CUBO A Mathematical Journal
Vol.16, No¯ 02, (53–69). June 2014

Existence of blow-up solutions for quasilinear elliptic
equation with nonlinear gradient term. 1

Fang Li

Institute of Mathematics, School of

Mathematical Sciences,

Nanjing Normal University,

Jiangsu Nanjing 210023, China.

lifang101216@126.com

Zuodong Yang

School of Teacher Education,

Nanjing Normal University,

Jiangsu Nanjing 210097, China

zdyang jin@263.net.

ABSTRACT

In this paper, we consider the quasilinear elliptic equation in a smooth bounded

domain. By using the method of lower and upper solutions, we study the existence,

asymptotic behavior near the boundary and uniqueness of the positive blow-up solutions

for quasilinear elliptic equation with nonlinear gradient term.

RESUMEN

En este art́ıculo consideramos la ecuación eĺıptica cuasilineal en un dominio acotado

suave. Usando el método de sub y súper soluciones, estudiamos la existencia, com-

portamiento asintótico cerca de la frontera y la unicidad de soluciones explosivas para

ecuaciones eĺıpticas cuasilineales con término del gradiente nolineal.

Keywords and Phrases: Quasilinear elliptic equation; Blow-up solutions; Asymptotic behavior

of solutions; Lower and upper solutions.

2010 AMS Mathematics Subject Classification: 35J65, 35J50.

1Project Supported by the National Natural Science Foundation of China(No.11171092); the Natural Science
Foundation of the Jiangsu Higher Education Institutions of China(No.08KJB110005)



54 Fang Li & Zuodong Yang CUBO
16, 2 (2014)

1 Introduction and main results

We shall establish the results on the existence, asymptotic behavior near the
boundary and uniqueness near the boundary for the following quasilinear elliptic
equation

{
△mu = b(x)u

p(1 + |∇u|q), x ∈ Ω,

u = ∞, x ∈ ∂Ω,
(1.1)

where Ω is a C2 bounded domain with smooth boundary ∂Ω in RN, △mu :=
div(|∇u|m−2∇u), m ≥ 2, p, q > 0, b(x) ∈ Cµ(Ω̄) for some 0 < µ < 1.

Problems like (1.1) are usually known in the literature as a boundary blow-up
problems and its solutions are named ”blow-up solutions” or ”explosive solutions”
or ”large solutions” of Eq. (1.1). Precisely, by a solution of (1.1) we mean a solution
of (1.1) satisfying u(x) → ∞ as d(x,∂Ω) → 0.

Semilinear elliptic problems involving a gradient term with boundary blow-up
interested many authors. Namely Bandle and Giarrusso[1] developed existence and
asymptotic behavior results for large solutions of ∆u+|∇u(x)|a = g(u) in a bounded
domain. In the case g(u) = p(x)uγ, a > 0, and γ > max(1, a). Ghergu et al.[2]
considered more general equation ∆u + q(x)|∇u(x)|a = p(x)g(u), where 0 ≤ a ≤
2, p and q are Hölder continuous functions on (0, ∞). More results about some
extensions to this problems, we can see in[24]-[25].

Recently, Goncalves et al. [11] showed the existence of nonnegative solutions of
the boundary blow-up problem

{
△u = ψ(x, u, ∇u), x ∈ Ω,

u = ∞, x ∈ ∂Ω
(1.2)

under the condition

a(x)g(t) ≤ ψ(x, t,ξ) ≤ h(t)(1 + Λ|ξ|2),

where Λ > 0 is a constant, a, g and h are continuous functions, a(x) > 0 in Ω, g
and h are non-decreasing and satisfying g(0) = 0, g(t) > 0 for t > 0, h(0) ≥ 1, and
g satisfies the so called Keller-Osserman condition, namely

∫
∞

1

1
!

G(t)
dt < ∞, G(t) =

∫t

0

g(s)ds.

The study of the following equation:
{

△mu = g(x)f(u), in Ω

u(x) → ∞, as x → ∂Ω
(1.3)



CUBO
16, 2 (2014)

Existence of blow-up solutions for quasilinear elliptic equation . . . 55

also has been many results, see [3],[4],[15]-[19] and the references therein. Gladiali
and Porru [15] studied boundary asymptotic of solutions of this equation under some
condition on f and when g(x) ≡ 1. Related problems on asymptotic behavior and
uniqueness were also studied in [16]. Ahmed Mohammed in [17] established bound-
ary asymptotic estimate for solution of this equation under appropriate conditions
on g and the nonlinearity f. They still allowed g to be unbounded on Ω or to vanish
on ∂Ω. Diaz and Letelier [18] proved the existence and uniqueness of large solutions
to the problem (1.3) both for f(u) = uγ,γ > m−1(super-linear case) and ∂Ω being
of the class C2. Lu, Yang and E.H.Twizell [4] proved the existence of Large solutions
to the problem (1.1) both for f(u) = uγ,γ > m − 1,Ω = RN or Ω being a bounded
domain (super-linear case) and γ ≤ m − 1,Ω = RN(sub-linear case) respectively.
Z.Yang et.al. [19] also established an explosive sub-supersolution method for the
existence of solutions to (1.3).

