PROJECT NAME

Corrected Quadrature Methods And Their Applications.

SUBMITTED BY

MR./MS. _____________________

COURSE:

SEMESTER/YEAR:

ENROLLMENT NUMBER:

SUBMITTED TO
SINGHANIA UNIVERSITY
Corrected Quadratures for Singular Integrals
1.Introduction
Singular and Hypersingular integrals play an important role in differential equations, numerical analysis,
mathematical physics and engineering. Such integrals mostly occur in modeling many physical situations like
aerodynamics, elastic theory, diffraction theory, electron optics, fracture analysis, acoustic, fluid mechanics etc.
Several methods such as direct definition, method of subtraction of singularity, method of regularization, sigmoidal
method etc., are available for the computation of singular and hypersingular integrals. Also, many quadrature rules
of Newton-Cote and Gauss type have been developed for the evaluation of these integrals. Hunter [18]
constructed Gauss-Legendre quadrature formula for Cauchy Principal Value Integral (CPVI) with zeros of Legendre
polynomial as mesh points. When the nodes and poles coincide with each other, quadrature rule is derived by
applying Cauchy’s Residue theorem or density function is expanded using Laurent expansion and the quadrature
rule is obtained by taking the limit of resulting equation. Chawla and Ramkrishnan [50] derived the quadrature
1
rules for CPVI containing Jacobi weight function w (t)=¿ for α =β=± . They only considered the case where
2
no pole coincides with zero of the approximating polynomial. In [69 ,67 , 47] quadrature rules are obtained for
z +h f (z )
complex CPVI of the form ∫ z −h
0
dz , along the line segment [ z 0−h , z 0 +h ] assuming that the function f (z)
0
z −z 0
is analytic in the domain, that contains the prescribed line segment. In [69], eight point quadrature rule is
1
constructed using nodes z 0 , z 0 ± h , z 0 ±ih , z 0 ± αh and z 0 ±iαh , where α ∈(0 ,1) is a parameter. For α = ,
√3
the degree of precision of the rule is 10. From this quadrature rule, Hota, Das and Bej [47] obtained some six
point quadrature rules by assigning particular values to the parameter α . Das and others [67], constructed
quadrature rules of precision eight, twelve and sixteen. All these quadratures with degree of precision n requires
x f (x )
n+1 nodes. Nyíry and Baranyi [3], converted generalized CPVI ∫ x −a +a dx , into Riemann integral
0
g( x)−g ( x 0 )
by suitable change of variable. Then the Riemann integral is evaluated by Romberg integration. Elliott and
Venturino [20] approximated the density function using Sigmoidal transformation and then applying Euler-
Maclaurin's expansion, quadrature rule for CPVI is obtained.

In fracture mechanics, crack problems are formulated as Fredholm integral equations of the form

∫ ❑ K (x , t) ϕ (t)dt=g(x ), c ∈(a , b),

a

where K ( x , t) is the kernel function of the form

K 1 (x , t) containing the singularity,
ϕ (t ) is the density
¿¿
function which is to be determined and g(x ) is the known function that corresponds to the load on crack
surfaces. In [89], singular integral

1
w(s)ψ (s )
∫❑ ¿¿
−1
where w (s) is the weight function given by w (s)=¿ is obtained. Since Chebyshev polynomials are orthogonal
with respect to this weigh w (s) ,ψ ( s) is approximated by Chebyshev polynomials of first and second kind. Ther
applying the inversion formulae, quadrature rules are constructed.

In [21],Yu derived generalized Trapezoidal rule for the hypersingular integral: of the form b ϕ (t ) and
∫a
¿¿
ϕ (t)
∫Γ dt
(
2 (t−s)
4 π sin ⁡
2 ) , where Γ is a unit circle. The quadra ture formulae are obtained by approximating the

density function by its Lagrang, interpolation. These rules are applied to Galerkin boundary element method, collo
cation boundary element method and adaptive boundary element method. But thest quadrature rules are less
accurate when mesh point coincides with the singular point
Many researchers developed Gaussian quadrature rules to evaluate hypersingula: integral [29 ,61 , 15 , 87].
Application of Gauss quadratures for the evaluation of hy persingular integral of the form
1 1 ϕ (t)
I (x)= ∫ −1 w(t) λ dt was first studied by Kutt [29] The formula uses Lagrange interpolation at equispaced
π t
points. This method do no give the solution if singularity is at endpoints. Kolm and Rokhlin [61] constructed the
quadrature rules for integrals where integrand contains algebraic and logarithmi singularity. The rule is derived by
approximating the density function by Legendr polynomial. Following the work of Kolm and Rokhlin, Carley [53]
developed th Gauss quadratures for both singular and hypersingular integrals. He improved the result by Kolm
using the approach of Brandão [52]. Advantage of this method is i computes hypersingular integrals where the
singularities are at endpoints. Hui and Shia [15] developed two types of Gauss quadrature rules for singular
integrals. Ther approximated the density function ϕ (t ) by using Legendre polynomial of first kind and Chebyshev
polynomial of second kind. Using these quadrature rules for singular inte grals and the relation between singular
and hypersingular integrals, Gauss-Legendr and Gauss-Chebyshev quadrature rules for hypersingular integrals are
1
derived. Esnaola and others [23] developed the Gauss quadrature rule corresponding to the weight . The
√1−t2
density function is nearly singular and is approximated by Chebyshev polynomial of first kind. Coefficients in the
series expansion are computed using fast Fourier transform. This rule is then applied to hypersingular integral. The
results so obtained are less accurate.

Linz [62] derived the Newton-Cote's quadratures for hypersingular integrals by approximating the density function
by piecewise polynomials at equispaced grid points. The method fails to give the solution when singular point and
mesh point coincides. This difficulty is overcome by Sun and Wu [87]. They developed the method using
Trapezoidal rule.

In [40] and references therein, interpolatory quadrature rule using Chebyshev polynomials of first, second, third
and fourth kind are derived.

b
In [10 ,11], Sidi derived Euler-Maclaurin's expansion for ∫ a f (x )dx under certain conditions. In [9] the more
general rule is developed for hypersingular integrals. He obtained the corrected Trapezoidal rule using Euler-
Maclaurin's expansion where f (x) contains algebraic and/or logarithmic singularities at the endpoints.
In this chapter, the corrected quadrature formulae for integral of the form
 1
1 ϕ (t)
I (x)− ∫ ❑ w(t )
π −1 ¿¿

with n=1 i.e. for singular integrals, are constructed where w (t) is weight function and
ϕ (t )∈ H r (α )(r ≥ 1 , 0<α ≤ 1) is a smooth function. H r (α ) is a class of functions that have continuous
derivatives up to order r −1, and the derivative of the r th order belongs to the Hölder class with an index α . The
1
corrected quadrature rules are constructed for the weight functions w (t)= 2 and √ 1−t . Orthogonal
2
√ 1−t
polynomials corresponding to these weight functions are Chebyshev polynomials of first and second kind
respectively. The function ϕ (t ) is approximated by Hermite interpolation polynomial where mesh points are zeros
of the corresponding Chebyshev polynomials. Properties of Chebyshev polynomials and their inversion formulae
are used to integrate the resulting expressions appearing in Hermite interpolation.

