-
Research ArticleThe Approximate Solution of Fredholm Integral
Equations withOscillatory Trigonometric Kernels
Qinghua Wu
School of Mathematics and Statistics, Central South University,
Changsha, Hunan 410083, China
Correspondence should be addressed to Qinghua Wu;
[email protected]
Received 9 March 2014; Accepted 4 April 2014; Published 17 April
2014
Academic Editor: Turgut Öziş
Copyright © 2014 Qinghua Wu.This is an open access article
distributed under the Creative Commons Attribution License,
whichpermits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
A method for approximating the solution of weakly singular
Fredholm integral equation of the second kind with highly
oscillatorytrigonometric kernel is presented. The unknown function
is approximated by expansion of Chebychev polynomial and
thecoefficients are determinated by classical collocation method.
Due to the highly oscillatory kernels of integral equation,
thediscretised collocation equation will give rise to the
computation of oscillatory integrals. These integrals are
calculated by usingrecursion formula derived from the fundamental
recurrence relation of Chebyshev polynomial. The effectiveness and
accuracy ofthe proposed method are tested by numerical
examples.
1. Introduction
The Fredholm integral equations of second kind,
𝑦 (𝑥) + 𝜆∫
𝑏
𝑎
𝐾 (𝑥, 𝑡)
|𝑥 − 𝑡|𝛼𝑦 (𝑡) 𝑑𝑡 = 𝑓 (𝑥) ,
𝑥 ∈ [𝑎, 𝑏] , 0 < 𝛼 < 1,
(1)
where 𝐾(𝑥, 𝑡) is a continuous function and 𝑓(𝑥) is a
givenfunction, have many applications in mathematical physicsand
engineering, such as heat conduction problem, potentialproblems,
quantum mechanics, and seismology image pro-cessing [1–4].
Particularly, when 𝐾(𝑥, 𝑥) ̸= 0 and 0 < 𝛼 < 1,(1) is called
weakly singular.
In most of the cases, the integral equation cannot bedone
analytically and one has to resort to numericalmethods.Many
numerical methods, such as collocation method andGalerkinmethod,
have been developed to solve (1); for detailssee [2, 5, 6]. These
methods are well-established numericalalgorithms; however, standard
version of these classicalmeth-ods may suffer from difficulty for
computation of (1), con-taining highly oscillatory kernels since
the computation of thehighly oscillatory integrals by standard
quadrature methodsis exceedingly difficult and the cost steeply
increases withthe frequency. Furthermore, Galerkin method requires
manydouble integrals when approximating the solution of
integral
equation. Specially, when the kernel is highly oscillatory,
itrequires the evaluation of many highly oscillatory
doubleintegrals, which can become computationally expensive.
Recently, for weakly singular Volterra integral equationsof the
second kind with highly oscillatory Bessel kernels, itwas found
that the collocationmethods aremuchmore easilyimplemented and can
get higher accuracy than discontinuousGalerkin methods under the
same piecewise polynomialsspace; for details see [7–10]. In
addition, collocation methodonly involves single integrals which
are a little easier to evalu-ate. Motivated by this fact, here we
investigate the applicationof collocation method for the solution
of Fredholm integralequation
𝑦 (𝑥) + 𝜆∫
𝑏
𝑎
𝑒𝑖𝜔(𝑥−𝑡)
|𝑥 − 𝑡|𝛼𝑦 (𝑡) 𝑑𝑡 = 𝑓 (𝑥) ,
𝑥 ∈ [𝑎, 𝑏] , 𝛼 < 1, 𝜔 ≫ 1.
(2)
Here 𝑦(𝑥) is the unknown function and 𝑓(𝑥) is a givenfunction
and assume that 𝜆 is not the eigenvalue of integralequation, since
for any finite interval [𝑎, 𝑏] it can be trans-formed to interval
[−1, 1] by linear transformation. In thispaper, we only consider
the case of [𝑎, 𝑏] = [−1, 1].
On the other hand, when the unknown function isapproximated
byChebychev polynomial, the discretization of
Hindawi Publishing CorporationJournal of Applied
MathematicsVolume 2014, Article ID 172327, 7
pageshttp://dx.doi.org/10.1155/2014/172327
-
2 Journal of Applied Mathematics
the integral equation (2) by collocation method will give riseto
highly oscillatory integral which can be formulated as
∫
𝑏
𝑎
𝑒(𝑖𝜔(𝑥−𝑡))
|𝑥 − 𝑡|𝛼𝑇𝑗 (𝑡) 𝑑𝑥, 𝑥 ∈ [𝑎, 𝑏] , 0 < 𝛼 < 1, (3)
where 𝑇𝑗(𝑥) denotes the Chebychev polynomials of the
firstkind.
