• 沒有找到結果。

SYMMETRY-BREAKINGS FOR SEMILINEAR ELLIPTIC-EQUATIONS OF FINITE CYLINDRICAL DOMAINS

N/A
N/A
Protected

Academic year: 2021

Share "SYMMETRY-BREAKINGS FOR SEMILINEAR ELLIPTIC-EQUATIONS OF FINITE CYLINDRICAL DOMAINS"

Copied!
9
0
0

加載中.... (立即查看全文)

全文

(1)

Volume 117, Number 3, March 1993

SYMMETRY-BREAKINGS

FOR SEMILINEAR ELLIPTIC EQUATIONS

ON FINITE CYLINDRICAL DOMAINS

SONG-SUN LIN

(Communicated by Barbara L. Keyfitz)

Abstract. We study the existence and multiplicity of asymmetric positive solu-tions of a semilinear elliptic equation on finite cylinders with mixed type bound-ary conditions. By using a Nehari-type variational method, we prove that the numbers of asymmetric positive solutions are increasing without bound when the lengths of cylinders are increasing. On the contrary, by using the blow up technique, we obtain an a priori bound for positive solutions and then prove that all positive solutions must be symmetric when the cylinders are short enough.

1. INTRODUCTION

Let co be a bounded smooth domain in W, n > 1, a > 0, and Qa = (-a, a) x co be a finite cylindrical domain in K"+1 = {(x, y) : x € E1 and y e W}. In this paper, we shall study the existence and multiplicity of positive solutions of the following semilinear elliptic equation with mixed type boundary

conditions

(1.1) Au + f{u) = 0 inQa,

(1.2) u = 0 on[-a, a]xdco,

(1.3) — = 0 on {-a, a} x co,

where / satisfies the following conditions:

(H-0) feCl(Rl),

f{u)>0 for w>0,

(H-l) /(0) = 0 and /'(0) = 0,

(H-2) there exists a > 0 such that uf'(u) > (1 + a)f(u) for all u > 0,

(H-3) f(u) < C(up + 1) for some C > 0 and all u > 0, where 1 < p < (n + 3)/(« - 1) if n > 2 and 1 < p < oo if n = 1 .

A solution u € C2(£la) of (1.1)-(1.3) is said to be symmetric (with respect to the x-axis) if u is constant along the x-axis, i.e., u = u(y) and satisfies

(1.4)

Au + f(u) = 0 in co,

(1.5) u = 0 on dco.

Received by the editors January 30, 1991 and, in revised form, July 5, 1991.

1991 Mathematics Subject Classification. Primary 35B32, 35J65, 35P30.

Key words and phrases. Symmetry breaking, cylinders.

Work partially supported by the National Science Council of the Republic of China.

© 1993 American Mathematical Society

0002-9939/93 $1.00+ $.25 per page

(2)

Otherwise, u is called an asymmetric solution. The following problem was

posed by Berestycki:

,p. For /(«) = up , taking the length a as a bifurcation parameter, ^ ' is there symmetry-breaking for (1.1)-(1.3)?

Previously, it has been known that there is an asymmetric solution when a is large enough. In this paper, we answer the problem (P) partially, and hope it will lead to more complete results of the problem. The main results are as

follows.

Theorem 1.1. Assume conditions (H-0)-(H-3) are satisfied. Then there exists an increasing sequence ak — ak(co, a) —► oo as k —> oo, such that for each k > 1 and a e (ak , oo), (1.1)—(1.3) have at least k-many asymmetric positive

solutions.

On the contrary, for short cylinders we have

Theorem 1.2. Assume f satisfies (H-0)-(H-2) and

(H-3)' lim^oo f(u)/uP = C > 0 for 1 < p < (n + 3)/(« - 1) if n>2 and

1 < p < oo if n = 1. Then there exists <z» = a*(co, p) > 0 such that for any ae(0,fl,), every positive solution o/(l.l)-(1.3) is symmetric.

The last two theorems indicate that there are asymmetric bifurcations at certain critical numbers a*., aj-»oo as k —> oc, and the bifurcation branches will go to the direction of increasing a . However, we still need more rigorous justifications.

