Volume 2008, Article ID 254593,10pages doi:10.1155/2008/254593
Research Article
Global Behavior of the Components for the Second
Order
m
-Point Boundary Value Problems
Yulian An1, 2 and Ruyun Ma1
1Department of Mathematics, Northwest Normal University, Lanzhou 730070, China 2Department of Mathematics, Lanzhou Jiaotong University, Lanzhou 730070, China
Correspondence should be addressed to Yulian An,an [email protected]
Received 9 October 2007; Accepted 16 December 2007
Recommended by Kanishka Perera
We consider the nonlinear eigenvalue problems u rfu 0, 0 < t < 1, u0 0, u1
m−2
i1 αiuηi, wherem ≥ 3,ηi ∈ 0,1, andαi > 0 fori 1, . . . , m−2, withim−21 αi < 1;r ∈ R;
f ∈ C1R,R. There exist two constantss
2 < 0 < s1 such that fs1 fs2 f0 0 and
f0 :limu→0fu/u∈0,∞,f∞:lim|u|→∞fu/u∈0,∞. Using the global bifurcation
tech-niques, we study the global behavior of the components of nodal solutions of the above problems.
Copyrightq2008 Y. An and R. Ma. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
1. Introduction
In 1, Ma and Thompson were concerned with determining values of real parameter r, for which there exist nodal solutions of the boundary value problems:
uratfu 0, 0< t <1,
u0 u1 0, 1.1
whereaandfsatisfy the following assumptions:
H1 f ∈CR,Rwithsfs>0 fors / 0;
H2 there existf0, f∞∈0,∞such that
f0 lim |s|→0
fs
s , f∞|slim|→∞ fs
s ; 1.2
H3 a:0,1→0,∞is continuous andat/≡0 on any subinterval of0,1.
Theorem 1.1. Let (H1), (H2), and (H3) hold. Assume that for somek∈N, either
λk
f∞ < r < λk
f0 1.3
or
λk
f0
< r < λk
f∞. 1.4
Then1.1have two solutionsuk andu−k such thatuk has exactlyk−1zeros in0,1and is positive near0, andu−k has exactlyk−1zeros in0,1and is negative near0. In1.3and1.4,λk is thekth eigenvalue of
ϕλatϕ0, 0< t <1, ϕ0 ϕ1 0. 1.5
Recently, Ma2extended this result and studied the global behavior of the components of nodal solutions of1.1under the following conditions:
H1 f ∈CR,Rand there exist two constantss2<0< s1, such thatfs1 fs2 f0
0 andsfs>0 fors∈R\ {0, s1, s2}; H4 f satisfies Lipschitz condition ins2, s1.
Using Rabinowitz global bifurcation theorem, Ma established the following theorem.
Theorem 1.2. LetH1, (H2), (H3), and (H4) hold. Assume that for somek∈N,
λk
f∞ < λk
f0. 1.6
Then
iifr ∈λk/f∞, λk/f0, then1.1have at least two solutionsu±k,∞, such thatuk,∞has exactly
k−1 zeros in0,1and is positive near0, anduk,−∞has exactlyk−1zeros in0,1and is negative near0,
iiifr ∈λk/f0,∞, then1.1have at least four solutionsu±k,∞andu±k,0, such thatuk,∞(resp., uk,0) has exactlyk−1zeros in0,1and is positive near0;u−k,∞(resp.,u−k,0) has exactlyk−1zeros in
0,1and is negative near0.
Remark 1.3. LetH1,H2,H3, andH4hold. Assume that for somek∈N,λk/f0< λk/f∞.
Similar results toTheorem 1.2have also been obtained.
Making a comparison between the above two theorems, we see that asf has two zeros
s1, s2 : s2 < 0 < s1, the bifurcation structure of the nodal solutions of 1.1 becomes more
complicated: two new nodal solutions are obtained whenr >max{λk/f0, λk/f∞}.
In 3, Ma and O’Regan established some existence results which are similar to Theorem 1.1of the nodal solutions of them-point boundary value problems
ufu 0, 0< t <1,
u0 0, u1
m−2
i1 αiu
ηi
under the following condition:
H1 f ∈C1R,Rwithsfs>0 fors /0.
Remark 1.4. For other results about the existence of nodal solution of multipoint boundary value problems, we can see4–7.
