See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/370340849

Thesis - A study of different types of interpolating polynomials and their

convergence behaviour on non-uniformly distributed set of nodes on the unit
circle.

Thesis · April 2023

DOI: 10.13140/RG.2.2.26547.71202

CITATIONS READS

0 789

1 author:

Varun ..
University of Lucknow
20 PUBLICATIONS 12 CITATIONS

SEE PROFILE

All content following this page was uploaded by Varun .. on 28 April 2023.

The user has requested enhancement of the downloaded file.

 A STUDY OF DIFFERENT TYPES OF INTERPOLATING
POLYNOMIALS AND THEIR CONVERGENCE BEHAVIOUR
ON NON-UNIFORMLY DISTRIBUTED SET OF NODES ON
THE UNIT CIRCLE

Ph.D. THESIS
SUBMIT T ED T O T HE
UN IVERSIT Y OF LUCKN OW
FOR T HE AWARD OF T HE DEGREE OF

Doctor of Philosophy
IN
MATHEMATICS

BY
VARUN
M.Sc., CSIR(UGC)-NET(JRF)

UN DER
T HE SUPERVISION OF
Dr. SWARNIMA BAHADUR

DEPARTMENT OF MATHEMATICS AND ASTRONOMY,

UNIVERSITY OF LUCKNOW LUCKNOW, PIN–226 007, U.P.,
INDIA
2022
 Interpolation on non-uniform nodes of the unit circle

Self Declaration

I, Varun certify that the work embodied in this Ph.D. thesis entitled:
“A study of different types of interpolating polynomials and their
convergence behaviour on non-uniformly distributed set of nodes
on the unit circle” is my own bona-fide work, which was carried out by
me under the supervision of Dr. Swarnima Bahadur for a period of
January 2017 to October 2022 at the University of Lucknow. The content
embodied in this Ph.D. thesis has not been submitted for the award of any
other degree/diploma. I declare that I have faithfully acknowledged, given
credit to and referred to the researchers wherever their works have been cited
in the text and the body of the thesis. I further certify that I have not wilfully
used any other person’s work, para, text, data, results, etc. reported in the
journals, books, magazines, reports, dissertations, theses, etc., or available
at web-sites/ Internet and included them in this Ph.D. thesis and cited that
as my own work.

Varun
Department of Mathematics
and Astronomy, University of
Lucknow, Lucknow–226 007,
Date:
Uttar Pradesh, India.
Place: Lucknow, India.

University of Lucknow, Lucknow–226007 ii

 Interpolation on non-uniform nodes of the unit circle

Certificate

This is to certify that the thesis entitled: “A study of different types

of interpolating polynomials and their convergence behaviour on
non-uniformly distributed set of nodes on the unit circle” submitted
by Mr. Varun to the University of Lucknow, Lucknow, for the award
of Doctor of Philosophy (Ph.D.) degree in Mathematics, is absolutely
based upon his own work and is a record of bonafide research work carried out
by him under my supervision and guidance. The thesis contents have been
submitted neither partially nor fully to any other University or Institution
for the award of any degree or diploma.
His thesis has fulfilled all the requirements for submitting the thesis for
the degree of Doctor of Philosophy of University of Lucknow, Lucknow,
U.P., India.

Dr. Swarnima Bahadur

(Supervisor)
Department of Mathematics
and Astronomy, University of
Date: Lucknow, Lucknow–226 007,
Place: Lucknow, India. Uttar Pradesh, India.

University of Lucknow, Lucknow–226007 iii

 Interpolation on non-uniform nodes of the unit circle

Dedicated
to my
Father Shri Prakash Kanaujia
Mother Mrs. Prabha
and
Sister Dr. Rimjhim

University of Lucknow, Lucknow–226007 iv

 Interpolation on non-uniform nodes of the unit circle

Preface
The present thesis, which is being submitted for the award of degree of Doc-
tor of Philosophy of the University of Lucknow, Lucknow, is the outcome
of the research work carried by the author since 2017 as a Junior Research
Fellow (CSIR UGC-JRF) and since 2019 as a Senior Research Fellow
(CSIR UGC-SRF) in University of Lucknow.

 Organization of Thesis
This thesis is organized in eight chapters as follows:
Chapter 1: Introduction
This chapter presents an overview of some basic definitions, theorems and
historical survey of various types of interpolatory polynomials and their con-
vergence behaviour on some set of nodes, which we will use in this thesis.

Chapter 2: Convergence of Lagrange-Hermite Interpo-

lation on the unit circle
In this chapter, we study a Lagrange-Hermite interpolation on the nodes,
(0,0)
which are obtained by projecting vertically the zeroes of the (1 − x2 )Pn (x)
(0,0)
on the unit circle, where Pn (x) stands for the Legendre polynomial of the
degree n. This chapter includes the proof of regularity of the problem as well
as the explicit forms of the fundamental polynomials of the interpolatory
polynomial. Estimates of the fundamental polynomials as well as conver-
gence theorem with their proof is a key highlight of this chapter.