To prove Theorem 1.1, we need to study the problems of asymmetric insta-bility of symmetric positive solutions and then use a Nehari-type variational method to prove that there exist asymmetric positive solutions when all

sym-metric positive solutions are unstable with respect to certain asymsym-metric mode.

To prove Theorem 1.2, we need to obtain some a priori bounds for posi-tive solutions of (1.1)—(1.3) and then prove that asymmetric posiposi-tive solutions cannot exist when the cylinders are short enough.

As for the related problems, the problems of symmetry-breaking of positive radial solutions on a ball were studied in [4, 6, 15-17], on annular domains in

[3, 5, 8-11, 18, 19], and on sectorial domains in [2, 12].

The paper is organized as follows. In §2 we study the linear eigenvalue prob-lems and obtain some asymmetric instability results for symmetric positive so-lutions for long cylinders. In §3 we prove the existence of asymmetric positive solutions for long cylinders. In §4 we prove the nonexistence of asymmetric positive solutions for short cylinders.

2. Linearized eigenvalue problems

Let g e C(co). Consider the following linear eigenvalue problem:

(2.1) Av + g(y)v = -vv ina;,

(2.2) v = 0 on dco.

Then there is the sequence {v\,v2,...} = {v\ {g), v2{g), ...} of eigenvalues of (2.1) and (2.2) satisfying v\ < v2 < 1/3 < •• • , where each distinct eigenvalue

(3)

is repeated according to its multiplicity. By separation of variables, the linear eigenvalue problem

(2.3) Aw + g(y)w = -Xw in ila,

(2.4) w = 0 on [-a, a] x dco,

(2.5) -£— = 0 on {-a, a} x co,

is equivalent to the problems

(2.6) Av + g(y)v = - (A + p)v m co,

(2.7) v = 0 on dco,

and

(2.8)

<p"{x)

= n<p{x), x£(-a,a),

(2.9)

<p'(-a) = 0 = <p'(a),

with w(x, y) = cp(x)v{y). It is clear that the eigenvalues of (2.8) and (2.9) are given by

(2.10) fik(a) = -(kn/2a)2,

k = 0, 1,2, ... , and the associated eigenfunctions are

sin —x if k is odd,

(2.11)

<Pk(x)={

H

cos -r— x if k is even.

*> 2a

Therefore, X is an eigenvalue of (2.3)—(2.5) if and only if X-\-fik = v\ for some k and /. Hence, we denote

(2.12)

h,i=h,i{g,a)

= vt-nk = vt{g)+ l-^J

where k = 0, 1, 2, ... and / = 1,2,.... Note that when k = 0 it corre-sponds to symmetric modes and k > 1 to asymmetric modes.

Let Mo be any symmetric positive solution and v\ = u\(uo) be the least eigenvalue of (2.1) and (2.2) with g(y) = f'(uo(y)). Then we have the following estimate for ^i(wn).

Lemma 2.1. Assume condition (H-2) is satisfied. Then for any symmetric positive solution uq we have

(2.13) "i(«o) < -ov\ ,

where V\ = u\{co) > 0 is the least eigenvalue of the Laplacian -A on co with Dirichlet boundary conditions.

Proof. It is well known that v\(uo) can be characterized by

vx = inf \Q{v)l j v2 : v e 7/0V)\{0}

J ,

where

Q(v)= f \Vv\2-f'(u0)v2.

(4)

Since Uq is a solution of (1.4) and (1.5), we have / |Vwo|2 = / u0f{u0).

Jw Jw

Hence, by (H-2), we have

Q{u0) = / |Vw0|2

- f'(u0)ul = / u0f{u0) - f'(u0)ul

Jw J(0

< - a / u0f(u0) = -o /

|V«o|2-Jco Jw

Therefore, by using the Poincare inequality, (2.13) follows. The proof is

com-plete.

Note that the upper bound -avx of vx(uq) is independent of symmetric positive solutions «o ■

A consequence of Lemma 2.1 is the following asymmetric instability results

for long cylinders.

Theorem 2.2. Assume condition (H-2) is satisfied. For each k > 1, let

(2.14)

a* = ^W"1/2.

Then for any a e (ak , oo) and any symmetric positive solution uq we have

4,i("o, a) < 0.

Proof. By(2.12), Xk \{u$, a) = v\{u0) + fik(a); the result follows from Lemma

2.1.

