Open Access

Bifurcation from interval at infinity for discrete eigenvalue problems which are not linearizable

Advances in Difference Equations20142014:125

https://doi.org/10.1186/1687-1847-2014-125

Received: 3 October 2013

Accepted: 4 April 2014

Published: 6 May 2014

Abstract

In this paper, we are concerned with the bifurcation from infinity for a class of discrete eigenvalue problems with nonlinear boundary conditions which are not linearizable and give a description of the behavior of the bifurcation components.

MSC:34B10, 34B15.

Keywords

difference equationnonlinear boundary conditionbifurcation from intervalpositive solutions

1 Introduction

For over a decade, there has been significant interest in positive solutions and multiple positive solutions for boundary value problems for finite difference equations; see, for example, [111]. Much of this interest has been spurred on by the applicability of the topological method such as the upper and lower solutions technique [2], a number of new fixed point theorems and multiple fixed point theorems [37] as applied to certain discrete boundary value problems. Quite recently, Rodríguez [9] and Ma [10] have given a topological proof and used a bifurcation theorem to study the structure of positive solutions of a difference equation. In this paper, we demonstrate a bifurcation technique that takes advantage of dealing with the discrete eigenvalue problem with nonlinear boundary conditions
Δ [ p ( k 1 ) Δ y ( k 1 ) ] + q ( k ) y ( k ) = λ a ( k ) f ( y ( k ) ) , k I , Δ y ( 0 ) + α g ( y ( 0 ) ) = 0 , Δ y ( N ) + β g ( y ( N + 1 ) ) = 0 ,
(1.1)

which has different asymptotic linearizations at infinity. Here I = { 1 , 2 , , N } , Δ is a forward difference operator with Δ y ( k ) = y ( k + 1 ) y ( k ) , α , β 0 are constants, λ > 0 is a parameter, the functions p : { 0 , 1 , , N } ( 0 , ) , q , a : I [ 0 , ) with a ( k ) > 0 on k I and functions f, g satisfy f C 1 ( [ 0 , ) ) and g C 1 ( [ 0 , ) ) .

Since problem (1.1) has different linearizations at infinity, the standard global bifurcation results [12] are not immediately applicable. However, Schmitt [13] and Peitgen and Schmitt [14] obtained a theorem on bifurcation from intervals at infinity. We can use this theorem to discuss the bifurcation from infinity for problem (1.1) and obtain further information on the location and behavior of the bifurcating sets of solutions.

We will make the following assumptions:

  • (H1) p : { 0 , 1 , , N } ( 0 , ) , q , a : I [ 0 , ) with a ( k ) > 0 on k I , q 0 and q 0 .

  • (H2) f C 1 ( [ 0 , ) ) and there exist constants f , f ( 0 , ) and functions h 1 , h 2 C 1 ( [ 0 , ) ) such that
    f = lim inf s f ( s ) s , f = lim sup s f ( s ) s ,
    (1.2)
    f s + h 1 ( s ) f ( s ) f s + h 2 ( s ) , s [ 0 , )
    (1.3)
    with
    h j ( s ) = o ( | s | ) as  s , j = 1 , 2 .
  • (H3) g C 1 ( [ 0 , ) ) and there exist constants g , g ( 0 , ) and functions γ 1 , γ 2 C 1 ( [ 0 , ) ) such that
    g = lim inf s g ( s ) s , g = lim sup s g ( s ) s , g s + γ 1 ( s ) g ( s ) g s + γ 2 ( s ) , s [ 0 , )
    (1.4)
    with
    γ j ( s ) = o ( | s | ) as  s , j = 1 , 2 .

Let I ˆ = { 0 , 1 , , N + 1 } and X = { y | y : I ˆ R } be the space of all real-valued functions on  I ˆ . Then it is a Banach space with the norm y = max { | y ( k ) | | k I ˆ } .

Definition 1.1 ([[13], p.450], [14])

A solution set S of (1.1) is said to bifurcate from infinity in the interval [ a , b ] if
  1. (i)

    the solutions of (1.1) are a priori bounded in X for λ = a and λ = b .

     
  2. (ii)

    there exists { ( μ n , y n ) } S such that { μ n } [ a , b ] and y n .

     
Let λ 1 and μ 1 be the first eigenvalues of
Δ [ p ( k 1 ) Δ φ ( k 1 ) ] + q ( k ) φ ( k ) = λ a ( k ) φ ( k ) , k I , Δ φ ( 0 ) + α g φ ( 0 ) = 0 , Δ φ ( N ) + β g φ ( N + 1 ) = 0 ,
(1.5)
and
Δ [ p ( k 1 ) Δ ψ ( k 1 ) ] + q ( k ) ψ ( k ) = μ a ( k ) ψ ( k ) , k I , Δ ψ ( 0 ) + α g ψ ( 0 ) = 0 , Δ ψ ( N ) + β g ψ ( N + 1 ) = 0 ,
(1.6)

respectively, φ 1 and ψ 1 be the corresponding eigenfunctions of λ 1 and μ 1 , respectively, and φ 1 , ψ 1 X be positive and normalized as φ 1 = 1 and ψ 1 = 1 .

