By the Picard iteration method,
BR BR
there is t0,m > 0 depending on |ajm | such that problem (3.15) admits a unique local solution gjm ∈
C 1 ([0, t0,m ]).
Hereafter, the letters c, ci , C, Ci , i = 1, 2, . . . , denote positive constants which vary from line to line,
but they are independent of m.

Multiplying the j th equation in (3.13) by gjm (t) and summing over j from 1 to m, afterward integrating
over (0, t), we find
t
(um )s (s)22 ds + E(um (t)) = E(u0m ), t ∈ [0, t0,m ]. (3.16)
0
1,p(x) 1,p(x)
Since u0m converges to u0 strongly in W0,rad (BR ), the continuity of E on W0,rad (BR ) ensures that
E(u0m ) → E(u0 ), as m → +∞.
From the assumption E(u0 ) < d, we have E(u0m ) < d for sufficiently large m. This combined with (3.16)
yields
E(um (t)) < d, t ∈ [0, t0,m ], (3.17)
for sufficiently large m. We will show that t0,m = +∞ and
um (t) ∈ W, ∀t ≥ 0, (3.18)
for sufficiently large m. Suppose by contradiction that um (t1 ) ∈
/ W for some t1 ∈ [0, t0,m ]. Let t∗ ∈ [0, t0,m ]
be the smallest time for which um (t∗ ) ∈
/ W . Then, by continuity of um (t), we get um (t∗ ) ∈ ∂W . Hence,
it turns out that
E(um (t∗ )) = d (3.19)
or
J(um (t∗ )) = 0. (3.20)
It is clear that (3.19) could not occur by (3.17) while if (3.20) holds, then by the definition of d,
E(um (t∗ )) ≥ inf E(u) = d,
u∈N

which also is a contradiction with (3.17). Consequently, (3.18) is true.

Combining (3.17) and (3.18),



1 1
d > E(um (t)) > − |∇um (t)|p(x) dx, t ∈ [0, t0,m ],
p+ q−
BR

for sufficiently large m. Moreover, by (3.16),

t 

1 1
(um )s (s)22 ds + − |∇um (t)|p(x) dx < d, t ∈ [0, t0,m ], (3.21)
p+ q−
0 BR
ZAMP Existence of solution for a class of heat equation. . . Page 11 of 18 2

from where it follows that t0,m = +∞. Thus, (3.21) ensures the existence of a function u ∈ L∞ ((0, +∞),
W0,rad (BR )) with ut ∈ L2 ((0, +∞), L2 (BR )) such that for a subsequence of {um }∞
1,p(x)
m=1 , still denoted by
{um }∞ ,
m=1
⎧
∗
⎪
⎨ um  u in L∞ ((0, +∞), W0,rad (BR )),
1,p(x)

(um )t  ut in L2 ((0, +∞), L2 (BR )), (3.22)

⎪
⎩ ∗ −1,p (x)
−div(|∇um | p(x)−2
∇um )  χ in L∞ ((0, +∞), Wrad (BR )).
Moreover, from [21] [Proposition 1.3]for all T > 0 we have
um → u in L2 (0, T, L2 (BR )) (3.23)
and
um → u a.e. (x, t) ∈ BR × (0, T ). (3.24)
The next step is to prove that
− div(|∇u|p(x)−2 ∇u) = χ. (3.25)
It is well known that Au = −div(|∇u| ∇u) is bounded, monotone and hemicontinuous from
p(x)−2
1,p(x) −1,p (x)
W0,rad (BR ) → Wrad (BR ) (see [8] [Section 13.4, p. 421]). In addition,

1,p(x)
Au, u = |∇u|p(x) dx, ∀u ∈ W0,rad (BR ).
BR

For this purpose, for each T > 0 fixed, we can apply [3][Proposition 2.6] to get
T  T 
|um |
q(x)−2
um v dxdt → |u|q(x)−2 uv dxdt, ∀v ∈ Lq(x) (BR ). (3.26)
0 BR 0 BR