Of course an interesting question is, as for m-point boundary value problems, when
f possesses zeros in R\ {0}, whether we can obtain some new results which are similar to Theorem 1.2.
We consider the eigenvalue problems
urfu 0, 0< t <1, 1.8
u0 0, u1 m−2
i1 αiu
ηi
, 1.9
where m ≥ 3, ηi ∈ 0,1, andαi > 0 for i 1, . . . , m−2. Also using the global bifurcation techniques, we study the global behavior of the components of nodal solutions of1.8,1.9 and give a positive answer to the above question. However, when m-point boundary value condition1.9is concerned, the discussion is more difficult since the problem is nonsymmetric and the corresponding operator is disconjugate.
In the following paper, we assume that
H0 αi>0 fori1, . . . , m−2,with 0<mi−12αi<1;
H1 f ∈ C1R,R and there exist two constantss
2 < 0 < s1, such that fs1 fs2
f0 0;
H2 there existf0, f∞∈0,∞such that
f0 lim |s|→0
fs
s , f∞|slim|→∞ fs
s . 1.10
The rest of the paper is organized as follows.Section 2contains preliminary definitions and some eigenvalue results of corresponding linear problems of 1.8, 1.9. In Section 3, we give two Rabinowize-type global bifurcation theorems. Finally, in Section 4, we consider two bifurcation problems related to1.8,1.9, and use the global bifurcation theorems from Section 3to analyze the global behavior of the components of nodal solutions of1.8,1.9.
2. Preliminary definitions and eigenvalues of corresponding linear problems
LetY C0,1with the norm
u∞ max
t∈0,1
ut. 2.1
Let
X
u∈C10,1|u0 0, u1
m−2
i1 αiu
ηi
,
E
u∈C20,1|u0 0, u1 m−2
i1 αiu
ηi
with the norm
uX max
u∞,u∞}, uEmax{u∞,u∞,u∞
, 2.3
respectively. DefineL:E→Y by setting
Lu:−u, u∈E. 2.4
ThenLhas a bounded inverseL−1:Y → Eand the restriction ofL−1toX, that is,L−1 :X→X
is a compact and continuous operator, see3,4,8.
LetER×Eunder the product topology. As in9, we add the points{λ,∞|λ∈R}
to our space E. For anyC1 functionu, ifux0 0, thenx0is a simple zero of uifux0/0.
For any integer k ≥ 1 and any ν ∈ {±}, define sets Sνk, Tkν ⊂ C20,1 consisting of functions
u∈C20,1satisfying the following conditions:
Sνk:
iu0 0, νu0>0;
iiuhas only simple zeros in0,1and has exactlyk−1 zeros in0,1;
Tkν:
iu0 0,νu0>0,andu1/0;
iiuhas only simple zeros in0,1and has exactlykzeros in0,1;
iiiuhas a zero strictly between each two consecutive zeros ofu.
Remark 2.1. Obviously, ifu∈Tkν, thenu∈Sνk oru∈Sνk1. The setsTkνare open inEand disjoint.
Remark 2.2. The nodal properties of solutions of nonlinear Sturm-Liouville problems with sep-arated boundary conditions are usually described in terms of sets similar toSν
k, see1,2,5,9– 11. However, Rynne 4 stated that Tν
k are more appropriate than Sνk when the multipoint boundary condition1.9is considered.
Next, we consider the eigenvalues of the linear problem
Luλu, u∈E. 2.5
We call the set of eigenvalues of2.5the spectrum ofL, and denote it byσL. The following lemmas can be found in3,4,12.
Lemma 2.3. Let (H0) hold. The spectrum σL consists of a strictly increasing positive sequence of eigenvaluesλk,k1,2, . . . ,with corresponding eigenfunctionsϕkx sin
λkx. In addition,
ilimk→∞λk ∞;
iiϕk ∈Tk, for eachk ≥1,andϕ1is strictly positive on0,1.
We can regard the inverse operator L−1 : Y → E as an operatorL−1 : Y → Y. In this
setting, eachλk,k1,2, . . . ,is a characteristic value ofL−1, with algebraic multiplicity defined to be dim∞j1NI−λkL−1j, whereNdenotes null-space andIis the identity onY.
