CORRECTED QUADRATURE METHODS

AND THEIR APPLICATIONS

A thesis submitted to

Shivaji University, Kolhapur

for the degree of Doctor of Philosophy

in

Mathematics

Under the Faculty of Science and Technology

by

MISS. SUTAR INDRAYANI YUVRAJ

Under the Guidance of

Prof. (Mrs.) S. H. THAKAR

M.Sc., M.Phil., Ph.D

Department of Mathematics,
Shivaji University,
Kolhapur-416 004.
Maharashtra( India).
September 2020
 CONCLUSIONS

th
The corrected Simpson’s 38 formula constructed in chapter 2 is of
th
order 7. The error E in corrected Simpson’s 38 formula is obtained by
constructing polynomial functions. Error bounds for corrected quadra-
ture rule Qc are obtained. Optimality of error bounds is demonstrated
by constructing upper and lower bounds for function f (x) = x8 .
th
To evaluate integral of function f (x), Simpson?s 83 formula is cor-
rected by including the values of derivatives of f (x) at mesh points.
The quadrature rule is derived by using the method of undetermined
1 f (8) (ξ)
coefficients. The approximate error in the integration formula is 17010 8! .
The double integral inequalities are used to obtain better error bounds.
By suitable example it is observed that the error bounds obtained by
double integral inequalities are the best possible error bounds.
In chapter 3, a third order corrected Radau quadrature rule is
derived by method of undetermined coefficients. Upper and lower er-
ror bounds are obtained by defining piecewise continuous polynomials.
With suitable example optimality of the error bounds is proved. A
third order numerical method for first order ODE is proposed. The
method is suggested in the light of corrected quadrature formula. It is
proved that the proposed method is optimal. By the test example it is
shown that the numerical results obtained from the method give better
approximations as compared to third order R-K method. Method of
approximate solutions for FIE, FIDE and VIDE are also derived from
corrected Radau quadrature rule. Proposed methods are illustrated by
some examples.
Corrected quadrature rules √ for singular integrals, corresponding to
1
the weight functions √1−t2 and 1 − t2 , are constructed in chapter 4.
Hermite interpolating polynomial is used to approximate the unknown
function. Zeros of Chebyshev polynomials of first, second, third and
fourth kind are chosen as grid points for interpolation. It is observed
that Hermite interpolation, obtained by using zeros of Chebyshev poly-

ii
nomials 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 cor-
responding 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 − t2 quadrature rule con-
structed by using zeros of Chebyshev polynomial of first kind, produce
exact value of integration at all points in (−1, 1).
In chapter 5 We have developed four corrected quadrature
√ rules
1 2
corresponding to the weight function √1−t2 as well as 1 − t , for hy-
persingular integral, using Hermite interpolation. Mesh points for Her-
mite interpolation are chosen to be the zeros of Chebyshev polynomials
of first, second, third and fourth kind. Advantage of choosing zeros of
Chebyshev polynomials lies in the fact that the quadrature rules ob-
tained in this chapter coincides with derivatives of all the rules obtained
in chapter 4. Rules differ when x is any of the grid points. These cor-
rected quadrature rules provide reliable results when singular point
coincides with one of the grid point.
In chapter 6, corrected quadrature rules for singular integrals, de-
rived in chapter 4, are applied to SIE. Grid points are chosen to be zeros
of the Chebyshev polynomial which are orthogonal to corresponding
weight functions. From the numerical results, it is observed that error
in the approximate solution and the exact solution of SIE, reduces as
n increases.

iii
 SUMMARY

This thesis deals with study of evaluation of definite integration of

the form Z b
I(x) = w(x)f (x)dx, (1)
a
where w(x) is weight function and f (x) is either Riemann integrable
function or a function containing Cauchy type singularity. Such inte-
grals became subject of research due to its wide range of applications in
various areas like mathematics, statistics, physics, engineering, biology
etc. Though several analytical methods are available for the evaluation
of equation (1), some functions can not be integrated by these meth-
ods. In such cases numerical integration/quadrature method plays a
vital role.