Moreover, we also claim that

T T
lim sup Aum , um  dt = χ, u dt. (3.27)
m→∞
0 0

Indeed, integrating (3.13) over (0, T ),

T  T T 
(um )s wj dxds + Aum , wj  ds = |um |q(x)−2 um wj dxds, j ∈ {1, . . . , m}. (3.28)
0 BR 0 0 BR

Therefore, for any fixed j, (3.22), (3.26) and (3.28) give

T  T T 
us wj dxds + χ, wj  ds = |u|q(x)−2 uwj dxds, j ∈ {1, . . . , m}.
0 BR 0 0 BR

1,p(x) 1,p(x)
Accordingly, from the density of V in W0,rad (BR ), it follows that for all ϕ ∈ W0,rad (BR )

T  T T 
us ϕ dxds + χ, ϕ ds = |u|q(x)−2 uϕ dxds. (3.29)
0 BR 0 0 BR
 2 Page 12 of 18 C. O. Alves and T. Boudjeriou ZAMP

On the other hand, multiplying the jth equation in (3.13) by gjm (t) and summing over j from 1 to m,
afterward integrating over (0, T ), we get
T T 
1 1
um (T )22 − u0m 22 + Aum , um  dt = |um |q(x) dxdt. (3.30)
2 2
0 0 BR

From (3.2), (3.22) and [26][Corollary 4, p. 85],

um → u in C([0, T ], Lq(x) (BR )),
and so,
T T 
1 1
lim sup Aum , um  = − u(T )22 + u0 22 + |u|q(x) dxdt. (3.31)
m→∞ 2 2
0 0 BR
Taking u = ϕ in (3.29) and using [21][Proposition 1.2], we obtain
T T 
1 1
χ, u dt = − u(T )22 + u0 22 + |u|q(x) dxdt. (3.32)
2 2
0 0 BR

Combining (3.31) and (3.32), it follows that (3.27) holds. Now, [25][Lemma 3.2.2, p. 117] together with
(3.27) yields (3.25) occurs. Hence, from (3.29),
T  T  T 
ut (t)ϕ dxdt + |∇u(t)| p(x)−2
∇u(t)∇ϕ dxds = |u(t)|q(x)−2 u(t)ϕ dxds,
0 BR 0 BR 0 BR
1,p(x)
for all ϕ ∈ W0 (BR ). Therefore,
  
ut (t)ϕ dx + |∇u(t)|p(x)−2 ∇u(t)∇ϕ dx = |u(t)|q(x)−2 u(t)ϕ dx, a.e. in t ∈ (0, +∞), (3.33)
BR BR BR
1,p(x)
for all ϕ ∈ W0,rad (BR ).
From (3.22) and [25][Lemma 3.1.7],
um (0) → u(0) weakly in L2rad (BR ).
1,p(x)
However, by (3.12), we know that um (0) → u0 in W0,rad (BR ), in particular um (0) → u0 in L2rad (BR ),
and so, u(0) = u0 in L2rad (BR ), from where it follows that u(0) = u0 . This shows that u satisfies the
initial condition.
1,p(x)
Finally, arguing as in Sect. 3.1, we must have that for all ϕ ∈ W0 (BR )
  
ut (t)ϕ dx + |∇u(t)|p(x)−2 ∇u(t)∇ϕ dx = |u(t)|q(x)−2 u(t)ϕ dx, a.e. in t ∈ (0, +∞). (3.34)
BR BR BR

4. Existence of solution for the non-radial case

In this section, we will assume the following condition on Ω :

There are 0<r<R such that AR,r ⊂ Ω where AR,r = B R (0) \ Br (0). ((Ω)R,r )
It is easy to see that Ω is not necessarily a ball centered at origin.
ZAMP Existence of solution for a class of heat equation. . . Page 13 of 18 2

