Upload
jinyuan
View
212
Download
0
Embed Size (px)
Citation preview
Acta Mathematica Scientia 2014,34B(1):125–140
http://actams.wipm.ac.cn
STABILITY OF DISPLACEMENT TO THE
SECOND FUNDAMENTAL PROBLEM IN
PLANE ELASTICITY∗
Juan LIN (��)1,2 Jinyuan DU (���)1†
1. School of Mathematics and Statistics, Wuhan University, Wuhan 430072, China
2. Department of Foundation, Fujian Commercial College, Fuzhou 350012, China
E-mail: [email protected]; [email protected]
Abstract In this article, by using the stability of Cauchy type integral when the smooth
perturbation for integral curve and the Sobolev type perturbation for kernel density hap-
pen, we discuss the stability of the second fundamental problem in plane elasticity when
the smooth perturbation for the boundary of the elastic domain (unit disk) and the Sobolev
type perturbation for the displacement happen. And the error estimate of the displacement
between the second fundamental problem and its perturbed problem is obtained.
Key words elastic domain; Cauchy type integral; displacement; perturbation
2010 MR Subject Classification 30E20; 30E25; 45E99
1 Introduction
It is well known that complex variable method is one of the effective methods for solving
various problems in plan elasticity mechanics (see, e.g. [1–4]). By introducing two analytic
functions, the fundamental problem can be transferred to a boundary value problem for analytic
function and then further reduced to a integral equation called Sherman-Lauricella equation.
When the elastic domain is a unit disk, we can get the closed form of solution for Sherman-
Lauricella equation. In general case, it is difficult to get the closed form of solution for Sherman-
Lauricella equation. So, the research on the stability of the second fundamental problem on the
elastic domain aroused the interest of all. In such discussion, the theory for boundary value of
the Cauchy type integral and when the smooth perturbation for integral curve and the Sobolev
type perturbation for the kernel density happen the stability of the Cauchy type integral play
an important role (see, e.g. [5–9]).
In [10], the authors discussed the stability of the second fundamental problem when the
smooth perturbation for the boundary circle of the elastic domain disk and the Sobolev type
∗Received September 24, 2012; revised January 25, 2013. This work was supported by NNSF of China
(11171260), RFDP of Higher Education of China (20100141110054), NSF of Fujian Province, China (2008J0187)
and STF of Education Department of Fujian Province, China (JA11341).†Corresponding author: Jinyuan DU.
126 ACTA MATHEMATICA SCIENTIA Vol.34 Ser.B
perturbation for the displacement given in the boundary circle happen by the results in [7]. But
the discussion in [10] was based on a closed set in the elastic domain and displacement with the
trivial perturbation. In the present article, we remove the limitation of a closed set and obtain
a better conclusion under the displacement with the Sobolev perturbation.
We assume that the elastic domain S is a bounded simply connected region with boundary
Γ, a smooth closed contour, oriented counter-clockwisely. And let O = (0, 0) ∈ S. Assume
all functions (and their derivatives) given in S or on Γ are Holder continuous. We introduce
holomorphic functions ϕ(z) and ψ(z), also called complex stress functions. The displacement
u+ iv at the point z in S could be expressed by ϕ(z) and ψ(z) [1, 2]
2μ(u+ iv) = κϕ(z)− zϕ′(z)− ψ(z), z ∈ S, (1.1)
where κ, μ are the elastic constants and 1 < κ < 3. So the second fundamental problem about
S can be reduced to the boundary problem for analytic functions [1, 2]: for given displacement
function g(t) = u(t) + iv(t) (t ∈ Γ), to find two holomorphic functions ϕ(z) and ψ(z) in S,
satisfying the boundary value condition
κ ϕ(t)− tϕ′(t)− ψ(t) = 2μg(t), t ∈ Γ (1.2)
and
ϕ(0) = 0 (or ψ(0) = 0). (1.3)
We introduce a new unknown function ω(t) defined on Γ. Let
ϕ(z) =1
2πi
∫Γ
ω(t)
t− zdt, z ∈ S, (1.4)
ψ(z) = − κ
2πi
∫Γ
ω(t)
t− zdt− 1
2πi
∫Γ
t ω′(t)
t− zdt, z ∈ S. (1.5)
By (1.4), we can obtain
ϕ′(z) =1
2πi
∫Γ
ω(t)
(t− z)2dt =
1
2πi
∫Γ
ω′(t)
t− zdt, z ∈ S. (1.6)
According to the Plemelj formula and (ϕ+)′(t) = (ϕ′)+(t), we can get ϕ(t), ϕ′(t), ψ(t) from
(1.4), (1.6) and (1.5), substituting them into (1.2), we gain the following integral equation [1, 2]
which be called the Sherman-Lauricella equation
κ ω(t) +κ
2πi
∫Γ
ω(τ) d
[ln
τ − t
τ − t
]+
1
2πi
∫Γ
ω(τ) d
[τ − t
τ − t
]= 2μg(t), t ∈ Γ. (1.7)
Remark 1.1 It is true that the ϕ and ψ given in (1.4) and (1.5) satisfy (1.2) while ω is
the solution of the Sherman-Lauricella equation (1.7) and ω′ ∈ H(Γ).
Assume Γ is the unit circle L : t = eiθ, 0 ≤ θ ≤ 2π, then S is the unit disk U = {z : |z| <1}. Under this case, we have
ττ = 1,τ − t
τ − t= −τt, d
[τ − t
τ − t
]= −t dτ, d
[ln
τ − t
τ − t
]=1
τdτ, while τ, t ∈ L, (1.8)
No.1 J. Lin & J.Y. Du: STABILITY OF DISPLACEMENT TO FUNDAMENTAL PROBLEM 127
so, by (1.8), equation (1.7) (Γ = L) can be changed into
ω(t) = a+ bt+2μ
κg(t), (1.9)
where
a = − μ
2κπi
∫L
g(τ)
τdτ, b =
2μ
κ2 − 1
[1
2πi
∫L
g(τ)dτ − 1
2κπi
∫L
g(τ)dτ
]. (1.10)
Remark 1.2 (1.9) show that g ∈ C(n)(L) is equivalent to ω ∈ C(n)(L) and g(n) ∈ H(L)
is equivalent to ω(n) ∈ H(L).
Remark 1.3 (1.9) and (1.10) yield∣∣a∣∣ ≤ μ
κ
∥∥g∥∥L,
∣∣b∣∣ ≤ 2μ
κ(κ− 1)
∥∥g∥∥L, (1.11)
and ∥∥ω∥∥L≤ (3κ− 1)μ
κ(κ− 1)
∥∥g∥∥L,
∥∥ω′∥∥L≤ 2μ
κ(κ− 1)
∥∥g∥∥L+2μ
κ
∥∥g′∥∥L, (1.12)
where ‖ · ‖L = maxL| · |.
