Blow-Up of Solutions for a Class of Reaction-Diffusion Equations with a Gradient Term under Nonlinear Boundary Condition
Junping Zhao
College of Science, Xi’an University of Architecture & Technology, Xi’an 710055, China Reprint requests to J. P. Z.; E-mail:junpingzhao@yeah.net
Z. Naturforsch.67a,479 – 482 (2012) / DOI: 10.5560/ZNA.2012-0048 Received February 2, 2012 / revised May 4, 2012
The blow-up of solutions for a class of quasilinear reaction-diffusion equations with a gradient term ut=div(a(u)b(x)∇u) +f(x,u,|∇u|2,t)under nonlinear boundary condition∂u/∂n+g(u) =0 are studied. By constructing a new auxiliary function and using Hopf’s maximum principles, we obtain the existence theorems of blow-up solutions, upper bound of blow-up time, and upper estimates of blow-up rate. Our result indicates that the blow-up timeT∗ may depend ona(u), while being independent ofg(u)and f.
Key words:Reaction-Diffusion Equation; Gradient Term; Nonlinear Boundary Conditions;
Maximum Principles; Blow-Up of Solutions.
Mathematics Subject Classification 2000:35K20, 35K55, 35K65
1. Introduction and Main Results
There are many works on the existence and non- existence of global solutions to the parabolic equa- tions and systems with various boundary conditions (see [1–7] and the references therein). In this paper, we consider the following reaction-diffusion equation with a gradient term under nonlinear boundary condi- tion:
ut=div(a(u)b(x)∇u)
+f(x,u,|∇u|2,t), in Ω×(0,T),
∂u
∂n+g(u) =0, on ∂ Ω×(0,T), u(x,0) =u0(x)>0, in ¯Ω,
(1)
whereΩ is a smooth bounded domain ofRN,N≥2, n denotes the outer unit normal on the boundary,a∈ C2(R+),b∈C1(Ω¯),f∈C1(Ω¯×R×R+×[0,T]),g∈ C2(R+), f>0, andg>0.
Models such as (1) are worthy to study not only because of the importance to the qualitative theory of nonlinear partial differential equations but also be- cause of possible applications in the field of mechan- ics, physics, and biology (gas flow in porous media, semiconductor, spread of biological populations, etc.);
see [3–13] and the references therein.
Some special cases of (1) have been discussed by many authors. The blow-up phenomenon of (1) with linear boundary conditions have been extensively stud- ied over the past decades. In [14,15], Souplet dis- cussed the following equations with Dirichlet bound- ary conditions
ut=∆u+f(x,u,|∇u|p,t), in Ω×(0,T), u(x,t) =0, on ∂ Ω×(0,T),
u(x,0) =u0(x)>0, in ¯Ω,
(2)
where re-scaling arguments, the self-similar subsolu- tion method, and the energy functional method are in- troduced to investigate the sufficient conditions and the qualitative properties for blow-up.
In [16], Ding studied the semilinear reaction diffu- sion equation
ut=∆u+f(x,u,|∇u|2,t) (3) with mixed linear boundary conditions. By using the Hopf maximum principle, the author obtained the nonexistence theorems of the global solutions and gave the bound of blow-up time.
In [17], Hu and Yin considered the heat equation ut=∆uin an unbounded domainΩ⊂RNwith Dirich- let conditionu(x,t) =0 on partial boundary ofΩ and nonlinear condition∂u/∂n=upon the other part of
c
2012 Verlag der Zeitschrift f¨ur Naturforschung, T¨ubingen·http://znaturforsch.com
480 J. P. Zhao·Blow-Up of Solutions for a Class of Reaction-Diffusion Equations the boundary, where p>1. It has been shown that the
nonnegative solutions blow up under certain conditions depending on the exponentp.
Recently, Ding and Guo [18] studied the blow- up and global solutions for the following reaction- diffusion equation with a gradient term and nonlinear boundary condition:
(g(u))t=∆u+f(x,u,|∇u|2,t), in Ω×(0,T),
∂u
∂n+g(u) =0, on ∂ Ω×(0,T), u(x,0) =u0(x)>0, in ¯Ω.
By constructing suitable auxiliary functions and using maximum principles, they proved the sufficient condi- tions for the existence of a blow-up solution, an up- per bound for the blow-up time, an upper estimate of the blow-up rate, the sufficient conditions for the exis- tence of the global solution, and an upper estimate of the global solution. However, as in [17], the blow-up timeT∗still depends ong(u).
In this paper, we extend the result in [16,18] to the quasilinear reaction-diffusion equation with a gradi- ent term under nonlinear boundary condition. By con- structing a new auxiliary function, we indicate that the blow-up timeT∗may depend ona(u), while be inde- pendent ofg(u)andf. Hence, our results in the present paper are also better than those obtained in [19], in whichT∗≤ 1
β R+∞
M0
a(s)
f(s)depend on botha(u)andf. It is known that, by the classical theory for parabolic equations and Hopf’s maximum principle [20], the problem (1) has a unique local solution, andu>0 since
f>0.