Related to the functions p, q : Ω → R, we assume that they are continuous functions and satisfy the
following conditions:
2 ≤ p− = min p(x) ≤ max p(x) = p+ < N and 1 < q− = min q(x) (H5 )
x∈Ω x∈Ω x∈Ω
and
p(x) = p(|x|) and q(x) = q(|x|), ∀x ∈ AR,r , (H6 )

1,p(x)
4.1. Existence of local solution in W0 (Ω)

In order to show the existence of local solution, in addition to conditions (H5 ) − (H6 ), we also assume
 
p∗ (x)
1 < q− ≤ q(x) ≤ min p∗ (x), 2 + , ∀x ∈ Ω \ AR,r . (H7 )
2
We point out that we are not assuming any growth condition on q in the annulus AR,r ; hence, as in
the previous section q can have a triple regime.
1,p(x)
Hereafter, we will consider the following subspace of W0 (Ω) given by
1,p(x)
X = {u ∈ W0 (Ω) : u(x) = u(|x|) a.e. x ∈ AR,r }.
1,p(x)
Arguing as in Sect. 2, the compact embedding (3.2) still holds with W0,rad (BR ) replaced by X.

Theorem 4.1. (Local weak solution) Assume (H5 ) − (H7 ), (Ω)R,r and u0 ∈ X. Then, there exist T > 0
and a function u ∈ L∞ ([0, T ], W0
1,p(x)
(Ω)) with ut ∈ L2 ((0, T ), L2 (Ω)) that satisfies the initial condition
1,p(x)
u(0) = u0 and W0 (Ω),
1,p(x)
and for each v ∈ W0 (Ω)
  
ut (t)v dx + |∇u(t)|p(x)−2 ∇u(t)∇v dx = |u(t)|q(x)−2 u(t)v dx, a.e. t ∈ (0, T ). (4.1)
Ω Ω Ω

1,p(x)
Proof. The existence of local solution in the present case follows as in Sect. 3.1, because W0,rad (BR ) and
X have the same continuous and compact embedding. Therefore, there is a u that verifies the equality
below   
ut (t)v dx + |∇u(t)| p(x)−2
∇u(t)∇v dx = |u(t)|q(x)−2 u(t)v dx, (4.2)
Ω Ω Ω
a.e. in t ∈ (0, T ) for all v ∈ X.
Our goal is proving that
  
ut (t)v dx + |∇u(t)|p(x)−2 ∇u(t)∇v dx = |u(t)|q(x)−2 u(t)v dx, (4.3)
Ω Ω Ω
1,p(x)
a.e. in t ∈ (0, T ) for all v ∈
W0 (Ω).
For this purpose, we cannot use the Palais principle used in Sect. 3, because Ω is not a ball. Hereafter,
we will use the approach found in Alves and Rădulescu [1] [Section 3]. Hereafter, we fix J ⊂ [0, T ] with
med(J c ) = 0 such that (4.3) holds for all t ∈ J .
To begin with, for all ϕ ∈ X0 (AR,r ) = {u ∈ X : u = 0 on ∂(AR,r )} we have
  
ut (t)ϕ dx + |∇u(t)|p(x)−2
∇u(t)∇ϕ dx = |u(t)|q(x)−2 u(t)ϕ dx.
AR,r AR,r AR,r
 2 Page 14 of 18 C. O. Alves and T. Boudjeriou ZAMP

By using regularity theory, it is possible to show that for all ψ ∈ E0 (AR,r )

  
ut (t)ϕ dx + |∇u(t)|p(x)−2
∇u(t)∇ϕ dx = |u(t)|q(x)−2 u(t)ϕ dx, (4.4)
AR,r AR,r AR,r
1,p(x)
where E0 (AR,r ) = {u ∈ W0 (Ω)
: u = 0 on ∂(AR,r )}.
Using the above information, we are ready to prove (4.3). Have this in mind, let us set an even function
φ ∈ C ∞ (R, R) satisfying
0 ≤ φ(s) ≤ 1 ∀s ∈ R, φ(s) = 0 ∀s ∈ [−1, 1] and φ(s) = 1 ∀s ∈ (−∞, −2] ∪ [2, +∞).
Thereby, for  > 0 small enough and v ∈ C0∞ (BR ) ⊂ E0 (BR ), where
1,p(x)
E0 (BR ) = {u ∈ W0 (Ω) : u = 0 on ∂(BR )},
we define the function
v (x) = φ((|x| − r)/)v(x), ∀x ∈ Ω.
From the definition of v , we see that v ∈ E0 (Br (0)) ⊂ X and v ∈ E0 (AR,r ). Thus, by (4.2) and (4.4),
  
