Existence result for (k,n-k) conjugate boundary-values

problems with integral boundary conditions at resonance

with dim ker L = 2.
Abstract
We shall study the existence of solution for (k,n-1) conjugate
boundary-values problems with integral boundary conditions at
resonance with dim ker L = 2 in this paper. The boundary value
problems shown as follows:

(−1)𝑛−𝑘 𝜑(𝑛) (𝑋) = 𝑓(𝑋, 𝜑(𝑋), 𝜑′ (𝑋), … … 𝜑𝑛−1 (𝑋)), 𝑋 ∈ [0,1],

𝜑(𝑖) (0) = 𝜑(𝑗) (1) = 0, 1 ≤ 𝑖 ≤ 𝑘 − 1, 1 ≤ 𝑗 ≤ 𝑛 − 𝑘 − 1,

1 1
𝜑(0) = ∫0 𝜑(𝑋)𝑑𝐴(𝑋) , 𝜑(1) = ∫0 𝜑(𝑋)𝑑𝐵(𝑋).

We can obtain that this boundary-value problem has at least one

solution under the conditions we provide through Mawhin’s
continuation theorem, and an example also provided for our new
result.

KEYWORDS: boundary value problem; resonance; Fredholm operator;

Mawhin’s continuation theorem.

1 Introduction

Conjugate boundary-value problems at non-resonance have

aroused considerable attention in recent years, and there is also much
research on boundary-value problems at resonance. However, there
are very few papers involving (k, n – k) conjugate boundary-value
problems at resonance, especially with dim ker L = 2. For example, Jiang

investigated the following boundary-value problem at resonance with

dim ker L = 2:

(−1)𝑛−𝑘 𝑦 (𝑛) (𝑡) = 𝑓 (𝑡, 𝑦(𝑡), 𝑦 ′ (𝑡), … … 𝑦 (𝑛−1) (𝑡)) + 𝜀 (𝑡), 𝑡[0,1],

𝑦 (𝑖) (0) = 𝑦 (𝑗) (1) = 0, 0 ≤ 𝑖 ≤ 𝑘 − 1, 0 ≤ 𝑗 ≤ 𝑛 − 𝑘 − 3,

𝑚 𝑙

𝑦 (𝑛−1) (1) = ∑ 𝛼𝑖 𝑦 (𝑛−1) (𝜉𝑖 ), 𝑦 (𝑛−2) (1) = ∑ 𝛽𝑗 𝑦 (𝑛−2) (𝜂𝑗 ),

𝑖=1 𝑖=1

1 ≤ 𝑘 ≤ 𝑛 − 3, 0 < 𝜉1 < 𝜉2 < ⋯ < 𝜉𝑚 < 1, 0 < 𝜂1 < 𝜂2 <. … . . < 𝜂𝑙 < 1.

We shall study the following (k, n – k) conjugate boundary-value

problem in the situation of resonance with dim ker L = 2:

(−1)𝑛−𝑘 𝜑(𝑛) (𝑥 ) = 𝑓(𝑥, 𝜑(𝑥 ), 𝜑′ (𝑥 ), … … 𝜑𝑛−1 (𝑥 )), 𝑥 ∈ [0,1],

𝜑(𝑖) (0) = 𝜑(𝑗) (1) = 0, 1 ≤ 𝑖 ≤ 𝑘 − 1, 1 ≤ 𝑗 ≤ 𝑛 − 𝑘 − 1,

1 1
𝜑(0) = ∫0 𝜑(𝑥 )𝑑𝐴(𝑥) , 𝜑(1) = ∫0 𝜑(𝑥 )𝑑𝐵(𝑥 ).

where 1 ≤ k ≤ n-1, n ≥ 2, A(x), B(x) are left continuous at x = 1, right

1 1
continuous on [0,1); ∫0 𝑢(𝑥 )𝑑𝐴(𝑥 ) 𝑎𝑛𝑑 ∫0 𝑢(𝑥 )𝑑𝐵(𝑥 ) denote the
Riemann-Stieltjes integrals of u with respect to A and B, respectively.

However, there are great differences between this article and the
above results, the boundary conditions we study are 𝜑(0) =
 1 1
∫0 𝑢(𝑥 )𝑑𝐴(𝑥 ) 𝑎𝑛𝑑 𝜑(0) = ∫0 𝑢(𝑥 )𝑑𝐵(𝑥 ). As is well known, it is an
original case to study conjugate boundary-value problems with integral
boundary conditions in the situation of resonance.

The organization of this paper is as follows. In Section 2, we provide

a definition and a theorem which will be used to prove the main results.
In Section 3, we will give some lemmas and prove the solvability of

problem (1)-(3).