2.Preliminaries
Integral in equation (1.1.1) is obtained by approximating ϕ (t ) by Hermite interpolation polynomial where mesh
points are zeros of Chebyshev polynomial. This section contains definition of Hermite and Chebyshev polynomials,
some properties of Chebyshev polynomials, inversion formulae, differential equations of Chebyshev polynomials
and some recurrence relations.

2.1 Hermite interpolating polynomial [30]: Let f ∈ C(n+1) (I , K) and t 0 ≤ t 1 ≤ t 2 ≤ … ≤ t n in I be the prescribed
points then Hermite interpolating polynomial for the function f (t) is given by

n n
H 2 n (f :t )=∑ ❑hi (t )f ( t i ) + ∑ ❑ h́ i (t)f ' ( t i ) ,
i=0 i=0

[ '
] 2
where hi (t )= 1−2 ( t−t i ) l i ( t i ) l i (t),

2
h́i (t )=( t−t i ) l i (t )

and l i (t) are fundamental Lagrange polynomials.

2.2 Chebyshev polynomials: Chebyshev polynomials of first, second, third and fourth kind are denoted and
defined respectively as

a. T n (x)=cos ⁡(nθ)
sin ⁡(n+1) θ
b. U n (x )−
sin ⁡θ
c. C n (x)=
cos ⁡ n+( 12 ) θ
θ
cos ⁡
2

d. Sn (x )=
( )
sin ⁡ n+ θ
1
2
θ
sin ⁡
2
where x=cos ⁡θ and n=0 , 1, 2 , …

2.3 Zeros of Chebyshev Polynomials:

a. For T n (x), xi =cos ⁡( )
2 i−1
2n
π ,i=1, 2 , … , n.

U (x ), x =ens ⁡(
n+ 1 )
iπ
b. For n i ,i=1 , 2 , … , n.

C (x), x =cos ⁡(
2 n+1 )
2 i−1
c. For n i π , i=1 ,2 , … , n .

S (x ), x =cos ⁡(
2 n+1 )
2i
d. For n i π ,i=1 , 2 , … ,n .

2.4 Properties of Chebyshev Polynomials:

a. T n (1)=1 ,T n (−1)=¿ U n (1)=n+ 1 ,U n (−1)=(n+1)¿
1
b. T n (x)= [ U (x)−U n−2 (x)],
2 n

¿ 2 x T n−1 (x)−T n−2 (x)

¿ x T n−1 (x)−( 1−x 2 ) U n−2 (x)
¿ U n (x)−x U n −1 (x )

1
2
c. T n (x)=
2
[ 1+T 2 n (x) ],

1
¿ [ U (x)−U 2 n−2 (x)+2 ] .
4 2n
'
d. T n (x)=nU n−1 (x).
e. U n (x )=x U n−1 (x)+T n (x),
 ¿ 2 x U n −1 (x )−U n−2 (x )
x T n+1 (x )−T n+2 (x )
¿ 2
1−x

sin ⁡( 2n+ 2)o

f. U 2 n+ 1( x)= ,
sin ⁡θ

sin ⁡2(n+ 1)θ

¿
sin ⁡θ
2sin ⁡(n+1)θ cos ⁡( n+1)θ
¿
sin ⁡θ
¿ 2 U n (x )T n+1 (x )

Replacing n by n−1, we get

U 2 n−1 ( x)=2U n−1( x)T n (x ).

' x U n (x )−(n+1)T n+1 (x )

g. U n (x )= 2 .
1−x
1
2
h. U n (x )= [ U 2 n (x )−x U 2 n +2 ( x)+1 ] ,
2 ( 1−x 2 )

1−T 2n +2 (x)
¿ .
2 ( 1−x 2 )

1
i. T n (x)U n (x )= [ U ( x)+1 ] .
2 2n
j. 2 T n (x )T n +1( x)=T 2 n+1 (x )+T 1 (x ).
T n (x)+T n+1 (x )
k. C n (x)= .
1+ x

2 1+T 2 n+1 (x)

1 C n (x)= ,
1+ x
U 2 n+1 (x )−U 2 n−1 (x)+2
¿=
2(1+ x )
1
¿ ( C (x )+C 2 n+1 (x )+ 2 )
2(1+ x) 2 n
¿
 T n (x )−T n+1 (x )
m. Sn (x )=
1−x
2 1−T 2 n+1 (x)
n. Sn (x )= ,
1−x

1
¿ ( S2 n (x)−S 2 n+1 (x )+ 2 )
2(1−x )
¿

O. T 2 n−1 (x )−T 2 n+1 (x)=4 T n (x ) 1−x √ 2

√1−T ( x )
2
n

p. T 2 n−1 (x )+T 2 n +1 ( x)=2 x ( 2T n (x )−1 ).

2

q. U 2 n ( x )=
sin ⁡2 n+ ( 12 ) θ = 2sin ⁡(n+ 12 ) θ cos ⁡(n+ 12 ) θ ,
sin ⁡θ θ θ
2 sin ⁡ cos ⁡
2 2

¿ 2 Cn (x )S n (x)

2.5 If x i is root of T n (x) then,

a. T 2 n−1 ( x i )−T 2 n+1 ( x i) =0 ,
' '
b. T 2 n−1 ( x i )−T 2 n+1 ( x i) =4 n,

'' '' 4 n xi
c. T 2 n−1 ( x i )−T 2 n+1 ( x i) = 2 ,
xi −1

( )
2
'' ' '' ' 4n
2
xi + 2
d. T 2 n−1 ( x i )−T 2 n+1 ( x i) =−4 n + ,
1−x 2i ( 1−x 2i ) 2

[ ]
''
T 2 n−1 (x)−T 2 n+1 ( x ) −12n x i
e. 2
= 2,
x −1 x=x i ( x 2i −1 )
f. T 2 n−1 ( x i ) +T 2n +1 ( x i ) =−2 x i .
2.6 Inversion formula:
1 1 T n (t)
a. ∫ −1 dt =U n−1 ( x ),
π √1−t2 (t−x )
1 1 √ 1−t U n (t )
2
b. ∫ −1 dt =−T n+1 (x ),
π (t−x)

c. ∫ −1
π √
1 1 1+ t C n (t)
1−t (t−x)
dt=S n (x),

d. ∫ −1
π √
1 1 1−t S n (t)
1+ t (t−x)
dt=−Cn (x ).
Differentiating above integrals w.r.t. x we get,
 1 1 T n (t)
a. ∫ ,
π −1 √ 1−t 2 ¿ ¿
1
b. ∫ 1−1
√1−t2 U n (t ) ,
π ¿¿
1
c. ∫ 1−1
π √1+ t C n (t) ,
1−t ¿ ¿
1
d. ∫ 11
π √1−t S n (t ) .
1∣ t ¿ ¿
2.7 In particular for n=0 in 2.6 ( b)