For the other results of large solutions to quasilinear elliptic problems (1.1) with
nonlinear gradient terms, see [5]-[8] and the references therein.

Motivated by the results of the above cited papers, we shall attempt to treat
such equation (1.1), the results of the semilinear equations are extended to the
quasilinear ones. We can find the related results for m = 2 in [10].

To study (1.1),we first consider the existence of nonnegative solutions of the
generary boundary blow-up problem

{
△mu = ψ(x, u, ∇u), x ∈ Ω,

u = ∞, x ∈ ∂Ω.
(1.4)

Our main results are summarized in the following and to our best knowledge, they
are not covered by any of the ones referred to above.

Theorem 1.1. Let ψ ∈ Cµ(Ω × R × RN), 0 < µ < 1, and ū, u ∈ W1,∞(Ω) be
the ordered weak upper and lower solutions of (1.4) and be bounded on any closed
subdomain of Ω. Assume that there exist constants k > m − 1, c1 > 0, and two
functions h1 ∈ C

µ(Ω) and g ∈ L∞loc([0, +∞)) for some 0 < µ < 1, such that

|ψ(x, t,ξ)| ≤ h1(x) + g(t) + c1|ξ|
k−1a.e.x ∈ Ω, ∀ξ ∈ RN, t ∈ [u, ū]. (1.5)

Then there is a C1,β(Ω)-solution u of (1.4) for some 0 < β < 1 such that u ≤ u ≤ ū
in Ω.The ordered weak upper and lower solutions will be defined by Definition 2.1.

Theorem 1.2. Suppose that m ≥ 2, p, q > 0 and p + q > m − 1. If there exist
two constants γ ≥ 0 and β1 > 0 such that γ + m − q ≥ 0 and b(x) ≥ β1d

γ(x).
Then the problem (1.1) has at least one nonnegative C1-solution.

Theorem 1.3. Suppose that b(x) > 0 in Ω, m ≥ 2, p, q > 0 and p+q > m−1.



56 Fang Li & Zuodong Yang CUBO
16, 2 (2014)

If there are two constants β > 0 and γ ≥ 0 satisfying γ + m − q > 0, such that

lim
d(x)→0

b(x)

d(x)γ
= β.

Then the problem (1.1) possesses a nonnegative solution and any nonnegative solu-
tion u(x) satisfies

lim
d(x)→0

u(x)

d−α(x)
= (

αm−1−q(α + 1)(m − 1)

β
)

1
p+q+1−m , (1.6)

where α = γ+m−q
p+q−m+1

. Furthermore, if p ≥ m − 1, then the nonnegative solution of
(1.1) is unique.

This work is organized as follows: In Section 2, we give a comparison principle
and prove Theorem 1.1. In Section 3, we first find out the blow-up rate in the
radially symmetric case and then prove Theorems 1.2 and 1.3.

2 Proofs of Theorem 1.1

Firstly, we consider the second order quasilinear operator Q of the form:

Q(u,ϕ) =

∫

Ω

(A(x, u, ∇u).∇ϕ − b(x, u, ∇u)ϕ)dx

where x = (x1, ..., xn) is contained in the domain Ω of R
N,the functions A(x, z, p)

and b(x, z, p) are assumed to be defined for all values of (x, z, p) in the set Ω×R ×
RN,ϕ ∈ C∞0 (Ω).

From [23], we get the following comparison principle which plays an important
role in the proofs of Theorems 1.2 and 1.3.

Lemma 2.1.(Comparison principle) Let u, v ∈ C1(Ω) satisfy Qu ≥ 0 in Ω,
Qv ≤ 0 in Ω and u ≤ v on ∂Ω, where the functions A, b are continuously differen-
tiable with respect to the z, p variables in Ω × R × RN, the operator Q is elliptic
in Ω, and the function b is non-increasing in z for fixed (x, p) ∈ Ω × RN. The, if
either

(i) the vector function A is independent of z; or

(ii) the function b is independent of p.

It follows that u ≤ v in Ω.

Now, we consider the general equation

△mu − ψ(x, u, ∇u) = 0 x ∈ Ω. (2.1)



CUBO
16, 2 (2014)

Existence of blow-up solutions for quasilinear elliptic equation . . . 57

Definition 2.1. Let 1 < k ≤ +∞, functions ū, u ∈ W1,k(Ω) are called the
weak upper and lower solutions of (2.1), respectively, if

ψ(·, ū(·), ∇ū(·)) ∈ Lk
′

(Ω),ψ(·, u(·), ∇u(·)) ∈ Lk
′

(Ω)

with

k′ =

{
k

k−1
, k < ∞,

1, k = ∞,

and
∫

Ω

|∇ū|m−2∇ū∇vdx ≥ −

∫

Ω

ψ(x, ū, ∇ū)vdx, ∀v ∈ W1,k0 (Ω), v ≥ 0 a.e.in Ω

∫

Ω

|∇u|m−2∇u∇vdx ≤ −

∫

Ω

ψ(x, u, ∇u)vdx, ∀v ∈ W1,k0 (Ω), v ≥ 0 a.e.in Ω.