Theorem 2.1 (Mawhin’s continuation theorem) L : dom L∈→Y is a

Fredholm operator of index zero, and N is L-compact on Ω.The equation
L𝜑 = N𝜑 has atleast one solution in dom L ∩ Ω if the following
conditions are satisfied:

(1)𝐿𝜑 ≠ 𝜆𝑁𝜑 𝑓𝑜𝑟 𝑒𝑣𝑒𝑟𝑦 (𝜑, 𝜆) ∈ [(𝑑𝑜𝑚 𝐿 ker 𝐿) ∩ 𝜕Ω] ∗ (0,1);

(2)𝑁𝜑 ∉ 𝐼𝑚 𝐿 𝑓𝑜𝑟 𝑒𝑣𝑒𝑟𝑦 𝜑 ∈ ker 𝐿 ∩ 𝜕Ω;
(3)deg(𝑄𝑁|ker 𝐿 , 𝛺 ∩ ker 𝐿, 0) ≠ 0, 𝑤ℎ𝑒𝑟𝑒 𝑄: 𝑌 →
𝑌 𝑖𝑠 𝑎 𝑝𝑟𝑜𝑗𝑒𝑐𝑡𝑖𝑜𝑛 𝑠𝑢𝑐ℎ 𝑡ℎ𝑎𝑡 𝐼𝑚 𝐿 = ker 𝑄.

Let X = 𝐶 𝑛−1 [1,0] with norm ||𝑢|| =

max{||𝑢||∞ , ||𝑢′ ||∞ , … . , ||𝑢(𝑛−1) ||∞} ,in which ||𝑢||∞ =
1
𝑚𝑎𝑥𝑥∈[0,1] |𝑈(𝑥 )|, 𝑎𝑛𝑑 𝐿 = 𝐿1 [0,1]𝑤𝑖𝑡ℎ 𝑛𝑜𝑟𝑚 ||𝑥 ||1 = ∫0 |𝑥 (𝑡)|𝑑𝑡.
We define an operator L as follows:

(L𝜑)(x) = (−1)𝑛−𝑘 𝜑𝑛 (𝑥)

With
dom L
𝜑 ∈ 𝑋: 𝜑𝑖 (0) = 𝜑𝑖 (1) = 0,1 ≤ 𝑖 ≤ 𝑘 − 1,1 ≤ 𝑗 ≤ 𝑛 − 𝑘 − 1,
{ 1 1 }
𝜑(0) = ∫0 𝜑(𝑥 )𝑑𝐴(𝑥 ), 𝜑(1) = ∫0 𝜑(𝑥 ) 𝑑𝐵(𝑥)}

An operator N:X →Y is defined as

(𝑁𝜑)(𝑥 ) = 𝑓(𝑥, 𝜑(𝑥 ), 𝜑′ (𝑥 ), … … 𝜑 (𝑛−1) (𝑥 ))

So problem (1)-(3) become Lφ = Nφ.

3 Main Results

Assume that the following conditions hold in this paper :

1 1
(H1) ∫0 𝛷1 (𝑥 )𝑑𝐴(𝑥 ) = 1, ∫0 𝛷2 (𝑥 )𝑑𝐵(𝑥 ) = 1,

1 1
∫0 𝛷1 (𝑥 )𝑑𝐵(𝑥 ) = 1, ∫0 𝛷2 (𝑥 )𝑑𝐴(𝑥 ) = 1,

Where

(𝑛 − 1)! 1
𝛷1 (𝑥 ) = ∫ 𝑡 𝑘−1 (1 − 𝑡)𝑛−𝑘−1 𝑑𝑡,
(𝑘 − 1)! (𝑛 − 1 − 𝑘)! 𝑥

(𝑛 − 1)! 𝑥
𝛷2 (𝑥 ) = ∫ 𝑡 𝑘−1 (1 − 𝑡)𝑛−𝑘−1 𝑑𝑡.
(𝑘 − 1)! (𝑛 − 1 − 𝑘)! 0
𝑒1 𝑒2
(H2) 𝑒 = |𝑒 𝑒4 | ≠ 0,
3

Where
1 1 1 1
𝑒1 = ∫0 ∫0 𝑘(𝑥, 𝑦)𝛷1 (𝑥 )𝑑𝑦𝑑𝐴(𝑥 ), 𝑒2 = ∫0 ∫0 𝑘(𝑥, 𝑦)𝛷1 (𝑥 )𝑑𝑦𝑑𝐵 (𝑥 ),
 1 1 1 1
𝑒3 = ∫0 ∫0 𝑘(𝑥, 𝑦)𝛷2 (𝑥 )𝑑𝑦𝑑𝐴(𝑥 ), 𝑒4 = ∫0 ∫0 𝑘(𝑥, 𝑦)𝛷2 (𝑥 )𝑑𝑦𝑑𝐵 (𝑥 ).
1 𝑥(1−𝑦) 𝑘−1
∫
(𝑘−1)!(𝑛−𝑘−1)! 0
𝑡 (𝑡 + 𝑦 − 𝑥)𝑛−𝑘−1 𝑑𝑡, 0 ≤ 𝑥 ≤ 𝑦 ≤ 1;
K(x,y)={ 1 𝑦(1−𝑥) 𝑛−𝑘−1
∫
(𝑘−1)!(𝑛−𝑘−1)! 0
𝑡 (𝑡 + 𝑦 − 𝑥)𝑘−1 𝑑𝑡, 0 ≤ 𝑦 ≤ 𝑥 ≤ 1.

(H3) :[0,1] X 𝑅 𝑛 → 𝑅 satisfies carathaodory conditions.