Remark 2.3. Assume conditions (H-0)-(H-3) are satisfied.

(i) If u e C2(fta)nC°(fia) is a solution of (1.1)-(1.3), then u is positive in Q. Indeed, by (H-0) and the maximum principle, the minimum of u in Q.a is achieved on dQa ■ On the other hand, by (1.3) and the strong maximum prin-ciple, the minimum cannot be achieved on {-a, a) xco. Hence the minimum will be achieved on [-a, a] x dco, which implies u is positive in Qa.

(ii) There exist positive solutions of (1.1)—(1.3) and (1.4) and (1.5),

respec-tively, see, e.g., [1, 14].

3. Long cylinders

In this section, we shall prove that if the cylinders are long enough, a > ak , then there exist at least /c-many asymmetric positive solutions. We shall use a Nehari-type variational method.

For any function u defined on Q.a , it can be extended to u on £1^ = (-oo, oo) x co. In fact, we first define

u(x, y) — u(2a - x, y) for x e (a, 3a) and y £ co,

a reflection with respect to x - a. Then « is extended to the whole of Q^o by 4a periodically along x-axis. Due to the Neumann boundary conditions on {-a, a] x co, the eigenfunctions wkj and solutions of (1.1)—(1.3) can be extended smoothly on ^lx in this way.

(5)

We need some notation. Let V — {u e Hl(Cla) : u = 0 on [-a, a] x dco}.

Define the functional

J(u)= f Uvu\2-F{u) and M(u) = f \Vu\2-uf(u)

on V , where F(u) = /„" /(/) dt.

For each k > 1, the submanifolds Vk and Foo of V are defined by

^ = ^t,« = <"€K:fifjc + -^,yJ =u{x,y)

for (x, y) € Qoo and M(u) = 0 >

and

Foo = {W€/f0'(w):M(M) = 0}.

Let

Ik = Ik{a) = inf{/(w) : u £ Vk}, /<*, = /^(a) = inf{/(w) : u 6 K^}. Let «o be a symmetric positive solution, and wq = i>i(y) > 0 and wk = <pk{x)v\(y) be eigenfunctions with respect to eigenvalues Ao,i = Xoy\{uo,a)

and Afc>i = Afc)1(«o, a) with /fi ty^ = fa w\ = 1. Note that for k > 1, we have

(3.1)

f wk=0.

Then we have the following results concerning the change of J(u) along the direction of asymmetric mode wk .

Lemma 3.1. Assume conditions (H-0)-(H-2) are satisfied. Then there exist e > 0 and a smooth function 6: (-e, e) -» R1 w/?/z r5(0) = S'{0) = 0 5mc/z that

(3.2)

M(M0

+ <5(0^o + ^t) = 0

/or a// t e (—e , e). Furthermore, we have

(3.3) /(ao + <$(Ot"o + ftofc) = J(u0) + \X0JS(t)2 + {XkAt2 + 0(t4),

as t ~ 0.

Proof. The proofs are similar to Lemmas 6.1 and 6.2 in [11]. (3.2) is proved by using the Implicit Function Theorem and (3.3) is proved by straightforward computations; the details are omitted. Note that (3.1) is used repeatedly in the

proofs.

To prove Theorem 1.1, we need the following result, which indicates that asymmetric minimizers are different for different modes.

Lemma 3.2. Assume conditions (H-0)-(H-3) are satisfied. If Ikm < /^ then

h < hm for k = 1 > 2, ... and m = 2, 3, ... .

Proof. Under the assumptions of (H-0)-(H-3), it is rather standard to verify that the Palais-Smale condition holds, see, e.g., [1, 14]. Therefore, the mini-mizers of Ik and /oo are achieved by some functions uk e Vk and «o € K»

(6)

Let m(x, y) be a minimizer of J(u) on Vkm . Since Ikm < 1^ ,

(3.4) du/dx^O onQfl.

Let

u (— , y J for x e (-a, a) when km is odd,

f(*»)0 = < (x + a \

u -, y ] for x e (-a, a) when /cm is even.

( \ m J

Then for km odd we have

(3.5)

v(x + f,y^=u^

+ ^,y^=u(^,y)=v(x,y).