ut (t)v dx + |∇u(t)| p(x)−2
∇u(t)∇v dx = |u(t)|q(x)−2 u(t)v dx
Br Br Br

and
  
ut (t)v dx + |∇u(t)|p(x)−2 ∇u(t)∇v dx = |u(t)|q(x)−2 u(t)v dx,
AR,r AR,r AR,r

which leads to
  
ut (t)v dx + |∇u(t)|p(x)−2 ∇u(t)∇v dx = |u(t)|q(x)−2 u(t)v dx.
Ω Ω Ω
After some calculus, taking the limit  → 0, we get
   
ut (t)v dx + |∇u(t)|p(x)−2 ∇u(t)∇v dx + u(t)v dx = |u(t)|q(x)−2 u(t)v dx.
Ω Ω Ω Ω

Now, we are going to show that the above equality holds for any w ∈ C0∞ (Ω) ⊂ W0
1,p(x)
(Ω). The idea
is as above; we set the function
w (x) = φ((|x| − R)/)w(x), ∀x ∈ Ω,
which belongs to E0 (BR ) and E0 (Ω \ B R ) ⊂ W 1,p(x) (Ω). Since w |Ω\B R ∈ W 1,p(x) (Ω),
  
ut (t)w dx + |∇u(t)| p(x)−2
∇u(t)∇w dx = |u(t)|q(x)−2 u(t)w dx.
Ω Ω Ω
Taking the limit  → 0, we obtain
  
ut (t)w dx + |∇u(t)|p(x)−2 ∇u(t)∇w dx = |u(t)|q(x)−2 u(t)w dx.
Ω Ω Ω
Again by density,
  
ut (t)w dx + |∇u(t)|p(x)−2 ∇u(t)∇w dx = |u(t)|q(x)−2 u(t)w dx,
Ω Ω Ω
1,p(x)
for all w ∈ W0 (Ω). This shows that u satisfies (4.3). 
ZAMP Existence of solution for a class of heat equation. . . Page 15 of 18 2

Before concluding this subsection, we would like point out that by Lemma 3.2,
1,p(x)
u ∈ C([0, T ], W0 (Ω)) ∩ C([0, T ], Lq(x) (Ω))
and
t
us (s)2L2 (Ω) + E(u(t)) = E(u0 ), t ∈ [0, T ]. (4.5)
0
These information will be used later on.

4.2. Existence of global solution

In order to prove the existence of global solution, we replaced (H7 ) by

A
p + < q− = min q(x) ≤ q+
A
= max q(x) < min p∗ (x). (H8 )
x∈Ω\AR,r x∈Ω\AR,r x∈Ω

1,p(x)
Hereafter, W and J are defined as in (3.9) and (3.10), respectively, with W0,rad (BR ) replaced by X.