The main results are the following.

Theorem 1.1 Assume that (H1)-(H3) hold. Then, for any σ ( 0 , μ 1 f ) , the interval [ μ 1 f σ , λ 1 f + σ ] is a bifurcation interval from infinity of (1.1), and there exists no bifurcation interval from infinity of (1.1) in the set ( 0 , ) [ μ 1 f σ , λ 1 f + σ ] . More precisely, there exists an unbounded, closed and connected component in ( 0 , ) × X , consisting of positive solutions of (1.1) and bifurcating from [ μ 1 f σ , λ 1 f + σ ] × { } .

Theorem 1.2 Assume that (H1)-(H3) hold.
  1. (i)
    If
    lim inf u h 1 ( u ) > f g lim sup u γ 2 ( u ) ,
    (1.7)
     
then the component obtained by Theorem  1.1 bifurcates into the region λ < λ 1 f .
  1. (ii)
    If
    lim sup u h 2 ( u ) < f g lim inf u γ 1 ( u ) ,
    (1.8)
     

then the component obtained by Theorem  1.1 bifurcates into the region λ > μ 1 f .

Remark 1.1 Notice that μ 1 λ 1 . Indeed, let ( λ 1 , φ 1 ) and ( μ 1 , ψ 1 ) satisfy (1.5) and (1.6), respectively. Multiplying (1.5) by ψ 1 and (1.6) by φ 1 , summing from k = 1 to k = N and subtracting, we have that
( λ 1 μ 1 ) k = 1 N a ( k ) φ 1 ( k ) ψ 1 ( k ) = k = 1 N [ Δ [ p ( k 1 ) Δ ψ 1 ( k 1 ) ] φ 1 ( k ) Δ [ p ( k 1 ) Δ φ 1 ( k 1 ) ] ψ 1 ( k ) ] = k = 1 N [ p ( k ) [ ψ 1 ( k + 1 ) φ 1 ( k ) ψ 1 ( k ) φ 1 ( k + 1 ) ] + p ( k 1 ) [ ψ 1 ( k 1 ) φ 1 ( k ) ψ 1 ( k ) φ 1 ( k 1 ) ] ] = p ( N ) [ ψ 1 ( N + 1 ) φ 1 ( N ) ψ 1 ( N ) φ 1 ( N + 1 ) ] + p ( 0 ) [ ψ 1 ( 0 ) φ 1 ( 1 ) ψ 1 ( 1 ) φ 1 ( 0 ) ] = [ α p ( 0 ) φ 1 ( 0 ) ψ 1 ( 0 ) + β p ( N ) φ 1 ( N + 1 ) ψ 1 ( N + 1 ) ] ( g g ) 0 .

This together with k = 1 N a ( k ) φ 1 ( k ) ψ 1 ( k ) > 0 implies μ 1 λ 1 .

2 Bifurcation theorem and reduction to a compact operator equation

Our main tools in the proof of Theorems 1.1-1.2 are topological arguments [15] and the global bifurcation theorem for mappings which are not necessary smooth [13, 14].

Let V be a real Banach space and F : R × V V be completely continuous. Let us consider the equation
u = F ( λ , u ) .
(2.1)

Lemma 2.1 ([[13], Theorem 1.3.3])

Let V be a Banach space, F : R × V V be completely continuous and a , b R ( a < b ) be such that the solutions of (2.1) are a priori bounded in V for λ = a and λ = b , i.e., there exists R > 0 such that
u F ( a , u ) , u F ( b , u ) for all u : u R .
Furthermore, assume that
d ( I F ( a , ) , B R ( 0 ) , 0 ) d ( I F ( b , ) , B R ( 0 ) , 0 )
for R > 0 large. Then there exists a closed connected set C of solutions of (2.1) that is unbounded in [ a , b ] × V , and either
  1. (i)

    C is unbounded in λ direction, or else

     
  2. (ii)

    there exists an interval [ c , d ] such that ( a , b ) ( c , d ) = and C bifurcates from infinity in [ c , d ] × V .

     

To establish Theorem 1.1, we begin with the reduction of (1.1) to a suitable equation for a compact operator and give some preliminary results.

Let u α ¯ ( k ) , v β ¯ ( k ) be the solutions of the initial value problems
Δ [ p ( k 1 ) Δ u α ¯ ( k 1 ) ] + q ( k ) u α ¯ ( k ) = 0 for  k I , u α ¯ ( 0 ) = 1 , Δ u α ¯ ( 0 ) = α ¯ ,
and
Δ [ p ( k 1 ) Δ v β ¯ ( k 1 ) ] + q ( k ) v β ¯ ( k ) = 0 for  k I , v β ¯ ( N + 1 ) = 1 , Δ v β ¯ ( N ) = β ¯ ,
respectively, here α ¯ , β ¯ [ 0 , ) . It is easy to compute and show that
  1. (i)

    u α ¯ ( k ) = 1 + α ¯ s = 0 k 1 p ( 0 ) p ( s ) + s = 1 k 1 ( j = s k 1 1 p ( j ) ) q ( s ) u α ¯ ( s ) > 0 , and u is increasing on I ˆ ;

     
  2. (ii)

    v β ¯ ( k ) = 1 + β ¯ s = k N p ( N ) p ( s ) + s = k + 1 N ( j = k + 1 N 1 1 p ( j ) ) q ( s ) v β ¯ ( s ) > 0 , and v is decreasing on I ˆ .