Furthermore, we have

(36) L}^--LLW)+{^2}dxdy

--mLlZ{X)1+(v'u]1)dxdy<Lml-Similarly, if km is even then (3.5) and (3.6) can also be proved. Since u and

v are periodic in x of periods km and k, respectively, it can be proved that

(3.7)

/ vf(v) = f uf(u) and / F(v) = I F(u).

Jo.a Jna Jiia Jaa

Therefore, (3.6) and (3.7) imply M(v) < 0. Since f'(0) = 0, we have M(tv) > 0 for t > 0 and sufficiently small. This implies that there exists a t\ e (0, 1) such that M(tiv) = 0.

Now, by (H-2), we have

j-t ^tuf(tu)

- F(tu)} = \{tuf'{tu) - /(/«)} > 0

for m > 0. Therefore, we have

Ik<J{t\v)= f l-t2\Vv\2-F(tiv)

jQa l

= / \txvf(txv)-F(txv)<

f Lf(v)-F(v)

= / ^uf(u)-F(u) = Ikm.

The proof is complete.

Proof of Theorem 1.1. For each k > 1 and a e (ak, oo), ak given by (2.14), the minimizers of Ij(a) and /oo(a) are achieved by some functions Uj e Vj

and Mo e Foo , ;' = 1, ... , k . By (3.3), /,-(a) < /^(a), ;'= 1,..., k.

Now let 1 < i < j < k and / be the least common multiple of i and j. If

j = I then, by Lemma 3.2, m, and Uj are different. If j < I then VjDVj — Vj. If m, = Uj■■ = u € ^/ then

(7)

By (3.8) and Lemma 3.2, we obtain /,(a) < //(a), a contradiction to (3.8).

Hence, m, and Uj are different. The proof is complete.

4. Short cylinders

In this section, we shall prove that there is no asymmetric positive solution of (1.1)—(1.3) when the cylinders are short enough.

We need the following a priori estimates.

Lemma 4.1. Assume conditions (H-0)-(H-3)' are satisfied. Then there exists a constant C = C(co) > 0 such that for any a > 0 and any positive solution ua

o/(l.l)-(1.3), we have

(4.1) II"alloc = max{|Ma(x)| : x e fia} < C.

Proof. Since the extension ua of ua is smooth in Qoo , i.e., ua e C2(Qoo), the

boundary {-a, a} x co is now an interior part of Qoo • Since ua > 0 in Q^o

and satisfies

Aua + f{ua) = 0 in Qoo,

ua = 0 on <9Qoo ,

by using the technique of "blow up" as in [7], we can prove (4.1). The details

are omitted.

Proof of Theorem 1.2. For a > 0, if ua is an asymmetric positive solution of

(l.l)-( 1.3), then

(4.2) wa = dua/dx £ 0

and wa satisfies

(4.3) Awa + f'(ua)wa = 0 inQa,

(4.4) wa = 0 on <9Qa.

Let Ai(a) > 0 be the least eigenvalue of the Laplacian -A on Qa with Dirichlet boundary condition. Then it is easy to verify that

(4.5) lim Ai(a) = oo.

a—>0+

On the other hand, the least eigenvalue t]{ (a) of the linearized eigenvalue problem Aw + f'{ua)w — -nw inQa,

w = 0

on <9Qj

can be characterized by

nx{a) = inf [qM/J

v2 : v e //0'(Qfl)\{0}} ,

where

Q(v)= f \Vv\2-f'(ua)v2.

By Lemma 4.1, there exists a constant Cx > 0 such that

(8)

By the Poincare inequality, we have

(4.7) / |Vu|2 >A,(a) / v2

for all v e H^(£la). Hence, by (4.5), (4.6), and (4.7), there exists a* > 0 such

that

(4.8)

m(a)>0

for all a e (0, a*). Therefore, (4.2), (4.3), (4.4), and (4.8) imply that there is

no asymmetric positive solution for a 6 (0, a*). The proof is complete. Remark 4.2. The results are still valid if f(u) is replaced by f(y, u), which satisfies similar conditions as (H-0)-(H-3) and (H-3)'.

ACKNOWLEDGMENT

The author is grateful for several useful conversations with Professor H. Berestycki during his stay in Taiwan in April 1988.