Theorem 4.2. (Global weak solution) Assume (H5 )−(H6 ), (H8 ), (ΩR,r ) and u0 ∈ W \{0}. Then, there is
a function u ∈ L∞ ((0, +∞), W0
1,p(x)
(Ω)) with ut ∈ L2 ((0, +∞), L2 (Ω)) that satisfies the initial condition
1,p(x)
u(0) = u0 ∈ W0 (Ω),
1,p(x)
and for each v ∈ W0 (Ω)
  
ut (t)v dx + ∇u(t)∇v dx = |u(t)|q(x)−2 u(t)v dx a.e. t ∈ (0, +∞).
Ω Ω Ω

1,p(x)
Proof. Since W0,rad (BR ) and X have the same compact embedding, we can repeat the arguments ex-
1,p(x)
plored in Sect. 3.2 to guarantee the existence of a function u such that u(0) = u0 in W0 (Ω) and for
1,p(x)
all ϕ ∈ W0 (Ω)
  
ut (t)ϕ dx + |∇u(t)|p(x)−2 ∇u(t)∇ϕ dx = |u(t)|q(x)−2 u(t)ϕ dx, a.e. in t ∈ (0, +∞).
Ω Ω Ω
1,p(x)
Now, we repeat the same approach used in Sect. 4.1 to conclude that for all ϕ ∈ W0 (Ω)
  
ut (t)ϕ dx + |∇u(t)|p(x)−2 ∇u(t)∇ϕ dx = |u(t)|q(x)−2 u(t)ϕ dx, a.e. in t ∈ (0, +∞). (4.6)
Ω Ω Ω

5. Blow-up of the solution

In order to prove the blow-up phenomena, we assume that (Ω)R,r holds and that the functions p and q
satisfy (H5 ) − (H6 ) and the following condition
 
∗ p∗ (x)
2 ≤ p+ < q− ≤ q(x) ≤ min p (x), 1 + , ∀x ∈ Ω \ AR,r . (H9 )
2
The main result this section is the following
 2 Page 16 of 18 C. O. Alves and T. Boudjeriou ZAMP

Theorem 5.1. (Blow-up phenomena) Assume (H5 ) − (H6 ), (H9 ), (Ω)R,r and u0 ∈ X with E(u0 ) < 0.
Then, there exist Tmax > 0 and a function u ∈ L∞ ([0, T ], W0
1,p(x)
(Ω)) with ut ∈ L2 ((0, T ), L2 (Ω)) for all
0 < T < Tmax that satisfies the initial condition
1,p(x)
u(0) = u0 ∈ W0 (Ω),
1,p(x)
and for each v ∈ W0 (Ω)
  
ut (t)v dx + |∇u(t)|p(x)−2 ∇u∇v dx = |u(t)|q(x)−2 u(t)v dx, a.e. t ∈ (0, T ).
Ω Ω Ω

Moreover, limt→Tmax u(t)L2 (Ω) = +∞.

Proof. Let u be the local solution of problem (1.1). Set
1
g(t) = u(t)22 .
2
By [21][Proposition 1.2],

g  (t) = ut (t)u(t) dx.
Ω

Taking u = v in (4.1) and μ ∈ (p+ , q− ), it follows that

 
g  (t) = − |∇u|p(x) dx + |u|q(x) dx
Ω Ω


μ
≥ −μE(u) + 1 − |u|q(x) dx ≥ G(t), (5.1)
q−
Ω

where G(t) = −μE(u). From (4.5,

G (t) = μut (t)2L2 (Ω) , (5.2)
then by Hölder inequality
⎛ ⎞2

μ μ μ  2
g(t)G (t) = ut (t)22 u(t)22 ≥ ⎝ ut (t)u(t) dx⎠ = [g (t)] .
2 2 2
Ω

Since E(u0 ) < 0 and E(u(t)) ≤ E(u0 ) for all t ≥ 0, it follows that G(t) > G(0) > 0 for all t > 0. Hence,
from (5.1),
μ
g(t)G (t) ≥ g  (t)G(t),
2
or equivalently
G (t) μ g  (t)
≥ .
G(t) 2 g(t)
Integrating over (0, t) and using (5.1), we get
µ G(0)
g  (t)[g(t)]− 2 ≥ µ . (5.3)
[g(0)] 2
Since μ > 2, inequality (5.3) implies that
 

− µ−2
2

1− µ μ−2 G(0)
g(t) ≥ [g(0)] 2 − µ t .
2 [g(0)] 2
ZAMP Existence of solution for a class of heat equation. . . Page 17 of 18 2

Therefore,
2g(0)
lim− u(t)2 = +∞, with Tmax = ,
t→Tmax (μ − 2)G(0)
showing the desired result. 