2 Perturbation of the Circle
If the function δ(τ) (τ ∈ L) has the continuous derivative on L up to order n, we denote
it by δ ∈ Cn(L) and call it a n-smooth perturbation of the circle L. If δ ∈ Cn(L), its Sobolev
norm reads ∥∥δ∥∥n= max
{∥∥δ∥∥L,∥∥δ′∥∥
L, · · · ,
∥∥δ(n)∥∥L
}, (2.1)
where∥∥δ(k)∥∥
L= max
{∣∣δ(k)(τ)∣∣, τ ∈ L}(k = 0, 1, · · · , n) denotes the Chebyshev norm of the
function δ(k) defined on the curve L.
We set
σ(t) = t+ δ(t), t ∈ L, (2.2)
then
Lδ : λδ(θ) = σ(eiθ)= eiθ + δ
(eiθ), 0 ≤ θ ≤ 2π (2.3)
is also a simple and closed smooth contour under a small perturbation δ while ‖δ‖1 < 2π. And
we also call Lδ the perturbation of L with δ. Let Uδ be the region bounded by Lδ. With the
length of minor arc between t1 and t2 denoted by [t1, t2]Lminarc, then
2
π
[t1, t2
]Lminarc
≤∣∣t1 − t2
∣∣ ≤ [t1, t2]Lminarc for any t1, t2 ∈ L (2.4)
by the Newton-Leibniz formula, we get
∣∣δ(t1)− δ(t2)∣∣ = ∣∣∣∣ ∫[ �
t1t2
]L
minarc
δ′(τ)dτ
∣∣∣∣ ≤ ∥∥δ′∥∥L
[t1, t2
]Lminarc
≤ π
2
∥∥δ∥∥1
∣∣t1 − t2∣∣, (2.5)
so we have ∣∣σ(t1)− σ(t2)∣∣ ≥ [1− π
2
∥∥δ∥∥1
] ∣∣t1 − t2∣∣ while
∥∥δ∥∥1
<2
π. (2.6)
128 ACTA MATHEMATICA SCIENTIA Vol.34 Ser.B
This inequality shows that σ(t) is 1-1 continuous mapping from L to Lδ, namely, there exist
ξ = σ(t), t ∈ L (2.7)
and
t = σ−1(ξ), ξ ∈ Lδ, (2.8)
which are inverse to each other.
We denote the interior region with the boundary curve Lδ as Uδ. The second fundamental
problem about Uδ can be reduced to the boundary problem for analytic functions: to find two
holomorphic functions ϕ∗(z) and ψ∗(z) in Uδ, satisfying the boundary value condition
κϕ∗(ξ) − ξϕ′∗(ξ)− ψ∗(ξ) = 2μg∗(ξ), ξ ∈ Lδ. (2.9)
As before, we introduce a new function ω∗(ζ), such that
ϕ∗(z) =1
2πi
∫Lδ
ω∗(ζ)
ζ − zdζ, z ∈ Uδ, (2.10)
ψ∗(z) = − κ
2πi
∫Lδ
ω∗(ζ)
ζ − zdζ − 1
2πi
∫Lδ
ζω′∗(ζ)
ζ − zdζ, z ∈ Uδ. (2.11)
By (2.10) and (2.11), (2.9) can be changed into the integral equation on Lδ
κ ω∗(ξ) +κ
2πi
∫Lδ
ω∗(ζ) d
[ln
ζ − ξ
ζ − ξ
]+
1
2πi
∫Lδ
ω∗(ζ) d
[ζ − ξ
ζ − ξ
]= 2μg∗(ξ), ξ ∈ Lδ, (2.12)
which is the Sherman-Lauricella equation of (2.9). Denoting
ω∗(ξ) = ω∗(σ(t)) = ωδ(t), g∗(ξ) = g∗(σ(t)) = gδ(t), (2.13)
(2.12) can be expressed as
κωδ(t) +κ
2πi
∫L
ωδ(τ)d
[ln
σ(τ) − σ(t)
σ(τ) − σ(t)
]+
1
2πi
∫L
ωδ(τ)d
[σ(τ) − σ(t)
σ(τ) − σ(t)
]= 2μgδ(t), t ∈ L,
(2.14)
then
κ ωδ(t) +κ
2πi
∫L
ωδ(τ) d
[ln
τ − t
τ − t
]+
1
2πi
∫L
ωδ(τ) d
[τ − t
τ − t
]= 2μgδ(t) +G(t), t ∈ L, (2.15)
where
G(t) =κ
2πi
∫L
ωδ(τ) d
⎡⎣ln (τ − t) (
σ(τ) − σ(t))
(τ − t
)(σ(τ) − σ(t)
)⎤⎦+ 1
2πi
∫L
ωδ(τ) d
[τ − t
τ − t− σ(τ) − σ(t)
σ(τ) − σ(t)
].
(2.16)
Remark 2.1 G is called the term of perturbation for equation (2.15). In the sequel, we
will see that its stability play an important role.
From (2.15) and (1.8), we get
ωδ(t) = aδ + bδt+2μ
κgδ(t) +
1
κG(t), (2.17)
No.1 J. Lin & J.Y. Du: STABILITY OF DISPLACEMENT TO FUNDAMENTAL PROBLEM 129
where
aδ = − μ
2κπi
∫L
gδ(τ)
τdτ − 1
4κπi
∫L
G(τ)
τdτ, (2.18)
bδ =2μ
κ2 − 1
[1
2πi
∫L
gδ(τ)dτ −1
2κπi
∫L
gδ(τ)dτ
]+
1
κ2 − 1
[1
2πi
∫L
G(τ)dτ − 1
2κπi
∫L
G(τ)dτ
]. (2.19)
Remark 2.2 By (2.17), we get
ω′∗(ξ) =bδ
1 + δ′(t)+2μ
κg′∗(ξ) +
G′(t)
κ(1 + δ′(t)), t = σ−1(ξ), ξ ∈ Lδ, (2.20)
and we will prove that ω′∗ ∈ H(Lδ) while g′∗ ∈ H(Lδ) and δ(3) ∈ C(L) in Section 4.
3 Stability of the Term of Perturbation
Suppose that Γ is a smooth curve and f ∈ Cn(Γ). We set its difference quotient function
of n-th order by
(Dn[f ]
)(τ, t) =
⎧⎪⎪⎪⎨⎪⎪⎪⎩1
(τ − t)n
[f(τ)−
n−1∑j=0
f (j)(t)
j!(τ − t)j
], τ �= t,
f (n)(t)
n!, τ = t,
(τ, t) ∈ Γ× Γ. (3.1)
Lemma 3.1 [11] If f ∈ Cn+m(Γ), Γ is smooth, its natural equation is t = φ(s), t ∈Γ, s ∈ [0, �Γ], �Γ is the length of Γ, then Dn[f ] ∈ Cm(Γ × Γ). More precisely, all its partial
derivatives up to order m inclusive are continuous and independent of the order of τ and t,
∥∥(Dn[f ])(m)∥∥
Γ×Γ≤ 2m
n!