Our main result of this paper reads as follows:
Theorem 1. Let u be a C3(Ω ×(0,T))∩C2(Ω¯ × [0,T))solution of (1). Suppose the following:
(1) For(x,s,d,t)∈Ω×R+×R+×R+, a(s)>0, a0(s)≥0, a(s)−a0(s)≥0,
b(s)>0, (4)
g(s)>0, g0(s)≥0, ag≥(ag)0, (5) ft0(x,s,d,t)≥0,
fd0(x,s,d,t)≥ −a(s)b(x)/2, fs0(x,s,d,t)−f(x,s,d,t)≥0.
(6)
(2) The constant β=min
Ω¯
a(u0)e−u0[div(a(u0)b(x)∇u0) +f(x,u0,|∇u0|2,0)] >0.
(7)
(3) The integration Z +∞
M0
a(s)
es−1ds<+∞, where M0=u0(x0) =max
Ω¯
u0(x)for some u0∈Ω¯. Then, u(x,t)must blow up in finite time T∗and
T2∗≤ 1 β
Z +∞
M0
a(s)
es−1ds, (8)
as well as
u(x,t)≤G−1(β(T−t)), (9) where
G(z) = Z +∞
z
a(s)
es−1ds, z>0. (10) 2. Proof of the Main Result
In this section, we will give a detailed proof of our result. We denoteq=|∇u|2below for convenience. We have the following Lemma:
Lemma 1. Let the conditions of Theorem1hold, then
ut≥0 (11)
for all(x,t)∈Ω×(0,T).
Proof. Set
w=aut (12)
then we have
∇w=a0ut∇u+a∇ut, (13) 4w=a00utq+2a0∇u·∇ut+a0ut∆u+a∆ut, (14) and
wt=a0u2t+a(ab∆u+a0bq+a∇b·∇u+f)t
=a0u2t+a2b∆ut+aa0but∆u+aa00butq +2aa0b∇u·∇ut+aa0ut∇b·∇u
+a2∇b·∇ut+a fu0ut+2a fq0∇u·∇ut+a ft0. (15)
Combining (14) with (15), we have
wt−ab4w=a0u2t+aa0ut∇b·∇u+a2∇b·∇ut +a fu0ut+2a fq0∇u·∇ut+a ft0. (16)
J. P. Zhao·Blow-Up of Solutions for a Class of Reaction-Diffusion Equations 481
From (12), we obtain
∇ut=−1
a2a0∇uw+1 a∇w. With (15) and (16), we have
wt−ab4w− a∇b+2fq0∇u
·∇w +
2
aa0fq0q−fu0
w=a0u2t+a ft0. (17) Assumptions (4) and (5) guarantee that the right side in equality (17) is nonnegative. Therefore, we have
wt−ab4w−(a∇b+2fq0∇u)·∇w +
2
aa0fq0q−fu0
w≥0, in Ω×(0,T),
∂w
∂n +(ag)0
a w=0, on ∂ Ω×(0,T), w(x,0) =w0(x)≥0, in ¯Ω.
(18)
Combining (18) and (6) with maximum principles, it follows thatwcannot attain its minimum inΩ×(0,T).
Therefore, we have in ¯Ω ×[0,T),w(x,t)≥0, then ut≥0.
Proof of Theorems1. Set
v=−a(u)ut+βeu=−w+βeu. Then we find
∇v=−∇w+βeu∇u, (19) 4v=−4w+βeuq+βeu4u, (20) and
vt=−wt+βeuut. (21) Combining (19) – (21), we have
vt−ab4v
=−wt+ab4w+βeuut−abβeuq−abβeu4u
=−(a∇b+2fq0∇u)·∇w+ 2
aa0fq0q−fu0
w−a0u2t
−a ft0+βeuut−abβeuq−abβeu4u
= (a∇b+2fq0∇u)·(∇v−βeu4u)−2
aa0fq0q(v−βeu) +fu0v−βeufu0−a0u2t −a ft0+βeuut
−abβeuq−abβeu4u.
(22)
Therefore,
vt−ab4v−(a∇b+2fq0∇u)·∇v+
2
aa0fq0q−fu0
v
=−aβeu∇b·∇u−2βeufq0q+2a0 a βeufq0q
−a0u2t−a ft0+βeuut−abβeuq−abβeu4u. (23)
From (1), we obtain
ab4u=ut−a0bq−a∇b·∇u−f. Therefore, we have
vt−ab4v−(a∇b+2fq0∇u)·∇v+
2
aa0fq0q−fu0
v
=−a0u2t −2βeuqa−a0 a
fq0+ab
2
−a ft0 +aβeu(f−fu0).
(24)
Assumptions (4) – (6) guarantee that the right side in equality (24) is nonpostive, i.e.,
vt−ab4v−(a∇b+2fq0∇u)·∇v +
2
aa0fq0q−fu0
v≤0. (25)
It follows from (7) that maxΩ¯
v(x,0) =max
Ω¯
−a(u0)[div(a(u0)b(x)∇u0) +f(x,u0,q0,0)] +βeu0 =0.