     
Lemma 2.2 Let h : I R . Then the linear boundary value problem
Δ [ p ( k 1 ) Δ y ( k 1 ) ] + q ( k ) y ( k ) = h ( k ) , k I , Δ y ( 0 ) + α ¯ y ( 0 ) = 0 , Δ y ( N ) + β ¯ y ( N + 1 ) = 0
(2.2)
has a solution
y ( k ) = s = 1 N G α ¯ , β ¯ ( k , s ) h ( s ) , k I ˆ ,
(2.3)
where
G α ¯ , β ¯ ( k , s ) = { u α ¯ ( s ) v β ¯ ( k ) , 1 s k N + 1 , u α ¯ ( k ) v β ¯ ( s ) , 0 k s N .

Moreover, if h ( k ) 0 and h 0 on I, then y ( k ) > 0 on I ˆ .

Proof It is a consequence of Atici [[5], Section 2], so we omit it. □

Let ω 1 , ω 2 ( 0 , ) . Then the linear boundary value problem
Δ [ p ( k 1 ) Δ y ( k 1 ) ] + q ( k ) y ( k ) = 0 , k I , Δ y ( 0 ) + α ¯ y ( 0 ) = w 1 , Δ y ( N ) + β ¯ y ( N + 1 ) = w 2
(2.4)
has a solution
y ( k ) = ω 2 u α ¯ ( k ) ( 1 + β ¯ ) u α ¯ ( N + 1 ) u α ¯ ( N ) + ω 1 v β ¯ ( k ) ( 1 + α ¯ ) v β ¯ ( 0 ) v β ¯ ( 1 ) , k I ˆ .
From the properties of u α ¯ ( k ) , v β ¯ ( k ) , it follows that
y ( k ) > 0 , k I ˆ .
Define the operator T : X X as follows:
T [ h ] ( k ) = s = 1 N G α ¯ , β ¯ ( k , s ) h ( s ) , k I ˆ .

By a standard argument of compact operator, it is easy to show that T is a compact operator and it is strong positive, meaning that T h > 0 on I ˆ for any h X with the condition that h 0 and h 0 on I; see [2, 4, 5].

Let R [ ω 1 , ω 2 ] : R 2 X be defined as
R [ ω 1 , ω 2 ] ( k ) = ω 2 u α ¯ ( k ) ( 1 + β ¯ ) u α ¯ ( N + 1 ) u α ¯ ( N ) + ω 1 v β ¯ ( k ) ( 1 + α ¯ ) v β ¯ ( 0 ) v β ¯ ( 1 ) , k I ˆ .
(2.5)

Then R [ ω 1 , ω 2 ] is a linear bounded function.

From Lemma 2.2, let T , T : X X denote the resolvent of the linear boundary value problems
Δ [ p ( k 1 ) y ( k 1 ) ] + q ( k ) y ( k ) = h ( k ) , k I , Δ y ( 0 ) + α g y ( 0 ) = 0 , Δ y ( N ) + β g y ( N + 1 ) = 0 ,
(2.6)
and
Δ [ p ( k 1 ) y ( k 1 ) ] + q ( k ) y ( k ) = h ( k ) , k I , Δ y ( 0 ) + α g y ( 0 ) = 0 , Δ y ( N ) + β g y ( N + 1 ) = 0 ,

respectively. Taking into account α ¯ = α g ( α ¯ = α g ), β ¯ = β g ( β ¯ = β g ), one can repeat the argument of the operator T with some obvious changes. It follows that T , T are linear mappings of X compactly into X and they are strong positive.

Let R , R be the solutions of linear boundary value problem (2.4) with α ¯ , β ¯ in place of α ¯ = α g , β ¯ = β g and α ¯ = α g , β ¯ = β g , respectively. Repeating the argument of R [ ω 1 , ω 2 ] with some minor changes, we have that R , R : R 2 X is a linear, bounded mapping and
R [ ω 1 , ω 2 ] ( k ) = ω 2 u α g ( k ) ( 1 + β g ) u α g ( N + 1 ) u α g ( N ) + ω 1 v β g ( k ) ( 1 + α g ) v β g ( 0 ) v β g ( 1 ) , R [ ω 1 , ω 2 ] ( k ) = ω 2 u α g ( k ) ( 1 + β g ) u α g ( N + 1 ) u α g ( N ) + ω 1 v β g ( k ) ( 1 + α g ) v β g ( 0 ) v β g ( 1 ) .

