This article was downloaded by: [Northeastern University]

On: 13 October 2014, At: 07:14

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Applicable Analysis: An International

Journal
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/gapa20

Blow-up in a semilinear parabolic

problem with variable source under
positive initial energy
a b
Ali Khelghati & Khadijeh Baghaei
a
Sharif University of Thechnology, Tehran, Iran.
b
IUST, Narmak, Tehran, Iran.
Published online: 13 Oct 2014.

To cite this article: Ali Khelghati & Khadijeh Baghaei (2014): Blow-up in a semilinear parabolic
problem with variable source under positive initial energy, Applicable Analysis: An International
Journal

To link to this article: http://dx.doi.org/10.1080/00036811.2014.956733

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
 Applicable Analysis, 2014
http://dx.doi.org/10.1080/00036811.2014.956733

Blow-up in a semilinear parabolic problem with variable source under

positive initial energy
Ali Khelghatia∗ and Khadijeh Baghaeib
a Sharif University of Thechnology, Tehran, Iran; b IUST, Narmak, Tehran, Iran

Communicated by R. Magnanini
(Received 20 April 2014; accepted 18 August 2014)
Downloaded by [Northeastern University] at 07:14 13 October 2014

In this paper, we consider a class of semilinear parabolic equations with variable

source
u t (x, t) = u(x, t) + (u(x, t)) p(x) , x ∈ , t > 0,
under homogeneous Dirichlet boundary conditions in a bounded domain  ⊂
Rn , n ≥ 1, with smooth boundary and 1 < p− = inf x∈ p(x) ≤ p(x) ≤ p+ =
supx∈ p(x) < +∞. For all p− > 1, we prove that the classical solutions to the
above equation blow-up in finite time with arbitrary positive initial energy and
suitable large initial values. This result extends a recent result by Wu et al. [Appl.
Math. Lett. 2013;26:539–543]
 which asserts the blow-up of solutions for n ≥ 3,
provided that 2 p+ − 1 < p− ≤ p+ ≤ n+2 n−2 .

Keywords: Parabolic equation; variable source; blow-up; positive initial energy

AMS Subject Classifications: 35K55; 35B44

1. Introduction
In this paper, we study the blow-up phenomena for the following semilinear parabolic
problem with a variable source:
⎧
⎨ u t (x, t) = u(x, t) + (u(x, t)) p(x) , x ∈ , t > 0,
u(x, t) = 0, x ∈ ∂, t > 0, (1.1)
⎩
u(x, 0) = u 0 (x) ≥ 0, x ∈ ,
where  ⊂ Rn and n ≥ 1 is a bounded domain with smooth boundary, and u 0 (x) is the
initial value. Moreover, we assume that the continuous function p :  → (1, +∞) satisfies

1 < p− = inf p(x) ≤ p(x) ≤ p + = sup p(x) < +∞.

x∈ x∈

Problems like (1.1) can be used as models of heat transfer, chemical reactions and population
dynamics. For more information, we refer the reader to [1] and the references therein.
When p(x) ≡ p > 1 is a constant, the solutions of (1.1) can become unbounded in
finite time if u 0 is sufficiently large or E(u 0 ) < 0, where E(u(t)) is the energy functional.
∗ Corresponding author. Email: khelghatibashkand@gmail.com

© 2014 Taylor & Francis

 2 A. Khelghati and K. Baghaei

In these cases, we have

u(., t)
L ∞ () → ∞ as t  Tmax ,
where Tmax denotes the maximal existence time. This phenomenon is called blow-up in the
literature. We refer the reader to [3] for details.
Pinasco [1] proved the local existence in time and regularity of solutions to problem
(1.1). Moreover, by applying the eigenfunction argument of Kaplan [4], he showed that for
p− > 1, there exist positive initial values such that the corresponding solutions blow-up in
finite time. In [5], the authors proved that if p + ≤ 1, then every solution of problem (1.1)
is global. In addition, they showed that for p + > 1, there exist functions p(x) and domains
 such that all solutions of problem (1.1) blow-up in finite time. On the other hand, if  is
small enough, then there exist global nontrivial solutions regardless of the exponent p(x).
The authors in [2] constructed a control function and applied suitable embedding
Downloaded by [Northeastern University] at 07:14 13 October 2014

theorems to prove that for n ≥ 3, the nonnegative classical solutions to the problem (1.1)
blow-up in finite time provided that
  