(26)
On∂ Ω×(0,T), we have
∂v
∂n−(ag)0 a v=−
g−(ag)0 a
βeu≤0. (27) Combining (25) – (27) and Hopf’s maximum principle, we have in ¯Ω×[0,T),
v≤0 and
a(u)
eu ut≥β. (28)
Without loss of generality, we let u0(x0) =M0, inte- grating (28) fromM0tou(x0,t), we have
1 β
Z u(x0,t) M0
a(s) es ds≥t. Since
1 β
Z +∞
M0
a(s)
es ds<+∞,
482 J. P. Zhao·Blow-Up of Solutions for a Class of Reaction-Diffusion Equations thenumust blow-up within finite timet=T2∗, and
T2∗≤ 1 β
Z +∞
M0
a(s) es ds.
By the same argument as in [19], we can obtain (9), the upper estimates of blow-up rate.
3. Application
In this section, we will give a example to which the main theorem obtained in this paper may be applied.
Example 1. Let u be a C3(Ω×(0,T))∩C2(Ω¯ × [0,T))solution of the problem
ut=div
eu/2
1−∑3
i=1
x2i
∇u
+7exp
2u+2q+2t−8
3
∑
i=1
x2i
, in Ω×(0,T),
∂u
∂n+eu/2=0, on ∂ Ω×(0,T), u(x,0) =u0(x) =1−∑3
i=1
x2i, in ¯Ω, where Ω =
x= (x1,x2,x3)
3
∑
i=1
xi2< 12
, 0<T <
+∞. Then
a(u) =g(u) =eu/2, b(x) =u0(x) =1−
3 i=1
∑
x2i,
f(x,u,q,t) =7 exp u
2+2q+2t−8
3
∑
i=1
x2i
.
It is easy to check that (4) – (6) hold. By (7), we find β=min
Ω¯
a(u0)e−u0
a(u0)b(x)∆u0+a0(u0)b(x)|∇u0|2 +a(u0)∇b(x)·∇u0+f(x,u0,q0,0)
=min
Ω¯
b(x)∆u0+1
2b(x)|∇u0|2+∇b(x)·∇u0 +7 exp
2q0−8
3 i=1
∑
x2i
=min
Ω¯
12
3
∑
i=1
x2i −2 3
∑
i=1
x2i 2
+1
= min
0≤s≤12
{12s−2s2+1}=1.
It follows from Theorem1thatu(x,t)must blow-up in finite timeT2∗and
T2∗≤ 1 β
Z +∞
M0
a(s) es−1ds=
Z +∞
1
es/2 es−1ds
=ln √
e+1
√e−1
.
Acknowledgements
This work was supported by the Scientific Research Program of the Shannxi Higher Education Institutions (Grant No. 12JK0868).
[1] J. Ding and B.-Z. Guo, J. Dyn. Control Syst. 17, 273 (2011).
[2] J. Ding and B.-Z. Guo, Appl. Math. Lett. 24, 936 (2011).
[3] F. Liang, Appl. Math. Comput.218, 3993 (2011).
[4] H. Lu, F. Wang, and Q. Jiang, Ann. Polon. Math.102, 187 (2011).
[5] J. T. Ma, Math. Comput. Model.54, 2551 (2011).
[6] F. Sun and M. Wang, J. Part. Diff. Equat.17, 71 (2004).
[7] Y. Zhou and W. Han, J. Math. Anal. Appl. 374, 585 (2011).
[8] S. B. Huang, Q. Y. Tian, S. Z. Zhang, and J. H. Xi, Non- lin. Anal.74, 2342 (2011).
[9] J. Dong, Electron. J. Diff. Equat.123, (2005).
[10] A. J¨ungel, Quasi-Hydrodynamic Semiconductor Equa- tions, Progress in Nonlinear Differential Equations and their Applications, 41, Birkh¨auser, Basel 2001.
[11] J. I. Diaz and F. de Th´elin, SIAM J. Math. Anal.25, 1085 (1994).
[12] P. Souplet, Math. Method. Appl. Sci.19, 1317 (1996).
[13] J. P. Zhao and P. Guan, J. Southe. Univ. Nat. Sci.33, 226 (2003).
[14] P. Souplet, Electron. J. Diff. Equat.10, (2001).
[15] P. Souplet and F. B. Weissler, J. Math. Anal. Appl.212, 60 (1997).
[16] J. Ding, Appl. Math. Lett.15, 159 (2002).
[17] B. Hu and H.-M. Yin, J. Math. Anal. Appl.209, 683 (1997).
[18] J. Ding and B.-Z. Guo, Appl. Math. Lett. 24, 936 (2011).
[19] J. Ding and S. Li, Comput. Math. Appl.49, 689 (2005).
[20] M. H. Protter and H. F. Weinberger, Maximum Princi- ples in Differential Equations, Prentice Hall, Engle- wood Cliffs, NJ 1967.