References

1. A. Ambrosetti and P. Rabinowitz, Dual variational methods in critical point theory and

applications, J. Funct. Anal. 14 (1973), 349-381.

2. H. Berestycki and F. Pacella, Symmetry properties for positive solutions of elliptic equations with mixed boundary conditions, preprint.

3. H. Brezis and L. Nirenberg, Positive solutions of nonlinear elliptic equations involving critical Sobolev exponents, Comm. Pure Appl. Math. 36 (1983), 437-477.

4. G. Cerami, Symmetry breaking for a class of semilinear elliptic problems, Nonlinear Anal. Theory, Methods, and Applications 10 (1986), 1-14.

5. C. V. CofFman, A nonlinear boundary value problem with many positive solutions, J. Differ-ential Equations 54 (1984), 429-437.

6. E. N. Dancer, On non-radially symmetric bifurcation, J. London Math. Soc. 20 (1979),

287-292.

7. B. Gidas and J. Spruck, A priori bounds for positive solutions of nonlinear elliptic equations. Comm. Partial Differential Equations 6 (1981), 883-901.

8. Y. Y. Li, Existence of many positive solutions of semilinear elliptic equations on annulus, J. Differential Equations 83 (1990), 348-367.

9. S. S. Lin, On non-radially symmetric bifurcation in the annulus, J. Differential Equations

80(1989), 251-279.

10. _, Positive radial solutions and non-radial bifurcation for semilinear elliptic equations in annular domains, J. Differential Equations 86 (1990), 367-391.

11. _, Existence of positive nonradial solutions for elliptic equations in annular domains. Trans. Amer. Math. Soc. (to appear).

12. _, Symmetry breaking for semilinear elliptic equations on sectorial domains in R2 , Proc.

Roy. Soc. Edinburgh Sect. A 118 (1991), 327-353.

13. Z. Nehari, On a class of nonlinear second order differential equations. Trans. Amer. Math.

Soc. 95(1960), 101-123.

14. W.-M. Ni, Some aspects of semilinear elliptic equations, lecture notes, National Tsing Hua University, Hsinchu, Taiwan, May 1987.

15. M. Ramaswamy and P. N. Srikanth, Symmetry breaking for a class of semilinear elliptic problems, Trans. Amer. Math. Soc. 304 (1987), 839-845.

16. J. Smoller and A. Wasserman, Symmetry-breaking for solutions of semilinear elliptic equa-tions, Arch. Rational Mech. Anal. 95 (1986), 217-225.

(9)

17. _, Symmetry-breaking for solutions of semilinear elliptic equations with general boundary

conditions, Comm. Math. Phys. 105 (1986), 415-441.

18. T. Suzuki and K. Nagasaki, Lifting of local subdifferentiations and elliptic boundary value problems on symmetry domains. I, Proc. Japan Acad. 64 (1988), 1-4.

19. _, On the nonlinear eigenvalue problem Am + Xe" = 0 , Trans. Amer. Math. Soc. 309

(1988), 591-608.

Department of Applied Mathematics, National Chiao Tung University, Hsin-chu, Taiwan, Republic of China

參考文獻

相關文件

Wang, Unique continuation for the elasticity sys- tem and a counterexample for second order elliptic systems, Harmonic Analysis, Partial Differential Equations, Complex Analysis,

Quantitative uniqueness of solutions to second order elliptic equations with singular lower order terms.. Quantitative uniqueness of solutions to second order elliptic equations

† Institute of Applied Mathematical Sciences, NCTS, National Taiwan University, Taipei 106, Taiwan.. It is also important to note that we obtain an inequality with exactly the

In this paper we prove a Carleman estimate for second order elliptic equa- tions with a general anisotropic Lipschitz coefficients having a jump at an interface.. Our approach does

This paper presents (i) a review of item selection algorithms from Robbins–Monro to Fred Lord; (ii) the establishment of a large sample foundation for Fred Lord’s maximum

Abstract In this paper, we consider the smoothing Newton method for solving a type of absolute value equations associated with second order cone (SOCAVE for short), which.. 1

Chen, Conditions for error bounds and bounded level sets of some merit func- tions for the second-order cone complementarity problem, Journal of Optimization Theory and

Due to the important role played by σ -term in the analysis of second-order cone, before developing the sufficient conditions for the existence of local saddle points, we shall