Chapter 3: Convergence of Interpolatory polynomial

between Lagrange and Hermite
In this chapter, we study an interpolation problem ,which is an intermediate
problem between Lagrange and Hermite. We considered this problem on the
(0,0)
nodes obtained by projecting vertically the zeroes of the (1−x2 )Pn (x) onto
(0,0)
the unit circle, where Pn stands for nth Legendre polynomial. We prove
the regularity of the problem, give explicit forms and establish a convergence
theorem for the same.

Chapter 4: A note on Hermite-Fejér Interpolation on

the non-uniform nodes of the unit circle
This chapter deals with the Hermite-Fejér interpolation problem on the unit
circle with the nodal system containing the vertically projected zeros of Ja-
cobi’s polynomial with boundary points on the unit circle. We worked upon
three nodal structures throughout this chapter and obtained rate of conver-
gence for each case. Moreover, we did a comparison of all the three cases
and provided some important conclusions.

University of Lucknow, Lucknow–226007 v

 Interpolation on non-uniform nodes of the unit circle

Chapter 5: Extension of Pál-type Hermite-Fejér Inter-

polation onto the unit circle
This chapter is devoted to the study of a Pál type (0;1) interpolation problem
on the unit circle considering two disjoint sets of nodes. The nodal points are
(α,β)
obtained by projecting vertically the zeros of the Jacobi polynomial Pn (x)
(α,β)0
and its derivative Pn (x), together with ±1 onto the unit circle. The La-
grange data are prescribed on the first set of nodes, the Hermite data are
prescribed on the second one and generalized Hermite-Fejér boundary con-
ditions are prescribed at ±1. An explicit representation of the interpolatory
polynomial is given and the convergence is studied for analytic functions on
the unit disk. The results are of interest to approximation theory.

Chapter 6: Revisiting Pál-type Hermite-Fejér Interpo-

lation onto the unit circle
In this chapter, we brought into consideration the set of non-uniformly dis-
tributed nodes on the unit circle to investigate a Pál-type (1;0) interpolation
problem in account with Hermite-Fejér boundary condition. These nodes are
obtained by projecting vertically the zeros of Jacobi polynomial as well as
zeros of its derivative onto the unit circle. Explicitly representing the inter-
polatory polynomial as well as establishment of convergence theorem are the
key highlights of this manuscript. The field of approximation theory enter-
tains the results proved.

Chapter 7: Higher order Hermite-Fejér Interpolation

on the unit circle.
In this chapter, we study the approximation of functions using a higher-order
Hermite-Fejér interpolation process on the unit circle. The system of nodes
is composed of vertically projected zeros of Jacobi polynomials onto the unit
circle with boundary points at ±1. Values of the polynomial and its first four
derivatives are fixed by the interpolation conditions at the nodes. Conver-
gence of the process is obtained for analytic functions on a suitable domain,
and the rate of convergence is estimated.

Chapter 8: Convergence of Lagrange Interpolation on

the non-uniform nodes on the unit circle.
This chapter aims to staunchly study the approximation using Lagrange
interpolation on the unit circle. Nodal system constitutes the vertically pro-
jected zeros of Jacobi polynomial onto the unit circle with boundary points
at ±1. Moreover, convergence is obtained by considering analytic functions
on a suitable domain accompanied by some numerical experiments.

University of Lucknow, Lucknow–226007 vi

 Interpolation on non-uniform nodes of the unit circle

Acknowledgements
First and foremost, I am incredibly grateful to my supervisor, Dr. Swarn-
ima Bahadur, for their invaluable advice, continuous support, and patience
during my research work. Her knowledge and experience have encouraged
me in my academic research and daily life. Her insightful feedback pushed
me to sharpen my thinking and brought my work to a higher level. Without
her persistent help, the goal of this work would not have been realized.
I want to express my deep and sincere gratitude towards Prof. Pankaj
Mathur and Prof. Meena Sahai for their valuable suggestions and support
on my research works.
I am grateful to Prof. Bhaskar Srivastava, Prof. Poonam Sharma
and Prof. Vivek Sahai, Head, Department of Mathematics and As-
tronomy, University of Lucknow, Lucknow, for providing the necessary
research facilities during my study. I am thankful to University and Depart-
mental library for providing me the required facilities.
I want to offer my special thanks to my friends Dr. Harendra Verma,
Dhanesh Kumar, Shivam Agarwal, Satish Kumar and Vivek Kumar Yadav
for their unwavering support during all times. Also, I would like to thank my
friends outside of the Mathematics world Er. Mayank Agarwal, Er. Sachin
Nair, Er. Shashank Gaurav and C.A. Apurva Patel for their valuable support
in my good and bad time and always encouraging me to achieve my goal and
move forward in the life.
This acknowledgement would be incomplete without my sincere gratitude
to my parents, Shri Prakash Kanaujia, Mrs. Prabha and my sister
Dr. Rimjhim for their love and understanding, and continuous support to
complete my research work.

University of Lucknow
Lucknow, U.P. Mr. Varun
2021

University of Lucknow, Lucknow–226007 vii

 Interpolation on non-uniform nodes of the unit circle

Publications Related To Thesis

 Research Papers published/accepted

1. Convergence of Lagrange-Hermite Interpolation on Unit Circle, Inter-

national Journal Of Mathematical Archive, 8(11), 2017, 167-170.
http://www.ijma.info/index.php/ijma/article/view/5174.