Lemma 2.4. Let (H0) hold. For eachk ≥ 1, the algebraic multiplicity of the characteristic valueλk,
3. Global bifurcation
Letg∈C1R,Rand satisfy
g0 g0 0. 3.1
Consider the following bifurcation problem:
Luμugu, μ, u∈R×X. 3.2
Obviously,u≡0 is a trivial solution of3.2for anyμ∈R. About nontrivial solutions of3.2, we have the following.
Lemma 3.1see4, Proposition 4.1. Let (H0) hold. Ifμ, u∈ Eis a nontrivial solution of 3.2, thenu∈Tkνfor somek, ν.
Remark 3.2. From Lemmas 2.3and 3.1, we can see that Tkν are more effectual than the setSνk
when the multipoint boundary condition1.9is considered. In fact, eigenfunctionsϕkx sinλkx, k 1,2, . . . , of 2.5do not necessarily belong to Sk. In 3, 4, there were some special examples to show this problem.
Also, in4, Rynne obtained the following Rabinowitz-type global bifurcation result for
3.2.
Lemma 3.3 see4, Theorem 4.2. Let (H0) hold. For eachk ≥ 1andν, there exists a continuum Cν
k ⊂Eof solution of 3.2with the following properties:
1o λk,0∈ Cνk;
2o Cνk\ {λk,0} ⊂R×Tkν;
3o Cν
k is unbounded inE.
Now, we consider another bifurcation problem
Luμuhu, μ, u∈R×X, 3.3
where we suppose thath∈C1R,Rand satisfy
lim
|x|→∞
hx
x 0. 3.4
TakeΛ⊂Ras an interval such thatΛ∩ {λj |j ∈N}{λk}andMas a neighborhood of
λk,∞whose projection onRlies inΛand whose projection onEis bounded away from 0.
Lemma 3.4. Let (H0) and3.4hold. For eachk≥1andν, there exists a continuumDνk ⊂Eof solution of 3.3which meetsλk,∞and either
1o Dν
k\ Mis bounded inEin which caseDνk\ Mmeets{λ,0|λ∈R}or
Moreover, if2ooccurs andDν
k\ Mhas a bounded projection onR, thenDνk\ Mmeetsμ, ∞, whereμ∈ {λj |j∈N}withμ /λk.
In every case, there exists a neighborhoodO ⊂ M ofλk,∞such thatμ, u ∈ Dνk ∩ O and
μ, u/ λk,∞impliesμ, u∈R×Tkν.
Remark 3.5. A continuum Dkν ⊂ E of solution of 3.3meets λk,∞ which means that there exists a sequence{λn, un} ⊂ Dkνsuch thatunE→ ∞andλn→λk.
Proof. Obviously,3.3is equivalent to the problem
uμL−1uL−1hu, μ, u∈R×X. 3.5
Note thatL−1 :X →Xis a compact and continuous linear operator. In addition, the mapping
u→ L−1huis continuous and compact, and satisfiesL−1hu ou
Xatu∞; moreover, u2
XL−1hu/u2Xis compactsimilar proofs can be found in9. Hence, the problem3.3is of the form considered in9, and satisfies the general hypotheses imposed in that paper. Then by 9, Theorem 1.6 and Corollary 1.8 together with Lemmas 2.3and 2.4 in Section 2, there exists a continuumDνk ⊂R×Xof solutions of3.3which meetsλk,∞and either
1o Dνk\ Mis bounded inR×Xin which caseDνk\ Mmeets{λ,0|λ∈R}or
2o Dν
k\ Mis unbounded inR×X.
Moreover, if(2o)occurs andDν
k \ Mhas a bounded projection onR, thenD ν
k \ Mmeets
μ, ∞whereμ∈ {λj |j ∈N}withμ /λk.
In every case, there exists a neighborhoodO ⊂ Mof λk,∞such that μ, u ∈ Dνk ∩ O andμ, u/ λk,∞impliesμ, u∈R×Tkν.
On the other hand, by3.5and the continuity of the operatorL−1:Y →E, the setDνklies inEand the injectionDνk →Eis continuous. Thus,Dνk is also a continuum inEand the above properties hold inE.
Now, we assume that
h0 0. 3.6
Lemma 3.6. Let (H0) and3.6hold. Ifμ, u ∈ Eis a nontrivial solution of 3.3, then u ∈Tν k for somek, ν.
Proof. The proof ofLemma 3.6is similar to the proof of Lemma 3.14, Proposition 4.1; we omit it.