(H4) : There Exist function 𝑟(𝑥 ), 𝑞𝑖 (𝑥 ) ∈ 𝐿1 [0,1]𝑤𝑖𝑡ℎ ∑𝑛𝑖=1||𝑞𝑖 ||1 < 1

Such that |𝑓 (𝑥, 𝑞1 , 𝑞2 , … 𝑞𝑛 )| ≤ ∑𝑛𝑖=1 𝑞1 (𝑥 )|𝜑𝑖 | + 𝑟(𝑥 ),

Where 𝑥 ∈ [0,1], 𝜑𝑖 ∈ 𝑅.

(H5) There exist a constant M>0 such that if |𝜑(𝑥)| + |𝜑 (𝑛−1) (𝑥)| > 𝑀
for all x∈[1,0],

Then
1 1
∫ ∫ 𝑘 (𝑥, 𝑦)𝑓 (𝑦, 𝜑(𝑦), 𝜑′ (𝑦), … , 𝜑(𝑛−1) (𝑦)) 𝑑𝑦𝑑𝐴(𝑥) ≠ 0,
0 0

Or,
1 1
∫ ∫ 𝑘(𝑥, 𝑦)𝑓 (𝑦, 𝜑(𝑦), 𝜑′ (𝑦), … , 𝜑 (𝑛−1) (𝑦)) 𝑑𝑦𝑑𝐵(𝑥) ≠ 0,
0 0

(H6) There are constant a,b>0 such that one of the following two
conditions hold:
1 1
𝑐1 ∫0 ∫0 𝑘(𝑥, 𝑦)𝑁(𝑐1 𝛷1 (𝑦) + 𝑐2 𝛷2 (𝑦))𝑑𝑦𝑑𝐴(𝑥 ) < 0, (3)
1 1
𝑐2 ∫0 ∫0 𝑘(𝑥, 𝑦)𝑁(𝑐1 𝛷1 (𝑦) + 𝑐2 𝛷2 (𝑦))𝑑𝑦𝑑𝐵 (𝑥 ) < 0, (4)

If |𝑐1 |>a and |𝑐2 |>b.

 1 1
𝑐1 ∫0 ∫0 𝑘(𝑥, 𝑦)𝑁(𝑐1 𝛷1 (𝑦) + 𝑐2 𝛷2 (𝑦))𝑑𝑦𝑑𝐴(𝑥 ) > 0, (5)
1 1
𝑐2 ∫0 ∫0 𝑘(𝑥, 𝑦)𝑁(𝑐1 𝛷1 (𝑦) + 𝑐2 𝛷2 (𝑦))𝑑𝑦𝑑𝐵 (𝑥 ) > 0, (6)

If |𝑐1 |>a and |𝑐2 |>b.

Then we can present the following theorem.

Theorem 3.1 : Suppose (H1) and (H6) are satisfied,then there must be
at least one solution of problem (1)-(3) in X.

To prove this theorem, we need the following lemmas.

Lemmas 3.1 : Assume that (H1) and (H2) hold, then L:dom Lϵ X→Y is a
Fredholm operator with index zero. And a linear continuous projector

Q : Y→Y can be defined by

(𝑄𝑢)(𝑥 ) = (𝑄1 𝑢)𝛷1 (𝑥 ) + (𝑄2 𝑢)𝛷2 (𝑥 ),

Where,
1 1
(𝑄1 𝑢) = (𝑒4 𝑇1 𝑢 − 𝑒3 𝑇2 𝑢), (𝑄1 𝑢) = (𝑒4 𝑇1 𝑢 − 𝑒3 𝑇2 𝑢),
𝑒 𝑒
1 1
𝑇1 𝑢 = ∫0 ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 ),
1 1
𝑇1 𝑢 = ∫0 ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 ),

Furthermore, define a linear operator 𝑘𝑝 ∶ 𝐼𝑚 𝐿 → 𝑑𝑜𝑚 𝐿 ∩ ker 𝑝 as

follows:
1
(𝑘𝑝 𝑢)(𝑥 ) = ∫0 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦 + 𝛷1 (𝑥 )𝑇1 𝑢 + 𝛷2 (𝑥 )𝑇2 𝑢

Such that 𝑘𝑝 = (𝐿|𝑑𝑜𝑚 𝐿∩ker 𝑝 )−1

Proof: it follows from (H1) that

(−1)𝑛−1 𝛷1(𝑛) (𝑥 ) = 0, (−1)𝑛−1 𝛷2(𝑛) (𝑥 ) = 0, 𝑥 ∈ [0,1],

(𝑖) (𝑗) (𝑖) (𝑗)
𝛷1 (0) = 𝛷1 (1) = 0, 𝛷2 (0) = 𝛷2 (1) = 0,

1≤i≤k-1, 1≤i≤n-k-1,

𝛷1 (0) = 1, 𝛷1 (1) = 0, 𝛷2 (0) = 0, 𝛷2 (1) = 1.