[1
CΓ
]n+m ∥∥f (n+m)∥∥Γ, (3.2)
ω((
Dn[f ])(m)
;h)≤[1
CΓ
]n+m+1 [2m
n!ω
(f (n+m);
h
CΓ
)+
2m+2
(n− 1)!
∥∥f (n+m)∥∥Γω
(φ′;
h
CΓ
)],
(3.3)
where CΓ is a chord-arc ratio of Γ.
Remark 3.1 This lemma generalizes and improves the results in some earlier literatures
[12–15]. In particular, if f (n) ∈ H(Γ), φ′ ∈ H , then Dn[f ] ∈ H(Γ× Γ).
Lemma 3.2 [11] If f ∈ Cn+1(Γ) and Γ is smooth, then
∂Dn[f ]
∂τ= Dn[f
′]− nDn+1[f ],∂Dn[f ]
∂t= nDn+1[f ]. (3.4)
Lemma 3.3 [11] Suppose that h is defined in the unit circle L. If h′(t) exists, then[h(t)
]′= −t2h′(t),
[Re[h(t)
]]′= i t Im
[h′(t)t
],[Im[h(t)
]]′= −i tRe
[h′(t)t
]. (3.5)
If h′′(t) exists, then[Re[h(t)
]]′′= t3 h′(t) + t2Re
[h′′(t)t2
],[Im[h(t)
]]′′= it3 h′(t) + t2Im
[h′′(t)t2
]. (3.6)
130 ACTA MATHEMATICA SCIENTIA Vol.34 Ser.B
By Lemma 3.2 and Lemma 3.3, we have
d
[ln(τ − t)(σ(τ) − σ(t))
(τ − t)(σ(τ) − σ(t))
]= d
[ln(D1[σ]
)(τ, t)− ln
(D1[σ]
)(τ, t)
]= τ G1(τ, t) dτ, (3.7)
where
G1(τ, t) = 2Re
[(D2[δ]
)(τ, t)−
(D1[δ
′])(τ, t)
1 +(D1[δ]
)(τ, t)
τ
], (3.8)
and
d
[τ − t
τ − t− σ(τ) − σ(t)
σ(τ) − σ(t)
]= t
[(D1[σ]
)(τ, t)(
D1[σ])(τ, t)
− 1
]dτ + τtd
[(D1[σ])(τ, t)(D1[σ]
)(τ, t)
](3.9)
with (D1[σ]
)(τ, t)
(D1[σ])(τ, t)− 1 =
(D1[δ])(τ, t) − (D1[δ])(τ, t)
1 +(D1[δ]
)(τ, t)
=: G2(τ, t), (3.10)
d
[(D1[σ])(τ, t)
(D1[σ])(τ, t)
]=(D1[σ])(τ, t)
(D1[σ])(τ, t)d
[ln(D1[σ])(τ, t)
(D1[σ])(τ, t)
]= −τ G1(τ, t)
(G2(τ, t) + 1
)dτ. (3.11)
Lemma 3.4 Suppose that G is given in (2.16). If δ ∈ C2(L) and∥∥δ′∥∥
L≤ ρ < 2
πthen∥∥G∥∥
L≤ M0
∥∥ωδ
∥∥L
∥∥δ∥∥2
with M0 =2(2κ+ 3)π
2− πρ. (3.12)
If δ ∈ C3(L) and∥∥δ∥∥
2≤ ρ < 2
πthen∥∥G′∥∥
L≤ M1
∥∥ωδ
∥∥L
∥∥δ∥∥3
(3.13)
with
M1 =1
2− πρ
[π3
6(κ+ 1) +
π2
2(κ+ 2) + 6π
]+(κ+ 3)π3ρ
[2− πρ]2 . (3.14)
If δ ∈ C4(L) and∥∥δ∥∥
2≤ ρ < 2
πthen∥∥G′′∥∥
L≤ M2
∥∥ωδ
∥∥L
∥∥δ∥∥4
(3.15)
with
M2 =π2
2− πρ
[(1 + κ)π2
16+(5 + 2κ)π
6+
κ+ 7
2
]+(6 + κ)π5ρ2
2 [2− πρ]3
+π3ρ
[2− πρ]2
[(3 + κ)π2
12+(51 + 14κ)π
24+ 11 + κ
]. (3.16)
Proof By (3.7) and (3.9), we can rewrite (2.16) as the following
G(t) = κ I1(t) + t I2(t)− t I3(t), (3.17)
where
I1(t) =1
2πi
∫L
ωδ(τ)τ G1(τ, t)dτ, (3.18)
I2(t) =1
2πi
∫L
ωδ(τ)G2(τ, t)dτ. (3.19)
I3(t) =1
2πi
∫L
ωδ(τ)(G2(τ, t) + 1)G1(τ, t)dτ. (3.20)
No.1 J. Lin & J.Y. Du: STABILITY OF DISPLACEMENT TO FUNDAMENTAL PROBLEM 131
Let
Υ(τ, t) =(D2[δ])(τ, t)− (D1[δ
′])(τ, t)
(D1[σ])(τ, t), (3.21)
by (3.2) and ‖δ′‖L ≤ ρ, we have
∥∥Υ∥∥L×L
≤ 2
2− π‖δ′‖L
[π
2+(π
2
)2 12!
]∥∥δ′′∥∥L≤ 2π
2− πρ
∥∥δ′′∥∥L, (3.22)
and by (3.8), we get ∥∥G1
∥∥L×L
≤ 4π
2− πρ
∥∥δ′′∥∥L, (3.23)
which results in ∥∥I1∥∥L≤ 4π
2− πρ
∥∥ωδ
∥∥L
∥∥δ′′∥∥L,∥∥I3∥∥L
≤ 4π
2− πρ
∥∥ωδ
∥∥L
∥∥δ′′∥∥L. (3.24)
(3.10) and (3.2) imply ∥∥G2
∥∥L×L
≤ 2π
2− πρ
∥∥δ′∥∥L, (3.25)
which results in ∥∥I2∥∥L≤ 2π
2− πρ
∥∥ωδ
∥∥L
∥∥δ′∥∥L. (3.26)
(3.17), (3.24) and (3.26) yield (3.12).
