• Tidak ada hasil yang ditemukan

Directory UMM :Journals:Journal_of_mathematics:GMJ:

N/A
N/A
Protected

Academic year: 2017

Membagikan "Directory UMM :Journals:Journal_of_mathematics:GMJ:"

Copied!
6
0
0

Teks penuh

(1)

Volume 10 (2003), Number 3, 543–548

ON THE ASYMPTOTICS OF SOLUTIONS OF ELLIPTIC EQUATIONS IN A NEIGHBORHOOD OF A CRACK WITH

NONSMOOTH FRONT

V. A. KONDRAT’EV AND V. A. NIKISHKIN

Abstract. Two terms of asymptotics near crack are obtained for solutions of the Dirichlet boundary value problem for second-order elliptic equations in divergent form. The front of a crack is fromC1+s

and the coefficients of the equations belong toCs

(0.5< s <1).

2000 Mathematics Subject Classification: 35J25.

Key words and phrases: Asymptotics, solution of elliptic equation, crack, nonsmooth front, Dirichlet boundary value problem, divergent form.

A second-order elliptic equation in G\Ω is considered, where G ⊂ Rn is a domain with a smooth boundary, Ω is an (n−1)-dimensional manifold with a boundary fromC1+s, 0< s <1.

We study solutions of the equation n

X

i,j=1

∂ ∂xi

µ aij(x)

∂u ∂xj

= 0, x∈G\Ω, (1)

that belong to the Sobolev space W21(G) and satisfy the condition

u(x) = 0, x∈Ω. (2)

Let L be the boundary of Ω, r be the distance from x to L, xL be a point on L nearest to x, ϕ be the polar angle in the plane normal to L and passing through xL.

Singularities of solutions of elliptic equations near a nonsmooth boundary were studied by many authors, see, e.g., [1]–[4].

In [4], the following representation was obtained for the plane (n− 1)-dimen-sional domain Ω:

u(x) =C(xL)r1/2Φ(ϕ) +u1(x), where Φ(ϕ) is a smooth function,

|u1| ≤C0r1/2+ε, 0< ε <min{s,0.5},

|C(xL)|+|C0| ≤const· kukW1 2(G). The following result is the main theorem of our paper.

Theorem 1. Let u(x)∈W21(G) be a solution of problem (1), (2), and let aij ∈Cs(G) (i, j = 1, . . . , n), Ω∈C1+s,

where 0< s <1.

(2)

Then:

if 0< s≤0.5, then

u(x) =C(xL)r1/2Φ(ϕ) +u1(x),

where Φ(ϕ) is a smooth function,

|u1| ≤C0r1/2+ε, 0< ε < s,

|C(xL)|+|C0| ≤const· kukW1 2(G);

if 0.5< s <1, then

u(x) =C1(xL)r1/2Φ1(ϕ) +C2(xL)rΦ2(ϕ) +u1(x),

where Φ1(ϕ), Φ1(ϕ) are smooth functions

|u1| ≤C0r1+ε, 0< ε < s−0.5,

|C1(xL)|+|C2(xL)|+|C0| ≤const· kukW1 2(G).

Straightening of the boundary. Let P be an arbitrary point of the set L. Consider a neighborhood U of P in which Ω admits a one-to-one projection to the tangent plane. Assume that P is the origin and that in this neighborhood we have

Ω ={x|x1 =F(x2, . . . , xn), (x2, . . . , xn)∈Ω1}, O ∈∂Ω1.

∂Ω1 is given in a neighborhood of the origin by the equation

x2 =h(x3, . . . , xn)∈C1+s.

Let us extend F(x2, . . . , xn) ∈ C1+s in a neighborhood of the origin so that the class of smoothness be prescribed. Let F(0),∇F(0) = 0.

Introduce an averaging kernelK(τ) such thatK(τ)∈C∞

(R1),K(τ) is even,

K(τ)≡0 for |τ |≥1, and

1

Z

1

K(τ)dτ = 1.