If u ≤ ū, we call that they are the ordered weak upper and lower solutions of (2.1).

Firstly, we consider the existence of weak solution to the problem
{

△mu − ψ(x, u, ∇u) = 0, x ∈ Ω,

u = φ(x), x ∈ ∂Ω,
(2.2)

where ψ(·, u(·), ∇u(·)) ∈ Lk
′

(Ω),φ ∈ W1,k(Ω). Assume that ū ∈ W1,k(Ω) is a weak
upper solution (u ∈ W1,k(Ω) is a weak lower solution) of (2.1). Here by ū ≥ φ(u ≤
φ) on ∂Ω, we mean (φ − u)+ := max{φ − u, 0} ∈ W1,k0 (Ω) ((u − φ)

+ ∈ W1,k0 (Ω)).
If ū ≥ φ(u ≤ φ) on ∂Ω, we call that ū(u) is a weak upper solution (lower solution)
of (2.2). If u ≤ ū a.e. in Ω, we call that they are ordered.

Lemma 2.2.([9,Theorem 4.9]). Let ū, u ∈ W1,k(Ω) be the ordered weak upper
and lower solutions of (2.2), respectively, and u ≤ ū a.e. in Ω. Assume that there
exists a positive constant C1 and a function h1 ∈ L

k′(Ω) with k′ = k/(k − 1), such
that

|ψ(x, t,ξ)| ≤ h1(x) + C1|ξ|
k−1, a.e. x ∈ Ω, ∀ξ ∈ RN, t ∈ [u, ū]. (2.3)

Then there is a weak solution u ∈ W1,k(Ω) of the problem (2.2) such that u ≤ u ≤
ū a.e. in Ω.

Lemma 2.3. Let ψ ∈ Cµ(Ω × R × RN), and ū, u ∈ W1,∞(Ω) be the ordered
weak upper and lower solutions of (2.2), φ ∈ C1+µ(Ω), 0 < µ < 1, and u ≤ φ ≤
ū a.e. in Ω. Assume that there exists constants k > 1, c1 > 0, and a function
h1 ∈ C

µ(Ω)
"

L∞(Ω), such that (2.3) holds. Then for some 0 < β < 1, there is a
C1,β-solution u of (2.2) such that u ≤ u ≤ ū in Ω.



58 Fang Li & Zuodong Yang CUBO
16, 2 (2014)

The proof of the lemma 2.3 is similar to [10], so we omit it here.

Definition 2.2. A domain Ω is called satisfying the uniform outside spherical
condition: if there exists a constant r > 0 such that there exists a sphere B whose
radius is r in RN for any z ∈ ∂Ω, such that B ∩ Ω = {z}.

Noticing that any C2 bounded domain satisfies the uniform outside spherical
condition.

Lemma 2.4.( Theorem 4.2 in [20]) Assume that Ω is a bounded domain in RN

satisfying the uniform outside spherical condition, then there exists a series of C∞

domains {Ωn}
∞

1 , such that Ωn ⊂ Ωn+1 ⊂ Ω,
#

∞

n=1 Ωn = Ω.

Proof of Theorem 1.1. Since Ω is a C2 bounded domain, from Lemma 2.4
we know that there exists a series of C∞ domains {Ωn}

∞

1 , such that Ωn ⊂ Ωn+1 ⊂
Ω,

#

∞

n=1 Ωn = Ω. Now we consider the problem
{

△mu = ψ(x, u, ∇u), x ∈ Ωn,

u = ū, x ∈ ∂Ωn.
(2.4)

Since ū ∈ W1,∞(Ωn) and u ∈ W
1,∞(Ωn), by (1.5) we see that there is a constant

C2 = C2(n) > 0 such that |ψ(x, t,ξ)| ≤ h1(x) + C2 + C1|ξ|
k−1, a.e. x ∈ Ωn, ∀ξ ∈

RN, t ∈ [u, ū]. It is obvious that ū|Ω and u|Ω are the ordered upper and lower
solutions of (2.4), and h1 ∈ C(Ωn). By Lemma 2.3, there exists a solution un ∈
C1,β(Ωn) of (2.4) for some 0 < β < 1 such that u ≤ un ≤ ū in Ωn.

Now, we want to apply elliptic interior estimates together with a diagonal process
to conclude: {un : n ≥ 1} has a subsequence {uni : ni ↑ ∞} such that {uni} converges
to a function u in Ω(pointwise) and this convergence is in C1 on every compact set
in Ω. (Therefore, u ∈ C1 and div(|∇u|m−2∇u) = ψ(x, u, ∇u) with u(x) ≤ u(x) ≤
u(x), and this concludes the proof.)

Step 1. On Ω2, {un : n ≥ 2} is uniformly bounded by u(x) and u(x). Since
both u(x) and u(x) are bounded functions on Ω2, there exists M > 0 such that
∥u(x)∥L∞(Ωn) ≤ M, for all n ≥ 2.