It is obvious that
1 1
𝛷1 (0) = ∫0 𝛷1 (𝑥 )𝑑𝐴(𝑥 ), 𝛷2 (0) = ∫0 𝛷2 (𝑥 )𝑑𝐵(𝑥 ),
1 1
𝛷1 (1) = ∫0 𝛷1 (𝑥 )𝑑𝐵(𝑥 ) = 0 𝛷2 (0) = ∫0 𝛷2 (𝑥 )𝑑𝐴(𝑥 ) = 0,

Thus we have

ker 𝐿 = {𝑐1 𝛷1 (𝑥 ) + 𝑐2 𝛷2 (𝑥 ), 𝑐1 , 𝑐2 ∈ 𝑅 }.
Moreover, we can obtain there that
1 1
𝐼𝑚 𝐿 = {𝑢 ∈ 𝑌: ∫ ∫ 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 )
0 0
1 1
= ∫ ∫ 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐵 (𝑥 ) = 0}
0 0

On the one hand, suppose u∈Im L,then there exist φ ∈ dom L such that

U = L φ∈Y.

Then we have
1
𝜑(𝑥 ) = ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦 + 𝜑(0) 𝛷1 (𝑥 ) + 𝜑(1)𝛷2 (𝑥 ).

Furthermore, for φ ∈ dom L,then

 1
𝜑(0) = ∫0 𝜑(𝑥 )𝑑𝐴(𝑥)
1 1
=∫0 [∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦 + 𝜑(0) 𝛷1 (𝑥 ) + 𝜑(1)𝛷2 (𝑥 )] 𝑑𝐴(𝑥)
1 1
=∫0 ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 ) +
1 1
𝜑(0) ∫0 𝛷1 (𝑥 )𝑑𝐴(𝑥 ) + ∫0 𝛷2 (𝑥 )𝑑𝐵(𝑥 ),

Using this together with (H1), we can get

1 1
𝜑(0) = ∫0 ∫0 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 ) + 𝜑(0),
1 1
It means ∫0 ∫0 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 ) = 0 and
1
𝜑(1) = ∫0 𝜑(𝑥 )𝑑𝐵 (𝑥 )
1 1
=∫0 [∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦 + 𝜑(0) 𝛷1 (𝑥 ) + 𝜑(1)𝛷2 (𝑥 )] 𝑑𝐵(𝑥)
1 1
=∫0 ∫0 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐵 (𝑥 ) +
1 1
𝜑(0) ∫0 𝛷1 (𝑥 )𝑑𝐵(𝑥 ) + ∫0 𝛷2 (𝑥 )𝑑𝐵(𝑥 ).

So we obtain that
1 1
∫ ∫ 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 ) = 0
0 0

Thus
1 1
𝐼𝑚 𝐿 = {𝑢: ∫ ∫ 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 )
0 0
1 1
= ∫ ∫ 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐵 (𝑥 ) = 0}
0 0

On the other hand, if u∈Y satisfies

 1 1 1 1
∫0 ∫0 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 ) = ∫0 ∫0 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐵 (𝑥 ) = 0

We let
1
𝜑(𝑥 ) = ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦 + 𝛷1 (𝑥 ) + 𝛷2 (𝑥 ).

Then we conclude that

(𝐿𝜑)(𝑥 ) = (−1)𝑛−𝑘 𝜑(𝑛) (𝑥 ) = 𝑢(𝑥 ),

𝜑(𝑖) (0) = 𝜑(𝑗) (1) = 0, 1≤i≤k-1, 1≤i≤n-k-1,

And
1
𝜑(0) = ∫0 𝑘(0, 𝑦)𝑢(𝑦)𝑑𝑦 + 𝛷1 (0) + 𝛷2 (0) = 1.
1
𝜑(1) = ∫0 𝑘(1, 𝑦)𝑢(𝑦)𝑑𝑦 + 𝛷1 (1) + 𝛷2 (1) = 1.

Besides
1 1 1 1
∫0 𝜑(𝑥 )𝑑𝐴(𝑥 ) = ∫0 ∫0 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 ) + ∫0 𝛷1 (𝑥 )𝑑𝐴(𝑥 ) +
1
∫0 𝛷2 (𝑥 )𝑑𝐴(𝑥 ) = 1,

And
1 1 1 1
∫0 𝜑(𝑥 )𝑑𝐵 (𝑥 ) = ∫0 ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐵 (𝑥 ) + ∫0 𝛷1 (𝑥 )𝑑𝐵(𝑥 ) +
1
∫0 𝛷2 (𝑥 )𝑑𝐵 (𝑥 ) = 1.

Therefore
1 1
𝜑(0) = ∫0 𝜑(𝑥 )𝑑𝐴(𝑥 ), 𝜑(1) = ∫0 𝜑(𝑥 )𝑑𝐵(𝑥 ),

That is

That is φ∈ dom L, hence u∈ Im L.In conclusion,

 1 1
𝐼𝑚 𝐿 = {𝑢 ∈ 𝑌: ∫ ∫ 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 )
0 0
1 1
= ∫ ∫ 𝑘 (𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐵 (𝑥 ) = 0}
0 0

We define a linear operator P : X → X as

(𝑃𝜑)(𝑥 ) = 𝛷1 (𝑥 )𝜑(0) + 𝛷2 (𝑥 )𝜑(1),

Then

(𝑃2 𝜑)(𝑥 ) = (𝑃(𝑃𝜑))(𝑥)

= 𝛷1 (𝑥 )[(𝑃𝜑)(0)] + 𝛷2 (𝑥 )[(𝑃𝜑)(1)]

=𝛷1 (𝑥 )[𝛷1 (0)𝜑(0) + 𝛷2 (0)𝜑(1)] + 𝛷2 (𝑥 )[𝛷1 (0)𝜑(0) +

𝛷2 (0)𝜑(1)]
=[𝛷1 (𝑥 )𝜑(0) + 𝛷2 (𝑥 )𝜑(1)]

It is obvious that 𝑃2 𝜑 = 𝑃𝜑 and lm P = ker L. For any 𝜑 = (𝜑 − 𝑃𝜑) +

𝑃𝜑 we have X = ker P+ker L. it is easy to obtain that ker L∩ ker 𝑃 = {0},

Which implies

𝑋 = 𝑘𝑒𝑟𝑃 ⊕ 𝑘𝑒𝑟 𝐿.