Lemma 2.3 Let (H1)-(H3) hold. If [ α , β ] is a bifurcation interval from infinity of the set of nonnegative solutions of (1.1), then we have ( α , β ) [ μ 1 f , λ 1 f ] . Moreover, there exist constants ϵ > 0 small enough and M > 0 large enough such that any nonnegative solution u of (1.1) is positive on I ˆ whenever dist ( λ , [ μ 1 f , λ 1 f ] ) < ϵ and u M .

Proof Let ( μ j , y j ) be a nonnegative solution of (1.1) with λ = μ j [ α , β ] such that
y j , and μ j μ ˆ as  j .
(2.7)
If
w j = y j y j ,
(2.8)
then we have
w j = μ j T [ a f ( y j ) y j ] + R [ τ ( g w j g ( y j ) y j ) ] in  X ,
(2.9)
here τ : { y ( 0 ) , y ( 1 ) , , y ( N + 1 ) } { y ( 0 ) , y ( N + 1 ) } is a linear operator and
R [ τ ( g w j g ( y j ) y j ) ] ( k ) = g w j ( N + 1 ) g ( y j ( N + 1 ) ) / y j ( 1 + β g ) u α g ( N + 1 ) u α g ( N ) u α g ( k ) + g w j ( 0 ) g ( y j ( 0 ) ) / y j ( 1 + α g ) v β g ( 0 ) v β g ( 1 ) v β g ( k ) .
From conditions (1.3) and (1.4), for any ϵ > 0 , there exist constants d ϵ , c ϵ > 0 such that
f ( s ) f s ϵ s + d ϵ , u 0 , f ( s ) f s ϵ s d ϵ , u 0 ,
(2.10)
g ( s ) g s ϵ s + c ϵ , u 0 , g ( s ) g s ϵ s c ϵ , u 0 .
(2.11)
Those imply that both f ( y j ) / y j and g ( y j ) / y j are bounded. By the compactness of T and R , it follows from (2.9) that there exist a function w 0 X and a subsequence of { w j } , still denoted by { w j } , such that
w j w 0 in  X  as  j .
(2.12)
By (2.7), it follows from (2.10)-(2.11) that
lim sup j max k I ( f ( y j ) y j f w j ) ϵ , lim inf j min k I ( f ( y j ) y j f w j ) ϵ , lim sup j max k I ( g ( y j ) y j g w j ) ϵ , lim inf j min k I ( g ( y j ) y j g w j ) ϵ .
(2.13)
Since ϵ is arbitrary, it follows that
lim sup j f ( y j ) y j f w 0 in  X , lim inf j f ( y j ) y j f w 0 in  X , lim sup j g ( y j ) y j g w 0 in  X , lim inf j g ( y j ) y j g w 0 in  X .
(2.14)
Let μ j μ ˆ , f ( y j ) y j ν 0 and g ( y j ) y j z 0 as j . Then, in view of (2.9),
w 0 = μ ˆ T [ a ν 0 ] + R [ τ ( g w 0 z 0 ) ] .
(2.15)
We claim that
μ ˆ [ μ 1 f , λ 1 f ] .
Since
f w 0 ν 0 f w 0 and g w 0 z 0 g w 0 ,
it follows from (2.15) that
μ ˆ T [ a f w 0 ] + R [ τ ( ( g g ) w 0 ) ] w 0 μ ˆ T [ a f w 0 ] + R [ τ ( ( g g ) w 0 ) ] .
Moreover, we have
w 0 μ ˆ T [ a f w 0 ] .
(2.16)

Since w 0 = 1 and w 0 0 , the strong positivity of T ensures that w 0 > 0 on I ˆ .