2. Convergence Of Interpolatory Polynomial Between Lagrange And Her-

mite, Annals Of Pure And Applied Mathematics, 17, 2018, 23-28
http://dx.doi.org/10.22457/apam.v17n1a3

3. A note on Hermite-Fejér Interpolation on the non-uniform nodes of the

Unit Circle, South East Asian J. Of Mathematics And Mathematical
Sciences, 17(2), 2021, 83-92.
http://rsmams.org/journals/articleinfo.php?articleid=570&tag=
seajmams

4. Extension of Pál type Hermite-Fejér interpolation onto the unit circle,

Applied Mathematics E-Notes, 22, 2022, 371-382.
https://www.emis.de/journals/AMEN/

5. Revisiting Pál-type Hermite-Fejér Interpolation Onto The Unit Circle,

Ganita, 71(1), 2021, 145-153.
https://bharataganitaparisad.com/wp-content/uploads/2021/10/
711-ch014.pdf

6. Higher order Hermite-Fejér interpolation on the unit circle, TWMS J.

App. Eng. Math., 2022 {accepted}
http://jaem.isikun.edu.tr/web/index.php/archive

7. Convergence of Lagrange Interpolation on the non uniform nodes on

the unit circle, Advanced Studies : Euro-Tblisi Mathematical Journal,
15(2), 2022, 101-111.
http://tcms.org.ge/Journals/ASETMJ/Volume15/Volume15_2/Abstract/
abstract15_2_7.html

University of Lucknow, Lucknow–226007 viii

Contents

Page No.

1 Introduction 1
1.1 Best approximation . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Orthogonal Polynomials . . . . . . . . . . . . . . . . . . . . . 4
1.3 Classical Orthogonal Polynomials . . . . . . . . . . . . . . . . 4
1.3.1 Hermite Polynomials . . . . . . . . . . . . . . . . . . . 4
1.3.2 Laguerre Polynomials . . . . . . . . . . . . . . . . . . . 5
1.3.3 Jacobi Polynomials . . . . . . . . . . . . . . . . . . . . 5
1.4 Approximations . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4.1 Types of Approximations . . . . . . . . . . . . . . . . . 6
1.5 Polynomial Interpolation . . . . . . . . . . . . . . . . . . . . . 8
1.6 Nodal systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6.1 Uniform Nodal System . . . . . . . . . . . . . . . . . . 10
1.6.2 Non-Uniform Nodal System . . . . . . . . . . . . . . . 10
1.7 Interpolation on the unit circle . . . . . . . . . . . . . . . . . . 11
1.8 Different Types of Polynomial Interpolation . . . . . . . . . . 12
1.8.1 Lagrange Interpolation . . . . . . . . . . . . . . . . . . 12
1.8.2 Hermite Interpolation . . . . . . . . . . . . . . . . . . . 16
1.8.3 Hermite-Fejér Interpolation . . . . . . . . . . . . . . . 20
1.8.4 Lacunary Interpolation . . . . . . . . . . . . . . . . . . 21
1.8.5 Pál-type Interpolation . . . . . . . . . . . . . . . . . . 27
1.9 What does notation (0,1) and (0,1;2) type Interpolation
means? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.10 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2 Convergence of Lagrange-Hermite Interpolation on unit cir-

cle 36
2.1 The Problem and the Regularity . . . . . . . . . . . . . . . . . 36
2.2 Explicit Representation of Interpolatory Polynomials . . . . . 37
2.3 Estimation of Fundamental Polynomials . . . . . . . . . . . . 42
2.4 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

3 Convergence of Interpolatory polynomial between Lagrange

and Hermite 49
3.1 The Problem and the Regularity . . . . . . . . . . . . . . . . . 49

ix
 Interpolation on non-uniform nodes of the unit circle

3.2 Explicit Representation of Interpolatory Polynomials . . . . . 50

3.3 Estimation of Fundamental Polynomials . . . . . . . . . . . . 53
3.4 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

4 A note on Hermite-Fejér Interpolation on the non-uniform

nodes of the unit circle 61
4.1 The Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.2 Explicit Representation of Interpolatory Polynomial . . . . . . 62
4.3 Estimates of Fundamental Polynomials . . . . . . . . . . . . . 66
4.4 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

5 Extension of Pál-type Hermite-Fejér Interpolation onto the

unit circle 73
5.1 The Problem & Explicit Representation of Interpolatory Poly-
nomial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.2 Estimates of Fundamental Polynomials . . . . . . . . . . . . . 79
5.3 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

6 Revisiting Pál-type Hermite- Fejér Interpolation onto the

unit circle 88
6.1 The Problem and the Explicit Representation of Interpolatory
Polynomial . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.2 Estimates of Fundamental Polynomials . . . . . . . . . . . . . 92
6.3 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

7 Higher order Hermite-Fejér Interpolation on the unit circle. 98

7.1 The Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.2 Explicit Representation of Interpolatory Polynomial . . . . . . 99
7.3 Estimation of the fundamental polynomials . . . . . . . . . . . 101
7.4 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

8 Convergence of Lagrange Interpolation on the non-uniform

nodes on the unit circle 107
8.1 The Problem and Explicit Representation of Interpolatory Poly-
nomial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
8.2 Estimation of the Fundamental Polynomial . . . . . . . . . . . 108
8.3 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
8.4 Numerical Experiments . . . . . . . . . . . . . . . . . . . . . . 112
8.5 Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
8.6 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . 116