Next, an operator Q : Y →Y is defined as follows:

(𝑄𝑢)(𝑥 ) = (𝑄1 𝑢)𝛷1 (𝑥 ) + (𝑄2 𝑢)𝛷2 (𝑥 )

Where
1 1
𝑄1 𝑢 = (𝑒4 𝑇1 𝑢 − 𝑒3 𝑇2 𝑢), 𝑄2 𝑢 = (−𝑒2 𝑇1 𝑢 − 𝑒1 𝑇2 𝑢),
𝑒 𝑒
 1 1
𝑇1 𝑢 = ∫0 ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐴(𝑥 ),
1 1
𝑇1 𝑢 = ∫0 ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦𝑑𝐵 (𝑥 ).

Obviously, 𝑒1 = 𝑇1 (𝛷1 (𝑥 )), 𝑒2 = 𝑇2 (𝛷1 (𝑥 )), 𝑒3 = 𝑇1 (𝛷2 (𝑥 )), 𝑒4 =

𝑇2 (𝛷2 (𝑥 )), Nothing that

(𝑄 2 𝑢)(𝑥 ) = (𝑄1 (𝑄𝑢))(𝑥 )𝛷1 (𝑥 ) + (𝑄2 (𝑄𝑢))(𝑥 )𝛷2 (𝑥 )

=[𝑄1 ((𝑄1 (𝑄𝑢))𝛷1 (𝑥 ) + (𝑄2 (𝑄𝑢))𝛷2 (𝑥 ))]𝛷1 (𝑥 )

+[𝑄2 ((𝑄1 (𝑄𝑢))𝛷1 (𝑥 ) + (𝑄2 (𝑄𝑢))𝛷2 (𝑥 ))] 𝛷2 (𝑥 ),

Since
1
𝑄1 ((𝑄1 (𝑢))𝛷1 (𝑥)) = (𝑒4 𝑇1 (𝛷1 (𝑥 )) − 𝑒3 𝑇2 (𝛷1 (𝑥))𝑄`1 𝑢
𝑒
1
= (𝑒4 𝑒1 − 𝑒3 𝑒2 )𝑄1 𝑢 = 𝑄1 𝑢
𝑒
1
𝑄1 ((𝑄2 (𝑢))𝛷2 (𝑥 ) = (𝑒4 𝑇1 (𝛷2 (𝑥 )) − 𝑒3 𝑇2 (𝛷2 (𝑥))𝑄`2 𝑢
𝑒
1
= (𝑒4 𝑒3 − 𝑒3 𝑒4 )𝑄2 𝑢 = 0
𝑒
1
𝑄2 ((𝑄2 (𝑢))𝛷2 (𝑥 ) = (−𝑒2 𝑇1 (𝛷1 (𝑥 )) + 𝑒1 𝑇2 (𝛷1 (𝑥))𝑄1 𝑢
𝑒
1
= (−𝑒2 𝑒1 + 𝑒1 𝑒2 )𝑄1 𝑢 = 0
𝑒
1
𝑄2 ((𝑄2 (𝑢))𝛷2 (𝑥 ) = (−𝑒2 𝑇1 (𝛷1 (𝑥 )) + 𝑒1 𝑇2 (𝛷1 (𝑥))𝑄2 𝑢
𝑒
1
= (−𝑒2 𝑒3 + 𝑒1 𝑒4 )𝑄2 𝑢 = 𝑄2 𝑢,
𝑒

So,
 (𝑄 2 𝑢)(𝑥 ) = (𝑄1 𝑢)𝛷1 (𝑥 ) + (𝑄2 𝑢)𝛷2 (𝑥 ) = (𝑄𝑢)(𝑥 ),

And since 𝑢 ∈ ker 𝑄, we have 𝑒4 𝑇1 𝑢 − 𝑒3 𝑇2 𝑢 = 0, −𝑒2 𝑇1 𝑢 − 𝑒1 𝑇2 𝑢 =

0, it follows from (H2) that 𝑇1 𝑢 = 𝑇2 𝑢 = 0, so 𝑢 ∈ 𝐼𝑚 𝐿, and obviously,

𝐼𝑚 𝐿 ∈ ker 𝑄. So ker 𝑄 = 𝐼𝑚 𝐿.For any 𝑢 ∈ 𝑌, because 𝑢 =

(𝑢 − 𝑄𝑢) + 𝑄𝑢, we have Y = Im L + Im Q. moreover, together with
𝑄 2 𝑢 = 𝑄𝑢, we can get 𝐼𝑚 𝑄 ∩ 𝐼𝑚 𝑙 = {0}. Above all, Y= Im L⊕Im Q.