Obviously, w 0 satisfies the following boundary value problem:
Δ [ p ( k 1 ) w 0 ( k 1 ) ] + q ( k ) w 0 ( k ) = μ ˆ a ( k ) ν 0 , k I , Δ w 0 ( 0 ) + α z 0 = 0 , Δ w 0 ( N ) + β z 0 = 0 .
(2.17)
This combined with φ 1 , ψ 1 satisfying (1.5) and (1.6) can get that
( μ ˆ f λ 1 ) k = 1 N a ( k ) φ 1 ( k ) w 0 ( k ) k = 1 N a ( k ) φ 1 ( k ) ( μ ˆ ν 0 ( k ) λ 1 w 0 ( k ) ) = k = 1 N [ Δ [ p ( k 1 ) Δ w 0 ( k 1 ) ] φ 1 ( k ) + Δ [ p ( k 1 ) Δ φ 1 ( k 1 ) ] w 0 ( k ) ] = p ( N ) [ φ 1 ( N + 1 ) w 0 ( N ) w 0 ( N + 1 ) φ 1 ( N ) ] + p ( 0 ) [ φ 1 ( 0 ) w 0 ( 1 ) φ 1 ( 1 ) w 0 ( 0 ) ] = p ( N ) β φ 1 ( N + 1 ) [ z 0 g w 0 ( N + 1 ) ] + p ( 0 ) α φ 1 ( 0 ) [ z 0 g w 0 ( 0 ) ] p ( N ) β φ 1 ( N + 1 ) [ g w 0 ( N + 1 ) g w 0 ( N + 1 ) ] + p ( 0 ) α φ 1 ( 0 ) [ g w 0 ( 0 ) g w 0 ( 0 ) ] = 0
(2.18)
and
( μ ˆ f μ 1 ) k = 1 N a ( k ) ψ 1 ( k ) w 0 ( k ) k = 1 N a ( k ) ψ 1 ( k ) ( μ ˆ ν 0 ( k ) μ 1 w 0 ( k ) ) = k = 1 N [ Δ [ p ( k 1 ) Δ w 0 ( k 1 ) ] ψ 1 ( k ) + Δ [ p ( k 1 ) Δ ψ 1 ( k 1 ) ] w 0 ( k ) ] = p ( N ) [ ψ 1 ( N + 1 ) w 0 ( N ) w 0 ( N + 1 ) ψ 1 ( N ) ] + p ( 0 ) [ ψ 1 ( 0 ) w 0 ( 1 ) ψ 1 ( 1 ) w 0 ( 0 ) ] = p ( N ) β ψ 1 ( N + 1 ) [ z 0 g w 0 ( N + 1 ) ] + p ( 0 ) α ψ 1 ( 0 ) [ z 0 g w 0 ( 0 ) ] p ( N ) β ψ 1 ( N + 1 ) [ g w 0 ( N + 1 ) g w 0 ( N + 1 ) ] + p ( 0 ) α ψ 1 ( 0 ) [ g w 0 ( 0 ) g w 0 ( 0 ) ] = 0 .
(2.19)
Thus
μ 1 f μ ˆ λ 1 f .

Since w 0 > 0 on I ˆ , (2.12) implies that w j > 0 on I ˆ for j large enough, and so is y j from (2.8). This leads to the latter part of assertions of this proposition. □

3 Existence of a bifurcation interval from infinity

This section is devoted to studying the existence of a bifurcation interval from infinity for (1.1). To do this, we associate with (1.1) a nonlinear mapping Φ ( λ , y ) : ( 0 , ) × X X as follows:
Φ ( λ , y ) : = y λ T [ a f ( | y | ) ] R [ τ ( g y g ( | y | ) ) ] .
(3.1)

We note that a nonnegative y X attains (1.1) if and only if Φ ( λ , y ) = 0 .

In this section, we shall apply Lemma 2.1 to show that for any σ ( 0 , μ 1 f ) , the interval [ μ 1 f σ , λ 1 f + σ ] is a bifurcation interval from infinity for (3.1) and, consequently, [ μ 1 f σ , λ 1 f + σ ] is a bifurcation interval from infinity for the nonnegative solutions of (1.1).

In fact, if [ μ 1 f σ , λ 1 f + σ ] is a bifurcation interval from infinity for (3.1), then, according to Definition 1.1, we have
  1. (i)

    the solutions of (3.1) are a priori bounded in X for λ = μ 1 f σ and λ = λ 1 f + σ .

     
  2. (ii)

    there exists a sequence { ( μ n , y n ) } S such that { μ n } [ μ 1 f σ , λ 1 f + σ ] and y n .

     
Let { μ n j } be any convergent subsequence of { ( μ n , y n ) } , and let
lim j μ n j = μ and lim j y n j = .
We claim that
μ [ μ 1 f , λ 1 f ] and y n j > 0 on  I ˆ  if  j  is large enough .
(3.2)
Indeed, as in the proof of Lemma 2.3, we have the same conclusion that there exist some ν 0 , z 0 X and μ such that
w 0 = μ T ( a | ν 0 | ) + R [ τ ( g w 0 | z 0 | ) ] .
(3.3)
Since
f w 0 | ν 0 | f w 0 and g w 0 | z 0 | g w 0 ,
it follows from the strong positivity of T and the positivity of R that
μ T [ a f w 0 ] w 0 μ T [ a f w 0 ] + R [ τ ( g g ) w 0 ] .
(3.4)
This together with the strong positivity of T implies that
w 0 > 0 on  I ˆ .
(3.5)
By using (2.18) and (2.19) with obvious changes, it follows that
μ 1 f μ λ 1 f .
(3.6)

From (3.5), it follows that w j > 0 on I ˆ for j large enough and so is y j from (2.8). Therefore [ μ 1 f σ , λ 1 f + σ ] is actually an interval of bifurcation from infinity for (1.1).

In what follows, we shall apply Lemma 2.1 to show that [ μ 1 f σ , λ 1 f + σ ] is a bifurcation interval from infinity for (3.1), two lemmas on the nonexistence of solutions will be first shown. Let Φ χ : ( 0 , ) × X X be defined as
Φ χ ( λ , u ) : = u λ T [ a f ( | u | ) ] χ ( λ ) R [ τ ( g u g ( | u | ) ) ] .
(3.7)
Here χ : [ 0 , μ 1 f ] [ 0 , 1 ] is a smooth cut-off function such that
χ ( λ ) = { 0 near  λ = 0 , 1 near  λ = μ 1 f .
(3.8)
Lemma 3.1 Let (H1)-(H3) hold and Λ R + be a compact interval with Λ [ μ 1 f , λ 1 f ] = . Then there exists a constant r > 0 such that
Φ χ ( λ , y ) 0 , λ Λ , y X : y r .
(3.9)
Proof Assume on the contrary that there exist λ j 0 , y j X and λ 0 Λ such that
Φ χ ( λ j , y j ) = 0 , λ j λ 0 , y j  as  j .