By (3.21) and (3.4), we obtain
∂Υ
∂t(τ, t) =
2(D3[δ]
)(τ, t)−
(D2[δ
′])(τ, t)−
(D2[δ]
)(τ, t)Υ(τ, t)
1 +(D1[δ]
)(τ, t)
, (3.27)
while ‖δ‖2 ≤ ρ, (3.22) and (3.2) result in∥∥∥∥∂Υ
∂t
∥∥∥∥L×L
≤ 1
2
[π2
2− π‖δ′‖L
(π
6+1
2
)+
π3
(2− π‖δ′‖L)2
∥∥δ′′∥∥L
]∥∥δ∥∥3≤ 1
2C1
∥∥δ∥∥3, (3.28)
where
C1 =π2
2− πρ
[π
6+1
2
]+
π3ρ
[2− πρ]2. (3.29)
By (3.5), we have ∥∥∥∥∂G1
∂t
∥∥∥∥L×L
≤ C1
∥∥δ∥∥3, (3.30)
so, we get ∥∥I ′1∥∥L≤ C1
∥∥ωδ
∥∥L
∥∥δ∥∥3. (3.31)
From (3.10)
∂G2
∂t(τ, t) =
∂
∂t
{(D1[σ])(τ, t)
(D1[σ])(τ, t)
}=
t(D1[σ])(τ, t)
(D1[σ])(τ, t)2Re
[(D2[δ])(τ, t)
1 + (D1[δ])(τ, t)t
], (3.32)
by (3.2) and ‖δ′‖L ≤ ρ, we get∥∥∥∥∂G2
∂t
∥∥∥∥L×L
≤ π2
2(2− π‖δ′‖L
)∥∥∥∥δ′′∥∥∥∥L
≤ C2
∥∥∥∥δ′′∥∥∥∥L
, (3.33)
132 ACTA MATHEMATICA SCIENTIA Vol.34 Ser.B
where
C2 =π2
2(2− πρ
) . (3.34)
Thus, ∥∥I ′2∥∥L≤ C2
∥∥ωδ
∥∥L
∥∥δ′′∥∥L. (3.35)
(3.20), (3.23), (3.30) and (3.33) result in∥∥I ′3∥∥L≤ C3
∥∥ωδ
∥∥L
∥∥δ∥∥3, (3.36)
where
C3 =π2
2− πρ
[π
6+1
2
]+
3π3ρ
[2− πρ]2. (3.37)
By (3.17), (3.24), (3.26), (3.31), (3.35) and (3.36), we get
∥∥G′∥∥L≤[κ C1 + C2 + C3 +
6π
2− πρ
] ∥∥ωδ
∥∥L
∥∥δ∥∥3= M1
∥∥ωδ
∥∥L
∥∥δ∥∥3, (3.38)
where M1 is just that given in (3.14).
By (3.27), we have
∂2Υ
∂t2=6D4[δ]− 2D3[δ
′]− 2D3[δ]Υ
1 +D1[δ]− 2D2[δ]
1 +D1[δ]
∂Υ
∂t. (3.39)
(3.22), (3.28), (3.2) and (3.39) result in∥∥∥∥∂2Υ
∂t2
∥∥∥∥L×L
≤ 1
2C∗4∥∥δ∥∥
4while ‖δ‖2 ≤ ρ, (3.40)
where
C∗4 =π3
2− πρ
[π
16+1
6
]+
π4ρ
[2− πρ]2
[π
12+
7
12
]+
π5ρ2
2 [2− πρ]3 . (3.41)
Thus, by (3.6), (3.28) and (3.40),∥∥∥∥∂2G1
∂t2
∥∥∥∥L×L
≤ 2
[∥∥∥∥∂Υ
∂t
∥∥∥∥L×L
+
∥∥∥∥∂2Υ
∂t2
∥∥∥∥L×L
]≤ C4‖δ‖4, (3.42)
where
C4 =π2
2− πρ
[π2
16+
π
3+1
2
]+
π3ρ
[2− πρ]2
[π2
12+7π
12+ 1
]+
π5ρ2
2 [2− πρ]3. (3.43)
By (3.18) and (3.42), we get ∥∥I ′′1 ∥∥L≤ C4
∥∥ωδ
∥∥L
∥∥δ∥∥4. (3.44)
Let
Λ(τ, t) =(D2[δ])(τ, t)
(D1[σ])(τ, t), (3.45)
then, by Lemma 3.2,
∂Λ
∂t(τ, t) = − [(D2[δ])(τ, t)]
2
[(D1[σ])(τ, t)]2 +
2(D3[δ])(τ, t)
(D1[σ])(τ, t)= −
[Λ(τ, t)
]2+2(D3[δ])(τ, t)
(D1[σ])(τ, t), (3.46)
No.1 J. Lin & J.Y. Du: STABILITY OF DISPLACEMENT TO FUNDAMENTAL PROBLEM 133
and by (3.2), we get∥∥∥∥∂Λ
∂t
∥∥∥∥L×L
≤[
π3
12 (2− π‖δ′‖L)+
π4
16 (2− π‖δ′‖L)2
∥∥δ′′∥∥L
]∥∥δ∥∥3≤ C∗5
∥∥δ∥∥3
while ‖δ‖2 ≤ ρ,
(3.47)
where C∗5 =π3
12[2−πρ] +π4ρ
16[2−πρ]2. By (3.32), (3.5), (3.33), (3.47) and (3.2), we obtain∥∥∥∥∂2G2
∂t2
∥∥∥∥L×L
≤ C5
∥∥δ∥∥3, (3.48)
where
C5 =π2
2− πρ
[π
6+1
2
]+
3π4ρ
8 [2− πρ]2, (3.49)
which results in ∥∥I ′′2 ∥∥L≤ C5
∥∥ωδ
∥∥L
∥∥δ∥∥3. (3.50)
By (3.20), (3.23), (3.30), (3.33), (3.42) and (3.48), we have∥∥I ′′3 ∥∥L≤ C6
∥∥ωδ
∥∥L
∥∥δ∥∥4
while∥∥δ∥∥
2≤ ρ, (3.51)
where
C6 =π2
2− πρ
[π2
16+
π
3+1
2
]+
π3ρ
[2− πρ]2
[π2
4+7π
4+ 5
]+
3π5ρ2
[2− πρ]3 . (3.52)
By (3.17), (3.35), (3.36), (3.44), (3.50) and (3.51), we get∥∥G′′∥∥L≤[κ C4 + C5 + C6 + 2C2 + 2C3
]∥∥ωδ
∥∥L
∥∥δ∥∥4= M2
∥∥ωδ
∥∥L
∥∥δ∥∥4, (3.53)
where M2 is just that given in (3.16). �
4 Stability of the Displacement
Denote
AΦ,υ = supt1,t2∈L
{ |Φ(t1)− Φ(t2)||t1 − t2|υ
, t1 �= t2
}, AΦ∗,υ = sup
ξ1,ξ2∈Lδ
{ |Φ∗(ξ1)− Φ∗(ξ2)||ξ1 − ξ2|υ
, ξ1 �= ξ2
},
(4.1)
which are respectively the Holder semi-norm of Φ and the Holder semi-norm of Φ∗.