1 1
dx < E¯1 ,
∇u 0
22 > α1 ,
p(x)+1
E(u 0 ) = |∇u 0 |2 − u0
 2 p(x) + 1
 n+2
2 p+ − 1 < p− ≤ p+ ≤ ,
n−2
where E¯1 and α1 are positive constants which are introduced in their work. They proposed
one open question
 about occurrence of blow-up under positive initial energy in the case
1 < p − < 2 p+ − 1.
In this paper, we give one answer to this question. We prove that for all p − > 1 and n ≥
1, the nonnegative classical solutions to the problem (1.1) blow-up in finite time when the
initial energy is positive and initial data is suitably large. In addition, as a by-product result,
we also prove that the nonnegative classical solutions to the problem (1.1) with negative
energy must blow-up in a finite time. In the next section we will prove the above result.

2. Blow-up in finite time

In this section, we will prove the blow-up of solutions to problem (1.1) with suitable initial
data and arbitrary positive initial energy. The main method employed in this paper is based
on the calculation of the energy functional and concavity argument developed by Levin
[6]. The concavity argument was also used in the work of Fila [7] from which the uniform
boundedness of global solutions is proved.
In order to prove our main result, we will use the following lemma:

Lem m a 2.1 ([8, Lemma 1.1]) Suppose that a positive, twice-differentiable function θ (t)
satisfies the inequality
θ (t)θ (t) − (1 + β)θ 2 (t) ≥ 0, t > 0,

where β > 0 is some constant. If θ (0) > 0 and θ (0) > 0, then there exists t1 ∈ 0, βθθ (0)
(0)

such that θ (t) tends to infinity as t → t1 .

Here, we have our main result.

 Applicable Analysis 3

Theorem 2.2 Let u(x, t) be the nonnegative classical solution to the problem (1.1) in a
bounded domain  ⊂ Rn , n ≥ 1. Then for all p − > 1 and E(u 0 ) > 0, the solution u(x, t)
blows up in finite time, provided that

4 p+ − 1

u 0
22 > E(u 0 ) + − ||. (2.1)
p− −1 p +1

Proof In order to prove that the solution blows up in finite time under the assumption
of the theorem, we assume the solution u(x, t) is global to get a contradiction. We now
consider the energy functional E : H01 () ∩ L p(x)+1 () → R which is in the following
form:

  
1 1
E(u(t)) = |∇u(t)|2 − u p(x)+1 (t) dx. (2.2)
Downloaded by [Northeastern University] at 07:14 13 October 2014

 2 p(x) + 1

Differentiating (2.2) and making use of (1.1), we obtain

  
d d 1 1
E(u(t)) = |∇u(t)| −
2
u p(x)+1
(t) dx
dt dt 2 p(x) + 1
 

= ∇u(t) · ∇u t (t) − u p(x) (t)u t (t) dx



=− u t (t) u(t) + u p(x) (t) dx


=− u 2t (t) dx. (2.3)


Integrating (2.3) over (0, t), we get

 t
E(u(t)) = E(u 0 ) −
u τ (τ )
22 dτ. (2.4)
0

Multiplying Equation (1.1) by u and integrating over , we get

 
1 d
u (t) dx =
2
u(t)u t (t) dx
2 dt  

= u(t) u(t) + u p(x) (t) dx

 
=− |∇u(t)|2 dx + u p(x)+1 (t) dx. (2.5)
 

In the rest of the proof, we consider the following two cases:

Case 1 E(u(t)) ≥ 0, for all t > 0.