University of Lucknow, Lucknow–226007 x

Chapter 1

Introduction

This thesis is about interpolation on the non-uniformly distributed set of

nodes on the unit circle. Before presenting our new contributions to the
interpolation problems in Chapters 2–8, here we review the basic notions from
interpolation theory and give some basic definitions, theorems and historical
survey of various types of interpolatory polynomials and their behaviour on
some set of nodes that we will need throughout.
We define two sets T = {z : |z| = 1} and D = {z : |z| < 1} and these sets
will be frequently used throughout the thesis.
Lipschitz Class of Functions: A function f (x) defined on an interval [a, b]
is said to satisfy Lipschitz condition with the exponent α and the coefficient
K if it satisfies

| f (y) − f (x) | ≤ K | y − x |α ,α > 0

for all pairs of x, y of this interval. We write it as f (x) ∈ LipK α or simply

f (x) ∈ Lip α

Modulus of Continuity: In mathematical analysis, a modulus of conti-

nuity is a function used to measure quantitatively the uniform continuity of
functions.

1
 Interpolation on non-uniform nodes of the unit circle

For a function, f : R → R, we shall use a special form of modulus of conti-

nuity ω(f ; δ) introduced by G. Freud.

ω(f ; δ) = sup || w(x + t) f (x + t) − w(x) f (x) || + || τ (δ(x) ω(x) f (x)) ||,

where 

| x |, if | x |< 1

τ (x) =

1,
 if | x |> 1.

and || . || denotes the Sup-norm in C(R).

If f ∈ C(R) and
lim ω(x)f (x) = 0,
|x|→∞

then
lim ω(f ; δ) = 0.
δ→∞

1.1 Best approximation

It was introduced by P.L. Tchebycheff. It is associated with the following

theorem.

Theorem 1.1.1. For every bounded function f(x) defined on the finite inter-
val [a,b] and for any integer n, there exists an ordinary polynomial

Pn (f, x) = a0 + a1 x + a2 x2 + · · · + an xn ,

of degree not greater than n, which deviates least from it on this interval.
The least value of its deviation from f(x) is given by the quantity.

En (f ) =| f (x) − Pn (x) |,

University of Lucknow, Lucknow–226007 2

 Interpolation on non-uniform nodes of the unit circle

En (f ) ≤ max | f (x) − Pn (f, x) |,

a≤x≤b

which is called best approximation.

Approximation Theorems

Theorem 1.1.2. Jackson’s First Theorem

Let En (f ) be the approximation by algebraic polynomials Pn (x) of degree ≤
n and ω(f ; t) be the modulus of continuity of f (x) on [a, b], then for any
arbitary function f (x)  C[a, b], we have

 b − a
En (f ) ≤ c ω f ; ,
n

where c is an absolute constant.

Theorem 1.1.3. Jackson’s Second Theorem

If the function f (x) defined on the finite segment [a, b] has the rth continuous
derivative on the segment, then for n ≥ r, we have

 b − a r 
(r) b − a

En (f ) ≤ Mr ω f ; ,
n n

where Mr is a constant depending only on r.

Theorem 1.1.4. O Kiŝ inequality

Let f (z) be continous for | z |≤ 1 and analytic for | z |< 1. Then there exist
a polynomial Fn (z) satisfying

| Fn(m) (z) |≤ c nm ω(f, n−1 ) , m  I +,

where c is a constant independent of n and z.

University of Lucknow, Lucknow–226007 3

 Interpolation on non-uniform nodes of the unit circle

1.2 Orthogonal Polynomials

If f (x) and g(x) are two real functions with x ∈ [a, b], such that their
Lebesgue-Stieltjes integral is zero, i.e.

Z b
(f, g) = f (x)g(x) dα(x) = 0 (1.1)
a

where f (x), g(x) ∈ Lα (a, b) and dα(x) is the continous or discontinous dis-
tribution of mass in [a, b], then we say that “f (x) is orthogonal to g(x)”.
If we orthogonalize the set of non-negative powers of 1, x, . . . , xn . . . . We ob-
tain a set of polynomials p0 (x), p1 (x), . . . , pn (x), . . . satisfying the conditions:
(i) pn (x) is a polynomial of degree n, in which the coefficients of xn is posi-
tive.
(ii) The sequence {pn (x)} is orthonormal i.e.

Z b
pm (x)pn (x) dα(x) = δmn ; n, m = 0, 1, 2, . . .
a

Then, the polynomial pn (x) are said to be orthogonal polynomials associated

with distribution dα(x).

1.3 Classical Orthogonal Polynomials

The system of orthogonal polynomials associated with the names of Hermite,

Laguerre and Jacobi are called classical orthogonal polynomials.