The straightening of the boundary consists of two steps. The first transformation of the coordinates has the form:

x1 =x ′

1+H(x ′

), x2 =x ′

2, . . . , xn=x ′ n,

where

H(x′ ) =

Z

Rn−1

F(t) n

Y

l=2

µ

1 |x1|

K µ

tl−zl′

|x1|

¶¶ dt.

The second transformation of the coordinates is the same as that in [4].

(3)

Dirichlet problem in a dihedral angle for an equation with constant coefficients. Let G0 be a dihedral angle

G0 ={x|0< x21+x22 <∞, 0< ϕ < ω},

where ϕ is the polar angle in the plane (x1, x2). Set ρ= v u u t n X i=1 x2

i , r ′

=

p x2

1+x22

ρ .

We need the weighted Sobolev spaces ˙Wα,β0 and ˙Wα,β1 in which the norms are defined as follows:

kuk2W˙0 α,β =

Z

G0

u2ρα(r′ )βdx,

kuk2W˙1 α,β =

Z

G0

ρα(r′

)βgrad2udx+

Z

G◦

u2ρα−2 (r′

)β−2

dx.

Letu(x)∈W1

2(G0) be a generalized solution (here and below, in the sense of distributions) of the following Dirichlet problem:

∆u(x) = f0(x) + n

X

i=1

∂fi(x)

∂xi

, x∈G0, (3)

u(x) = 0, x∈∂G0. (4)

The following two assertions can be proved by the method developed in [1].

Theorem 2. Let u(x) ∈ W1

2(G0) be a generalized solution of problem (3), (4), where

f0 ∈W˙α,β0 (G0), fi ∈W˙α0−2,β2(G0) (i= 1, . . . , n),

α+ 2³π

ω −2 ´

+n >0, β+ 2³π

ω −2 ´

+n−1>0.

Then u∈W˙ 1

α−2,β2(G0).

Theorem 3. Let u(x) ∈ W21(G0) be a generalized solution of problem (3), (4), where

f0 ∈W˙α,β0 (G0), fi ∈W˙α0−2,β−2(G0) (i= 1, . . . , n),

α+ 2

µ

2π ω −2

+n <0, β+ 2³π

ω −2 ´

+n−1>0,

α+ 2

µ

3π ω −2

+n >0.

Then u(x) can be represented in the form

u(x) = C1ρ

π

ωΦ1(θ) +C2ρ 2π

(4)

where θ are the coordinates on the unit sphere, Φ1(θ),Φ1(θ) are the

eigenfunc-tions of the Beltrami operator, and

u1 ∈W˙α1−2,β2(G0).

Dirichlet problem in a dihedral angle for an equation with variable coefficients. Let u(x) ∈ W1

2(G0) be a generalized solution of the Dirichlet problem

n

X

i,j=1

∂ ∂xi

µ aij(x)

∂u(x)

∂xj

=f0(x) + n

X

i=1

∂fi(x)

∂xi

, x∈G0, (5)

u(x) = 0, x∈∂G0, (6)

where aij ∈Cs(G0), aij(0) =δij (without loss of generality) (i, j = 1, . . . , n).

Lemma 1. Let u(x)∈W21(G0)be a generalized solution of problem (5), (6),

where

f0 ∈W˙20−2ks+ε0,2(G0), fi ∈W˙

0

2ks+ε0,0(G0) (i= 1, . . . , n),

2−2ks+ε0+ 2

³π

ω −2 ´

+n >0, 2−2(k+ 1)s+ε0+ 2

³π

ω −2 ´

+n <0,

k is a nonnegative integer, and ε0 >0 is sufficiently small.

Then

u∈W˙−12ks+ε0,0(G0).

The proof of Lemma 1 follows from Theorem 2 by induction onk1 (0≤k1 ≤

k) and is based on the representation of equation (5) in the form

∆u(x) =f0(x) + n

X

i=1

∂ ∂xi

Ã

fi(x)− n

X

j=1

(aij(x)−δij)

∂u ∂xj

! .

Using this representation, Lemma 1, and Theorem 3, we obtain the following assertion.