By considering (2.1), we can take α as follows:
 
p− − 1 p+ − 1
1<α<
u 0
22 − − || . (2.6)
4E(u 0 ) p +1
 4 A. Khelghati and K. Baghaei

We now return to (2.5) and use (2.4) to write

  
1 d
u 2 (t)dx = − |∇u(t)|2 dx + u p(x)+1 (t) dx
2 dt   
   
1 1 p− − 1
≥ −2 |∇u(t)| dx −
2
u p(x)+1
(t) dx + − u p(x)+1 (t) dx
 2 p(x) + 1 p +1 

p− − 1
= −2E(u(t)) + − u p(x)+1 (t) dx
p +1 

p− − 1
= 2(α − 1)E(u(t)) − 2α E(u(t)) + − u p(x)+1 (t) dx
p +1 
 t 
p− − 1
≥ −2α E(u 0 ) + 2α
u τ (τ )
22 dτ + − u p(x)+1 (t) dx. (2.7)
0 p + 1 
Downloaded by [Northeastern University] at 07:14 13 October 2014

Here, we have used the fact that the function ( p(x)−1)/( p(x)+1) is an increasing function
with respect to p(x). Now, we need a lower bound for the third term on the right hand side
of (2.7). In order to obtain this, for each t > 0, we divide  into two sets,
1 = {x ∈  : u(x, t) < 1} and 2 = {x ∈  : u(x, t) ≥ 1}.
Thus, we have
  
u p(x)+1 (t) dx = u p(x)+1 (t) dx + u p(x)+1 (t) dx
 1 2
 
+ −
≥ u p +1 (t) dx + u p +1 (t) dx. (2.8)
1 2

We now make use of Young’s inequality as follows:

 
− ( p − + 1) ( p − − 1)
u p +1 (t) dx ≥ u 2 (t) dx − |2 | (2.9)
2 2 2 2
and
 
p+ +1 ( p + + 1) ( p + − 1)
u (t)dx ≥ u 2 (t) dx − |1 |
1 2 1 2

( p− + 1) ( p − − 1) ( p+ − p− )
≥ u 2 (t) dx − |1 | − |1 |
2 1 2 2

( p− + 1) ( p − − 1) ( p+ − p− )
≥ u 2 (t) dx − |1 | − ||. (2.10)
2 1 2 2
By inserting (2.9) and (2.10) in (2.8), we obtain
 
( p − + 1) ( p + − 1)
u p(x)+1 (t) dx ≥ u 2 (t) dx − ||. (2.11)
 2  2
Substituting (2.11) into (2.7), we get
  t
d
u (t) dx ≥ −4α E(u 0 ) + 4α
2

u τ (τ )
22 dτ + ( p − − 1)
dt  0

( p + − 1)( p − − 1)
× u 2 (t) dx − ||. (2.12)
 p− + 1
 Applicable Analysis 5

Then,
 
d ( p+ − 1)( p − − 1)
u 2 (t) dx − ( p − − 1) u 2 (t) dx ≥ −4α E(u 0 ) − ||,
dt   p− + 1
which yields

u 2 (t)dx

 

1 ( p+ − 1)( p − − 1) − −
≥ −
4α E(u 0 ) + −
|| 1 − e( p −1)t +
u 0
22 e( p −1)t .
p −1 p +1
(2.13)
t
We define y(t) = 0
u(τ )
22 dτ. Since we have assumed the solution u(x, t) is global, then
the function y(t) is bounded for all t ≥ 0. Now we have
Downloaded by [Northeastern University] at 07:14 13 October 2014

 
d
y (t) = u (t) dx, y (t) =
2
u 2 (t) dx.
 dt 
By inserting (2.13) in (2.12), we can see
 t

y (t) ≥ 4α
u τ (τ )
22 dτ
 0

( p+ − 1)( p − − 1) −
+ ( p− − 1)
u 0
22 − || − 4α E(u 0 ) e( p −1)t . (2.14)
p− + 1
By considering (2.6), we can take ε > 0 small enough such that
 