To sum up , we can get that Im L is a closed subspace of Y; dim ker L =

codim Im L <+⧝; that is ,L is a Fredholm operator operator of index
zero.

We now define an operator 𝐾𝑝 : 𝑌 → 𝑌 as follows:

1
(𝐾𝑝 𝑢)(𝑥 ) = ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦 + 𝛷1 (𝑥 )𝑇1 𝑢 + 𝛷2 (𝑥 )𝑇2 𝑢.

For any 𝑢 ∈ 𝐼𝑚 𝐿, we have 𝑇1 𝑢 = 0, 𝑇2 𝑢 = 0, Consequently,

1
(𝐾𝑝 𝑢)(𝑥 ) = ∫0 𝑘(𝑥, 𝑦)𝑢(𝑦)𝑑𝑦, (𝐾𝑝 𝑢)(0) = 0, (𝐾𝑝 𝑢)(1) = 0.

So,
1
(𝐾𝑝 𝑢)(𝑥 ) ∈ ker 𝑃, (𝐾𝑝 𝑢)(0) = ∫0 (𝐾𝑝 𝑢)(𝑥 )𝑑𝐴(𝑥 ),
1
(𝐾𝑝 𝑢)(1) = ∫0 (𝐾𝑝 𝑢)(𝑥 )𝑑𝐵 (𝑥 ),

In addition, it is easy to know that

(𝑖) (𝑗)
(𝐾𝑝 𝑢) (0) = ,1 ≤ 𝑖 ≤ 𝑘 − 1; (𝐾𝑝 𝑢) (0) = ,1 ≤ 𝑗 ≤ 𝑛 − 𝑘 − 1;

Then (𝐾𝑝 𝑢)(𝑥 ) ∈ 𝑑𝑜𝑚 𝑙. 𝑡ℎ𝑒𝑟𝑒𝑓𝑜𝑟𝑒

(𝐾𝑝 𝑢)(𝑥 ) ∈ 𝑑𝑜𝑚 𝑙 ∩ ker 𝑃 , 𝑢 ∈ 𝐼𝑚 𝐿.

Next we will prove that 𝑘𝑝 is the inverse of 𝐿|𝑑𝑜𝑚 𝐿∩ker 𝑃. It is clear that

(𝐿𝐾𝑃 𝑢)(𝑥 ) = 𝑢(𝑥 ), 𝑢 ∈ 𝐼𝑚 𝐿.

For each 𝑣 ∈ 𝑑𝑜𝑚 𝐿 ∩ ker 𝑃,we have
1
(𝐾𝑝 𝐿𝑣)(𝑥 ) = ∫0 𝑘(𝑥, 𝑦)(−1)𝑛−𝑘 𝑣 (𝑛) (𝑦)𝑑𝑦 +
1 1
+𝛷1 (𝑥 ) ∫0 ∫0 𝑘 (𝑥, 𝑦)(−1)𝑛−𝑘 𝑣 (𝑛) (𝑦)𝑑𝑦𝑑𝐴(𝑥 ) +
1 1
+ 𝛷2 (𝑥 ) ∫0 ∫0 𝑘(𝑥, 𝑦)(−1)𝑛−𝑘 𝑣 (𝑛) (𝑦)𝑑𝑦𝑑𝐵 (𝑥 )
1
=𝑣(𝑥 ) − 𝑣(0)𝛷1 (𝑥 ) − 𝑣 (1)𝛷2 (𝑥 ) + 𝛷1 (𝑥) ∫0 (𝑣 (𝑥 ) − 𝑣 (0)𝛷1 (𝑥)
1
−𝑣 (1)𝛷2 (𝑥))𝑑𝐴(𝑥) + 𝛷2 (𝑥) ∫0 (𝑣 (𝑥 ) − 𝑣 (0))𝛷1 (𝑥 ) −
(𝑣 (1)𝛷2 (𝑥 ))𝑑𝐵(𝑥)
1 1
= 𝑣 (𝑥 ) + 𝛷1 (𝑥 ) ∫0 𝑣 (𝑥 )𝑑𝐴(𝑥 ) + 𝛷2 (𝑥 ) ∫0 𝑣(𝑥 )𝑑𝐵(𝑥 )

= 𝑣(𝑥 ) + 𝑣(0)𝛷1 (𝑥 ) + 𝑣(1)𝛷2 (𝑥 )

= 𝑣(𝑥 ).

It implies that 𝐾𝑝 𝐿𝑣 = 𝑣. So 𝐾𝑝 = (𝐿|𝑑𝑜𝑚 𝐿 ∩ker 𝑃 )−1 . Thus the lemma

holds.

Lemma 3.2 N is L-compact on 𝛺 if dom L ∩ ≠ 0,where 𝛀 is a bounded

open subset of X.

Proof: We can get easily that QN is bounded. From (H3) we know that
there 𝑀0 (𝑥) ∈ 𝐿′ such that |(𝐼 − 𝑄)𝑁𝜑| ≤ 𝑀0 (𝑥 ), a.e. x∈[0,1], φ∈𝛀.