1
∫
1
❑
√1−t dt=−x
2

π −1 (t−x )

Differentiating above integral w.r.t. x we get,

1
1

∫ ❑
√1−t 2

π −1 ¿¿

2.8 For x=1 ,−1 , property 2.7 reduces to

1
∫
1
❑
√1−t dt=± 1
2

π −1 (t ± 1)

1 1 dt
2.9 a.∫ ❑−1 for r =1, 2.
π √ 1−t 2 ¿ ¿
1 1 dt
b. ∫ ❑−1 =0 ; for r =1, 2.
π √ 1−t ( t−x ) ( t−x i )r
2

1 1 dt
c. ∫ ❑−1 for k , r =1, 2.
π √ 1−t ( 1−t 2) ¿ ¿
2

1 1 dt
d. ∫ ❑−1 for k , r =1, 2
π √ 1−t (1 ±t)¿ ¿
2

1 1 1−x i t
e. ∫ ❑−1 for k =1 ,2.
π √ 1−t 2 ( 1−t 2) ¿ ¿
f.
1 √ 1−t 2
∫ −1 (1± t)¿ ¿ for k , r =1, 2.
❑
1

π
1
g. ∫ −1 ❑
1 √ 1−t 2
=−1 ; for r =1,
r
π ( t−x ) ( t−x i)
¿ 0 ; for r =2
2.10 Differential Equations: Chebyshev polynomial of first and second kind satisfy the following differential
equations.
a. ( x 2−1 ) T ''n ( x ) =n2 T n ( x )−x T 'n ( x ) ,
b. ( x 2−1 ) U 'n' ( x )=n ( n+ 2 ) U n ( x )−3 x U 'n ( x ) .
2.11 Recurrence relations:
a. T n+1 ( x )=2 x T n ( x )−T n−1 ( x ) where n=1 ,2 , 3 , …
b. U n +1 ( x )=2 x U n ( x )−U n−1 ( x ) where n=1 ,2 , 3 ,
c. C n+1 (x )=2 x C n (x)−C n−1 (x) where n=1 ,2 , 3 , …
d. Sn +1 ( x)−2 x S n (x)−S n−1( x) where n−1 ,2 , 3 , …

3. Corrected Quadratures for Singular Integrals

In this section singular integral (4.1.1) with n=1 is evaluated by using zeros of Chebyshev polynomials as mesh
points and using Hermite interpolation to approximate the ınknown density function ϕ (t ).

1
3.1 w (t)= :
√ 1−t 2
In this section, we evaluate

1
1 ϕ(t)
I (x)= ∫ ❑ dt (3.1)
π −1 √ 1−t 2 (t−x)

3.1.1 Mesh points x i 's are zeros of T n (x)

Hermite interpolation at zeros of first kind Chebyshev polynomial T n (x) , i.e. at x i=¿

cos ⁡ ( 2i−1
2n )
π , i=1 ,2 , ⋯ , n approximates ϕ (t ), in the equation (3.1), as

[ (1−x i ) T n (t) '

]
2 2 2
( 1−x i t ) T n (t)
n
H 2 n (ϕ , t)=∑ ❑ 2 2
ϕ ( x i ) + 2
ϕ ( x i ) ( 3.2)
i=1 n ( t−x i) n ( t−x i )

Therefore equation (3.1) is approximated by

n
1
I (x)=∑ ❑
i=1 n
2 [ 2 '
]
I 1 i (x ) ϕ ( x i ) + ( 1−xi ) I 2 i ( x)ϕ ( x i ) (3.3)

where

2
1
1
( 1−xi t ) T n (t)
I 1i (x )= ∫ ❑ dt
π −1 √ 1−t 2 (t−x ) ( t−x )2
i

and
 1 2
1 T n (t)
I 2i (x )= ∫ ❑ dt
π −1 √ 1−t (t−x ) ( t−x )
2
i

From property 2.4 (c ) we have

1
1
( 1−xi t ) [ T 2 n (t)+1 ]
I 1i (x ) ¿ ∫ ❑
2 π 1 √ 1−t 2 (t−x) ( t−x ) 2
dt
i
¿ ¿

Applying the method of partial fractions and using inversion formulae, we get

{ }
1 2

( )
1 T (t) x x i−1 1 1 x i −1
I 1i (x ) ¿ ∫ ❑ 2n 2 − +
2 π −1 √ 1−t ( x−x i ) t−xi t−x ( t−xi )2 ( x−x i )
2
dt ,

¿ ¿

From 2.4(f), we have U 2 n−1 (x)=2U n−1( x)T n (x ).

Therefore at x=x i (roots of T n (x) ), U 2 n−1 ( xi ) =0.

1 ( xi ) .

Similarly, with the help of properties 2.4 (c ), 2.4 (f ) and using inversion formula

1 2
1 T n (t)
I 2i (x ) ¿ ∫❑ dt ,
π −1 √ 1−t 2 (t−x) ( t−x )
i
1
¿ ¿
2 ( x−x i)
[ U 2 n−1 ( x)−U 2 n−1 ( x i ) ] ,
¿ ¿

Thus from equation (3.4) and (3.5) ēquation (3.3) bēcomès

[ ]
2
n
ϕ ( x i ) 1−x x i x −1 '
I (x)=¿ ∑ ❑ U 2 n−1 (x)+ i U (x )
i=1 2 n2 ( x−x i )
2
x −xi 2 n−1 i
n
U 2 n−1(x )
+∑ ❑ ϕ ( x i) ( 1−x i ) (3.6)
' 2
2
i=1 2 n ( x−x i )

When x coincides with one of the grid point say x i then

I 1i ( x i )=lim I 1 i( x),
x→ xi
 { }
2
1−x x i x i −1
¿ lim U 2 n−1 (x)+ U '2 n−1 ( x i )
x→ x i 2 ( x−x i ) 2
2 ( x−x i )

Substitute x−x i=u . Then u → 0 as x → x i. Thus we get

{ 1−xi ( x i +u )
}
2
x −1 '
I 1i ( x i )=lim i U 2n−1 ( x i ) + 2
U 2n −1 ( x i+ u ) .
u →0 2u 2u

Expanding U 2 n−1 ( xi +u ) using Taylor's series and using 2.4 (f ), we get

{
2
x i −1 '
I 1 i ( x i )=¿ lim U 2 n−1 ( x i )