Definition 4.1 Let δ ∈ Sn(ρ) = {δ| ‖δ‖n ≤ ρ, δ ∈ Cn(Γ)}, Γδ is the perturbation for the
curve Γ by δ, Φ ∈ Hυ(Γ) and Φ∗ ∈ Hυ(Γδ
)(0 < υ ≤ 1). If there are constants A+(Φ, n, ρ, υ)
and R(Φ, n, ρ, υ) independent of the perturbation δ, sometime simply denoted as A+Φ,υ and
RυΦ,n, such that
max{AΦ∗,υ, AΦ,υ} ≤ A+(Φ, n, ρ, υ) = A+Φ,υ , (4.2)
and ∥∥Φ∗ ◦ σ − Φ∥∥
L≤ R(Φ, n, ρ, υ)
∥∥δ∥∥υ
n= Rυ
Φ,n
∥∥δ∥∥υ
n, (4.3)
then, the function Φ∗ is called a Sobolev type perturbation of n-th order for Φ by δ with the
index υ. In particular, we simply say that Φ∗ is the Sobolev type perturbation of Φ by δ while
134 ACTA MATHEMATICA SCIENTIA Vol.34 Ser.B
n = 0. The constant A+(Φ, n, ρ, υ) is called a Holder super norm of Φ∗ and Φ. The constant
R(Φ, n, ρ, υ) is called a Sobolev perturbation coefficient of Φ∗ for Φ.
Remark 4.1 Let Ω is a region and Γ∪Γδ ⊂ Ω. If f ∈ Hυ(Ω) then f |Γδis just a Sobolev
type perturbation of f |Γ by δ, since we may take A+(f |Γ, 0, ρ, υ) = R(f |Γ, 0, ρ, υ) = Af,υ where
Af,υ is the Holder semi-norm of f on Ω. Under this case, we call f |Γδthe trivial perturbation
for f |Γ.Lemma 4.1 [11] Suppose that δ ∈ Sn(ρ), Γδ is the perturbation for the curve Γ by δ,
Φ ∈ Hυ(Γ)(0 < υ ≤ 1) and Φ∗ is a Sobolev type perturbation of n-th order for Φ by δ.
D+(D−)and D+
δ
(D−δ)denote respectively the interior (exterior) region bounded by Γ and the
interior (exterior) region bounded by Γδ. Let the Cauchy integral operators
(S[Φ]
)(z) =
⎧⎪⎪⎨⎪⎪⎩1
2πi
∫Γ
Φ(τ)
τ − zdτ, z �∈ Γ,
1
πi
∫Γ
Φ(τ)
τ − tdτ, z = t ∈ Γ,
(4.4)
(Sδ[Φ∗]
)(z) =
⎧⎪⎪⎨⎪⎪⎩1
2πi
∫Γδ
Φ∗(ζ)
ζ − zdζ, z �∈ Γδ,
1
πi
∫Γδ
Φ∗(τ)
τ − tdτ, z = t ∈ Γδ,
(4.5)
and their projection operators
(S±[Φ]
)(z) =
⎧⎪⎪⎨⎪⎪⎩1
2πi
∫Γ
Φ(τ)
τ − zdτ, z ∈ D±,
1
2
[(S[Φ]
)(t)± Φ(t)
], z = t ∈ Γ,
(4.6)
(S±δ [Φ∗]
)(z) =
⎧⎪⎪⎨⎪⎪⎩1
2πi
∫Γδ
Φ∗(ζ)
ζ − zdζ, z ∈ D±δ ,
1
2
[(Sδ[Φ∗]
)(t)± Φ∗(t)
], z = t ∈ Γδ,
(4.7)
then, for 0 < ν < υ,∥∥∥S±[Φ]− S±δ [Φ∗]∥∥∥
D±∩D±
δ
≤ C(υ, ν, ρ, �Γ, CΓ
)∥∥δ∥∥υ−ν
max{n,1}, (4.8)
where �Γ denotes the length of the curve Γ, CΓ is the Chord-Arc ratio for Γ,
C(υ, ν, ρ, �Γ, CΓ) = C1(υ, ν, ρ, �Γ, CΓ) + C2(υ, ν, ρ, �Γ, CΓ) + C3(υ, ν, ρ, �Γ, CΓ) (4.9)
with
C1(υ, ν, ρ, �Γ, CΓ) =�νΓ
2νπνC1−νΓ
{[(1 +
ρ
CΓ
)υ
+ 1
]A+Φ,υ
} νυ [2Rυ
Φ,n
]1− νυ
+�υΓ
2υπυ
[1 +
1
CΓ
]ρ1−υ+ν
[CΓ − ρ]1−υA+Φ,υ +Rυ
Φ,nρν , (4.10)
C2(υ, ν, ρ, �Γ, CΓ) = 3ρν
[22υ(1 + ρ)
πυ(CΓ − ρ)+ 2υ
]A+Φ,υ , (4.11)
No.1 J. Lin & J.Y. Du: STABILITY OF DISPLACEMENT TO FUNDAMENTAL PROBLEM 135
C3(υ, ν, ρ, �Γ, CΓ) =
⎧⎪⎪⎨⎪⎪⎩22+υ32−υ(1 + ρ)ρν
π(1− υ)(CΓ − ρ)A+Φ,υ, 0 < υ < 1,
24
π
{ρν ln [1 + (1 + ρ)�ΓCΓ] +
1
ν e
}1 + ρ
CΓ − ρA+Φ,υ , υ = 1.
(4.12)
Theorem 4.1 Assume that δ ∈ C3(L) and ‖δ‖3 ≤ ρ ≤ 1.92κ(κ−1)(3κ−1)(2κ+3)π+0.96κ(κ−1)π . If g∗
is the Sobolev type perturbation of g by δ, then ω∗ is the the Sobolev type perturbation of
2-nd order for ω by δ, where ω and ω∗ are respectively the normal solutions for the Sherman-
Lauricella equations (1.7) (Γ = L) and (2.12).