1 − ( p+ − 1)( p − − 1)
ε< ( p − 1)
u

0 2
2
− || − 4α E(u 0 ) .
2α
u 0
22 p− + 1
Now, we pick c > 0 large enough to have
1 −2
ε
u 0
42 ,
c> (2.15)
4
then, we define the auxiliary function ϕ as follows:
ϕ(t) = y 2 (t) + ε −1
u 0
22 y(t) + c.
Therefore,

ϕ (t) = 2y(t) + ε −1
u 0
22 y (t), (2.16)

ϕ (t) = 2y(t) + ε −1
u 0
22 y (t) + 2(y (t))2 . (2.17)

Set δ = 4c − ε −2
u 0
42 , because of (2.15), δ > 0. Now, from (2.16), we obtain

2
(ϕ (t))2 = 2y(t) + ε −1
u 0
22 (y (t))2

= 4y 2 (t) + 4ε −1
u 0
22 y(t) + ε −2
u 0
42 (y (t))2

= 4y 2 (t) + 4ε −1
u 0
22 y(t) + 4c − δ (y (t))2
= (4ϕ(t) − δ)(y (t))2 . (2.18)
 6 A. Khelghati and K. Baghaei

Then,
4ϕ(t)(y (t))2 = (ϕ (t))2 + δ(y (t))2 . (2.19)
By integrating the following identity from 0 to t,

1 d

u(t)
2 =
2
u(t)u t (t) dx,
2 dt 
we obtain  t
1

u(t)
22 −
u 0
22 = u(τ )u τ (τ ) dxdτ.
2 0 
Therefore,  t

u(t)
22 =
u 0
22 + 2 u(τ )u τ (τ ) dxdτ.
0 
Downloaded by [Northeastern University] at 07:14 13 October 2014

This equality along with the Hölder and Young inequalities gives

(y (t))2 =
u(t)
42 (2.20)
  t 2
=
u 0
22 + 2 u(τ )u τ (τ ) dxdτ
0 
   12   12 2
t t
≤
u 0
22 +2
u(τ )
22 dτ
u τ (τ )
22 dτ
0 0
 t  12  t  12  t
=
u 0
42 + 4
u 0
22
u(τ )
22 dτ
u τ (τ )
22 dτ + 4y(t)
u τ (τ )
22 dτ
 t 0 0
 t 0
−1
≤
u 0
2 + 4y(t)
4

u τ (τ )
2 dτ + 2ε
u 0
2 y(t) + 2ε
u 0
2
2 2 2

u τ (τ )
22 dτ.
0 0
(2.21)
From (2.17) and (2.19), we get

2ϕ(t)ϕ (t) = 2 2y(t) + ε −1

u 0
22 y (t) + 2(y (t))2 ϕ(t)

= 2 2y(t) + ε −1
u 0
22 y (t)ϕ(t) + 4ϕ(t)(y (t))2

= 2 2y(t) + ε −1
u 0
22 y (t)ϕ(t) + (ϕ (t))2 + δ(y (t))2 . (2.22)

Now, from (2.22), (2.18), (2.14) and (2.20), we can write

2ϕ (t)ϕ(t) − (1 + α)(ϕ (t))2

= 2 2y(t) + ε −1
u 0
22 y (t)ϕ(t) + δ(y (t))2 − α(ϕ (t))2

= 2 2y(t) + ε −1
u 0
22 y (t)ϕ(t) + δ(y (t))2 − α(4ϕ(t) − δ)(y (t))2

= 2 2y(t) + ε −1
u 0
22 y (t)ϕ(t) − 4αϕ(t)(y (t))2 + δ(1 + α)(y (t))2

 ( p+ − 1)( p − − 1)