In last few years many efficient methods have beendevised for
the evaluation of oscillatory integral, such asasymptotic method
[11], Filon-type method [12, 13], Levin’scollocation method [14],
modified Clenshaw-Curtis method,Clenshaw-Curtis-Filon-type method
[15], and generalizedquadrature rule [16], although some of these
methods do notinvolve Chebyshev polynomial. Piessens and Poleunis
[17]consider a simpler case of 𝛼 = 0 and use Chebyshev polyno-mials
to evaluate the integral by somewhat indirect methodinvolving a
truncated infinite series of Bessel functions. Theevaluation of
this infinite summation suggests a defect of thismethod. An
alternative procedure which avoids this infiniteseries is the
original Bakhvalo and Vasil’eva-Legendre workinvolved in the
evaluation of the integral
𝑀𝑖 (𝜔) = ∫
1
−1
𝑃𝑖 (𝑥) 𝑒𝑖𝜔𝑥𝑑𝑥, (4)
by recurrence relation, where𝑃𝑖(𝑥) denotes Legendre polyno-mial.
Sadly, this relation proved to be unstable in the forwardsdirection
for small 𝜔. In [18], Alaylioglu et al proposeda simple alternative
approach analogous to Newton-Cotesbased formulae. This method
avoids the instability to acertain degree compared with these two
methods mentionedabove. We used the same idea and generalized it to
the caseof weakly singular and calculated oscillatory integral
(3).
This paper is organized as follows: in Section 2 we derivesome
basic formulae and introduce some mathematicalpreliminaries of the
proposed method. In Section 3 we dis-cuss the evaluation of the
integrals occurring in collocationequation. In Section 4 numerical
experiments are conductedto illustrate the performance of the
proposed method.
2. Fundamental Relations
Firstly, by separation of real and imaginary part of 𝑦(𝑥)
and𝑒𝑖𝑤(𝑥−𝑡), we transform the integral equation (2) into
equivalentsystems of two linear integral equations of Fredholm in
theforms
𝑦1 (𝑥) + 𝜆∫
1
−1
cos (𝜔 (𝑥 − 𝑡))|𝑥 − 𝑡|
𝛼𝑦1 (𝑡) 𝑑𝑡
− 𝜆∫
1
−1
sin (𝜔 (𝑥 − 𝑡))|𝑥 − 𝑡|
𝛼𝑦2 (𝑡) 𝑑𝑡 = 𝑓1 (𝑥) ,
𝑦2 (𝑥) + 𝜆∫
1
−1
sin (𝜔 (𝑥 − 𝑡))|𝑥 − 𝑡|
𝛼𝑦1 (𝑡) 𝑑𝑡
+ 𝜆∫
1
−1
cos (𝜔 (𝑥 − 𝑡))|𝑥 − 𝑡|
𝛼𝑦2 (𝑡) 𝑑𝑡 = 𝑓2 (𝑥) ,
(5)
where 𝑦(𝑥) = 𝑦1(𝑥) + 𝑖𝑦2(𝑥), 𝑓(𝑥) = 𝑓1(𝑥) + 𝑖𝑓2(𝑥), and 𝑖
=√−1.
Assume that y(𝑥) = (𝑦1(𝑥), 𝑦2(𝑥))𝑇 and f(𝑥) = (𝑓1(𝑥),
𝑓2(𝑥))𝑇; then, systems of linear integral equations (5) can
be
written in the matrix form
Iy (𝑥) + 𝜆∫1
−1
K (𝑥, 𝑡) y (𝑡) 𝑑𝑡 = f (𝑥) , (6)
where
I = (1 00 1) ,
K (𝑥, 𝑡) = (
cos (𝜔 (𝑥 − 𝑡))|𝑥 − 𝑡|
𝛼
− sin (𝜔 (𝑥 − 𝑡))|𝑥 − 𝑡|
𝛼
sin (𝜔 (𝑥 − 𝑡))|𝑥 − 𝑡|
𝛼
cos (𝜔 (𝑥 − 𝑡))|𝑥 − 𝑡|
𝛼
).