]} { [ ]
u →0 2u 2 2 2
' x −1 1−x i −x i u +1−x i −x i u ' '
( x i ) i2 u + U 2 n−1 ( x
[
2
+1−x i −x i u 2
u '' lim U 2 n−1 ¿
'
U 2 n−1 ( x i ) +u U 2 n−1 ( xi ) + U 2n −1 ( x i ) +⋯ u →0 2u 4
2u 2
2!
¿

Similarly,

I 2i ( x i ) ¿ lim I 2i (x )
x→ x i

lim U 2n−1 ( x i+u )

u→0
¿ ¿
2u
1 '
¿ ¿ U 2 n−1 ( x i) .(3.8)
2

Thus x is one of the grid point say x j from equations (3.3)-(3.8) we get

{ }
2
n
ϕ ( xi ) 1−x i x j x −1
I ( x j) ¿ ∑ ❑ U '
2 n−1 ( x j ) + x i−x U '2 n−1 ( x i )
i=1 2n 2 ( x j −xi )2 j i
i≠ j
2
+1−x i ' '
¿ 2
ϕ ( x j ) U 2 n−1 ( x j ) .(3.9)
2n

3.1.2 Mesh points x i 's are zeros of U n (x )

Approximate ϕ (t ) by Hermite interpolating polynomial where mesh points are zeros of second kind Chebyshev

polynomial U n (x ), i.e. at x i=cos ⁡ ( n+1i π ) ,i=0 ,1 , 2 ,⋯ , n , then H 2 n (ϕ , t) is given by

2
n
( x 2i −1 ) U 2n (t )
H 2 n (ϕ , t)=∑ ❑ ¿¿
i=1
Then approximate value of equation (3.1) becomes

Integral I 2i (x ) is evaluated by using method of partial fraction, inversion formula 2.6 (a) and 2.4 (a)

1 2
1 U n (t)
I 2i (x )=¿ ∫ ❑
π −1 √ 1−t 2 (t−x ) ( t−x )
dt 1
∫
1
❑
[ 1−T 2 n+2 (t )] dt ¿=¿−
1
∫
1
❑
T 2 n+2 (t )
i 2 π −1 ( 1−t 2 ) √ 1−t 2(t−x ) ( t−x ) 2 π −1 √ 1−t 2 ( 1−t 2 ) (t−x ) ( t
¿ i

Since U 2 n+ 1( x)=2 U n ( x )T n+1 ( x ), at x=x i ¿ root of U n (x ) ) we get

Ú 2 n+ 1 ( x i )=2 Ú n ( x i ) T́ n +1 ( x i )=0.

Thus,

(n+1) ( x x i +1 ) U 2n +1 (x)
I 2i (x )= + (3.11).
( x −1 ) ( x −1 )
2
i
2
2 ( x−x i ) ( x 2−1 )

Consider

( )√
1
1 1 3 xi U 2n (t)
I 1i (x ) ¿ ∫ ❑ 2
+ 2 dt
π −1 ( t−x i ) ( x i −1 ) ( t−x i ) 2
1−t (t−x )
3 xi 1
1
[ 1−T 2 n+2 (t)]
¿ ¿ 2 I 2i (x )+ ∫ ❑ dt
x i −1 2 π −1 √ 1−t ( 1−t 2 ) (t−x ) ( t−x i )2
2

¿ ¿

Applying the method of partial fractions, using inversion formulae 2.6 (a) and 2.4 (a),
we get

{
1
3 xi 1 T 2 n+2 (t) 1
I 1i (x )=¿ 2 I 2i (x )− ∫ ❑
x i −1 2 π −1 √ 1−t ( x i −1 ) ( x−x i ) ( t−x i )2
2 2

+1 1

{ (
2
− 2 3 xi n+1 n+1 +1
2 ( x i−1 ) (x−1)(t−1) 2 ( x i+1 ) ( x+1)(t+ 1) + ¿
x −1 2 ( x i−1 ) ( x−1) 2 ( xi +1 ) ( x+1) 2 ( x−x i )
2

}
2 i
+3 x −2 x x i−1
i 1
2
− dt
( x −1 ) ( x−x i ) ( t−x i )
2
i
2
( x −1 ) ( x−x i )2 (t−x)
2

Thus from equations (3.11) and (3.12), equation (3.10) becomes

 2
n
(1−xi2)
I ( x )=∑ ❑ ¿¿
i=1

(n+ 1) ( 2 x x 2i −x + x i )
}
'
−U 2 n+1 ( xi )
+ 2
2 ( x i −1 ) ( x−x i) ( x 2i −1 ) ( x 2−1 )
2

2
n
( x 2i −1 )
+∑ ❑
i=1
¿¿
¿

When x coincides with one of the grid point say x i then

I 1 i ( x i ) =lim I 1 i (x)
x → xi

¿ lim
{ U 2 n+1 (x ) 1
x → x 2 ( x −1 ) ( x−x ) x−x i
i
2
i
(
3 xi
+ 2 −
U '2n +1 ( x i )
)
x i −1 2 ( x 2i −1 ) ( x −xi )

+(n+1) ( 2 x x 2i −x+ x i )
2
( x 2i −1 ) ( x 2−1 ) }
−x i=u then u → 0 as x → x i . Thus we get

{ U '2n +1 ( x i ) (n+1) (2 x i ( x i +u ) −u )
}
2

( )
U 2 n +1 ( x i +u )
1 3 xi
U i )=lim + − +
(( i ) )
u →0 2 x +u −1 u u
2
x 2
i −1 2 ( x
2
i −1 ) u ( x 2i −1 ) (( x i +u ) 2−1 )
2

If x−x i=u then u → 0 as x → x i. Thus we get

{ U 2 n+1 ( x i ) (n+1) ( 2 x i ( x i+u )−u )

}
2

( )
'
U 2 n+1 ( x i+ u )
1 3 xi
I 1 i ( x i )=lim + 2 − +
( ( i )2 )
u→ 0 2 x +u −1 u u xi −1 2 ( x 2i −1 ) u ( x2i −1 ) ( ( x i+ u )2−1 )
2

Expanding U 2 n+1 ( x i+u ) using Taylor's series about x i and substituting the limits we
get,
' ''
xi U 2 n+ 1 ( x i ) U 2 n +1 ( x i ) 2( n+1) x 3i
I 1 i ( xi ) ¿ + + .(3.14)
4 ( x 2i −1 ) ( x 2i −1 )
2 3
2 ( x i −1 )
2

Similarly ¿ lim I 2 i (x)

x → xi

(n+1) ( x +1 ) U '2 n+1 ( x i )

2
i
¿ ¿ + .(3.15)
2 ( x i −1 )
2 2
( x 2i −1 )
Similarly
Thus from equations (3.14) and (3.15), equation (3.13) at x=x j becomes
 ¿
¿