−1 −
≥ 2ϕ(t) 2y(t) + ε
u 0
2 2
( p − 1)
u 0
2 −
2
|| − 4α E(u 0 )
p− + 1
 t 
−
e( p −1)t + 4α
u τ (τ )
22 dτ
0
 Applicable Analysis 7
  t
− 4αϕ(t)
u 0
42 + 4y(t)
u τ (τ )
22 dτ + 2ε
u 0
22 y(t)
0
 t 
+2ε −1
u 0
22
u τ (τ )
22 dτ .
0
− −1)t
Now, by considering the values of α and ε, and the facts that e( p ≥ 1 and ϕ > 0, we
obtain

2ϕ (t)ϕ(t) − (1 + α)(ϕ (t))2

  t 
≥ 4αϕ(t) 2y(t) + ε −1
u 0
22 2
u τ (τ )
22 dτ + ε
u 0
22
0
  t
− 4αϕ(t)
u 0
2 + 4y(t)
4

u τ (τ )
22 dτ + 2ε
u 0
22 y(t)
Downloaded by [Northeastern University] at 07:14 13 October 2014

0
 t 
+ 2ε −1
u 0
22
u τ (τ )
22 dτ = 0.
0

Thus,
 
α−1
ϕ (t)ϕ(t) − 1 + (ϕ (t))2 ≥ 0.
2

Since ϕ(0) > 0 and ϕ (0) > 0, therefore, we can apply Lemma 2.1 to conclude that
ϕ(t) → +∞ as t → t ∗ ≤ (α−1)ϕ 2ϕ(0)
(0) . Now, by considering the continuity of ϕ with respect
to y, we can conclude that y(t) tends to infinity at some finite time which is a contradiction.
Case2: In this case, we can assume that there exists t0 > 0 such that E(u(t0 )) < 0.
Thus, by giving v(x, t) = u(x, t + t0 ), we have E(v(0)) = E(u(t0 )) < 0.
Now, we make use of (2.5) to obtain
  
1 d
v 2 (t) dx = − |∇v(t)|2 dx + v p(x)+1 (t) dx
2 dt   
   
− 1 1
= −( p + 1) |∇v(t)| dx − −
2
v p(x)+1
(t) dx
2  p +1 

p− − 1
+ |∇v(t)|2 dx
2 

− p− − 1
≥ −( p + 1)E(v(t)) + |∇v(t)|2 dx. (2.23)
2 

This inequality along with (2.4) and E(v(0)) < 0 yields

  t
1 d − −
v (t)dx ≥ −( p + 1)E(v(0)) + ( p + 1)
2

vτ (τ )
22 dτ
2 dt  0
 t
> ( p− + 1)
vτ (τ )
22 dτ. (2.24)
0
t
Now, we define ψ(t) = 0
v(τ )
22 dτ. From (2.3), we have

E(v(t)) ≤ E(v(0)) < 0. (2.25)

 8 A. Khelghati and K. Baghaei

We now make use of (2.25) and the Poincaré inequality in (2.23) to get

ψ (t) ≥ −2( p − + 1)E(v(t)) + ( p − − 1) |∇v(t)|2 dx
 
−
> ( p − 1) |∇v(t)| dx
2


> λ( p− − 1) v 2 (t) dx = λ( p − − 1)ψ (t).