The same argument as in the proof of Lemma 2.3 gives a contradiction that μ 1 f λ 0 λ 1 f . This is a contradiction. The proof of Lemma 3.1 is complete. □

Lemma 3.2 Let (H1)-(H3) hold. Then for any λ > λ 1 f fixed, there exists a constant r > 0 such that
Φ ( λ , y ) t φ 1 , t [ 0 , 1 ] , y X : y r .
(3.10)
Proof Assume on the contrary that there exist μ 0 ( λ 1 f , ) , t 0 , t j [ 0 , 1 ] , and y j X can be taken such that
Φ ( μ 0 , y j ) = t j φ 1 , t j t 0 , y j  as  j .
Using the same argument as in the proof of Lemma 2.3, we can obtain a subsequence of { y j } , still denoted by { y j } , which may satisfy that y j > 0 on I ˆ for all j > 1 . It follows that
y j = μ 0 T [ a f ( y j ) ] + R [ τ ( g y j g ( y j ) ) ] + t j φ 1 , t j t 0 [ 0 , 1 ] , y j  as  j .
(3.11)
Thus
Δ [ p ( k 1 ) Δ y j ( k 1 ) ] + q ( k ) y j ( k ) = μ 0 a ( k ) f ( y j ) + t j λ 1 φ 1 , k I , Δ y j ( 0 ) + α g ( y j ( 0 ) ) = 0 , Δ y j ( N ) + β g ( y j ( N + 1 ) ) = 0 .
(3.12)
Moreover, it follows from φ 1 satisfies (1.5) and (3.12) that
k = 1 N [ μ 0 a ( k ) f ( y j ) φ 1 ( k ) + t j λ 1 φ 1 2 ( k ) λ 1 a ( k ) y j ( k ) φ 1 ( k ) ] = k = 1 N [ Δ [ p ( k 1 ) Δ y j ( k 1 ) ] φ 1 ( k ) + Δ [ p ( k 1 ) Δ φ 1 ( k 1 ) ] y j ( k ) ] = p ( N ) β φ 1 ( N + 1 ) [ g ( y i ( N + 1 ) ) g y j ( N + 1 ) ] + p ( 0 ) α φ 1 ( 0 ) [ g ( y j ( 0 ) ) g y j ( 0 ) ] .
(3.13)
This implies that
p ( N ) β φ 1 ( N + 1 ) [ g ( y i ( N + 1 ) ) g y j ( N + 1 ) ] + p ( 0 ) α φ 1 ( 0 ) [ g ( y j ( 0 ) ) g y j ( 0 ) ] ( μ 0 f λ 1 ) k = 1 N a ( k ) y j ( k ) φ 1 ( k ) + μ 0 k = 1 N a ( k ) h 1 ( y j ( k ) ) φ 1 ( k ) .
Hence assertion (2.10) gives
p ( N ) β φ 1 ( N + 1 ) [ g ( y i ( N + 1 ) ) g y j ( N + 1 ) ] y j + p ( 0 ) α φ 1 ( 0 ) [ g ( y j ( 0 ) ) g y j ( 0 ) ] y j f ( μ 0 λ 1 + μ 0 ε f ) k = 1 N a ( k ) y j ( k ) y j φ 1 ( k ) μ 0 a d ε y j k = 1 N φ 1 ( k ) .
(3.14)
Now use again for (3.12) the same procedure as in the proof of Lemma 2.3, then we see that some subsequence of { y j / y j } , still denoted by { y j / y j } , tends to a positive function w 0 in X. Take ϵ > 0 so small that μ 0 λ 1 + μ 0 ϵ f > 0 . Then combining (3.14) with (2.15) leads to a contradiction that
0 = p ( N ) β φ 1 ( N + 1 ) [ g w 0 ( N + 1 ) g w 0 ( N + 1 ) ] + p ( 0 ) α φ 1 ( 0 ) [ g w 0 ( 0 ) g w 0 ( 0 ) ] lim j { p ( N ) β φ 1 ( N + 1 ) [ g ( y j ( N + 1 ) ) g y j ( N + 1 ) ] y j + p ( 0 ) α φ 1 ( 0 ) [ g ( y j ( 0 ) ) g y j ( 0 ) ] y j } lim j { f ( μ 0 λ 1 + μ 0 ε f ) k = 1 N a ( k ) y j ( k ) y j φ 1 ( k ) μ 0 a d ε y j k = 1 N φ 1 ( k ) } = f ( μ 0 λ 1 + μ 0 ϵ f ) k = 1 N a ( k ) w 0 ( k ) φ 1 ( k ) > 0 .