3.1.3 Grid points x i 's are zeros of C n (x)

In this section we approximate the function ϕ (t ) hy Hermite interpolation at zerns of third kind Chebyshev

polynomial i.e. at x i=cos ⁡ ( 22 i−1

n+1 )
π , i=1 , 2 ,⋯ , n as

2 ( 1+ x i )2 ( 1−x i )
n
H 2 n (ϕ , t)=∑ ❑ ¿¿
i=1

By approximating ϕ (t ) by H 2 n (ϕ , t), equation (3.1) reduces to

1
1 H 2 n (ϕ , t)
I (x) ¿ ∫ ❑ dt
π −1 √ 1−t 2 (t−x)
¿ ¿

Where

1 2
1 C n (t)
I 2i (x )= ∫ ❑ dt .
π −1 2
√
1−t (t−x ) ( t−x i )

Applying the method of partial fractions and using the properties 2.4 (1), 2.4 (q) and
2.6 , we get

1
1
[T 2 n+1 (t )+ 1 ]
∫
{
1 1
I 2i (x )=¿ ❑ dt 1 12 n +1 (t) 1 T 2 n+1 (t ) −1
π −1 √ 1−t 2 (t−x ) ( t−x ) (1+t ) ∫ ❑
π −1 √ 1−t (1+t )(t−x) ( t−x i )
dt ¿=¿ ∫ ❑
¿
i 2 π −1 √ 1−t ( x i ∣ 1 )( x x i )(
2

−U 2 n ( x i ) U2n ( x ) U 2 n (−1 )
¿ + +
( x i+1 ) ( x−x i ) ( x +1 ) ( x−x i ) ( x i +1 ) ( x+ 1 )
U2n( x ) 2n+1
¿ + ( 3.18 )
( x +1 ) ( x−x i) ( x i+ 1 ) ( x +1 )

And

( )√
1 2
1 2 x i−1 C n (t)
I 1i (x )=¿ ∫ ❑ 1+ 2 [ T 2 n+1 (t)+1 ]
1
( t−x i ) dt 2 x i−1 1
π −1 x i −1 2
1−t ( t− x) ( t−xi )
2 I 2 i (x)+ ∫ ❑ dt ¿=¿
2
x i −1 π −1 √ 1−t (1+t )(t−x) ( t−x )2
2
¿ i
Thus from equations (3.18) and (3.19), equation (3.17) becomes

When x coincides with one of the grid point say x=x i then

I 1i ( x i ) ¿ lim I 1i (x )
x→ x i
' ''
xi U ( xi )
2n U (x )2n i (2 n+1)x i
¿ ¿ + + (3.21)
( x i+ 1 ) ( x i−1 ) 2 ( xi +1 ) ( x i+1 ) ( x i−1 )
2 3

and

I 2i ( x i ) ¿ lim I 2i (x )
x→ x i

¿ ¿
1
(U ' ( x )+
2n+ 1
x i+ 1 2 n i x i+1
(3.22)
)
Thus Thus when x = xj , one of the grid point, then from equations (3.17)-(3.22), we get

I ( x j) ¿

3.1.4 Interpolation points x i 's are zeros of Sn (x )

We approximate the function ϕ (t ) by Hermite interpolation polynomial H 2 n (ϕ , t) at zeros of fourth kind

Chebyshev polynomial i.e. at x i=cos ⁡ ( 2 n+1

2i
π ), i=1 , 2 ,⋯ , n as

2 ( 1−x i )2 ( 1+ x i )
n
H 2 n (ϕ , t)=∑ ❑ ¿¿
i=1

Then

1 1
1 ϕ( t) 1 H 2n (ϕ , t)
I (x) ¿ ∫ ❑
π −1 √ 1−t (t−x)
2
dt ≅ ∫ ❑
π −1 √ 1−t 2 (t−x )
dt

¿ ¿

where

1 2
1 S n (t)
I 2i (x )= ∫ ❑ dt
π −1 √ 1−t (t−x ) ( t−x )
2
i

and
 ( )
1 2
1 S n (t) 1 2 x i+1
I 1i (x ) ¿ ∫ ❑ + 2 dt
π −1 √ 1−t (t−x ) ( t−x i ) ( x i −1 ) ( t−xi )
2 2

¿ ¿

Consider

1 2
1 S n (t)
I 2i (x )= ∫ ❑
π −1 √ 1−t (t−x ) ( t−x )
2
dt
i

then from property 2.4 (n), inversion formula, 2.4 (a) and 2.4 (q)

1
1
[ 1−T 2n +1 (t)]
∫
{
1 1
I 2i (x )=¿ ❑ dt 1 T 2 n+1 (t ) 1 T 2 n +1(t)
π −1 √ 1−t (t−x ) ( t−x ) (1−t) − ∫ ❑
2 dt ¿=¿− ∫ ❑
i π −1 √ 1−t (1−t)(t−x ) ( t−x i )
2 π −1 √ 1−t ( x i−1 ) (
2
¿

Now

1 2
2 x i +1 1 S n (t)
I 1i (x )= 2
x i −1
I 2 i (x)+ ∫ ❑
π −1 √1−t (t−x ) ( t−x )2
2
dt
i

2 x i +1 1
1
[ 1−T 2 n+1 ( t ) ]
¿ 2 I 2i( x )+ ∫ ❑ dt
x i −1 π −1 √1−t (1−t )( t−x ) ( t−x i) 2
2

1
2 x i +1 1 T 2 n+1 (t )
¿ 2 I 2 i ( x )− ∫ ❑ dt
x i −1 π −1 √1−t 2 ( 1−t )( t−x ) ( t−x i )2
Applying the method of partial fractions and using inversion formulae and equation (3.25), we get

[ ]
'
2 x i +1
U 2 n (x) 2 n+1 U 2n ( xi )
I 1i (x )=¿ 2 + −
x i −1 ( x−1) ( x−x i ) ( x i−1 ) ( x+ 1) ( x i−1 ) ( x−x i )
−2 xi −x−1
2
( x i−1 ) ( x−x i )
2
U 2 n ( x i) −
U 2 n (1)
2
+
( xi −1 ) (x−1) (x−1) ( x−x i )
U 2 n (x )
2
U 2 n (x)
[
2 x i +1
+
1
+
] xi (2 n+1)
(x−1) ( x−x i ) x i −1 x−x i ( xi −1 )2 ( x i +1 ) (x −
2

Thus from equations (3.25), (3.26) and (3.24), we get

When x coincides with any of the grid point say x=x i then
 I 1 i ( x i )=lim I 1i (x )
x→ x i

¿ lim
x → xi { U 2 n (x)
(x −1) ( x−x i ) x −1 [ 2 xi +1
2
i
+
1
+
] x i( 2n+ 1)
x −xi ( x i−1 )2 ( xi +1 ) ( x−1)

}
'
−U 2 n ( x i )
(3.28)
( x i−1 ) ( x−x i )
' ''
x i (2 n+ 1) x i U 2n ( x i ) U 2 n ( xi )
and + +
( x i−1 ) ( x i +1 ) ( x i−1 ) ( x i +1 ) 2 ( x i−1 )
3 2

I 2 i ( x i )=lim I 2i (x )
x→ x i

¿ lim
x → xi { U 2 n (x) 2 n+1
+
(x −1) ( x−x i ) ( x i−1 ) (x−1) }
¿
1
( '
U ( x )+
2n+ 1
xi −1 2 n i xi −1 )
(3.29)
Thus from equations (3.24)-(3.29), we get

3.2 w (t)= √1−t2

In this section the following singular integral I (x) is evaluated.