(7)
Set 𝑦𝑖(𝑥) = T(𝑥)b𝑖, 𝑖 = 1, 2, where T(𝑥) = [𝑇0(𝑥),𝑇1(𝑥), . . . ,
𝑇𝑁(𝑥)], b𝑖 = [𝑏𝑖0, 𝑏𝑖1, . . . , 𝑏𝑖𝑁]
𝑇.Hence, unknown functions can be expressed by
y (𝑥) = T𝐵, (8)where
y (𝑥) = (𝑦1 (𝑥)𝑦2 (𝑥)) , T (𝑥) = (
T (𝑥) 00 T (𝑥)) ,
𝐵 = (b1b2) .
(9)
Then, the aim is to find Chebyshev coefficients, that is,the
matrix 𝐵. We first substitute the Chebyshev collocationpoints,
which are defined by 𝑥𝑖 = cos(𝑖𝜋/𝑁), 𝑖 = 0, . . . , 𝑁,into (6) and
then rearrange a new matrix form to determine𝐵:
IY + 𝜆K = F (10)
in whichK is the integral part of (6) and
I = (
I 0 ⋅ ⋅ ⋅ 00 I ⋅ ⋅ ⋅ 0...
... d...
0 0 ⋅ ⋅ ⋅ I
), Y =(
y (𝑥0)
y (𝑥1)...
y (𝑥𝑁)
),
F =(
f (𝑥0)
f (𝑥1)...
f (𝑥𝑁)
), K =(
IK (𝑥0)
IK (𝑥1)...
IK (𝑥𝑁)
).
(11)
By substituting (8) into (10), the unknown coefficients canbe
easily computed from this linear algebraic equations andtherefore
we find the solution of integral equation (2).
3. Evaluation of the Integral
𝐼[𝜔, 𝛼, 𝑗, 𝑥]= ∫1
−1(𝑒(𝑖𝜔(𝑥−𝑡))
/|𝑥−𝑡|𝛼)𝑇𝑗(𝑡)𝑑𝑡
The discretization of integral equation will lead to
thecalculation of integral 𝐼[𝜔, 𝛼, 𝑗, 𝑥]. In this section, we
willderive the recursion formula to compute it efficiently from
thefundamental recurrence relation of Chebyshev polynomial.
-
Journal of Applied Mathematics 3
100
10−2
10−4
10−6
10−8
10−10
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
x
Abso
lute
erro
rApproximate the solution of example 4.1 with
𝛼 = 0.5, = 1000, N = 4𝜔
(a)
100
10−2
10−4
10−6
10−8
10−10
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
x
Abso
lute
erro
r
Approximate the solution of example 4.1 with𝛼 = 0.5, = 1000, N =
8𝜔
(b)
100
10−2
10−4
10−6
10−8
10−10
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
x
Abso
lute
erro
r
Approximate the solution of example 4.1 with𝛼 = 0.5, = 1000, N =
16𝜔
(c)
Figure 1: Example 1: the absolute error versus the 𝑥-coordinate
for𝑁 = 4 (a),𝑁 = 8 (b), and𝑁 = 16 (c) with negative data
ignored.
The Chebyshev polynomials are of the form
𝑇0 (𝑥) = 1
𝑇1 (𝑥) = 𝑥
...𝑇𝑛 (𝑥) = 2𝑥𝑇𝑛−1 (𝑥) − 𝑇𝑛−2 (𝑥) , 𝑛 = 2, 3, . . . ,
(12)
and the coefficients 𝐶𝑖,j of 𝑥𝑗 in 𝑇𝑖(𝑥) can be easily
calculated
by the means of the recurrence relation
𝐶𝑖,𝑗 = 2𝐶𝑖−1,𝑗−1 − 𝐶𝑖−2,𝑗, 𝑖 ≥ 2, 𝑗 ≤ 𝑖. (13)
By expanding the Chebyshev polynomial 𝑇𝑖(𝑥) in terms ofpowers of
𝑥, the integral 𝐼[𝜔, 𝛼, 𝑗, 𝑥] would be transformedinto the form
of
𝐼 [𝜔, 𝛼, 𝑗, 𝑥] =
𝑗
∑
𝑘=0
𝐶𝑗,𝑘𝐼1 [𝜔, 𝛼, 𝑘, 𝑥] , (14)
where 𝐼1[𝜔, 𝛼, 𝑘, 𝑥] = ∫1−1(𝑒(𝑖𝜔(𝑥−𝑡))
/|𝑥 − 𝑡|𝛼)𝑡𝑘𝑑𝑡.