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

References

[1] Diening, L., Nägele, P., Råz̆ička, M.: Monotone operator theory for unsteady problems in variable exponent spaces.
Complex Var. Elliptic Equ. 57, 1209–1231 (2012)
[2] Kobayashi, J., Otani, M.: The principle of symmetric criticality for non-differentiable mappings. J. Funct. Anal. 214,
428–449 (2004)
[3] Liu, Y., Zhao, J.: On the potential wells and applications to semilinear hyperbolic equations and parabolic equations.
Nonlinear Anal. 64(12), 2665–2687 (2006)
[4] Alves, C.O., Rădulescu, V.: The Lane–Emden equation with variable double-phase and multiple regime. Proc. Am.
Math. Soc. 148, 2937–2952 (2020)
[5] Alves, C.O., Simsen, J., Simsen, M.: Parabolic problems in RN with spatially variable exponents. Asymptot. Anal. 93,
51–64 (2015)
[6] Lourêdo, A.T., Miranda, M.M., Clark, M.R.: Variable exponent perturbation of a parabolic equation with p(.)-Laplacian.
Electron. J. Qual. Theory Differ. Equ. 60, 1–14 (2019)
[7] Musielak, J.: Orlicz Spaces and Modular Spaces. Lecture Notes in Mathematics. Springer-Verlag, Berlin (1983)
[8] Pan, N., Zhang, B., Cao, J.: Weak solutions for parabolic equations with p(x)−growth. Electron. J. Differ. Equ. 209,
1–15 (2016)
[9] Rădulescu, V.D., Repovš, D.D.: Partial Differential Equations with Variable Exponents: Variational Methods and
Qualitative Analysis, CRC Press. Taylor & Francis Group, Boca Raton FL (2015)
[10] Rajagopal, K.R., Ruzicka, M.: Mathematical modelling of electrorheological fluids. Contin. Mech. Thermodyn. 13, 59–78
(2001)
[11] Showalter, E. R.: Monotone operators in Banach spaces and nonlinear partial differential equations (Mathematical
surveys and monographs vol. 49). American Mathematical Society, Providence
[12] Simon, J.: Compact sets in the space Lp (0, T, B). Ann. Math. Pura. Appl. 4(146), 65–96 (1987)
[13] Ruzicka, M.: Electrorheological Fluids: Modeling and Mathematical Theory. Lecture Notes in Mathematics. Springer-
Verlag, Berlin (2000)
[14] Strauss, W.A.: Existence of solitary waves in higher dimensions. Commun. Math. Phys. 55, 149–162 (1977)
[15] Alves, C.O., Ferreira, M.C.: Multi-bump solutions for a class of quasilinear problems involving variable exponents. Ann.
Math. Pura Appl. 194, 1563–1593 (2015)
[16] Antontsev, S., Shmarev, S.: Evolution PDEs with Nonstandard Growth Conditions, Atlantis Studies in Differential
Equations, vol. 4. Atlantics Press, Paris (2015)
[17] Autouri, G., Pucci, P.: Asymptotic stability for Kirchhoff systems in variable exponent Sobolev spaces. Complex Var.
Elliptic Equ. 56, 715–753 (2011)
[18] Brézis, H.: Opérateurs Maximaux Monotones et semi-groupes des contractions dans les espaces de Hilbert. North-
Holland/American Elsevier, Amsterdam/London/New York (1971)
[19] Galaktionov, V.A., Levine, H.A.: A general approach to critical Fujita exponents in nonlinear parabolic problems.
Nonlinear Anal. 34, 1005–1027 (1998)
[20] Galaktionov, V.A., Vázquez, J.L.: Regional blow up in a semilinear heat equation with convergence to a Hamilton–
Jacobi equation. SIAM J. Math. Anal. 24, 1254–1276 (1993)
[21] Gazzola, F., Weth, T.: Finite time blow-up and global solutions for semilinear parabolic equations with initial data at
high energy level. Differ. Integral Equ. 18(9), 961–990 (2005)
[22] Akagia, G., Matsuura, K.: Well-posedness and large-time behaviors of solutions for a parabolic equation involving
p(x)-Laplacian. Contin. Dyn. Syst. 1, 22–31 (2011)
[23] Alves, C.O., Boudjeriou, T.: Existence of solution for a class of nonvariational Kirchhoff type problem via dynamical
methods. Nonlinear Anal. 197, 1–17 (2020)
[24] Alves, C.O., Ferreira, M.C.: Existence of solutions for a class of p(x)- Laplacian equations involving a concave–convex
nonlinearity with critical growth in RN . Topol. Methods Nonlinear Anal. 45, 399–422 (2015)
 2 Page 18 of 18 C. O. Alves and T. Boudjeriou ZAMP