1
1
√ 1−t 2 ϕ (t ) dt .(3.31)
I (x)= ∫ (t−x )
π −1
❑

3.2.1 Mesh points x i 's are zeros of T n (x)

(2i−1)π
Interpolating ϕ (t ) by Hermite interpolation at roots of T n (t) i.e. at x i=cos ⁡ we get
2n

[ (1−x i ) T n (t) '

]
2 2 2
( 1−x i t ) T n (t)
n
H 2 n (ϕ , t)=∑ ❑ 2 2
ϕ ( x i ) + 2
ϕ ( xi )
i=1 n ( t−x i) n ( t−x i )

Then, equation (3.31) becomes,

n
1
I (x)=∑ ❑
i=1 n
2 [ 2 '
]
I 3 i (x ) ϕ ( x i ) + ( 1−x i ) I 4 i ( x )ϕ ( x i ) ,(3.32)

where
 1
1
√1−t ( 1−x i t ) T n (t) 2 2

I 3 i( x)= ∫ ❑ 2
dt
π −1 (t−x) ( t−x i )

1
1
√1−t 2 ( 1−x i t ) ( U 2 n(t)−U 2 n−2 (t)+2 )
¿ ∫
π −1
❑
4(t−x ) ( t−x i )2
dt

1
1
¿ ∫❑
√1−t ( 1−x i t ) ( U 2 n (t)−U 2 n−2 (t)) dt+ x i
2

π −1 4(t−x ) ( t−xi )2 2

and

I 4 i (x )
1
1

¿ ∫❑
√ 1−t 2 T 2n (t)
dt
π −1 (t− x) ( t−xi )
1
1
√ 1−t 2 ( U 2 n (t)−U 2n −2 (t )) 1
¿ ¿ ∫❑ dt−
π −1 4(t− x) ( t−xi ) 2

Applying the method of partial fractions to the integrals in I 3 i( x) and I 4 i (x) and using inversion formulae we
get,

2
xi x i −1
I 3 i (x)=¿ + (T ' ( x ) −T '2 n+ 1 ( x i ) )
2 4 ( x−x i ) 2 n−1 i
+1−x x i
4 ( x −xi )
2 [ (T 2 n−1 (x )−T 2 n+1 (x )) −( T 2 n−1 ( x i ) −T 2 n+ 1 ( x i ) ) ]
Using the properties (2.5 a) and (2.5 b) we get,

x i n ( x i −1 ) 1−x x i
2

I 3 i( x)= + + ( T (x)−T 2 n +1 ( x) ) .(3.33)

2 ( x−x i) 4 ( x−x i )2 2 n−1

Similarly,
1
I 4i (x )= ¿
4 ( x−x i )

Thus from equations (3.33) and (3.34) we get,

[ x i n ( x i −1 ) 1−x x i
]
2
n
ϕ ( xi )
I (x)=¿ ∑ ❑ + + 2 ( 2 n−1
T (x )−T 2 n+1 (x) ) 2
i=1 2 ( x−x i ) 4 ( x−x i ) n

(
( 1−x 2i ) ( T 2 n−1 (x)−T 2 n +1 (x))
)
n
+∑ ❑ −1 ϕ ' ( x i ) (3.35)
i=1 2n
2
2 ( x−x i )
Where x is not any of the grid points.

When x coincides with one of the grid points say x i then

I 3 i ( xi ) ¿ lim I 3i ( x i )
x→ x i
−x i
¿ ¿ (3 n−1)(3.36)
2

And

I 4 i ( xi ) ¿ lim I 4 i ( x i )
x → xi
1
¿ ¿ (2 n−1)(3.37)
2

Thus from equations (3.33), (3.34), (3.36) and (3.37), equation (3.32) becomes

[
x i n ( x i −1 ) 1−x j x i
]
2
n
ϕ ( xi )
I ( x j )=¿ ∑ ❑ + + 2 ( 2 n−1 ( j )
T x −T 2n +1 ( x j ) ) 2
i=1 2 ( x j−x i ) 4 ( x j−x i ) n
i≠j

[
(1−x 2i ) (T 2 n−1 ( x j ) −T 2n +1 ( x j ) )
]
n
+∑ ❑
'
−2 ϕ ( xi )
i=1 4n
2
( x j−x i )
i≠ j

−x j (1−x 2j ) '
2
(3 n−1)ϕ ( x j ) + 2
(2 n−1)ϕ ( x j ) .(3.38)
2n 2n

3.2.2 Mesh points x i 's are zeros of U n (x )

iπ
Interpolating ϕ (t ) in equation (3.31) by Hermite interpolation at roots of U n (t ) i.e. at x i=cos ⁡ we get
n+ 1
2
n
( x 2i −1 )
U n (t )
2

H 2 n (ϕ , t)=∑ ❑ ¿¿
i=1

Then, equation (4.3.31) is approximated by,

2
n
( x 2i −1 )
I (x)=∑ ❑ ¿¿
i=1

Where

1
1

I 4 i (x)= ∫ ❑
√ 1−t 2 U 2n (t)
dt
π −1 (t−x) ( t−x i )
and

∫ ❑ √t−x ( )
1 2
1 1−t 1 3 xi 1 2
I 3 i( x) ¿ 2
+ 2 U n (t) dt
π −1 ( t−x i) xi −1 ( t−x i)

¿ ¿

Consider,

1
1

I 4 i (x)= ∫ ❑
√ 1−t 2 U 2n (t)
dt .
π −1 (t−x) ( t−x i )

From 2.4(h) we have

1
1
√ 1−t 2 [ 1−T 2 n +2 (t)]
I 4 i (x ) ¿ ∫❑
2 π −1 ( 1−t 2) (t−x) ( t−x i )
dt
1
1 T 2 n+2 (t )
¿ ¿ 0− ∫ ❑
2 π −1 √ 1−t 2 (t−x ) ( t−x )
dt
i
1
¿ ¿− [ U (x)−U 2 n+1 ( xi ) ] .
2 ( x−x i ) 2n +1

we have

From the property 2.4 (q), we have

U 2 n+ 1( x)=2 U n ( x )T n+1 ( x ).

At x=x i ( root of U n (x ) ) we get

U 2 n+ 1 ( x i )=2 U n ( x i ) T n +1 ( x i )=0.