By separation of real and imaginary part of 𝑒𝑖𝑤(𝑥−𝑡), (14)
istransformed into
𝐼1𝑐 [𝜔, 𝛼, 𝑗, 𝑥] = ∫
1
−1
𝑡𝑗
|𝑥 − 𝑡|𝛼cos (𝜔 (𝑥 − 𝑡)) 𝑑𝑡,
𝐼1𝑠 [𝜔, 𝛼, 𝑗, 𝑥] = ∫
1
−1
𝑡𝑗
|𝑥 − 𝑡|𝛼sin (𝜔 (𝑥 − 𝑡)) 𝑑𝑡,
(15)
whose main difficulty now is turning to the evaluation of
thefollowing basic integrals:
𝑀1𝑐 [𝜔, 𝛼, 𝑗, 𝑥] = ∫
1
−1
𝑡𝑗
|𝑥 − 𝑡|𝛼cos (𝜔𝑡) 𝑑𝑡,
𝑀1𝑠 [𝜔, 𝛼, 𝑗, 𝑥] = ∫
1
−1
𝑡𝑗
|𝑥 − 𝑡|𝛼sin (𝜔𝑡) 𝑑𝑡.
(16)
-
4 Journal of Applied Mathematics
100
10−2
10−4
10−5
10−6
10−7
10−3
10−1
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
x
Abso
lute
erro
rApproximate the solution of example 4.2 with
𝛼 = 1/3, = 1000, N = 4𝜔
(a)
100
10−2
10−4
10−6
10−8
10−10
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
x
Abso
lute
erro
r
Approximate the solution of example 4.2 with𝛼 = 1/3, = 1000, N =
8𝜔
(b)
100
10−2
10−4
10−6
10−8
10−10
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
x
Abso
lute
erro
r
Approximate the solution of example 4.2 with𝛼 = 1/3, = 1000, N =
16𝜔
(c)
Figure 2: Example 2: the absolute error versus the 𝑥-coordinate
for𝑁 = 4 (a),𝑁 = 8 (b), and𝑁 = 16 (c) with negative data
ignored.
With substitution for definite integral and binomial the-orem,
(16) is equivalent to
𝑀1𝑐 [𝜔, 𝛼, 𝑗, 𝑥]
=
𝑗
∑
𝑘=0
𝐶𝑘
𝑗𝑥𝑗−𝑘[(−1)
𝑘∫
𝑥+1
0
𝑡𝑘−𝛼 cos (𝜔 (𝑥 − 𝑡)) 𝑑𝑡
+∫
1−𝑥
0
𝑡𝑘−𝛼 cos (𝜔 (𝑥 − 𝑡)) 𝑑𝑡] ,
(17)
where 𝐶𝑘𝑗is number of combination and𝑀1𝑠[𝜔, 𝛼, 𝑗, 𝑥] can
be formulated analogy.Efficient evaluation of (18) is based on
accurate calcu-
lation of integrals in the formula which can be computed
explicitly by the incomplete Gamma function Γ(𝑧, 𝛼) [19];that
is,
∫ 𝑡𝜇−1 cos (𝜔𝑡) 𝑑𝑡
=1
2[(𝑖𝜔)−𝜇Γ (𝜇, 𝑖𝜔𝑡) + (−𝑖𝜔)
−𝜇Γ (𝜇, −𝑖𝜔𝑡)] ,
(18)
∫ 𝑡𝜇−1 sin (𝜔𝑡) 𝑑𝑡
=𝑖
2[(𝑖𝜔)−𝜇Γ (𝜇, 𝑖𝜔𝑡) − (−𝑖𝜔)
−𝜇Γ (𝜇, −𝑖𝜔𝑡)] ,
(19)
in which 𝜇 > 0, 𝑥 > 0; for details see ([20], pp 215).Once
the integral 𝐼[𝜔, 𝛼, 𝑗, 𝑥] is obtained, by substituting
into (10), the coefficient matrix is derived; then, we
cancompute the solution of integral by solving these
linearalgebraic equations.