Proof First, we have, from (1.10), (2.18) and (2.19)∣∣aδ − a∣∣ ≤ μ
κ
∥∥gδ − g∥∥
L+
1
2κ
∥∥G∥∥L,∣∣bδ − b
∣∣ ≤ 2μ
κ(κ− 1)
∥∥gδ − g∥∥
L+
1
κ(κ− 1)
∥∥G∥∥L, (4.13)
so, by (1.9), (2.17) and (4.13), we get∥∥ωδ − ω∥∥
L≤∣∣aδ − a
∣∣+ ∣∣bδ − b∣∣+ 2μ
κ
∥∥gδ − g∥∥
L+1
κ
∥∥G∥∥L
≤ μ(3κ− 1)
κ(κ− 1)
∥∥gδ − g∥∥
L+
3κ− 1
2κ(κ− 1)
∥∥G∥∥L. (4.14)
(3.12) and (4.14) result in
(1− ρ N)∥∥ωδ − ω
∥∥L≤ μ(3κ− 1)
κ(κ− 1)
∥∥gδ − g∥∥
L+N
∥∥δ∥∥2
∥∥ω∥∥L, (4.15)
where N = 3κ−12κ(κ−1)M0, M0 is given in (3.12). Noting
ρ ≤ 1.92κ(κ− 1)
(3κ− 1)(2κ+ 3)π + 0.96κ(κ− 1)π, (4.16)
we get
ρ N =3κ− 1
2κ(κ− 1)
2(2κ+ 3)π
2− πρρ < 0.96. (4.17)
Obviously, there is a Sobolev perturbation coefficient Rυg,0 of g∗ for g such that∥∥gδ − g
∥∥L≤ Rυ
g,0
∥∥δ∥∥υ
L. (4.18)
(4.15), (4.18), (1.12) and (2.13) result in∥∥ω∗ ◦ σ − ω∥∥
L≤ (3κ− 1)μ
κ(κ− 1) (1− ρ N)
(Rυ
g,0 + ρ1−υN∥∥g∥∥
L
) ∥∥δ∥∥υ
2= Rυ
ω,2
∥∥δ∥∥υ
2. (4.19)
This inequality and (1.12) imply∥∥ωδ
∥∥L≤ (3κ− 1)μ
κ(κ− 1) (1− ρ N)
(ρυ Rυ
g,0 +∥∥g∥∥
L
), (4.20)
so, we have ∣∣bδ − b∣∣ ≤ 2μ
(Rυ
g,0 + ρ1−υN∥∥g∥∥
L
)κ (κ− 1) (1− ρN)
∥∥δ∥∥υ
2(4.21)
and ∣∣bδ
∣∣ ≤ 2μ(ρυRυ
g,0 +∥∥g∥∥
L
)κ (κ− 1) (1− ρN)
. (4.22)
136 ACTA MATHEMATICA SCIENTIA Vol.34 Ser.B
Second, for ξj = tj + δ(tj)(tj ∈ L
), by (2.17), we have
∣∣ω∗(ξ2)− ω∗(ξ1)∣∣ ≤ ∣∣bδ
∣∣∣∣t2 − t1∣∣+ 2μ
κ
∣∣g∗(ξ2)− g∗(ξ1)∣∣+ 1
κ
∣∣G(t2)−G(t1)∣∣
= l1 + l2 + l3. (4.23)
(2.6) implies ∣∣t2 − t1∣∣ ≤ 2
2− πρ
∣∣σ(t2)− σ(t1)∣∣ = 2
2− πρ
∣∣ξ2 − ξ1∣∣. (4.24)
(4.24), (4.22) and∣∣t2 − t1
∣∣ = ∣∣t2 − t1∣∣1−υ∣∣t2 − t1
∣∣υ result in
l1 ≤4μ
κ(κ− 1) (1− ρ N) (2− πρ)υ[ρυ Rυ
g,0 +∥∥g∥∥
L
] ∣∣ξ2 − ξ1∣∣υ. (4.25)
By (3.2), (3.13) and (4.20), we obtain
∣∣G(t2)−G(t1)∣∣ ≤ ∥∥ (D1[G]])
∥∥L
∣∣t2−t1∣∣ ≤ πρμ(3κ− 1)M1
2κ(κ− 1) (1− ρ N)
[ρυ Rυ
g,0 +∥∥g∥∥
L
] ∣∣t2−t1∣∣, (4.26)
so,
l3 ≤πρμ(3κ− 1)M1
κ2(κ− 1) (1− ρ N) (2− πρ)υ[ρυ Rυ
g,0 +∥∥g∥∥
L
] ∣∣ξ2 − ξ1∣∣υ. (4.27)
Obviously,
l2 ≤2μ
κAg∗,υ
∣∣ξ2 − ξ1∣∣υ. (4.28)
Hence, by (4.23), (4.25), (4.27) and (4.28), we get
Aω∗,υ ≤4κμ+ πρμ(3κ− 1)M1
κ2(κ− 1) (1− ρ N) (2− πρ)υ[ρυ Rυ
g,0 +∥∥g∥∥
L
]+2μ
κAg∗,υ. (4.29)
From (1.9), we have ∣∣ω(t2)− ω(t1)∣∣ ≤ ∣∣b∣∣ |t2 − t1|+
2μ
κ|g(t2)− g(t1)| . (4.30)
(4.30) and (1.11) result in
Aω,υ ≤22−υμ
κ(κ− 1)
∥∥g∥∥L+2μ
κAg,υ . (4.31)
From (4.29) and (4.31), we know that
max{Aω∗,υ, Aω,υ} ≤4κμ+ πρμ(3κ− 1)M1
κ2(κ− 1) (1− ρ N) (2 − πρ)υ[ρυ Rυ
g,0 +∥∥g∥∥
L
]+2μ
κA+
g,υ = A+ω,υ. (4.32)
�
Theorem 4.2 Assume that δ ∈ C4(L) and ‖δ‖4 ≤ ρ ≤ 1.92κ(κ−1)(3κ−1)(2κ+3)π+0.96κ(κ−1)π . If
g∗ and g′∗ are respectively the Sobolev type perturbation of g by δ and the Sobolev type
perturbation of g′ by δ, then ω′∗ is the Sobolev type perturbation of 3-rd order for ω′ by δ,
where ω and ω∗ are respectively the normal solutions for the Sherman-Lauricella equations
(1.7) (Γ = L) and (2.12).
Proof On the one hand, by (1.9) and (2.20), we have∣∣∣ω′∗(ξ)− ω′(t)∣∣∣ ≤ ∣∣∣∣ bδ
1 + δ′(t)− b
∣∣∣∣+ 1
κ
∣∣∣∣ G′(t)
1 + δ′(t)
∣∣∣∣+ 2μ
κ
∣∣∣g′∗(ξ)− g′(t)∣∣∣ = �1 + �2 + �3. (4.33)
No.1 J. Lin & J.Y. Du: STABILITY OF DISPLACEMENT TO FUNDAMENTAL PROBLEM 137
By (4.21) and (1.11),
�1 ≤∣∣∣∣ bδ − b
1 + δ′(t)
∣∣∣∣+ ∣∣∣∣ b δ′(t)
1 + δ′(t)
∣∣∣∣≤
2μ(Rυ
g,0 + ρ1−υN∥∥g∥∥
L
)κ (κ− 1) (1− ρN) (1− ρ)
∥∥δ∥∥υ
2+
2μ∥∥g∥∥
L
κ(κ− 1)(1− ρ)
∥∥δ′∥∥L
≤ 2μ
κ (κ− 1) (1− ρN) (1− ρ)
[Rυ
g,0 + ρ1−υ(1 +N − ρ N
)∥∥g∥∥L
]∥∥δ∥∥υ
2. (4.34)
By (3.13) and (4.20),
�2 ≤(3κ− 1)μM1
κ2(κ− 1) (1− ρ N) (1− ρ)
(ρυ Rυ
g,0 +∥∥g∥∥
L
) ∥∥δ∥∥3. (4.35)
Obviously, there is a Sobolev perturbation coefficient Rυg′,0 of g′∗ for g′ such that
�3 ≤2μ
κRυ
g′,0
∥∥δ∥∥υ
L. (4.36)
By (4.33), (4.34), (4.35) and (4.36), we get∥∥ω′∗ ◦ σ − ω′∥∥
L≤ Rυ
ω′,3
∥∥δ∥∥υ
3, (4.37)
where
Rυω′,3 =
μ [2κ+ (3κ− 1)ρM1]Rυg,0 + μρ1−υ
[2κ(1 +N − ρN
)+ (3κ− 1)M1
] ∥∥g∥∥L
κ2 (κ− 1) (1− ρN) (1− ρ)+2μ
κRυ
g′,0.