Here, λ is the best constant in the Poincaré inequality. The last inequality yields
− −1)t
ψ (t) > ψ (0)eλ( p . (2.26)

This inequality says that ψ (t) > 0 for all t > 0. Now, by using the Hölder inequality and
(2.24), we obtain
Downloaded by [Northeastern University] at 07:14 13 October 2014

 t  t
ψ (t) − ψ (0) = ψ (t) = 2 v(τ )vτ (τ )dxdτ
0 0 
 t   12  t   12
≤2 v (τ ) dxdτ
2
vτ2 (τ ) dxdτ
0  0 
  1  t   12   1
2 2 d 2
≤ −
v (τ ) dxdτ
2
v (t) dx
2
p +1 0  dt 
 1
2 2 1 1
= (ψ(t)) 2 (ψ (t)) 2 . (2.27)
p− + 1
In view of (2.26), we have ψ (t) > ψ (0). Therefore, (2.27) implies that
p− + 1  2
ψ (t) − ψ (0) ≤ ψ(t)ψ (t). (2.28)
2
From (2.26), we get
lim ψ (t) = +∞. (2.29)
t→+∞
p− −1
Set ρ = 4 ; it then follows from (2.28) that
 
ψ (0) 2 ψ(t)ψ (t)
(1 + 2ρ) 1 − ≤ . (2.30)
ψ (t) (ψ (t))2
By (2.29), there exists t¯ > 0 such that for all t ≥ t¯ :
ψ (0)
ψ (t) >  . (2.31)
1+ρ
1 − 1+2ρ

From now on, in what follows, we assume that t ≥ t¯. By considering (2.30) and (2.31), we
can write
ψ(t)ψ (t)
> 1 + ρ.
(ψ (t))2
Thus,
ψ(t)ψ (t) − (ψ (t))2 > ρ(ψ (t))2 .
 Applicable Analysis 9

Hence, we get
 2
ψ(t)ψ (t) − (ψ (t))2 ψ (t)
>ρ .
ψ 2 (t) ψ(t)
Now, we set ξ(t) = ψ (t)/ψ(t). Thus, the last inequality yields

ξ (t)ξ −2 (t) > ρ. (2.32)

By integration of (2.32) from t¯ to t, we obtain

1 1
0< < − ρ(t − t¯).
ξ(t) ξ(t¯)
 
But limt→+∞ (ξ(t¯))−1 − ρ(t − t¯) = −∞, which is a contradiction. This completes the
proof. 
Downloaded by [Northeastern University] at 07:14 13 October 2014

Remark 2.3 We note that case 2is proved without using the assumption (2.1). Also, by
considering case 2, we can prove that for all p − > 1, the nonnegative classical solutions
to the problem (1.1) blow-up in finite time when the initial energy is negative.

Acknowledgements
The authors would like to thank the anonymous referees for their careful reading and valuable
suggestions on this article.

References

[1] Pinasco JP. Blow-up for parabolic and hyperbolic problems with variable exponents. Nonlinear
Anal. TMA. 2009;71:1049–1058.
[2] Quittner P, Souplet P. Superlinear parabolic problems, Blow-up, global existence and steady states.
Birkhäuser, Basel: Birkhäuser Advanced Texts; 2007.
[3] Kaplan S. On the growth of the solutions of quasilinear parabolic equations. Comm. Pure Appl.
Math. 1963;16:305–330.
[4] Ferreira R, De Pablo A, Pérez-Llanos M, Rossi JD. Critical exponents for a semilinear parabolic
equation with variable reaction. Proc. Roy. Soc. Edinburgh: Sec. A. 2012;142:1027–1042.
[5] Wu X, Gua B, Gao W. Blow-up of solutions for a semilinear parabolic equation involving variable
source and positive initial energy. Appl. Math. Lett. 2013;26:539–543.
[6] Levin HA. Some nonexistence and instability theorems for solutions of formally parabolic
equations of the form Pu t = −Au + F(u). Arch. Rational Mech. Anal. 1973;51:371–386.
[7] Fila M. Boundedness of global solutions of nonlinear diffusin equations. J. Differ. Equ.
1992;98:226–240.
[8] Kalantarov VK, Ladyzhenskaya OA. The occurrence of collapse for quasilinear equations of
parabolic and hyperbolic type. Zap. Nauchn. Sem. LOMI. 1977;69:77–102.