1.3.1 Hermite Polynomials

2 dn −x2
Hn (x) = (−1)n ex [e ]. (1.2)
dxn

University of Lucknow, Lucknow–226007 4

 Interpolation on non-uniform nodes of the unit circle

1.3.2 Laguerre Polynomials

For arbitrary real α, the polynomial solutions of the differential equation

xy 00 + (α + 1 − x)y 0 + ny = 0.

are the Laguerre polynomials given by

−1 −α x dn −x n+α
L(α)
n (x) = (n!) x e [e x ].
dxn

1.3.3 Jacobi Polynomials

They are solutions to the Jacobi differential equation

(1 − x2 )y 00 + [β − α − (α + β + 2)x]y 0 + n(n + α + β + 1)y = 0,

and give some other special named polynomials as special cases.

For α, β > −1, we have

dn
Pn(α,β) (x) = (−2)−n (n!)−1 (1 − x)−α (1 + x)−β [(1 − x)n+α (1 + x)n+β ]. (1.3)
dxn

Special cases of Jacobi polynomials are

• Legendre Polynomial:- Put α = β = 0 in (1.3),

Pn (x) = Pn(0,0) (x).

• Tchebychev Polynomials of first kind:- Put α = β = − 12 in (1.3),

 −1
2n
2n (− 1 ,− 1 )
Tn (x) = 2 Pn 2 2 (x).
n

University of Lucknow, Lucknow–226007 5

 Interpolation on non-uniform nodes of the unit circle

• Tchebychev Polynomials of second kind:- Put α = β = 1

2
in (1.3),

 −1
2n + 1
2n (1,1)
Un (x) = 2 Pn 2 2 (x).
n+1

• Gegenbauer (or Ultraspherical) Polynomials:- Put α = β in

(1.3),
 −1  −1
2α n + 2α
Pn(λ) (x) = Pn(α,α) (x),
α α
1 1
where 6= α = λ − .
2 2

1.4 Approximations

Approximation theory is concerned with how functions can best be approx-

imated with simpler functions and with quantitatively characterizing the
errors introduced thereby.

1.4.1 Types of Approximations

• Linear Approximation
Let f (x) be the function, which we wish to approximate using the class
of functions {gn (x)}, n = 0, 1, 2, . . . , m. Suppose we approximate the
function f (x) by linear combination

f (x) ≈ a0 g0 (x) + a1 g1 (x) + a2 g2 (x) + · · · + am gm (x),

where a0i s (i = 0, 1, 2, . . . , m) are constants.

• Rational Approximation
Let f (x) be the function, which we wish to approximate using the class
of functions {gn (x)}, n = 0, 1, 2, . . . . We approximate the function

University of Lucknow, Lucknow–226007 6

 Interpolation on non-uniform nodes of the unit circle

f (x) in such a manner i.e.

a0 g0 (x) + a1 g1 (x) + a2 g2 (x) + · · · + an gn (x)

f (x) ≈ ,
b0 g0 (x) + b1 g1 (x) + b2 g2 (x) + · · · + bk gk (x)

where a0i s and b0j s are constants, i = 0(1)n and j = 0(1)k.

• Least Square Approximation

Here, we minimize the integral of the square of the difference between
f (x) and its approximated value over the interval of consideration.

• Minimum - Maximum Approximation

Here, the aim is to minimize the maximum of the difference between
f (x) and its approximated value over the interval of consideration.

• Interpolation
In the mathematical field of approximation theory and numerical anal-
ysis, one can define interpolation as the technique of obtaining the most
likely estimate of a certain assumptions. In simple words, if a set of
values of a function f (x) at a set of values of x are given, then the
process to determine either the values of x, for which f (x) has some
specified value or the value of f (x) at an intermediate value of set of
values of x is known as interpolation.
For e.g.:- Consider the data in the following table 1.

x 1 2 3 4

f (x) 0.5 2.5 -3 5

Then, estimating the function f (x) at an intermediate value, such as

x = 2.5 is a question that can be answered by means of interpolation.

University of Lucknow, Lucknow–226007 7

 Interpolation on non-uniform nodes of the unit circle

Figure 1.1: Different types of interpolation on the data given in table 1

There are many different kinds of interpolation methods such as piecewise

constant interpolation, linear interpolation, polynomial interpolation, spline
interpolation, mimetic interpolation and many more. Since, these different
kind of interpolation methods possess book-length treatments of their own,
we leave the reader to look up the details elsewhere.
We will immediately narrow our focus to the polynomial interpolation and
the types of polynomial interpolation, which we will be most concerned in
the later chapters, namely, Lagrange, Hermite, and Pál-type interpolation in
a single (complex) variable.

1.5 Polynomial Interpolation

Polynomial interpolation has remained a ceaseless topic of research since

decades due to its various numerical applications such as forming the basis
for algorithms in numerical derivation and numerical quadrature.

University of Lucknow, Lucknow–226007 8

 Interpolation on non-uniform nodes of the unit circle

The type of interpolation in which data set is interpolated using the polyno-
mial of lowest possible degree that passes through the points of the data set
is known as the polynomial interpolation. The theory of interpolation may
be traced from the well - known theorem of Weierstrass.

Theorem 1.5.1. Weierstrass Theorem

A continuous function f (x) can be uniformly approximated by a polynomial
Pn (x) of sufficiently high degree in the sense that

| f (x) − Pn (x) |→ 0 as n → ∞.