From (2.4), un satisfies
∫

Ω2

|∇un|
m ≤

∫

Ω2

ψun. (2.5)

Therefore,
∫

Ω2

|∇un|
m ≤ M(measΩ2)

1/q′C1∥∇un∥m, (2.6)

here 1/q′ +1/m = 1, and C1 is the Sobolev embedding constant. So, ∥un∥1,m ≤ C2.
When 1 < m < N, the embedding of W1,m0 (Ω2) in L

Nm/(N−m)(Ω2) implies that



CUBO
16, 2 (2014)

Existence of blow-up solutions for quasilinear elliptic equation . . . 59

uk ∈ L
Nm/(N−m)(Ω2). Applying Theorem 7.1 in [21, Page 286-287], we obtain the

estimate

sup{|un|; x ∈ Ω2} ≤ C3, (2.7)

here C3 = C3(∥ψ∥0). If m ≥ N, we get (2.7) from the Sobolev embedding theorem.
Using Theorem 1.1 in ([21], Page 251), we see that un belongs to C

α(Ω2) for some
0 < α < 1, and

∥un∥Cα ≤ C4, (2.8)

here C4 is determined by C3. By Proposition 3.7 in [22, Page 806], we also know
that un belongs to C

1,α(Ω2) and

∥un∥C1,α ≤ C5. (2.9)

here C5 is determined by C4.

From the arguments above we see that there exists C > 0 such that

∥un∥C1+α(Ω1) ≤ C, for all n ≥ 2. (2.10)

Since the embedding C1+α(Ω1) → C1(Ω1) is compact, there exists a sequence de-
noted by {un1j}j=1,2... (where n1j ↑ ∞), which converges in C

1(Ω1). Let u1(x) =
limj→∞ un1j(x), for x ∈ Ω1, then u1 is a solution of (2.1) with u(x) ≤ u1 ≤ u(x).

Step 2. Repeat Step 1 up to the existence of the sequence {un1j}j=1,2... to get a
subsequence {un2i}i=1,2... converging in C

1(Ω2) to a limit u2. Then, likewise, u2 is a
solution of (2.4) and u2|Ω1 = u1. Repeat Step 1 again on Ω3, ..., etc. In this way,
we obtain a sequence {unnj}j=1,2... which converges in C

1(Ωn) and is a subsequence
of {unn−1)j}j=1,2.... Let un = limj→∞ unnj, then, un is a solution of (2.4) in Ωn and
un|Ωn−1 = un−1.

Step 3. By a diagonal process, {unll}l=1,2... is a subsequence of {unlj}j=1,2... for
every l. Thus, on Ωn for each n we have

lim
l→∞

unll = un.

So, if we define u(x) = limn→∞ un(x), then u(x) satisfies

div(|∇u|m−2∇u) = ψ(x, u, ∇u),

and u ≤ u(x) ≤ u (since u ≤ un(x) ≤ u) for every n. This completes the proof of
Theorem 1.1.



60 Fang Li & Zuodong Yang CUBO
16, 2 (2014)

3 Proofs of Theorems 1.2 and 1.3

To get the existence of large solutions, we first find the blow-up lower and super
solutions. Furthermore, if the blow-up lower solution and upper solution have the
same blow-up rate near the boundary, we could get the asymptotic behavior of large
solutions near the boundary. The idea of this section mainly comes from [10],[12]-
[14].

3.1 blow-up rate

In order to get the asymptotic behavior of large solutions in the general domain,
we first study the radically symmetric case:

{
(Φm(v

′))′ + N−1
r
Φp(v

′) = a(r)(R − r)γvp(1 + |v′|q), r ∈ (0, R),

v′(0) = 0, lim
r→R

v(r) = ∞,
(3.1)

where Φm(u) = |u|
m−2u, m ≥ 2. To ascertain the blow-up rate of the solution of

(3.1) at R > 0, we first find out the blow-up rate of the following one-dimensional
problem

{
(Φm(u

′))′ = a(r)(R − r)γup(1 + |u′|q), r ∈ (0, R),

u′(0) = 0, lim
r→R

u(r) = ∞.
(3.2)

Set u(r) = (R − r)−αψ(r), r ∈ [0, R] for some positive constant α which will be
determined later, ψ(r) ∈ C2(0, R), then the problem (3.2) becomes

(|u′|m−2u′)′ = (m − 1)(α(R − r)−α−1ψ(r) + (R − r)−αψ′(r))m−2

sgn(α(R − r)−α−1ψ(r) + (R − r)−αψ′(r))

[α(α + 1)(R − r)−α−2ψ(r) + 2α(R − r)−α−1ψ′(r) + (R − r)−αψ′′(r)]

= (m − 1)(R − r)−α−2−(α+1)(m−2)(αψ(r) + (R − r)ψ′(r))m−2

sgn(α(R − r)−α−1ψ(r) + (R − r)−αψ′(r))

[α(α + 1)ψ(r) + 2α(R − r)ψ′(r) + (R − r)2ψ′′(r)]

= a(r)(R − r)γ−αpψp(r) + a(r)(R − r)γ−αp−(α+1)q

ψp(r)|αψ(r) + (R − r)ψ′(r)|q (3.3)

with the boundary condition ψ(0) = 0, and ψ(R) ∈ (0, ∞). Therefore, the constant
α provides us with the exact blow-up rate of u at R. Multiplying (3.3) by (R −
r)α+2+(α+1)(m−2) we have