Quadrature method was first introduced by Greek mathematician

Archimedes and later developed by many researchers. Quadrature
refers to any method that numerically approximates definite integral.
The goal is to attain a certain level of degree of precision with the
fewest possible function evaluations. Any quadrature method relies on
evaluation of integrand f (x) in set of points and processing these eval-
uations to produce an approximation to the value of integral. This is
done by taking weighted average at nodal points. With this approach
Newton provided a profound base to numerical integration. Following
Newton, many mathematician of the 18th and 19th centuries, Leon-
hard Euler, Joseph-Louis Lagrange, Carl Friedrich Gauss and others,
made major contributions to numerical analysis. Near around 1735,
Leonhard Euler and Colin Maclaurin independently discovered a for-
mula that involves higher derivatives of function at end points. In-
clusion of derivatives of function at some/all grid points improves the
degree of precision of a quadrature rule. Such rules are called corrected
quadrature rules.

iv
 In the present thesis we have focused on constructing corrected
quadrature rules for regular as well as singular integrals and applying
them to solve differential equations, integral equations and integro-
differential equations.
Chapter 1 starts with history of quadrature methods. Some ex-
isting and widely used methods are listed. Brief literature review of
quadrature methods is discussed. Some basic definitions are also in-
cluded in this chapter.

Chapter 2 includes literature review of Newton-Cote’s type quadra-

ture rules. In the second section we have constructed corrected Simp-
th
son’s 38 rule using method of undetermined coefficient as
Z 1     
31 81 1 2
f (x)dx = [f (0) + f (1)] + f +f
0 224 224 3 3
    
19 0 0 9 0 2 0 1
+ [f (0) − f (1)] + f −f .(2)
3360 1120 3 3

Error in the rule is

f (8) (ξ)
E= .
8! × 17010
Degree of precision of corrected quadrature rule (2) is seven. It is
th
observed that error in the corrected Simpson’s 38 rule obtained by
method of undetermined coefficients and that obtained by Hermite
interpolation coincides each other.
To derive upper and lower bounds for the error in corrected Simp-
th
son’s 83 rule, we have constructed piecewise continuous polynomials,
over the interval [0, 1], such that
1 2
Z 1 Z
3
Z
3
(8) (8)
P8 (x)f (x)dx = P8 (x)f (x)dx + P8 (x)f (8) (x)dx
1
0 0 3
Z 1
+ P8 (x)f (8) (x)dx
2
3

v
 Z 1
= f (x)dx − Qc .
0
th
where Qc is corrected Simpson’s 83 rule as defined in (2) and Pi (x) is
a polynomial of degree i, such that
0
Pi+1 (x) = Pi (x); i = 1, 2, 3, ..., 7. (3)
Using these polynomials error bounds are derived. Optimality of these
error bounds is tested with the help of test equation f (x) = x8 .

In Chapter 3, a new third ordered corrected quadrature rule is

constructed. Since this quadrature rule contains the value of function
and its derivative at initial point x = 0 of the interval [0, 1], we called
it as corrected Radau quadrature rule. Corrected Radau quadrature
rule is obtained by using method of undetermined coefficients is
Z 1  
11 16 3 1
f (x)dx = f (0) + f + f 0 (0). (4)
0 27 27 4 18
Error in formula (4) is given by

f (4) (ξ)
E = 0 < ξ < 1. (5)
80 × 4!
Applying the procedure described in Chapter 2, upper and lower
bounds are obtained for the error term (5) in the Corrected Radau
quadrature rule. The rule is used to develop numerical methods for
solving ordinary differential equation (ODE) of first ordered and first
degree, integral and integro differential equations.
To construct the numerical scheme for ODE, we have converted
ODE into an integral equation by integrating it within the limits xn and
xn+1 . An integral term in the resulting integral equation is evaluated
with the help of corrected Radau quadrature rule (4). Thus we get a
numerical scheme
11 1 16
yn+1 = yn + K1 + K2 + K3 ,
27 18 27
vi
where