Hence 𝐾𝑝 (𝐼 − 𝑄 )𝑁(𝛺) is bounded. By the lebesgue dominated

convergence theorem and condition (H3), we can obtain that
 1
𝐾𝑝 (𝐼 − 𝑄 )𝑁(𝛺) is continuous. In addition, for {∫0 𝑘 (𝑥, 𝑦) (𝐼 −
1 1
𝑄 )𝑁𝜑(𝑦)𝑑𝑦 + 𝛷1 (𝑥) ∫0 ∫0 𝑘(𝑥, 𝑦)(𝐼 − 𝑄 )𝑁𝜑(𝑦)𝑑𝑦𝑑𝐴(𝑥 ) +
1 1
+𝛷2 (𝑥) ∫0 ∫0 𝑘(𝑥, 𝑦)(𝐼 − 𝑄 )𝑁𝜑(𝑦)𝑑𝑦𝑑𝐵 (𝑥 )} is equi continuous, by
the Ascoli-Arzela theorem, we get 𝐾𝑝 (𝐼 − 𝑄 )𝑁: 𝜴 → 𝑿 is compact.
Thus, N is L-compact. The proof is completed.

Lemma 3.3: The set 𝛺1 = {𝜑 ∈ 𝑑𝑜𝑚 𝐿\𝑘𝑒𝑟 𝐿: 𝐿𝜑 = 𝝀 ∈ [𝟎, 𝟏]}

Is bounded if (H1)-(H5) are satisfied .

Proof: Take 𝜑 ∈ 𝛺1 , 𝑡ℎ𝑒𝑛 𝑁𝜑 ∈ 𝐼𝑚𝐿, 𝑡ℎ𝑢𝑠𝑤𝑒 ℎ𝑎𝑣𝑒

1 1
∫0 ∫0 𝐾 (𝑥, 𝑦)𝑓 (𝑦, 𝜑(𝑦), 𝜑′ (𝑦), … . . , 𝜑 (𝑛−1) (𝑦)) 𝑑𝑦𝑑𝐴(𝑥 ) = 0

And,
1 1
∫0 ∫0 𝐾 (𝑥, 𝑦)𝑓 (𝑦, 𝜑(𝑦), 𝜑′ (𝑦), … . . , 𝜑 (𝑛−1) (𝑦)) 𝑑𝑦𝑑𝐵 (𝑥 ) = 0

By this together with (H5) we know that there exists 𝑥0 ∈ [0,1] such
that

|𝜑(𝑥0 )| + |𝜑 (𝑛−1) (𝑥0 )| ≤ 𝑀.

And 𝜑 (𝑖) (0) = 𝜑(𝑗) (1), 1 ≤ 𝑖 ≤ 𝑘 − 1, 1 ≤ 𝑗 ≤ 𝑛 − 𝑘 − 1, hence there

exists at least a point 𝜃𝑖 ∈ [0,1] such that 𝜑(𝑖) (𝜃𝑖 )=0, i=1,2,…..,n-1.Thus
𝑥
we get 𝜑(𝑖) (𝑥 ) = ∫𝜃 𝜑 (𝑖+1) (𝑡)𝑑𝑡, 𝑖 = 1,2, … , 𝑛 − 2. So
𝑖

‖𝜑𝑖 ‖∞ ≤ ‖𝜑𝑖+1 ‖1 ≤ ‖𝜑𝑖+1 ‖∞′ 𝑖 = 1,2, … , 𝑛 − 1.

From,

‖𝜑𝑖+1 (𝑥)‖∞ = max |𝜑 (𝑛−1) (𝑥)|

𝑥∈[0,1]
And,
𝑥
𝜑(𝑛−1) (𝑥 ) = 𝜑(𝑛−1) (𝑥0 ) + ∫𝑥 𝜑 (𝑛) (𝑡)𝑑𝑡
0

𝑥
= 𝜑(𝑛−1) (𝑥0 ) + ∫𝑥 (−1)𝑛−𝑘 𝑓 (𝑡, 𝜑(𝑡), 𝜑′ (𝑡), … , 𝜑 (𝑛−1) (𝑡)) 𝑑𝑡,
0

It follows from (H4) and (10) that

𝑥
|𝜑 (𝑛−1) (𝑥)| ≤ |𝜑(𝑛−1) (𝑥0 )| + |∫𝑥 |𝜑(𝑛) (𝑡)|𝑑𝑡|
0

≤M+∑𝑛𝑖=1‖𝑞𝑖 ‖1 ‖𝜑(𝑖−1) ‖∞ + ‖𝑟‖1

≤𝑀1 + 𝑐′‖𝜑‖∞ + 𝑐"‖𝜑 (𝑛−1) ‖∞′

Where 𝑐 ′ = ‖𝑞1 ‖1, 𝑐" = ∑𝑛𝑖=2‖𝑞𝑖 ‖1, 𝑀1 = 𝑀 + ‖𝑟‖1

In addition, for
𝑥
𝜑(𝑥 ) = 𝜑(𝑥0 ) + ∫𝑥 𝜑′ (𝑡)𝑑𝑡,
0

From (10) we have

‖𝜑‖∞ ≤ 𝑀 + ‖𝜑 (𝑛−1) ‖ .
∞′

Besides, ‖𝜑‖ = max {‖𝜑‖∞ , ‖𝜑(𝑛−1) ‖∞ } . 𝑖𝑓 ‖𝜑‖∞ ≥ ‖𝜑 (𝑛−1) ‖∞ , by

(11) and (12) we have
𝑀1 +𝑐′‖𝜑‖∞
‖𝜑(𝑛−1) ‖∞ ≤
1−𝑐"