Since, then mathematicians have been trying to find classes of polyno-

mials, which approximate to f (x) in one sense or the other. The theorems
of convergence generally involve the modulus of continuity of the functions,
defined by the quantity:

ω(δ) = ω(f ; δ) = max |f (x) − f (y)|.

|x−y|≤δ

There are different types of polynomial interpolation such as Lagrange

interpolation, Hermite interpolation, which will be discussed in the later
sections.

1.6 Nodal systems

Choosing a set of nodal points plays an important role in the interpolation

process. Different nodal system means different set of points on which inter-
polation takes place, which in turn affects the convergence behaviour of the
interpolatory polynomial.

University of Lucknow, Lucknow–226007 9

 Interpolation on non-uniform nodes of the unit circle

1.6.1 Uniform Nodal System

If the arc length between the consecutive points of the nodal system on the
unit circle remains same, then the nodal system is called uniform nodal sys-
tem.
2kπ
For e.g. zk = exp(iθ) where θ = n
and k = 0, 1, . . . , n − 1.

Figure 1.2: Uniform Nodal System

1.6.2 Non-Uniform Nodal System

The nodal system, in which arc length between the consecutive points of the
nodal system on the unit circle can be same or not is a non-uniform nodal
system.
For e.g. zk = exp(iθk ), where k = 1(1)n. It is interesting to note that in
above example if θ1 = θ2 = ... = θn , then the set of non-uniform nodes
transforms to a uniform nodal system.

University of Lucknow, Lucknow–226007 10

 Interpolation on non-uniform nodes of the unit circle

Figure 1.3: Non-Uniform Nodal System

1.7 Interpolation on the unit circle

In the interpolation theory, there are many interesting problems that require
domains other than the real interval. Interpolation problems, in which nodes
of the interpolation lie on the unit circle is known as interpolation on the
unit circle.
Kiŝ [70] was the initiator of interpolation processes on the unit circle.
He considered the (0, 2) and (0, 1, ..., r − 2, r) interpolation for an integer
r ≥ 2 on the nth roots of unity. Darius [43] took the roots of complex
numbers with modulus one as the nodal points and using a suitable modulus
of continuity obtained a result about convergence of the interpolants for
continuous functions.
Recently, Berriochoa et.al [36] studied the Lagrange interpolation on the
unit circle (LIUC) for piecewise continuous functions with suitable properties
and zeros of the para-orthogonal polynomials were considered as the nodal

University of Lucknow, Lucknow–226007 11

 Interpolation on non-uniform nodes of the unit circle

points in accordance with analytic weights. Interpolation on the unit circle

have also been studied on some non-uniformly distributed nodes on the unit
circle (see [38] and [139]). Bahadur [13] studied LIUC by projecting vertically
the zeros of (1 − x2 )Pn (x), where Pn (x) denotes the nth Legendre polynomial.
A lot of research work on interpolation on the unit circle has been discussed
in later section of this chapter.

1.8 Different Types of Polynomial Interpola-

tion

This section discuss about the literature of different types of interpolation

methods.

1.8.1 Lagrange Interpolation

One of the profuse approaches to read polynomial interpolation is the La-

grange interpolation. Let us consider a triangular matrix Q of order n × n
and a function f (z), which is to be interpolated by a Lagrange interpolatory
polynomial Ln (f )

 
z11
 
 
 z21 z22 
 
 
Q=
 z31 z32 z33 

 .. .. ..
 
 . . .


 
zn1 zn2 . . . znn

Here, nth row is given by

−1 ≤ znn < · · · < zn2 < zn1 ≤ 1

University of Lucknow, Lucknow–226007 12

 Interpolation on non-uniform nodes of the unit circle

and Lagrange interpolating polynomial Ln (f, z) is of degree atmost n − 1.

The explicit form of such polynomial is given by

n
X
Ln (f, z) = f (zk )lk (z),
k=1

where lk (z) is known as the fundamental polynomial of Lagrange interpola-

tion given by

q(z)
lk (z) = , k = 1, 2, ..., n ,
(z − zkn )q 0 (zkn )

where
q(z) = C(z − z1n )(z − z2n ) . . . (z − znn ).

Here, C is an arbitrary constant (real or complex) independent of n and z.

0
Also, q (zkn ) denotes the first derivative of q(z) at z = zkn .
Grünwald [60] found out that there existed a continuous function such
that the sequence of Lagrange interpolation polynomials diverges correspond-
ing to the nodal system constituted of zeros of Tchebychef polynomials.
Korenin [72], showed that

n
X
lim sup lkn (z) = +∞,
n→∞
k=1

holds densely in a closed set of points that lie on the unit circle |z| = 1.
Erdös and Vertesi [48] in 1980 showed that for a proper function f ∈
[−1, 1], Ln (f, z) will diverge almost everywhere for any set of nodes. In the
same year, A. H. German [56] considered the Lagrange interpolation problem
in the region D̄ = D ∪ T.
Lagrange interpolation on an infinite interval was considered by J. Sz-
abadós and A. K. Varma [114] in 1985 and they also obtained the almost

University of Lucknow, Lucknow–226007 13

 Interpolation on non-uniform nodes of the unit circle

everywhere divergence for the same. In 1990, K. G. Ivanov and E. B. Saff

[61] considered Ln−1 (f, z) on |z| = 1 and analysed the size of the coeffi-
cients of Ln−1 (f, z) and asymptotic behaviour of the zeros of the polynomials
Ln−1 (f, z) and also studied the convergence and divergence behaviour of the
same.
In 1991, X. C. Shen and B. P. Shuai [106] considered Lagrange interpo-
lation on unit circle and obtained the results concerning the rate of approx-
imation of Lagrange interpolation.