(4.38)
On the other hand, by (2.17), we have∣∣∣ω′∗(ξ)− ω′∗(ζ)∣∣∣ ≤ ∣∣bδ
∣∣ ∣∣∣∣ 1
1 + δ′(t)− 1
1 + δ′(τ)
∣∣∣∣+ 1
κ
∣∣∣∣ G′(t)
1 + δ′(t)− G′(τ)
1 + δ′(τ)
∣∣∣∣+ 2μ
κ
∣∣∣g′∗(ξ)− g′∗(ζ)∣∣∣
= k1 + k2 + k3. (4.39)
By Lemma 3.1, we obtain∣∣∣∣ 1
1 + δ′(t)− 1
1 + δ′(τ)
∣∣∣∣ ≤ 1
(1 − ρ)2π
2
∣∣∣δ′′∣∣∣L
∣∣∣t− τ∣∣∣ ≤ πρ
2(1− ρ)2
∣∣∣t− τ∣∣∣, (4.40)
and ∣∣∣G′(t)−G′(τ)∣∣∣ ≤ ∥∥ (D1[G
′]])∥∥
L
∣∣∣t− τ∣∣∣ ≤ π
2
∥∥G′′∥∥L
∣∣∣t− τ∣∣∣. (4.41)
By (4.22), (4.40), (3.13), (4.20), (4.41), (4.24) and (3.15),
k1 ≤2πρμ
κ (κ− 1) (1− ρN) (1− ρ)2 (2− πρ)υ
[ρυRυ
g,0 +∥∥g∥∥
L
]∣∣ξ − ζ∣∣υ, (4.42)
and
k2 ≤1
κ
[∣∣G′(t)∣∣ ∣∣∣∣ 1
1 + δ′(t)− 1
1 + δ′(τ)
∣∣∣∣+ ∣∣∣∣ 1
1 + δ′(τ)
∣∣∣∣ |G′(t)−G′(τ)|]
≤ πρ
2κ(1− ρ)2[ρM1 + (1− ρ)M2
]∥∥ωδ
∥∥L
∣∣t− τ∣∣
≤ πρμ(3κ− 1) [ρM1 + (1− ρ)M2]
κ2(κ− 1) (1− ρN) (1 − ρ)2 (2− πρ)υ
[ρυRυ
g,0 +∥∥g∥∥
L
]∣∣ξ − ζ∣∣υ. (4.43)
138 ACTA MATHEMATICA SCIENTIA Vol.34 Ser.B
Obviously,
k3 ≤2μ
κAg′
∗,υ
∣∣ξ − ζ∣∣υ. (4.44)
So, (4.39), (4.42), (4.43) and (4.44) result in
Aω′
∗,υ ≤
πρμ[2κ+ (3κ− 1) (ρM1 + (1− ρ)M2)
]κ2 (κ− 1) (1− ρN) (1− ρ)2 (2− πρ)
υ
[ρυRυ
g,0 +∥∥g∥∥
L
]+2μ
κAg′
∗,υ. (4.45)
From (1.9), we have
|ω′(t2)− ω′(t1)| =2μ
κ|g′(t2)− g′(t1)| , (4.46)
obviously,
Aω′,υ =2μ
κAg′,υ. (4.47)
By (4.45) and (4.47), we get
max{Aω′
∗,υ, Aω′,υ} ≤ A+
ω′,υ, (4.48)
where
A+ω′,υ ≤
πρμ {2κ+ (3κ− 1) [ρM1 + (1− ρ)M2]}κ2 (κ− 1) (1− ρN) (1− ρ)2 (2− πρ)
υ
(ρυRυ
g,0 +∥∥g∥∥
L
)+2μ
κA+
g′,υ. (4.49)
�
Corollary 4.1 Assumptions are the same to that in Theorem 4.2, then �∗(ξ) = ξω′∗(ξ)
is the Sobolev type perturbation of 3-rd order for �(t) = tω′(t) by δ.
Proof We know that
|�∗(ξ)−�(t)| ≤ |ξ| |ω′∗(ξ)− ω′(t)|+ |ξ − t| |ω′(t)| , (4.50)
namely,
‖�∗ ◦ σ −�‖L ≤ (1 + ρ) ‖ω′∗ ◦ σ − ω′‖L + ‖δ‖L ‖ω′‖L . (4.51)
By (4.37) and (1.12), we get
‖�∗ ◦ σ −�‖L ≤ Rυ�,3
∥∥δ∥∥υ
3, (4.52)
where
Rυ�,3 = (1 + ρ)Rυ
ω′,3 +2μρ1−υ
κ(κ− 1)
∥∥g∥∥L+2μρ1−υ
κ
∥∥g′∥∥L
(4.53)
with Rυω′,3 is given in (4.38).
We also have∣∣�∗(ξ2)−�∗(ξ1)∣∣ ≤ ∣∣ξ2∣∣ |ω′∗(ξ2)− ω′∗(ξ1)|+
∣∣ξ2 − ξ1∣∣ |ω′∗(ξ1)| , (4.54)
namely, |�∗(ξ2)−�∗(ξ1)| ≤ (1 + ρ)Aω′
∗,υ |ξ2 − ξ1|υ + |ξ2 − ξ1|
[‖ω′δ ◦ σ − ω′‖
L+ ‖ω′‖L
]. By
(4.48), (4.37) and (1.12), we get
A�∗,υ ≤ (1 + ρ)A+ω′,υ + (2 + πρ)1−υ
[Rυ
ω′,3ρυ +
2μ
κ(κ− 1)
∥∥g∥∥L+2μ
κ
∥∥g′∥∥L
](4.55)
with A+ω′,υ and Rυ
ω′,3 are given in (4.49) and (4.38), respectively.