Remark 3.7. If3.6holds, Lemma 3.6guarantees thatDkν inLemma 3.4is a component of so-lutions of 3.3 in Tkν which meets λk,∞. Otherwise, if there exist η1, y1 ∈ Dνk ∩ Tkν and η2, y2∈ Dνk∩Thνfor somek /h∈N, then by the connectivity ofDνk, there existsη∗, y∗∈ Dνk
4. Statement of main results
We return to the problem1.8,1.9. LetH1,H2hold and letζ, ξ∈C1R,Rbe such that
fu f0uζu, fu f∞uξu. 4.1
Clearly
ζ0 0, ξ0 0,
lim
|u|→0 ζu
u ζ
0 0, lim |u|→∞
ξu
u 0.
4.2
Let us consider
Lu−rf0urζu 4.3
as a bifurcation problem from the trivial solutionu≡0, and
Lu−rf∞urξu 4.4
as a bifurcation problem from infinity. We note that4.3and4.4are the same, and each of them is equivalent to1.8,1.9.
The results of Lemma 3.3for4.3can be stated as follows: for each integerk ≥ 1,ν ∈
{, −}, there exists a continuumCνk,0of solutions of4.3joiningλk/f0,0to infinity, andCνk,0\
{λk/f0,0} ⊂R×Tkν.
The results of Lemma 3.4for4.4can be stated as follows: for each integerk ≥ 1,ν ∈
{, −}, there exists a continuumDνk,∞of solutions of4.4meetingλk/f∞,∞.
Theorem 4.1. Let (H0),H1,and (H2) hold. Then
iforr, u∈ Ck,0∪ C−k,0,
s2< ut< s1, t∈0,1; 4.5
iiforr, u∈ Dk,∞∪ D−k,∞,
max
t∈0,1ut> s1, or tmin∈0,1ut< s2. 4.6
Proof ofTheorem 4.1. Suppose on the contrary that there existsr, u∈ Ck,0∪ C−k,0∪ Dk,∞∪ Dk,−∞
such that either
maxut|t∈0,1s1 4.7
or
minut|t∈0, 1s2. 4.8
Sinceu∈Tν
k, byRemark 2.1,u∈Sνk oru∈ Sνk1. We assumeu∈Sνk. Whenu∈Sνk1, we
can prove all the following results with small modifications. Let
0τ0< τ1<· · ·< τk−1<1 4.9
Case 1max{ut|t∈0, 1}s1. In this case, there existsj ∈ {0, . . . , k−2}such that
maxut|t∈τj, τj1
s1 or max{ut|t∈τk−1, 1
s1,
0≤ut≤s1, t∈
τj, τj1
, or t∈τk−1, 1
.
4.10
Sinceu1 mi−12αiuηiandH0, we claimu1< s1.
Lett0∈τj, τj1 ort0∈τk−1, 1such thatut0 s1, thenut0 0. Note that
fut0fs1 0. 4.11
By the uniqueness of solutions of 1.8subject to initial conditions, we see that ut ≡ s1 on 0,1. This contradicts1.9andH0.
Therefore,
maxut|t∈0, 1/s1. 4.12
Case 2min{ut|t∈0,1}s2. In this case, the proof is similar toCase 1, we omit it.
Consequently, we obtain the resultsiandii.
Theorem 4.2. Let (H0),H1,and (H2) hold. Assume that for somek∈N,
λk
f∞ < λk
f0
resp., λk
f0 < λk
f∞
. 4.13
Then
iif r ∈ λk/f∞, λk/f0 (resp., r ∈ λk/f0, λk/f∞, then 1.8, 1.9 have at least two solutionsu±k,∞(resp.,u±k,0), such thatuk,∞ ∈ Tk andu−k,∞ ∈ Tk− (resp.,uk,0 ∈Tkandu−k,0 ∈ Tk−),
iiifr∈λk/f0,∞(resp., r ∈λk/f∞,∞, then1.8,1.9have at least four solutionsu±k,∞ andu±k,0, such thatuk,∞,uk,0∈Tk, andu−k,∞,u−k,0∈Tk−.