! !
1
||f − Ln (f )||∞ ≤ cω f, log(n + 1)
n+1

and 0 < p < ∞, !

1
||f − Ln (f )||p ≤ cω f, ,
n+1
!
1
where ω f, denotes the modulus of continuity of f and c is a constant
n+1
independent of n.
Also in 1991, Zhong Kai Li [77] considered Quasi-Lagrange interpolatory
polynomial for every f (x) ∈ [−1, 1] and constructed two sequences based on
Bernstein process and Freud – Sharma construction respectively.
In 1992, Y. G. Shi [107] compared Hermite-Fejér interpolation and La-
grange interpolation and investigates a triangular matrix whose entries are
the nodes.

−1 ≤ xn1 · · · < xnn ≤ 1, n = 1, 2, . . .

In 1993, K. Balázs and T. Kilgore [31] deal with the approximation of an

arbitrary function f ∈ C r [−1, 1] (and its derivatives) by use of certain inter-
polation polynomials Pn (f ) (and their derivatives).

University of Lucknow, Lucknow–226007 14

 Interpolation on non-uniform nodes of the unit circle

Also, Tian Zi Jiang [65] obtained two infinite asymptotic expansions for
the Lebesgue constant associated with Lagrange polynomial interpolation on
unit circle.In 1990, M. R. Akhlaghi, A. Jakjmovski and A. Sharma [3] con-
sidered Lagrange interpolating polynomial in the unit circle and extended
Walsh’s over convergence theorem to several complex interpolating polyno-
mial. In 1993, G. Mastroianni and P. Vertesi [78] established the mean
convergence of Lagrange interpolation on the general set of nodes.
Also in 1996, R. Brück [38] considered the case of Lagrange interpolation
on non-uniformly distributed nodes on the unit circle. He also studied its
convergence on the nodes, which are the images of the roots of unity, under
a Möbius transform of unit disc onto itself. In 2000, Darius [4] took the
roots of complex numbers with modulus one as the nodal points and using
a suitable modulus of continuity obtained a result about convergence of the
interpolants for continuous functions.
Trefethen [124] breaking the myths about the polynomial interpolation
quoted that polynomial interpolants always converge, if function is smooth
a little bit in Chebyshev points.Thus, function is required to be imposed by
some restrictions such as altering conditions on its modulus of continuity
in order that sequence of Lagrange interpolation polynomials attains finer
properties of the convergence.
In 2012, S. Bahadur [13] studied Lagrange interpolation on the unit circle
by projecting vertically the zeros of (1 − x2 )Pn (x), where Pn (x) denotes the
nth Legendre polynomial and studied the convergence for the same. Recently,
Berriochoa et.al [36] studied the Lagrange interpolation on the unit circle for
piecewise continuous functions with suitable properties and zeros of the para-
orthogonal polynomials were considered as the nodal points in accordance
with analytic weights.

University of Lucknow, Lucknow–226007 15

 Interpolation on non-uniform nodes of the unit circle

1.8.2 Hermite Interpolation

Hermite interpolation is the process of finding a polynomial, which coincides

with the continous function at certain pre-assigned points, called the nodes
of interpolation, and its successsive derivatives coinciding with arbritarily
chosen numbers.
In 1980, A.S. Cavaretta, A. Sharma and R.S. Varga [40] reinvestigated
the following theorem of Walsh:
“Let Aρ be the class of functions analytic in |z| < ρ, (ρ > 1) with at least
one singularity on |z| = ρ. If Pn−1 is the Lagrange interpolation polynomial
at the nth roots of unity to f ∈ Aρ and P̆n−1 , the (n − 1)th partial sum in the
Taylor’s expansion of f (z):

∞
X
f (z) = ak z k ,
k=0

then
lim (Pn−1 (z) − P̆n−1 (z)) = 0, (1.4)
n→∞

holds uniformly in every compact subset of {z : |z| < ρ2 }.”

According to D.J. Newman, if Lagrange interpolation is replaced by the best
approximating polynomial pn−1 (z, f ) of degree (n − 1) to f ∈ Aρ in |z| ≤ 1,
then there exist f0 ∈ Aρ , such that

lim (pn−1 (z, f0 ) − P̆n−1 (z, f0 )) = 0, for |z| < ρ (1.5)

n→∞

Further, S. A. N. Enduanya [47] obtained the explicit forms and conver-

(α)
gence for the roots of Ultraspherical polynomials Pn (x).
In 1990, Tian Liang Tu [125] obtained the divergence and mean conver-
gence of the Hermite interpolation operator.