And,
𝑀1 +𝑐′‖𝜑‖∞
‖𝜑‖∞ ≤ 𝑀 + ,
1−𝑐"

So
 1
‖𝜑‖∞ ≤ [(1 − 𝑐")𝑀 + 𝑀1 ].
1−𝑐 ′ −𝑐"

If ‖𝜑 (𝑛−1) ‖∞ > ‖𝜑‖∞′ , then by (11) and (12)

‖𝜑 (𝑛−1) ‖∞ ≤ 𝑀1 + 𝑐 ′ (𝑀 + ‖𝜑 (𝑛−1) ‖∞ ) + 𝑐"‖𝜑(𝑛−1) ‖∞

≤ 𝑀1 + 𝑐 ′ 𝑀 + (𝑐 ′ + 𝑐")‖𝜑(𝑛−1) ‖∞′

So,
1
‖𝜑 (𝑛−1) ‖∞ ≤ [(𝑀1 + 𝑐′𝑀)]. Above all, ‖𝜑‖ ≤ 𝑀𝑥,
1−𝑐 ′ −𝑐"
where
1 1
𝑀𝑥 = max { [(1 − 𝑐")𝑀 + 𝑀1 ], (𝑀1 + 𝑐′𝑀)}.
1−𝑐 ′ −𝑐" 1−𝑐 ′ −𝑐"

Above all we know 𝛺1 is bounded. The proof of the lemma is

completed.

Lemma 3.4 The set 𝛺2 = {𝜑: 𝜑 ∈ ker 𝐿, 𝑁𝜑 ∈ 𝐼𝑚 𝐿} is bounded if (H1)-

(H3),(H6) holds.

Proof : let φ∈𝛺2 , then 𝜑(𝑥 ) ≡ 𝑐1 𝜙1 (𝑥 ) + 𝑐2 𝜙2 (𝑥 ), and Nφ ∈ Im L, so

we can get
1 1 (𝑛−1)
𝑐2 ∫0 ∫0 𝑘(𝑥, 𝑦)𝑓 (𝑦, 𝑐1 𝜙1 (𝑦) + 𝑐2 𝜙2 (𝑦), … , 𝑐1 𝜙1 (𝑦) +
(𝑛−1)
+𝑐2 𝜙2 (𝑦)) 𝑑𝑦𝑑𝐴(𝑥 ) = 0

And,
1 1 (𝑛−1)
𝑐2 ∫0 ∫0 𝑘(𝑥, 𝑦)𝑓 (𝑦, 𝑐1 𝜙1 (𝑦) + 𝑐2 𝜙2 (𝑦), … , 𝑐1 𝜙1 (𝑦) +
(𝑛−1)
+𝑐2 𝜙2 (𝑦)) 𝑑𝑦𝑑𝐵 (𝑥 ) = 0.
According to (H6), we have |𝑐1 | ≤ 𝑎, |𝑐2 | ≤ 𝑏, that is to say, 𝛺2 is
bounded. We complete the proof.
Lemmas 3.5 The set 𝛺3 = {𝜑 ∈ ker 𝐿: 𝜆𝐽𝜑 + 𝛼 (1 − 𝜆)𝑄𝑁𝜑 = 0, 𝜆 ∈
[0,1]} is bounded if conditions (H1)-(H3), (H6) are satisfied,where
J:ker L →Im L is a linear isomorphism given by 𝐽(𝑐1 𝜙1 (𝑥 ) +
1 1
𝑐2 𝜙2 (𝑥 )) = (𝑒4 𝑐1 − 𝑒3 𝑐2 )𝜙1 (𝑥 ) + (−𝑒2 𝑐1 − 𝑒1 𝑐2 )𝜙2 (𝑥 ), and
𝑒 𝑒

−1 𝑖𝑓 (4) − (5)ℎ𝑜𝑙𝑑𝑠;
𝛼={ }
1, 𝑖𝑓 (6) − (7)ℎ𝑜𝑙𝑑𝑠.
Proof: suppose that 𝜑 ∈ 𝛺3 , we have φ(x)= 𝑐1 𝜙1 (𝑥 ) + 𝑐2 𝜙2 (𝑥 ), and
𝜆𝑐1 = −𝛼 (1 − 𝜆)𝑇1 𝑁𝜑, 𝜆𝑐2 = −𝛼 (1 − 𝜆)𝑇2 𝑁𝜑.
If λ = 0, by condition (H6) we have |𝑐1 | ≤ 𝑎, |𝑐2 | ≤ 𝑏. If λ = 1, then
𝑐1 = 𝑐2 = 0. if λ∈(0,1), we suppose |𝑐1 | ≥ 𝑎, |𝑐2 | ≥ 𝑏, then

𝜆𝑐12 = −𝛼 (1 − 𝜆)𝑐1 𝑇1 𝑁𝜑 < 0

Or,

𝜆𝑐22 = −𝛼 (1 − 𝜆)𝑐2 𝑇2 𝑁𝜑 < 0,

Which contradicts with 𝜆𝑐12 > 0, 𝜆𝑐22 > 0. so the lemma holds.