CUBO
16, 2 (2014)

Existence of blow-up solutions for quasilinear elliptic equation . . . 61

(m − 1)(αψ(r) + (R − r)ψ′(r))m−2sgn(α(R − r)−α−1ψ(r)

+(R − r)−αψ′(r))[α(α + 1)ψ(r) + 2α(R − r)ψ′(r) + (R − r)2ψ′′(r)]

= a(r)(R − r)γ−αp+α+2+(α+1)(m−2)ψp(r)

+a(r)(R − r)γ−αp−(α+1)q+α+2+(α+1)(m−2)ψp(r)|αψ(r) + (R − r)ψ′(r)|q.

Assuming lim
r→R

(R − r)2ψ′′(r) = lim
r→R

(R − r)ψ′(r) = 0, we obtain

α =
γ + m − q

p + q + 1 − m
, ψ(R) = (

αm−1−q(α + 1)(m − 1)

a(R)
)

1
p+q+1−m . (3.4)

Theorem 3.1. Assume that R > 0, a ∈ C([0, R]; (0, ∞)), m ≥ 2, γ ≥ 0, p, q >
0, and γ+m−q > 0, p +q > m−1. Let α and ψ(R) be defined by (3.4). Then for
each ϵ > 0, the problem (3.1) has at least one nonnegative C1-solution vϵ satisfying

1 − ϵ ≤ lim
r→R

inf
vϵ(r)

ψ(R)(R − r)−α
≤ lim

r→R
sup

vϵ(r)

ψ(R)(R − r)−α
≤ 1 + ϵ. (3.5)

Therefore, for each x0 ∈ R
N, m ≥ 2, the function

uϵ(x) := vϵ(r) with r := |x − x0|

provides us with a radially symmetric nonnegative solution of the problem
{

△mu = a(r)d
γ(x)up(1 + |∇u|q), x ∈ BR(x0),

u = ∞, x ∈ ∂BR(x0),

satisfying

1 − ϵ ≤ lim
d(x)→0

inf
uϵ(x)

ψ(R)d−α(x)
≤ lim

d(x)→0
sup

uϵ(x)

ψ(R)d−α(x)
≤ 1 + ϵ,

where d(x) := dist(x,∂BR(x0)) = R − |x − x0| = R − r.

Proof. Firstly, we show that, for each ϵ > 0 sufficiently small, there exists a
constant Aϵ > 0, for each A > Aϵ ,

v̄ϵ(r) := A + B+(
r

R
)2(R − r)−α (3.6)

provides us with a positive upper solution of (3.1), where α is defined in(3.4),

B+ = (1 + ϵ)(
αm−1−q(α + 1)(m − 1)

a(R)
)

1
p+q+1−m . (3.7)



62 Fang Li & Zuodong Yang CUBO
16, 2 (2014)

Indeed, v̄ϵ
′(0) = 0 and lim

r→R
v̄ϵ(r) = ∞. Thus, v̄ϵ is an upper solution of (3.1) if and

only if

(
B+
R2

)m−1[2(m − 1)(R − r)2 + 4(m − 1)αr(R − r) + (m − 1)α(α + 1)r2

+2(N − 1)(R − r)2 + α(N − 1)r(R − r)]|2r(R − r)2 + αr2|m−2

≤ a(r)(R − r)γ−αp+α+2+(α+1)(m−2)(A(R − r)α + B+(
r

R
)2)p((R − r)(α+1)p

+|
2B+
R2

r(R − r) + αB+(
r

R
)2|q). (3.8)

Note that γ − αp + α + 2 + (α + 1)(m − 2) = 0, at r = R, (3.8) becomes

(m − 1)αm−1(α + 1)Bm−1+ ≤ B
p+q
+ α

qa(R)

which is valid if and only if

B+ ≥ (
αm−1−q(α + 1)(m − 1)

a(R)
)

1
p+q+1−m .

Therefore, according to the choice of B+, inequality (3.8) is satisfied in a left neigh-
borhood of r = R, say (R − δ, R] for some δ = δ(ϵ) > 0. Finally, by choosing A
sufficiently large, it is clear that the inequality is satisfied in the whole interval [0, R]
since p > 0 and a is away from zero. This concludes the proof of the claim above.