-
Journal of Applied Mathematics 5
Table 1: Approximations for 𝑦(𝑥) + ∫1−1(𝑒𝑖𝜔(𝑥−𝑡)/|𝑥 − 𝑡|
0.5)𝑦(𝑡)𝑑𝑡 = 𝑒
𝑥 with 𝜔 = 103.
𝑥 cos(𝜋) cos(3𝜋/4) cos(0)𝑦4
10.351696035495271 0.459037927481918 0.924223909091303
𝑦8
10.351696947581043 0.459033811940511 0.924224547405338
𝑦16
10.351741610883315 0.459036527929146 0.924226969620880
𝑦4
20.0141286409694892 0.000146310591227839
−0.00107727880701281
𝑦8
20.0141193392298137 0.000145019893750479
−0.00107593754108643
𝑦16
20.0140583682955762 0.000144178555797097
−0.00106881635211410
Table 2: Approximations for 𝑦(𝑥) + ∫1−1(𝑒𝑖𝜔(𝑥−𝑡)/|𝑥 − 𝑡|
0.5)𝑦(𝑡)𝑑𝑡 = 𝑒
𝑥 with 𝜔 = 103.
𝑥 cos(𝜋/4) 1𝑦4
11.88249404678421 2.61059423521207
𝑦8
11.88248275737555 2.61059120470589
𝑦16
11.88248361661370 2.61073031676337
𝑦4
20.00323202027752064 −0.0995508315886191
𝑦8
20.00321634589900894 −0.0994917961267987
𝑦16
20.00319500282444326 −0.0993143176430846
4. Numerical Examples
In this section, we give some numerical examples to
illustratethe performance of proposed method. In all the
followingexamples,𝑁+1 is the number of mesh points, 𝑦𝑁
𝑖(𝑥) denotes
the approximate solution, where 𝑁 is the number of termsof the
Chebyshev series, and 𝑦(𝑥) denotes the exact solution,respectively.
All the computations have been performed byusing Matlab R2012a on a
2.5GHz PC with 2GB of RAM.
Example 1. We consider 𝑦(𝑥) + ∫1−1(𝑒𝑖𝜔(𝑥−𝑡)/|𝑥 − 𝑡|
𝛼)𝑦(𝑡)𝑑𝑡 =
𝑓(𝑥) and set 𝑓(𝑥) = 1 + ∫1−1(𝑒𝑖𝜔(𝑥−𝑡)/|𝑥 − 𝑡|
𝛼)𝑑𝑡; then, the
solution of which is
𝑦 (𝑥) = 1. (20)
We plot the absolute error |𝑦(𝑥) − 𝑦(𝑥)| by Matlab 2012ainternal
function ezplot. Specially, for 𝑁 = 4, with 𝜔 = 103and 𝛼 = 0.5,
solve linear algebraic equations (10); it can befound that the
approximate solutions are
𝑦1 (𝑥) = 0.9999999996 − 0.110 × 10−15𝑥 + 0.281 × 10
−9𝑥2
+ 0.323 × 10−15𝑥3+ 0.563 × 10
−10𝑥4,
𝑦2 (𝑥) = 0.460 × 10−14− 0.329 × 10
−9𝑥 + 0.554 × 10
−11𝑥2
+ 0.322 × 10−9𝑥3− 0.555 × 10
−11𝑥4.
(21)
It is easy to see from Figure 1 that the proposed method
isconverging and we only need𝑁 = 4 so we could achieve
highaccuracy.
Example 2. We consider 𝑦(𝑥) + ∫1−1(𝑒𝑖𝜔(𝑥−𝑡)/|𝑥 − 𝑡|
𝛼)𝑦(𝑡)𝑑𝑡 =
𝑓(𝑥) and set 𝑓(𝑥) = 𝑒𝑥 + ∫1−1(𝑒𝑖𝜔(𝑥−𝑡)/|𝑥 − 𝑡|
𝛼)𝑒𝑡𝑑𝑡; then, the
solution of which is𝑦 (𝑥) = 𝑒
𝑥. (22)
In this case, for 𝑁 = 4, with 𝜔 = 1000 and 𝛼 = 1/3,
theapproximate solutions are
𝑦1 (𝑥) = 1.000000001 + 0.9956819990𝑥 + 0.4992866904𝑥2
+ 0.1795192648𝑥3+ 0.04379395375𝑥
4,
𝑦2 (𝑥) = −0.530 × 10−7+ 0.189 × 10
−7𝑥 + 0.426 × 10
−6𝑥2
− 0.246 × 10−7𝑥3− 0.420 × 10
−6𝑥4.