From (1.9), we have
|�(t2)−�(t1)| ≤ |ω′(t2)− ω′(t1)|+ |t2 − t1| |ω′(t1)| . (4.56)
No.1 J. Lin & J.Y. Du: STABILITY OF DISPLACEMENT TO FUNDAMENTAL PROBLEM 139
(4.56), (4.48) and (1.12) result in
A�,υ ≤ A+ω′,υ +
22−υμ
κ(κ− 1)
∥∥g∥∥L+22−υμ
κ
∥∥g′∥∥L. (4.57)
By (4.55), (4.57) and (4.49), we get
max{A�∗,υ, A�,υ} ≤ (1 + ρ)A+ω′,υ + (2 + πρ)1−υ
[Rυ
ω′,3ρυ +
2μ
κ(κ− 1)
∥∥g∥∥L+2μ
κ
∥∥g′∥∥L
]= A+
�,υ. (4.58)
�
Theorem 4.3 Assume that δ ∈ C4(L) and ‖δ‖4 ≤ ρ ≤ 1.92κ(κ−1)(3κ−1)(2κ+3)π+0.96κ(κ−1)π . If
g∗ and g′∗ are respectively the Sobolev type perturbation of g by δ and the Sobolev type
perturbation of g′ by δ, then, for 0 < ν < υ ≤ 1,∥∥[u∗ + iv∗]− [u+ iv]∥∥
U∩Uδ≤ D(υ, ν, ρ)
∥∥δ∥∥υ−ν
3, (4.59)
where D(υ, ν, ρ) is the constant independent of δ. More precisely,
D(υ, ν, ρ) =κ
μLω(υ, ν, ρ) +
1
2μLω′(υ, ν, ρ) +
1
2μL�(υ, ν, ρ), (4.60)
where
Lω(υ, ν, ρ) = α(υ, ν, ρ)A+ω,υ + β(υ, ν, ρ)
(A+
ω,υ
) νυ(Rυ
ω,2
)1− νυ + ρν Rυ
ω,2, (4.61)
Lω′(υ, ν, ρ) = α(υ, ν, ρ)A+ω′,υ + β(υ, ν, ρ)
(A+
ω′,υ
) νυ (
Rυω′,3
)1− νυ + ρν Rυ
ω′,3, (4.62)
L�(υ, ν, ρ) = α(υ, ν, ρ)A+�,υ + β(υ, ν, ρ)
(A+
�,υ
) νυ(Rυ
�,3
)1− νυ + ρν Rυ
�,3 (4.63)
with
α(υ, ν, ρ) =
⎧⎪⎪⎪⎨⎪⎪⎪⎩3ρν
[22υ(1 + ρ)
υ(2− πρ)+ 2υ +
22+υ31−υ(1 + ρ)
(1 − υ)(2− πρ)+
ρ1−υ (π + 2)
6υ (2− πρ)1−υ
], 0 < υ < 1,
12ρν
[1 + ρ
2− πρ+
π + 14
24+2(1 + ρ) ln(5 + 4ρ)
2− πρ
]+
24(1 + ρ)
νe(2− πρ), υ = 1,
β(υ, ν, ρ) =1
ν2νυ−1
[ (1 +
π
2ρ)υ
+ 1
] νυ
,
Rυω,2, A
+ω,υ, Rυ
ω′,3, A+ω′,υ, Rυ
�,3 and A+�,υ are given in (4.19), (4.32), (4.38), (4.49), (4.53) and
(4.58), respectively.
Proof From (1.1), (1.4), (1.5), (1.6) and (4.6), we know
u+ iv =κ
2μϕ(z)− z
2μϕ′(z)− 1
2μψ(z),
and
ϕ(z) =(S+[ω]
)(z), ϕ′(z) =
(S+[ω′]
)(z), ψ(z) = −κ
(S+[ω]
)(z)−
(S+[�]
)(z), z ∈ U,
140 ACTA MATHEMATICA SCIENTIA Vol.34 Ser.B
analogously,
u∗ + iv∗ =κ
2μϕ∗(z)−
z
2μϕ′∗(z)−
1
2μψ∗(z),
and
ϕ∗(z) =(S+δ [ω∗]
)(z), ϕ′∗(z) =
(S+δ [ω
′∗])(z), ψ∗(z) = −κ
(S+δ [ω∗]
)(z)−
(S+δ [�∗]
)(z), z ∈ Uδ.
So, we have∥∥[u∗ + iv∗]− [u+ iv]∥∥
U∩Uδ≤ κ
μ
∥∥S+δ [ω∗]− S+[ω]∥∥
U∩Uδ+
1
2μ
∥∥S+δ [ω′∗]− S+[ω′]∥∥
U∩Uδ
+1
2μ
∥∥S+δ [�∗]− S+[�]∥∥
U∩Uδ.
By Theorem 4.1, Theorem 4.2, Corollary 4.1 and Lemma 4.1, we get⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩
∥∥S+δ [ω∗]− S+[ω]∥∥
U∩Uδ≤ Lω(υ, ν, ρ)
∥∥δ∥∥υ−ν
2,∥∥S+δ [ω′∗]− S+[ω′]
∥∥U∩Uδ
≤ Lω′(υ, ν, ρ)∥∥δ∥∥υ−ν
3,∥∥S+δ [�∗]− S+[�]
∥∥U∩Uδ
≤ L�(υ, ν, ρ)∥∥δ∥∥υ−ν
3,
where Lω(υ, ν, ρ), Lω′(υ, ν, ρ) and L�(υ, ν, ρ) are given in (4.61), (4.62) and (4.63).
Finally, (4.59) is obtained. �
References
[1] Lu J K. Complex Variable Methods in Plane Elasticity. Singapore: World Scientific, 1995
[2] Muskhelishvili N I. Some Basic Problems of Mathematical Theory of Elasticity. Groninge: Noordhoff,
1963
[3] Wen G C, Yin W P. Applicition of Complex Function. Beijing: Beijing Normal Univ, 1984 (in Chinese)
[4] Qian W Z, Ye K Y. Elasticity. Beijing: Science Publishing Company, 1980 (in Chinese)
[5] Lu J K. Boundary Value Problems for Analytic Functions. Singapore: World Scientific, 1993
[6] Muskhelishvili N I. Singular Integral Equations. 2nd ed. Groningen: Noordhoff, 1968
[7] Wang X L, Gong Y F. On stability of a class of singular integral with respect to path of integration. Acta
Math Sinica, 1999, 42(2): 343–350 (in Chinese)
[8] Wang C R, Zhang H M, Zhu Y C. The Riemann boundary value problem with respect to the perturbation
of boundary curve. Complex Var Elliptic Equ, 2006, 51(8/11): 831–845
[9] Lin J, Wang C R. On stability of Cauchy-type integral with kernel density of class H∗ with respect to
path of integration for an open arc. Complex Var Elliptic Equ, 2007, 52(10/11): 883–898
[10] Lin J, Wang C R, Yan K M. Stability of the second fundamental problem under perturbation of the
boundary curve in plane elasticity. Math Methods App Sci, 2010, 33(14): 1762–1770
[11] Lin J, Du J Y. Stability of the first fundamental problem under perturbation of boundary curve and
external load, to appear
[12] Du J Y. The quadrature formulas of singular integrals of higher order. Chinese Math Ann, 1985, 6A(5):
625–636
[13] Du J Y. The quadrature formulae of closed and transformation weight type for singular integrals of higher
order. J of Math (PRC), 1986, 6(4): 439–454
[14] Du J Y. On quadrature formulae for singular integrals of arbitrary order. Acta Math Sci, 2004, 24B(1):
9–27
[15] Lu J K. Singular integrals of high order and their applications to solving singular integral equations. Sci
Tech Wuhan Univ, 1977, 2: 106–122