Now we will construct a suitable lower solution for problem (3.1). We claim that
for each sufficiently small ϵ > 0, there exists C < 0 such that

vϵ(r) := max{0, C + B−(r/R)
2(R − r)−α} (3.9)

provides us with a nonnegative lower solution of (3.1), here

B− = (1 − ϵ)(
αm−1−q(α + 1)(m − 1)

a(R)
)

1
p+q+1−m . (3.10)

Indeed, vϵ is a lower solution of (3.1) where

C + B−(
r

R
)2(R − r)−α ≥ 0, (3.11)

which implies



CUBO
16, 2 (2014)

Existence of blow-up solutions for quasilinear elliptic equation . . . 63

(
B−
R2

)m−1[2(m − 1)(R − r)2 + 4(m − 1)αr(R − r) + (m − 1)α(α + 1)r2

+2(N − 1)(R − r)2 + α(N − 1)r(R − r)]|2r(R − r)2 + αr2|m−2

≥ a(r)(R − r)γ−αp+α+2+(α+1)(m−2)(C(R − r)α + B−(
r

R
)2)p(R − r)(α+1)p

+|
2B−
R2

r(R − r) + αB−(
r

R
)2|q). (3.12)

Now, for each C < 0, we can find the constant z = Z(C) ∈ (0, R), such that

C + B−(
r

R
)2(R − r)−α < 0 if r ∈ [0, Z(C)),

and

C + B−(
r

R
)2(R − r)−α > 0 if r ∈ (Z(C), R).

Moreover, Z(C) is decreasing and

lim
C→−∞

Z(C) = R, lim
C→0

Z(C) = 0.

At r = R, (3.12) becomes into

(m − 1)αm−1(α + 1)Bm−1− ≥ B
p+q
− α

qa(R),

which is valid if and only if

B− ≤ (
αm−1−q(α + 1)(m − 1)

a(R)
)

1
p+q+1−m .

Therefore, by making the choice (3.10), inequality (3.11) is satisfied in a left neigh-
borhood of r = R, say (R −δ, R] for some δ = δ(ϵ) > 0. Moreover, thanks to (3.12),
there exists C < 0, such that Z(C) = R − δ(ϵ). For this choice of C, vϵ provides us
with a weak lower solution of (3.1).

Since v̄ϵ(r), v(r) ∈ W
1,∞(0, R) are the ordered weak lower and upper solutions

of (3.1) and are bounded on any closed subdomain of [0, R), it is easy to see (1.5)
holds owing to p, q > 0. So the existence of a C1-solution u of (3.1) is followed by
Theorem 1.1, and

vϵ(r) ≤ u ≤ v̄ϵ(r) in Ω.

Finally, since

lim
r→R

v̄ϵ(r)

B+(R − r)−α
= lim

r→R

vϵ(r)

B−(R − r)−α
= 1, (3.13)

where B+ and B− are the constants defined through (3.7)and (3.10), one can easily
deduce the remaining assertions of Theorem 3.1. The proof is completed.



64 Fang Li & Zuodong Yang CUBO
16, 2 (2014)

3.2 Proofs of Theorems 1.2 and 1.3

Proof of Theorem 1.2. For n ≥ 1, we consider the following problem
{

△mu = b(x)u
p(1 + |∇u|q), x ∈ Ω,

u = n, x ∈ ∂Ω.
(3.14)

It is obvious that the function ψ(x, t,ξ) = b(x)tp(1 + |ξ|q) satisfies the condition
(2.3) since b(x) ∈ Cµ(Ω̄). The constant functions u(x) = 0 and u(x) = n are
the ordered lower and upper solutions of (3.14). By Lemma 2.3, we see that the
problem (3.14) has at least one nonnegative solution un(x) ∈ C

1(Ω) and it satisfies
0 ≤ un(x) ≤ n < n + 1.

By our assumption, b(x) ≥ β1d
γ(x) > 0 in Ω.So,u = un and ū = n + 1 are the

order lower and upper solutions of the following problem
{

△mu = b(x)u
p(1 + |∇u|q), x ∈ Ω,

u = n + 1, x ∈ ∂Ω.

Therefore un ≤ un+1. Now we fix a point x0 ∈ Ω and consider a small ball B centered
at x0 and contained properly in Ω. By Theorem 3.1, there exists a nonnegative C

1-
solution v(x) to the problem

{
△mu = β1(dist(x,∂B))

γup(1 + |∇u|q), x ∈ BR(x0),

u = ∞, x ∈ ∂BR(x0)

Since

△mun − b(x)u
p
n(1 + |∇un|

q)

≤ △mun − β1d
γ(x)upn(1 + |∇un|

q)

≤ △mun − β1(dist(x,∂B))
γ(x)upn(1 + |∇un|

q), x ∈ B,

and un(x) ≤ v(x) = ∞ on ∂B, by the Comparison principle (Lemma 2.1), we
have un(x0) ≤ v(x0) for all n. Since un increases in Ω as n increases, un(x)
is uniformly bounded on any compact subset of Ω. Standard elliptic regularity
arguments show that lim

n→∞
un(x) = u

∗(x) exists and u∗(x) satisfies the differential

equation of (1.1). To prove u∗(x) is a nonnegative solution of (1.1), it remains to
verify u∗(x)|∂Ω = ∞. If this is not true, then there exist a nonnegative constant M,
a sequence {xj} ⊂ Ω and x0 ∈ ∂Ω, such that xj → x0 and un(xj) ≤ M. For any
fixed k, note that un(xj) → u∗(xj) as n → ∞, it follows that there exists Nj > 0,
such that un(xj) ≤ 1 + M for all n ≥ Nj. Note that un is increasing in n, we have
un(xj) ≤ 1 + M for every n > 0. Now fix a n > 1 + M and let j → ∞ in the above



CUBO
16, 2 (2014)

Existence of blow-up solutions for quasilinear elliptic equation . . . 65

inequality, it follows un(x0) ≤ 1 + M < n since xj → x0, which is a contradiction
with un(x0) = n. The theorem is proved.