[25] Xu, R.: Initial boundary value problem for semilinear hyperbolic equations and parabolic equations with critical initial
data. Q. Appl. Math. 68(3), 459–468 (2010)
[26] Zheng, S.: Nonlinear Evolution Equations. Monographs and Surveys in Pure and Applied Mathematics, vol. 133.
Chapman & Hall/CRC, Boca Raton, FL (2004)
[27] Antontsev, S., Shmarev, S.: Anisotropic parabolic equations with variable nonlinearity. Publ. Math. 53, 355–399 (2009)
[28] Diening, L., Hästo, P., Harjulehto, P., Ruzicka, M.: Lebesgue and Sobolev Spaces with Variable Exponents, Springer
Lecture Notes, vol. 2017. Springer-Verlag, Berlin (2011)
[29] Galaktionov, V.A., Vázquez, J.L.: The problem of blow-up in nonlinear parabolic equations. Discrete Contin. Dyn.
Syst. 8, 399–433 (2002)
[30] Guo, W., Gao, J.: Finite-time blow-up and extinction rates of solutions to an initial Neumann problem involving the
p(x, t)-Laplace operator and a non-local term. Discrete. Contin. Dyn. Syst. 36, 715–730 (2016)
[31] Halsey, T.C.: Electrorheological fluids. Science 258, 761–766 (1992)
[32] Alves, C.O., Shmarev, S., Simsen, J., Simsen, M.: The Cauchy problem for a class of parabolic equations in weighted
variable Sobolev spaces: existence and asymptotic behavior. J. Math. Anal. Appl. 443, 265–294 (2016)
[33] Fan, X.L., Shen, J.S., Zhao, D.: Sobolev embedding theorems for spaces W k,p(x) (Ω). J. Math. Anal. Appl. 262, 749–760
(2001)
[34] Ishii, H.: Asymptotic stability and blowing up of solutions of some nonlinear equations. J. Differ. Equ. 26, 291–319
(1977)
[35] Mihǎilescu, M., Rădulescu, V.: Existence and multiplicity of solutions for quasilinear nonhomogeneous problems: an
Orlicz–Sobolev space setting. J. Math. Anal. Appl. 330, 416–432 (2007)
[36] Mihǎilescu, M., Rădulescu, V.: A multiplicity result for a nonlinear degenerate problem arising in the theory of elec-
trorheological fluids. Proc. R. Soc. Lond. Ser. A 462, 2625–2641 (2006)

Claudianor O. Alves
Unidade Acadêmica de Matemática
Universidade Federal de Campina Grande
Campina Grande PB58429-970
Brazil
e-mail: coalves@mat.ufcg.edu.br

Tahir Boudjeriou
Department of Mathematics, Faculty of Exact Sciences, Lab. of Applied Mathematics
University of Bejaia
6000 Bejaia
Algeria
e-mail: re.tahar@yahoo.com

(Received: April 29, 2020; revised: November 6, 2020; accepted: November 9, 2020)