Thus,

−U 2 n+1 (x)
I 4 i (x)= (3.40)
2 ( x−x i )

Now consider

I 3 i( x)=¿ 2
3 xi 1
I 4 i (x)+ ∫ ❑
1
√ 1−t 2 U 2n (t)
dt 3 x i I ( x)+ 1 ❑
1
[ 1−T 2n +2 (t)] 3 xi
x i −1 π −1 (t−x ) ( t−x i )2
2
x i −1
4i ∫
π −1 2 √ 1−t (t−x) ( t−x )
2 2
dt ¿=¿ 2
x i −1
I 4 i (x)+0−
i
¿

Substituting equations (3.40) and (3.41) in equation (3.39) we get

 ¿

Where x is not any of the grid points. When x is one of the grid points then

{ [ ] }
'
−U 2 n+1 ( x ) 3 x i 1 U 2 n+1 ( x i )
I 3 i ( x i )=lim + + .
x → xi 2 ( x−x i ) x i −1 x−x i 2 ( x−x i )
2

Put x−x i=u then as x → x i , u → 0.

{ ( ) }
'
−1 3 x i 1 U 2 n+1 ( x i )
+ [ U 2 n +1 ( x i ) +u U 2 n+1 ( x i ) +⋯ ] +
'
I 3 i ( x i ) ¿ lim 2
u →0 2 u x −1 u
i
2 ( x−x i )
3 xi ' 1 ''
¿ ¿− U 2n +1 ( x i ) − U 2 n+1 ( xi ) .(3.43)
2 ( x −1 )
2
i
4

Similarly,

I 4 i ( xi ) =lim I 4 i ( x)
x → xi

lim −U 2 n+1 ( xi +u )
u→0
¿
2u
lim −1
[ ]
2
u→0 ' u ''
¿ U 2 n+1 ( x i )+ uU 2 n+1 ( xi ) + U ( x ) +⋯
2u 2 ! 2n +1 i
1 '
¿− U 2 n +1 ( x i ) .(3.44)
2

Thus, from equations (3.40),(3.41),(3.43) and (3.44) in equation (3.42)

we get

I ( x j) ¿

3.2.3 Grid points x i 's are zeros of C n (x)

We approximate ϕ (t ) in equation (3.31) by Hermite interpolation at zeros of third kind Chebyshev polynomial i.e.

at x i=cos ⁡ ( 22 i−1
n+1 )
π , i=1 , 2 ,3 , … , n .

2
2 ( 1∣ xi ) ( 1 xi )
n
H 2 n (ϕ :t)=∑ ❑ ¿¿
i =1

Equation (3.31) now becomes

 2
2 ( 1+ x i ) ( 1−x i )
n
I (x)≈ ∑ ❑ ¿¿
i=1

where

√
( )
1
1 2 x i−1 1−t 2 C 2n (t)
I 3 i( x)= ∫ ❑ 1+ 2 ( t−x i ) 2
dt
π −1 x i −1 (t−x ) ( t−xi )

and

1
1
√ 1−t 2 C 2n (t) dt
I 4 i (x)= ∫ ❑
π −1 (t−x) ( t−x i )

Consider,

1
1
√ 1−t 2 C 2n (t) dt
I 4 i (x ) ¿ ∫ ❑
π −1 (t−x) ( t−x i )
1
1
√ 1−t 2 ( U 2 n+1 (t)−U 2 n−1 (t))
¿ ¿ ∫❑ dt
π −1 2(1+t)(t−x ) ( t−x i )

Applying the method of partial fractions and then using inversion formulae and the property 2.4 (a), we get

T 2 n (x)−T 2 n+ 2( x) T 2 n ( xi ) −T 2n +2 ( x i )
I 4 i (x)= − .(3.47)
2(x+1) ( x−x i ) 2 ( x i +1 ) ( x−x i )

Consider,

√ 1−t 2 C 2n (t) dt
( )
1
1 2 x i−1
I 3 i( x) ¿ ∫
π −1
❑ 1+
xi
2 ( t−x i)
(t− x) ( t−xi )
2

2 xi −1 1
1
√1−t ( U 2 n+1 (t )−U 2 n−1 (t))
2

¿ ¿ 2 I 4 i (x)+ ∫ ❑ 2
dt
x i −1 π −1 2(1+t)(t−x ) ( t−x i )

Applying the method of partial fractions and then using inversion formulae, the property (2.4 a) and equation
( 1−x x i ) [ T 2 n ( x i )−T 2 n+2 ( x i) ]
(4.3.47),we get
I 3 i( x)=¿
T 2n (x )−T 2 n+2 (x ) 2 x i−1
2(x +1) ( x−x i ) 2
(+
1
xi −1 x−x i
'
+
) 2
2 ( x i−1 ) ( x i+1 ) ( x−x i )
'
2

−T 2n ( x i )−T 2 n+2 ( x i )
(3.48)
2 ( x i+1 )( x −xi )
From equations (3.47)-(3.48), equation (3.46) becomes
 ¿

Where x is not any of the grid points. When x coincides with one of the grid point say x i,

I 3 i ( x i )=¿ lim I 3 i (x)

x → xi
¿
lim
x→x i {
T 2n (x )−T 2 n+2 (x ) 2 x i−1
2(x +1) ( x−x i ) 2
+
(
1
x i −1 x−x i
¿
) 2
2 ( x i−1 ) ( x i +1 ) ( x−x i )
2
−
2 ( x i+ 1 )( x−xi ) }
+ ( 1−x x i ) ( T 2 n ( x i ) −T 2 n +2 ( x i ) ) T '2n ( x i )−T '2 n+2 ( x i )
¿=

And

I 4 i ( xi ) ¿ lim I 4 i (x)= lim

x → xi x → xi { T 2 n (x )−T 2n +2 (x) T 2 n ( x i )−T 2 n+2 ( xi )
2(x +1) ( x−x i )
−
2 ( x i+1 ) ( x−x i ) }
¿ ¿

From equations (3.49)-(3.51) we get

I ( x j) ¿

'
3.2.4 Mesh points x i s are zeros of Sn (x )
We approximate ϕ (t ) in equation (4.3.31) by Hermite interpolation at zeros of fourth kind Chebyshev polynomial

i.e. at x i=cos ⁡ ( 2 n+1

2i
π ), i=1 , 2 ,3 , … , n .