Proof of Theorem 1.3. In view of lim
d(x)→0

b(x)
d(x)γ

= β > 0 and b(x) > 0 in Ω, it

is easy to see that
b(x) ≥ β1d

γ(x), x ∈ Ω

for some β1 > 0. Thus, the nonnegative C
1-solution of problem (1.1) exists by

Theorem 1.2.

Now we prove the limit (1.6). Since

lim
d(x)→0

b(x)

d(x)γ
= β > 0,

for any small ϵ > 0, there exists δ = δ(ϵ) > 0, such that for all x ∈ Ω with
d(x) < 2δ,

(β − ϵ)dγ(x) ≤ b(x) ≤ (β + ϵ)dγ(x).

Now we define Ωδ = {x ∈ Ω : d(x) < δ} with ∂Ωδ = {x ∈ Ω : d(x) = δ} and
{

u+(x) = B+(ϵ)(d(x) − σ)
−α, x ∈ D+σ = Ω2δ/Ω̄σ,

u−(x) = B−(ϵ)(d(x) + σ)
−α, x ∈ D−σ = Ω2δ−σ,

(3.15)

where 0 < σ < δ, and

B+(ϵ) = (1 + ϵ)(
αm−1−q(α + 1)(m − 1)

β − ϵ
)

1
p+q+1−m ,

B−(ϵ) = (1 − ϵ)(
αm−1−q(α + 1)(m − 1)

β + ϵ
)

1
p+q+1−m .

It is easy to prove that by diminishing δ > 0 if necessary, d(x) is a C2-function on
the domain Ω̄2δ and

{
△mu

+ − b(x)(u+)p(1 + |∇u+|q) ≤ 0, x ∈ D+σ,

△mu
− − b(x)(u−)p(1 + |∇u−|q) ≥ 0, x ∈ D−σ.

(3.16)

Let u be any nonnegative solution of (1.1) and

M1(δ) = max
d(x)≥2δ

u(x), M2(δ) = B−(2δ)
−α.

We see that {
u(x) ≤ u+(x) + M1(δ), x ∈ ∂D

+
σ,

u−(x) ≤ u(x) + M2(δ), x ∈ ∂D
−
σ.

(3.17)



66 Fang Li & Zuodong Yang CUBO
16, 2 (2014)

On the other hand, by p > 0, we have
⎧
⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎩

△m(u
+ + M1(δ)) − b(x)(u

+ + M1(δ))
p(1 + |∇(u+ + M1(δ))|

q)

< 0, x ∈ D+σ,

△m(u + M2(δ)) − b(x)(u + M2(δ))
p(1 + |∇(u + M2(δ))|

q)

< 0, x ∈ D−σ.

(3.18)

Note (3.15) − (3.18), it follows by Lemma 2.1 that
{

u(x) ≤ u+(x) + M1(δ), x ∈ D
+
σ,

u−(x) ≤ u(x) + M2(δ), x ∈ D
−
σ.

(3.19)

For any x ∈ Ω2δ, there is a σ
∗ > 0, such that x ∈ D+δ

"

D−δ for all 0 < σ ≤ σ
∗.

Letting σ → 0 , (3.19) yields

B−(ϵ)d
−α(x) ≤ u + M2(δ) ≤ B+(ϵ)d

−α(x) + M1(δ) + M2(δ),

which implies

B−(ϵ) ≤ lim
d(x)→0

inf
uϵ(x)

d−α(x)
≤ lim

d(x)→0
sup

uϵ(x)

d−α(x)
≤ B+(ϵ). (3.20)

Taking ϵ → 0,(3.20) yields

lim
d(x)→0

u(x)

d−α(x)
= (

αm−1−q(α + 1)(m − 1)

β
)

1
p+q+1−m ,

which is accordance with (1.6).

The final step is to prove the uniqueness. Let u1 and u2 be two nonnegative
solutions of(1.1), then by(1.6), we have

lim
d(x)→0

u1(x)

u2(x)
= 1.

Indeed, for θ > 0 arbitrary, set

ωi = (1 + θ)ui, for i = 1, 2.

It follows that

lim
d(x)→0

(u1 − ω2)(x) = lim
d(x)→0

(u2 − ω1)(x) = −∞.

When p ≥ m − 1, since q > 0, we have that

△mωi − b(x)ω
p
i (1 + |∇ωi|

q) < (1 + θ)m−1[△mui − b(x)u
p
i (1 + |∇ui|

q)], x ∈ Ω.



CUBO
16, 2 (2014)

Existence of blow-up solutions for quasilinear elliptic equation . . . 67

Therefore, by lemma 2.1, we infer that

u1 ≤ (1 + θ)u2, u2 ≤ (1 + θ)u1, x ∈ Ω. (3.21)

Passing to the limit θ → 0+ in (3.21), we get u1 = u2 in Ω. This completes the
proof of Theorem1.3.