(23)
It is obvious to see from Figure 2 that the proposedmethod is
efficient and could achieve high accuracy with littlenumber of
collocation points.
Example 3. For a general case, we consider𝑦(𝑥) + ∫
1
−1(𝑒𝑖𝜔(𝑥−𝑡)/|𝑥 − 𝑡|
𝛼)𝑦(𝑡)𝑑𝑡 = 𝑒
𝑥 and set 𝛼 = 0.5.Although the solution of this equation is
unknown, we canverify the calculation precision of the method to a
certaindegree by comparing the approximate evaluation at
meshpoints.
In this case, for 𝑁 = 4, with 𝜔 = 1000 and 𝛼 = 0.5,
theapproximate solutions are𝑦1 (𝑥) = 0.9242239091
+ 0.8836218495𝑥 + 0.4292470859𝑥2
+ 0.2458272504𝑥3+ 0.1276741404𝑥
4,
-
6 Journal of Applied Mathematics
Table 3: Approximations for 𝑦(𝑥) + ∫1−1(𝑒𝑖𝜔(𝑥−𝑡)/|𝑥 − 𝑡|
0.5)𝑦(𝑡)𝑑𝑡 = 𝑒
𝑥 with 𝜔 = 106.
𝑥 cos(𝜋) cos(3𝜋/4) cos(0)𝑦4
10.367419631536920 0.491835233010361 0.997500711679274
𝑦8
10.367419631541538 0.491835233006225 0.997500711679218
𝑦16
10.367419632527633 0.491835233064998 0.997500711678921
𝑦4
20.000458466865252363 0.150404190 × 10
−5−0.219327280 × 10
−5
𝑦8
20.000458466860662437 0.150403636 × 10
−5−0.219326739 × 10
−5
𝑦16
20.000458463698666909 0.150390826 × 10
−5−0.219326736 × 10
−5
Table 4: Approximations for 𝑦(𝑥) + ∫1−1(𝑒𝑖𝜔(𝑥−𝑡)/|𝑥 − 𝑡|
0.5)𝑦(𝑡)𝑑𝑡 = 𝑒
𝑥 with 𝜔 = 106.
𝑥 cos(𝜋/4) 1𝑦4
12.02304108762491 2.71487514654707
𝑦8
12.02304108760947 2.71487514654071
𝑦16
12.02304108780835 2.71487515460854
𝑦4
20.384042582 × 10
−5−0.00339813438891857
𝑦8
20.384040621 × 10
−5−0.00339813431478428
𝑦16
20.3840356665 × 10
−5−0.00339813322237100
𝑦2 (𝑥) = −0.001077278807 + 0.06120358877𝑥
+ 0.05269959347𝑥2− 0.1180433250𝑥
3
− 0.09433340997𝑥4.
(24)
It is also easy to see from Tables 1, 2, 3, and 4 that
thepresentedmethod is efficient and accurate, although the
exactsolution is unknown.
5. Conclusion
In this paper, we explore quadrature methods for weaklysingular
Fredholm integral equation of the second kindwith oscillatory
trigonometric kernels and present colloca-tion methods with
Chebyshev series for calculation of thesolution. For integral
equationwith highly oscillatory kernels,the standard collocation
methods with classical quadraturemethods are not suitable for the
numerical approximationof the solution of integral equation, since
the computationof the highly oscillatory integrals by standard
quadraturemethods is exceedingly difficult and the cost steeply
increaseswith the frequency. Based on the recursion formula
derivedin Section 3, we compute the highly oscillatory
integralsoccurring in collocation equation, directly and
efficiently.Numerical examples demonstrate the performance of
algo-rithm.
Conflict of Interests
The author declares that there is no conflict of
interestsregarding the publication of this paper.
Acknowledgment
This paper was supported by NSF of China (no. 11371376).
References
[1] V.M. Aleksandrov and E. V. Kovalenko, “Mathematical methodin
the displacement problem,” Inzhenernyj Zhurnal. MekhanikaTverdogo
Tela, vol. 2, pp. 77–89, 1984.
[2] H. Brunner, On the Numerical Solution of First-Kind
VolterraIntegral Equations with Highly Oscillatory Kernels, Highly
Oscil-latory Problems: From Theory to Applications, Isaac
NewtonInstitute, HOP 13-17, 2010.