Remark 4.3. Making a comparison between results in3and the above theorem, we see that as f has two zeros s1, s2 : s2 < 0 < s1, the bifurcation structure of the nodal solutions of 1.8,1.9becomes more complicated: the component of the solutions of1.8,1.9from the trivial solution atλk/f0,0and the component of the solutions of1.8,1.9from infinity at λk/f∞,∞are disjoint; two new nodal solutions are born whenr >max{λk/f0, λk/f∞}.
Proof ofTheorem 4.2. Since1.8,1.9have a unique solutionu≡0, we get
C
k,0∪ C−k,0∪ Dk,∞∪ D−k,∞
⊂μ, z∈E|μ≥0. 4.14
TakeΛ ⊂ Ras an interval such thatΛ∩ {λj/f∞ | j ∈ N} {λk/f∞}andMas a
neigh-borhood ofλk/f∞,∞whose projection onRlies inΛand whose projection onEis bounded
Dν
k,∞\ Msatisfies one of the following:
1o Dν
k,∞\ Mis bounded inEin which caseDνk,∞\ Mmeets{λ,0|λ∈R};
2o Dν
k,∞\ Mis unbounded inEin which case ProjRDk,∞\ Mis unbounded.
Obviously,Theorem 4.1iiimplies that1odoes not occur. SoDk,∞\ Mis unbounded inE. Thus
ProjRDk,∞⊃
λk
f∞,∞
,
ProjRD−k,∞⊃
λk
f∞,∞
.
4.15
ByTheorem 4.1, for anyr, u∈Ck,0∪ C−k,0,
u∞<maxs1,s2:s∗. 4.16
Equations4.16,1.8, and1.9imply that
uE <max
rmax
|s|≤s∗
fs, s∗, 4.17
which means that the sets{μ, z∈ Ck,0 | μ∈0, d}and{μ, z∈ C−k,0 | μ∈0, d}are bound-ed for any fixbound-edd∈0,∞. This, together with the fact thatCk,0resp.,C−k,0joinsλk/f0,0to
infinity, yields that
ProjRCk,0⊃
λk
f0,∞
,
ProjRC−k,0⊃
λk
f0,∞
.
4.18
Combining4.15with4.18, we conclude the desired results.
Acknowledgments
This paper is supported by the NSFCno. 10671158, the NSF of Gansu Provinceno. 3ZS051-A25-016, NWNU-KJCXGC-03-17, the Spring-sun program no. Z2004-1-62033, SRFDP no. 20060736001, the SRF for ROCS, SEM2006311, and LZJTU-ZXKT-40728.
References
1R. Ma and B. Thompson, “Nodal solutions for nonlinear eigenvalue problems,”Nonlinear Analysis: Theory, Methods & Applications, vol. 59, no. 5, pp. 707–718, 2004.
2R. Ma, “Global behavior of the components of nodal solutions of asymptotically linear eigenvalue problems,”Applied Mathematics Letters, 2007.
4B. P. Rynne, “Spectral properties and nodal solutions for second-order,m-point, boundary value prob-lems,”Nonlinear Analysis: Theory, Methods & Applications, vol. 67, no. 12, pp. 3318–3327, 2007.
5R. Ma, “Nodal solutions for a second-orderm-point boundary value problem,”Czechoslovak Mathe-matical Journal, vol. 56, no. 4, pp. 1243–1263, 2006.
6X. Xu, S. Jingxian, and D. O’Regan, “Nodal solutions for m-point boundary value problems using bifurcation methods,” to appear inNonlinear Analysis, 2008.
7X. Xu and D. O’Regan, “Multiplicity of sign-changing solutions for some four-point boundary value problem,” to appear inNonlinear Analysis, 2008.
8M. K. Kwong and J. S. W. Wong, “The shooting method and nonhomogeneous multipoint BVPs of second-order ODE,”Boundary Value Problems, vol. 2007, Article ID 64012, 16 pages, 2007.
9P. H. Rabinowitz, “On bifurcation from infinity,”Journal of Differential Equations, vol. 14, pp. 462–475, 1973.
10P. H. Rabinowitz, “Some global results for nonlinear eigenvalue problems,”Journal of Functional Anal-ysis, vol. 7, no. 3, pp. 487–513, 1971.
11B. P. Rynne, “Second-order Sturm-Liouville problems with asymmetric, superlinear nonlinearities,”
Nonlinear Analysis: Theory, Methods & Applications, vol. 54, no. 5, pp. 939–947, 2003.