Received: September 2012. Revised: March 2014.

References

[1] C. Bandle, E. Giarrosso, Boundary blow-up for semilinear elliptic equations
with nonlinear gradient terms, Adv.Differential Equations 1(1996), 133-150.

[2] M. Ghergu, C. Niculescu and V. Rădulescu, Explosive solutions of elliptic
equations with absorption and non-linear gradient term, Proceedings of the
Indian Academy of Sciences.Mathematical Sciences 112(3)(2002), 441-451.

[3] Zuodong Yang, Existence of explosive positive solutions of quasilinear elliptic
equations, Applied Mathematics and Computation 177(2006), 581-588.

[4] Qishao Lu, Zuodong Yang, E.H.Twizell, Existence of entire explosive positive
solutions of quasilinear elliptic equations, Applied Mathematics and Compu-
tation 148(2004), 359-372.

[5] Cunlian Liu and Zuodong Yang, Existence of large solutions for a quasilinear
elliptic problem via explosive sub-supersolutions, Applied Mathematics and
Computation 199(2008), 414-424.

[6] Cunlian Liu and Zuodong Yang, Existence of large solutions for quasilinear
elliptic problems with a gradient term, Applied Mathematics and Computa-
tion 192(2007),533-545.

[7] Cunlian Liu and Zuodong Yang, Boundary blow-up quasilinear elliptic prob-
lems of the Bieberbach type with nonlinear gradient terms, Nonlinear Anal.
69(2008), 4380-4391.

[8] Cunlian Liu and Zuodong Yang, A Boundary blow-up for a class of quasilinear
elliptic problems with gradient term. J.Appl. Math. Comput. 33(2010), 23-34.

[9] Y. H. Du, Order structure and topological methods in nonlinear PDEs, in:
Maximum Principle and Applications, vol. 1, World Scientific, Press, Singa-
pore, 2005.



68 Fang Li & Zuodong Yang CUBO
16, 2 (2014)

[10] Yujuan Chen and Mingxin Wang, Large solutions for quasilinear elliptic equa-
tion with nonlinear gradient term, Nonlinear Analysis: Real World Applica-
tions 12(2011), 455-463.

[11] J. V. Goncalves and A. Roncaili, Boundary blow-up solutions for a class of
elliptic equations on a bounded domain, Appl. Math. Comput. 182(2006),
13-23.

[12] F-C. Cirstea, Y. H. Du, General uniqueness results and variation speed for
blow-up solutions of elliptic equations, Proc. Lond. Math. Soc. 91(2)(2005),
459-482.

[13] M. Delgado, J. Lopez-Gomez and A. Suarez, Singular boundary value prob-
lems of a porous media logistic equation, Hiroshima Math. J. 34(2004), 57-80.

[14] J. Lopez, The boundary blow-up rate of large solutions, J.Differential Equa-
tions 195(2003), 25-45.

[15] F. Gladiali, G. Porru, Estimates for explosive solutions to p-Laplace equa-
tions, in: Progress in Partial Differential Equations, Pont--Mousson, 1997,
vol. 1, in: Pitman Res. Notes Math. Ser., vol. 383, Longman, Harlow, 1998,
pp. 117-127.

[16] Y. Du, Z. Guo, Boundary blow-up solutions and their applications in quasi-
linear elliptic equations, J. Anal. Math. 89 (2003) 277-302.

[17] Ahmed Mohammed,Boundary asymptotic and uniqueness of solutions to the
p-Laplacian with infinite boundary values.J. Math. Anal. Appl. 325 (2007)
480-489.

[18] G. Diaz, R. Letelier, Explosive solutions of quasilinear elliptic equation: ex-
istence and uniqueness, Nonlinear Anal. 20(1993), 97-125.

[19] Zuodong Yang, Bing Xu and Mingzhu Wu, Existence of positive boundary
blow-up solutions for quasilinear elliptic equations via sub and supersolutions,
Applied Mathematics and Computation 188(1)(2007), 492-498.

[20] A. C. Lazer, P. J. McKenna, On a problem of Bieberbach and Rademacher,
Nonlinear Anal. 21(1993), 327-335.

[21] O. A. Ladyzhenskaya,N. N. Ural’tseva, Linear and Quasilinear Elliptic Equa-
tions, Academic Press, New York, 1968.

[22] P. Tolksdorf, On the Dirichlet problem for quasilinear equations in do-
mains with conical boundary point, Comm. Partial Differential Equations,
8(7)(1983), 773-817.



CUBO
16, 2 (2014)

Existence of blow-up solutions for quasilinear elliptic equation . . . 69

[23] D. Gilbarg, N. S. Trudinger, Elliptic Partial Differential Equations of Second
Order,2nd ed., Springer-Verlag, Press, Berlin, 1983.

[24] T. Leonori, Large solutions for a class of nonlinear elliptic equations with
gradient terms, Adv. Nonlin. Stud. 7(2007), 237-269.

[25] S.Alarcon, J.Garcia-melian, A. Quaas, Keller-Osserman type conditions for
some elliptic problems with gradient terms, J.Diff.Eqns. 252(2012), 886-914.