2
2 ( xi −1 ) ( x i +1 )
n
H 2 n (ϕ , t)=∑ ❑ ¿¿
i=1

Equation (3.31) now becomes

n
2 ( x i−1 )2 ( x i+ 1 )
I (x)≈ ∑ ❑ ¿¿
i=1

where

√
( )
1
1 2 x i+ 1 1−t 2 S2n (t )
I 3 i( x)= ∫ ❑ 1+ 2 ( t−x i ) 2
dt
π −1 x i −1 (t−x ) ( t−x i )

and

1
1

I 4 i (x )= ∫ ❑
√ 1−t 2 S2n( t)
dt = 1 ❑ √ 1−t S n (t) dt ¿= 1 ❑ √ 1−t ( U 2 n−1 (t)−U 2 n+1 (t )+2 ) dt ¿= 1 ❑ √ 1−
1 2 2 1 2 1

π −1 (t−x) ( t−x i ) ∫
π −1 (t−x) ( t−x i )
∫
π −1 2(1−t )(t−x) ( t−x i )
∫
π −1
¿
Consider,
Applying the method of partial fractions and then using inversion formulae and the property 2.4 (a), we get

T 2 n+2 ( x i )−T 2 n ( x i ) T 2 n +2 (x)−T 2 n (x)

I 4 i (x)= − (3.54)
2 ( x i−1 ) ( x−x i ) 2(x−1) ( x−x i )

Consider,

√1−t S n (t) dt
( )
1 2 2
1 2 xi +1
I 3 i( x)= ∫
π 1
❑ 1+ 2
xi −1
( t−xi )
(t−x) ( t−x i )
2

2 x i +1 1
1
√1−t 2 ( U 2 n−1 (t)−U 2 n+1 (t )+ 2 )
¿ 2 I 4 i (x )+ ∫ ❑ dt
x i −1 π −1 2 (1−t)(t−x ) ( t−x i )2
2 x i+1 1
1
√1−t ( U 2 n−1 (t)−U 2n +1 (t)) dt
2

¿ I 4 i (x)+ ∫
π −1
❑
x 2i −1 2 (1−t)(t−x ) ( t−x i )2

Applying the method of partial fractions and then using inversion formulae, the property 2.4 (a) and equation
(3.54) , we get

( x x i−1 ) ( T 2 n+ 2 ( x i )−T 2 n ( xi ) ) T '2n +2 ( x i ) −T '2 n ( x i )

I 3 i( x)=¿ +
2 ( xi +1 ) ( x i−1 )2 ( x−x i )2 2 ( xi −1 ) ( x−x i )

( )
1 2 x i +1 T 2 n+2 (x)−T 2 n (x )
− + 2 .(3.55)
x−x i ( x i −1 ) 2(x−1) ( x−x i)

From equations (3.53)-(3.55) we get

When x coincides with one of the grid point say x i then

I 3 i ( x i )=¿ lim Í 3 i (x)

¿
x → xi lim
x→x i
{
( x x i−1 ) ( T 2 n+ 2 ( x i )−T 2 n ( xi ) ) T '2n +2 ( x i ) −T '2 n ( x i )
2
2 ( xi +1 ) ( x i−1 ) ( x−x i )
2
+
2 ( x i−1 ) ( x−x i )
¿−
1
( )
x−x i ( x i −1 ) 2(x−1) ( x −xi ) }
2 x i +1 T 2n +2 (x)−T 2 n (x )
+ 2 ¿

And

I 4 i ( xi ) =lim I 3 i (x )=lim ¿
x → xi x→ x i

1
¿
2 ( x i−1 )
{
2 ( 2 n +2 ( i )
T }
x −T 2 n ( xi ) )−( x i−1 ) ( T '2 n +2 ( x i )−T '2 n ( xi ) ) (3.58)
From equations (3.53)-(3.58) we get

I ( x j) ¿

I ( x j ) ¿ In this section we apply the corrected quadrature rules, derived in section 4.3 , to integral
1 1 w(t)ϕ (t)
∫ dt for different functions of ϕ (t ) and compare the results with exact integral.
π −1 (t−x )
4 Numerical examples
Example 4.1. Consider the Singular integral (3.1) with ϕ (t )=64 x 7−112 x 5 +¿ 56 x 3−7 x , i.e.

1
1 ϕ (t)
I= ∫❑ dt
π −1 √ 1−t 2 (t−x )

Exact solution of integral (4.1) is 64 x 6−80 x 4+ 24 x 2−1. Corrected quadrature rules obtained in sections 3.1.1 -
3.1.4 are applied to equation (4.1). Graphical solution of (4.1) is presented in figure 4.1.

1
¿
2 ( x i−1 )
{
2 ( 2 n +2 ( i )
T }
x −T 2 n ( xi ) )−( x i−1 ) ( T '2 n +2 ( x i )−T '2 n ( xi ) ) (3.58)

From equations (3.53)-(3.58) we get

I ( x j) ¿

I ( x j ) ¿ In this section we apply the corrected quadrature rules, derived in section 4.3 , to integral
1 1 w(t)ϕ (t)
∫ dt for different functions of ϕ (t ) and compare the results with exact integral.
π −1 (t−x )
4 Numerical examples
Example 4.1. Consider the Singular integral (3.1) with ϕ (t )=64 x 7−112 x 5 +¿ 56 x 3−7 x , i.e.

1
1 ϕ (t)
I= ∫ ❑
π −1 √ 1−t 2 (t−x )
dt (4.1)

Exact solution of integral (4.1) is 64 x 6−80 x 4+ 24 x 2−1. Corrected quadrature rules obtained in sections 3.1.1 -
3.1.4 are applied to equation (4.1). Graphical solution of (4.1) is presented in figure 4.1.
 Figure 4.1: Solutions of integral (4.1)
Fixample 4.2. Consider the equation (3.31) with ϕ (t )=128 x7 −192 x 5 +80 x 3−8 x , i.e.

1
1
√1−t2 ϕ (t) dt
∫ (t−x)
π −1
❑

Exact solution of integral (4.2) is −128 x 8+ 256 x6 −160 x 4 +32 x 2−1 . We apply corrected quadrature rules
derived in sections 3.2 .1−3.2 .4, to above equation and compare the results with exact solution. Graphical
representation of solution of (4.2) is depicted in figure 4.2.

Figure 4.2: Solutions of equation (4.2)

From the graph it is observed that the corrected quadrature rules (3.28), (3.45), (3.52) provide the accurate results
whereas (3.57) produces an error in evaluation of integral (4.2)

5. Conclusions

1
Corrected quadrature rules for singular integrals, corresponding to the weight functions
√1−t2
and √ 1−t 2, are
constructed. Hermite interpolating polynomial is used to approximate the unknown function. Zeros of Chebyshev
polynomials of first, second, third and fourth kind are chnsen as grid points for interpolation. It is ohserved that
Hermite interpolation, obtained by using zeros of Chebyshev polynomials of first and second kind provide more
accurate quadrature rule over the rest two. When any grid point coincides with the singular point, corrected
quadrature rules are obtained by taking limit of corresponding quadrature rule. These limits are evaluated using
Taylor's series expansion. Methods are illustrated with some suitable examples. It is observed that, corrected
quadrature rule obtained by using zeros of corresponding orthogonal polynomial produces better results, e.g. to
√
evaluate I (x) with respect to w (t)= 1−t 2 quadrature rule constructed by using zeros of Chebyshev
polynomial of first kind, produce exact value of integration at all points in (−1 , 1).