Lemma 2. Let u(x)∈W21(G0)be a generalized solution of problem (5), (6),

where

f0 ∈W˙α,20 (G0), fi ∈W˙α0−2,0(G0) (i= 1, . . . , n),

α=−4πω −n+ 4−ε1, 0< ε1 <2s−2πω.

Then u(x) can be represented in the form

u(x) = C1ρ

π ωΦ

1(θ) +C2ρ

2π ωΦ

2(θ) +u1,

where θ are the coordinates on the unit sphere, Φ1(θ),Φ1(θ) are the

eigenfunc-tions of the Beltrami operator, and

u1 ∈W˙α1−2,0(G0).

(5)

Bounds for |u1|. Consider the cones K and Kb

K ={x|x23+c· · ·+xn2 ≥k2(x21 +x22)}, b

K ={x|x23+· · ·+x2n≥bk2(x21+x22)} and the domains

G1 =G0\K, Gb1 =G0\K.b Obviously, Gb1 ⋐G1 forbk < k.

Lemma 3. Suppose that, in addition to the assumptions of Lemma 2, the following inequalities hold in the domain G1:

|f0(x)| ≤C0ρ

ω−2+ε, |fi(x)| ≤C0ρ 2π

ω−1+ε (i= 1. . . , n),

and let f0(x) and fi(x) be continuous in G1, 0< ε < s−ωπ.

Then

|u1| ≤C1ρ

2π ω+ε,

|gradu1(x)| ≤Cρ

2π ω−1+ε

in Gb1.

The proof of Lemma 3 is the same as that in [4].

Remark 2. The proof of Theorem 1 is obtained on the basis of the above assertions.

Acknowledgement

The work was carried out according to Project No. 01-00 of A. M. Lyapunov French-Russian Institute.

References

1. V. A. Kondrat’ev, Boundary value problems for elliptic equations in domains with conical or angular points. (Russian)Trudy Moscow. Mat. Obshch.16(1967), 209–292. 2. G. M. Verzhbinski˘ıandV. G. Maz’ja,Asymptotic behavior of the solutions of second

order elliptic equations near the boundary. II. (Russian)Sibirsk. Mat. Zh.13(1972), 1239– 1271.

3. V. A. Kondrat’ev, I. Kopachek, and O. A. Oleinik, Best H¨older exponents for generalized solutions of the Dirichlet problem for a second-order elliptic uquation. Mat. Sb.131(1986), 113–125.

4. V. A. Kondrat’evandV. A. Nikishkin,On the behavior of solutions of elliptic equa-tions in a neighborhood of a crack with nonsmooth front.Russian J. Math. Phys.9(2002), No. 1, 61–65.

(6)

Authors’ addresses: V. A. Kondrat’ev

Department of Mechanics and Mathematics Moscow State University,

Vorob’evy gory, Moscow 119899 Russia

V. A. Nikishkin

Department of Mathematics

Moscow State University of Economics, Statistics and Informatics, 7, Nezhinskaya, Moscow 119501

Referensi

Dokumen terkait

Lomtatidze, On certain boundary-value problems for second-order linear ordinary differential equations with singu- larities.. Lomtatidze, On positive solutions of

Lomtatidze , On positive solutions of boundary value problems for second order ordinary differential equations with singularities.. Malaguti , On a nonlocal boundary value problem

Yang, Multiple symmetric positive solutions of a class of boundary value problems for higher order ordinary differential equations, Proc.. Yang, A three point boundary value problem

The study of positive radial solu- tions for elliptic problems in annular regions can usually be transformed into that of positive solutions of two-point boundary value problems

Tian, Positive solutions of m-point boundary value problems for higher order ordinary differential equations, Nonlinear Anal.. Kong, Higher order boundary value problems

Abstract : In this paper, we prove the existence of positive solutions for Floquet boundary value problem concerning fractional functional differential equations with bounded

That is, some authors have used fixed point theorems to show the existence of pos- itive solutions to boundary value problems for ordinary differential equations, difference

In this paper, we shall establish sufficient conditions for the existence of solutions for a first order boundary value problem for fractional differential equations..