[3] R. Estrada and R. P. Kanwal, Singular Integral
Equations,Birkhäauser, Boston, Mass, USA, 2000.
[4] A. D. Polyanin and A. V. Manzhirov, Handbook of
IntegralEquations, CRC Press, 1998.
[5] H. Brunner, Collocation Methods for Volterra Integral
andRelated Functional Equations, Cambridge University
Press,Cambridge, Cambridge, UK, 2004.
[6] I. G. Graham, “Galerkin methods for second kind
integralequations with singularities,”Mathematics of Computation,
vol.39, no. 160, pp. 519–533, 1982.
[7] S. Xiang and H. Brunner, “Efficient methods for
Volterraintegral equations with highly oscillatory Bessel kernels,”
BIT.Numerical Mathematics, vol. 53, no. 1, pp. 241–263, 2013.
[8] S. Xiang and K. He, “On the implementation of
discontinuousGalerkin methods for Volterra integral equations with
highlyoscillatory Bessel kernels,” Applied Mathematics and
Computa-tion, vol. 219, no. 9, pp. 4884–4891, 2013.
[9] S. Xiang and Q. Wu, “Numerical solutions to Volterra
integralequations of the second kind with oscillatory
trigonometrickernels,” Applied Mathematics and Computation, vol.
223, pp.34–44, 2013.
[10] Q.Wu, “On gradedmeshes for weakly singular Volterra
integralequations with oscillatory trigonometric kernels,” Journal
of
-
Journal of Applied Mathematics 7
Computational and Applied Mathematics, vol. 263, pp.
370–376,2014.
[11] A. Iserles and S. P. Nørsett, “On quadrature methods for
highlyoscillatory integrals and their implementation,” BIT.
NumericalMathematics, vol. 44, no. 4, pp. 755–772, 2004.
[12] A. Iserles and S. P. Nørsett, “Efficient quadrature of
highlyoscillatory integrals using derivatives,” Proceedings of The
RoyalSociety of London A, vol. 461, no. 2057, pp. 1383–1399,
2005.
[13] S. Xiang, “Efficient Filon-type methods for ∫𝑏𝑎𝑓(𝑥)𝑒
𝑖𝜔𝑔(𝑥)𝑑𝑥,”
Numerische Mathematik, vol. 105, no. 4, pp. 633–658, 2007.[14]
D. Levin, “Fast integration of rapidly oscillatory functions,”
Journal of Computational and Applied Mathematics, vol. 67, no.1,
pp. 95–101, 1996.
[15] S. Xiang, Y. J. Cho, H. Wang, and H. Brunner,
“Clenshaw-Curtis-Filon-type methods for highly oscillatory Bessel
trans-forms and applications,” IMA Journal of Numerical Analysis,
vol.31, no. 4, pp. 1281–1314, 2011.
[16] G. A. Evans and K. C. Chung, “Some theoretical aspects
ofgeneralised quadrature methods,” Journal of Complexity, vol.
19,no. 3, pp. 272–285, 2003.
[17] R. Piessens and F. Poleunis, “A numerical method for
theintegration of oscillatory functions,” vol. 11, pp. 317–327,
1971.
[18] A. Alaylioglu, G. A. Evans, and J. Hyslop, “The use of
Chebyshevseries for the evaluation of oscillatory integrals,”The
ComputerJournal, vol. 19, no. 3, pp. 258–267, 1976.
[19] M. Abramowitz and I. A. Stegun, Handbook of
MathematicalFunctions, National Bureau of Standards, Washington,
DC,USA, 1964.
[20] I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals,
Series, andProducts, New York Academic Press, 1965.
-
Submit your manuscripts athttp://www.hindawi.com
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttp://www.hindawi.com
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
Probability and StatisticsHindawi Publishing
Corporationhttp://www.hindawi.com Volume 2014
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
OptimizationJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
CombinatoricsHindawi Publishing
Corporationhttp://www.hindawi.com Volume 2014
International Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing
Corporationhttp://www.hindawi.com Volume 2014
International Journal of Mathematics and Mathematical
Sciences
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
The Scientific World JournalHindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume
2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com
Volume 2014 Hindawi Publishing Corporationhttp://www.hindawi.com
Volume 2014
Stochastic AnalysisInternational Journal of