The proof of Lemma 3.2 is complete. □

Lemma 3.3 Let λ n = μ 1 f 1 n and λ n + = λ 1 f + 1 n , where n > f μ 1 is an integer. Assume that (H1)-(H3) hold. Then there exists a constant r n > 0 satisfying r n as n such that for any n large enough,
deg ( Φ ( λ n , ) , B r n , 0 ) = 1 ,
(3.15)
deg ( Φ ( λ n + , ) , B r n , 0 ) = 0 .
(3.16)
Proof First we show assertion (3.15). From Lemma 3.1, there exists r n > 0 such that r n as n satisfying
Φ χ ( λ , y ) 0 , λ [ 0 , λ n ] , y X : y = r n .
Since χ ( 0 ) = 0 and χ ( λ n ) = 1 for n large enough from (3.8), by the homotopy invariance and normalization of the topology degree, it follows that for any n large enough,
deg ( Φ ( λ n , ) , B r n , 0 ) = deg ( Φ χ ( λ n , ) , B r n , 0 ) = deg ( Φ χ ( 0 , ) , B r n , 0 ) = deg ( I X , B r n , 0 ) = 1 .
Next, we show assertion (3.16). We may derive from Lemma 3.2 that
Φ ( λ n + , y ) t φ 1 , t [ 0 , 1 ] , y X : y r n .
So for any n large enough, by the homotopy invariance, it follows that
deg ( Φ ( λ n + , ) , B r n , 0 ) = deg ( Φ ( λ n + , ) φ 1 , B r n , 0 ) = 0 .

 □

Proof of Theorem 1.1 For any fixed n N with μ 1 f 1 n > 0 , set α n = μ 1 f 1 n , β n = λ 1 f + 1 n . It is easy to verify that for any fixed n large enough, there exists r n such that r n as n satisfying that for any r r n , it follows from Lemmas 3.1-3.3 that all conditions of Lemma 2.1 are satisfied. So there exists a closed connected component C n of solutions (3.1) such that C n is unbounded in [ α n , β n ] × X and either
  1. (i)

    C n is unbounded in λ direction, or

     
  2. (ii)

    there exists an interval [ c , d ] such that ( α n , β n ) ( c , d ) = and C n bifurcates from infinity in [ c , d ] × X .

     

By Lemma 3.1, the case (ii) cannot occur. Thus C n is unbounded bifurcated from [ α n , β n ] × { } in R × X . Furthermore, set σ = 1 n for n large enough, we have from Lemma 3.1 that for any closed interval I [ μ 1 f σ , λ 1 f + σ ] [ μ 1 f , λ 1 f ] , if y { y X | ( λ , y )  is a solution of (3.1) , λ I } , then y in X is impossible. So C 1 / σ must be bifurcated from [ μ 1 f σ , λ 1 f + σ ] × { } . □

Next, we are devoted to the proof of Theorem 1.2, which characterized the bifurcation components of (1.1).

Proof of Theorem 1.2 Under condition (1.7), assume to the contrary that there exists a positive solution y j of (1.1) with λ = λ j λ 1 f , and
y j , j .
If w j = y j y j , then the same argument as in the proof of Lemma 2.3 shows the existence of a positive function w 0 X such that a subsequence of { w j } , still denoted by w j , tends to w 0 in X. It follows that for any j large enough, we have
w j ( k ) > min k I ˆ w 0 ( k ) 2 on  I ,
(3.17)
which implies that
min I ˆ u j as  j .
(3.18)
Set
h 1 = lim inf u h 1 ( u ) ( , ] , γ 2 = lim sup u γ 2 ( u ) [ , ) .
Note that we consider only the cases h 1 ( , ) and γ 2 ( , ) . Either the case h 1 = or the case γ 2 = can be dealt with in a similar way with a minor modification. It follows from (3.18) that, for any ϵ > 0 , there exists j 1 1 such that for any j j 1 ,
h 1 ϵ < h 1 ( u j ( k ) ) on  I ˆ , γ 2 ( u j ( k ) ) < γ 2 + ϵ on  I ˆ .
Thus, for any j j 1 ,
( λ 1 λ j f ) k = 1 N a ( k ) y j ( k ) φ 1 ( k ) λ j k = 1 N a ( k ) h 1 ( u j ( k ) ) φ 1 ( k ) α p ( 0 ) φ 1 ( 0 ) [ g ( u j ( 0 ) ) g u j ( 0 ) ] β p ( N ) φ 1 ( N + 1 ) [ g ( u j ( N + 1 ) ) g u j ( N + 1 ) ] λ j k = 1 N a ( k ) h 1 ( u j ( k ) ) φ 1 ( k ) α p ( 0 ) φ 1 ( 0 ) γ 2 ( u j ( 0 ) ) β p ( N ) φ 1 ( N + 1 ) γ 2 ( u j ( N + 1 ) ) λ 1 ( h 1 ϵ ) f k = 1 N a ( k ) φ 1 ( k ) ( γ 2 + ϵ ) [ α p ( 0 ) φ 1 ( 0 ) + β p ( N ) φ 1 ( N + 1 ) ] .
On the other hand, we have
k = 1 N a ( k ) φ 1 ( k ) = 1 λ 1 k = 1 N q ( k ) φ 1 ( k ) + g [ α p ( 0 ) φ 1 ( 0 ) + β p ( N ) φ 1 ( N + 1 ) ] λ 1 .
These two assertions combined, we obtain that for any j j 1 ,
( λ 1 λ j f ) k = 1 N a ( k ) y j ( k ) φ 1 ( k ) > h 1 ϵ f k = 1 N q ( k ) φ 1 ( k ) + ( λ 1 ( h 1 ϵ ) g f λ 1 ( γ 2 + ϵ ) ) [ α p ( 0 ) φ 1 ( 0 ) + β p ( N ) φ 1 ( N + 1 ) ] > 0 .
On the right-hand side, we see from (1.7) that
lim ϵ 0 + ( λ 1 ( h 1 ϵ ) g f λ 1 ( γ 2 + ϵ ) ) = h 1 g f γ 2 > 0 .
This means that for any j large enough,
( λ 1 λ j f ) k = 1 N a ( k ) y j ( k ) φ 1 ( k ) > 0 ,