K1 = hf (xn , yn ),
K2 = h2 (fx (xn , yn ) + f (xn , yn )fy (xn , yn )) ,
 
3h 3 9
K3 = hf xn + , yn + K1 + K2
4 4 32

and h = xn+1 − xn .
4
Error in the rule is h4! fy (xn , yn ) y 000 (xn , yn ).
It is observed that, numerical results obtained by proposed method
are better than that obtained by third ordered Runge-Kutta method.
We consider linear integral equation of second kind of the form
Z b
y(x) = f (x) + λ k(x, t)y(t)dt, (6)
a

and linear integro differential equation of the form

Z b
0
y (x) + αy(x) = f (x) + k(x, t)y(t)dt, (7)
a

where f (x) and kernel k(x, t) are known functions and y(x) is an un-
known function. Equations (6) and (7) are called of Fredholm type if
both a and b are constants while they are known as Voterra type if a
is constant and b = x.
Approximate solutions of equations (6) and (7) are obtained by ap-
plying corrected Radau quadrature rule (4) to integral term in these
equations.

Chapter 4 is devoted to construct the corrected quadrature rules

for the integrand containing Cauchy type singularities i.e. for integrals
of the form
1 1
Z
φ(t)
I(x) = w(t) dt x ∈ (−1, 1), (8)
π −1 (t − x)

vii
where w(t) is a weight function and φ(t) is density function which
satisfies the Hölder’s condition. These corrected quadrature
√ rules are
1
constructed for the weight functions w(t) = 1−t2 and 1 − t2 . For
√

each weight function four different corrected quadrature rules are de-
veloped depending on the choice of grid points. Zeros of Chebyshev
polynomial of first, second, third and fourth kind are considered as
grid points.
Density function is approximated by Hermite interpolating poly-
nomial and then using method of partial fractions, inversion formulae
and properties of Chebyshev polynomial, corrected quadrature rules for
singular integral in equation (8) are derived. The corrected quadra-
ture rules so obtained give better results when x do not coincides with
any other grid point. For the case x = xi , where xi is any of the grid
point, rules are obtained by taking the limit as x −→ xi of correspond-
ing quadrature rule. We used Taylor’s series expansion to evaluate
these limits. Methods are illustrated with some suitable examples.
From these examples , it is observed that, corrected quadrature rule
obtained by using zeros of corresponding orthogonal polynomial pro-
duces better results.

In the same vein, corrected quadrature rules for hypersingular in-

tegral of the form

1 1
Z
φ(t)
I(x) = w(t) dt x ∈ (−1, 1), (9)
π −1 (t − x)2
1
√
with the weight functions w(t) = √1−t 2
and 1 − t2 , are constructed
in Chapter 5. From these rules we conclude that when x 6= xi , cor-
rected quadrature rules obtained for hypersingular integral coincided
with derivatives of the corresponding rules derived in chapter 4.

Chapter 6 is devoted to study the applications of corrected quadra-

ture rules for singular integrals. Corrected quadrature rules for singular
integrals are applied to evaluate singular integral equations. Solution
of these integral equations are obtained by considering an additional

viii
condition,
n
X
φ(xi ) = 0. (10)
i=1
The assumption is necessary for the existence of unique solution of
the system. It is observed that given method provides better results
when zeros of the Chebyshev polynomial, which are orthogonal to cor-
responding weight functions, are chosen as grid points. From the nu-
merical results, we notice that error in the approximate solution and
the exact solution of SIE, reduces as n increases.

ix
 Future Scope of the Study

In this thesis we have developed corrected quadratures for singular

and hypersingular integrals
Z √
1 1 1 1 1 − t2 φ(t)
Z
φ(t)
√ dt and dt, r = 1, 2.
π −1 1 − t2 (t − x)r π −1 (t − x)r

The function φ(t) is approximated by Hermite interpolating polyno-

mial. The collocation points are zeros of Chebyshev polynomial of
first, second, third and fourth kind.
Corrected quadrature for singular and hypersingular integrals of
the type

1 1 1 − t φ(t) 1 1 1 + t φ(t)
Z r Z r
dt and dt, r = 1, 2,
π −1 1 + t (t − x)r π −1 1 − t (t − x)r

can be developed by similar methods. These methods also can be used

to solve singular and hypersingular integral equations, singular and
hypersingular integro differential equations.