which contradicts the assumption λ j λ 1 f . Case (1.7) has been proved. Case (1.8) can be also verified by the same arguments, and the proof of Theorem 1.2 is complete. □

Declarations

Acknowledgements

The authors are very grateful to the anonymous referees for their valuable suggestions. This work was supported by the NSFC (No. 11361054, No. 11201378), SRFDP (No. 20126203110004), Gansu provincial National Science Foundation of China (No. 1208RJZA258).

Authors’ Affiliations

(1)
Department of Mathematics, Northwest Normal University

References

  1. Lasota A: A discrete boundary value problem. Ann. Pol. Math. 1968, 20: 183–190.MathSciNetGoogle Scholar
  2. Kelley WG, Peterson AC: Difference Equations: An Introduction with Applications. 2nd edition. Harcourt/Academic Press, San Diego, CA; 2001.Google Scholar
  3. Graef JR, Qian C, Yang B: Positive solutions to boundary value problems for nonlinear difference equations. J. Differ. Equ. Appl. 2000, 6: 91–104.MathSciNetView ArticleGoogle Scholar
  4. Agarwal RP, O’Regan D: Fixed-point approach for nonlinear discrete boundary value problems. Comput. Math. Appl. 1998, 36(10–12):115–121.MathSciNetView ArticleGoogle Scholar
  5. Atici FM: Existence of positive solutions of nonlinear discrete Sturm-Liouville problems. Boundary value problems and related topics. Math. Comput. Model. 2000, 32(5–6):599–607.MathSciNetView ArticleGoogle Scholar
  6. Cabada A, Otero-Espinar V: Fixed sign solutions of second-order difference equations with Neumann boundary conditions. Comput. Math. Appl. 2003, 45(6–9):1125–1136.MathSciNetView ArticleGoogle Scholar
  7. Anderson DR, Avery RI, Henderson J, Liu X, Lyons JW: Existence of a positive solution for a right focal discrete boundary value problem. J. Differ. Equ. Appl. 2011, 17(11):1635–1642.MathSciNetView ArticleGoogle Scholar
  8. Rodríguez J: On the solvability of nonlinear discrete boundary value problems. J. Differ. Equ. Appl. 2003, 9(9):863–867.View ArticleGoogle Scholar
  9. Rodríguez J: Nonlinear discrete Sturm-Liouville problems. J. Math. Anal. Appl. 2005, 308(1):380–391.MathSciNetView ArticleGoogle Scholar
  10. Ma R: Bifurcation from infinity and multiple solutions for some discrete Sturm-Liouville problems. Comput. Math. Appl. 2007, 54: 535–543.MathSciNetView ArticleGoogle Scholar
  11. Henderson J, Ntouyas S, Purnaras I: Positive solutions for systems of nonlinear discrete boundary value problems. J. Differ. Equ. Appl. 2009, 15: 895–912.MathSciNetView ArticleGoogle Scholar
  12. Rabinowitz PH: Some global results for nonlinear eigenvalue problems. J. Funct. Anal. 1971, 7: 487–513.MathSciNetView ArticleGoogle Scholar
  13. Schmitt K: Positive solutions of semilinear elliptic boundary value problems. In Topological Methods in Differential Equations and Inclusions. Edited by: Garnas A, Frigon M. Kluwer Academic, Dordrecht; 1995:447–500.View ArticleGoogle Scholar
  14. Peitgen H, Schmitt K: Global analysis of two-parameter elliptic eigenvalue problems. Trans. Am. Math. Soc. 1984, 283: 57–95.MathSciNetView ArticleGoogle Scholar
  15. Deimling K: Nonlinear Functional Analysis. Springer, Berlin; 1988.Google Scholar

Copyright

© Ma and Lu; licensee Springer. 2014

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.