University of Lucknow, Lucknow–226007 16

 Interpolation on non-uniform nodes of the unit circle

Further in 1991, Zi Yu Wang and Shen Ji Tian [133] considered the zeros
0 0
of (1 − x2 )Pn−1 (x), where Pn−1 (x) is the derivative of (n − 1)th Legendre
polynomial and obtained the estimates for the same.
In 1991, R. Sakai [99] considered f ∈ C[−1, 1] and assumed

Sn = {x1n < x2n < · · · < xnn }

(α,β)
be the set of zeros of the Jacobi polynomial Pn (x); α, β > −1; with

 
n+α
Pn(α,β) (1) = and Sn∗ = {Sn } ∪ {±1}
n

and proved a convergence result for a certain sequence of interpolatory poly-

nomials operators on Sn∗ satisfying a certain differential equation (the Ultras-
pherical case) and showed that the convergence rate depends on the modulus
of continuity of f . In 1992, G. H. Min [90] obtained the mean convergence
of the derivatives of Hermite interpolation operator Hn (f, x) based on the
zeros of the Chebyshev polynomial of the first kind.
Further, J. Prasad [96] found the order of convergence and mean conver-
gence of a known Hermite interpolation polynomial on Chebyshev nodes of
the second kind. He [95] also studied the degree of approximation of the Her-
mite and Hermite - Fejér interpolation on the zeros of Legendre polynomial
of degree with normalization Pn (1) = 1 .
Also, in 1992, P. Vertesi and Y. Xu [132] considered the Hermite interpo-
lating polynomial Hnm (w, f ) be defined at the zeros of the Jacobi polynomi-
als, which are orthogonal on [−1, 1] with weight function

w(x) = (1 − x)α (1 + x)β , (α, β > −1).

University of Lucknow, Lucknow–226007 17

 Interpolation on non-uniform nodes of the unit circle

Also in 1993, J. Prasad [97] studied the Hermite and Hermite - Fejér
interpolation of higher order on the n distinct zeros of (1 − x2 )Pn−2 (x) where
Pn−2 (x) is the (n − 2)th Legendre polynomial.
In 1993, B. Della Vecchia and G. Mastroianni [131] considered the case
of Hermite interpolation based on the zeros of Jacobi and generalized Jacobi
polynomials in C - space and obtained the estimates in uniform and weighted
Lp −norm.
In 1996, T. N. T. Goodman, K. G. Ivanov and A. Sharma [58] considered
the behaviour of Hermite interpolant on the roots of unity.
In 1997, G. Y. Shi [108] obtained the convergence in Lp - norm of the trun-
cated Hermite interpolants on an arbitrary system of nodes, where truncated
Hermite polynomial is

r X
X n
Lnmr (f, x) = f (j) (xk )Ajk (x),
j=0 k=1

for a fixed r, 0 ≤ r ≤ m − 1, f ∈ [−1, 1] and Ajk (x) are polynomials of

degree mr − 1 defined by the conditions:

(p)
Ajk (x) = δjp δkq , j, p = 0(1)m − 1 and k, q = 1(1)n.

In 1998, S. Bahadur and K. K. Mathur [5] proved the convergence of

Quasi- Hermite interpolation on the nodes obtained by projecting vertically
the zeros of (1 − x2 )Pn (x) onto the unit circle, where Pn (x) stands for the
nth Legendre polynomial.
Later on convergence of Hermite interpolation was considered by P. Mathur
and Ashwini Kr Srivastava [88] on the same set of nodes.
In 2011, S. Bahadur [9] presented a method for computing the conver-
gence of Hermite interpolation polynomial based on the nodes obtained by

University of Lucknow, Lucknow–226007 18

 Interpolation on non-uniform nodes of the unit circle

Q
projecting vertically the zeros of n (x) on the unit circle.
Also, E. Berriochoa, A. Cachafeiro and E. Martinez Breyb [32] studied
the convergence of the Hermite - Fejér and the Hermite interpolation on
polynomials, which are constructed by taking equally spaced nodes on the
unit circle. Consequently, they achieved some improvements on Hermite in-
terpolation problems on the real line. In 2014, E. Berriochoa, A. Cachaferiro
and J. Diaz [35] deals with the order of convergence of the Laurent polyno-
mials of Hermite - Fejér interpolation on the unit circle with nodal system.
The supremum norm of the error of interpolation is obtained for analytic
functions as well as the corresponding asymptotic constants.
In the same year, S. Bahadur and M. Shukla [17] considered Hermite
(α,β)
interpolation on the vertically projected zeros of (1 − x2 )Pn (x), where
(α,β)
Pn (x) stands for Jacobi’s polynomial and proved the following Theorem
“Let f (z) be continuous in |z| ≤ 1 and analytic for |z| < 1 . Let the arbitrary
numbers βk0 s be such that:

|βk | = O(nω(f, n−1 )), k = 0(1)2n + 1

Then, {Hn } be defined by

2n+1
X 2n+1
X
Hn (z) = f (zk )Ak (z) + βk Bk (z),
k=0 k=0

satisfies the relation

|Hn (z) − f (z)| = O(ω(f, n−1 ) log n),

where ω(f, n−1 ) be the modulus of continuity of f (z).”

After that they [23] obtained the convergence theorem for the derivative of

University of Lucknow, Lucknow–226007 19

 Interpolation on non-uniform nodes of the unit circle

the Hermite interpolation on the above said nodes.They [18] also considered
the Hermite – interpolation on the vertically projected zeros of the derivative
of the Jacobi polynomial in place of Jacobi’s polynomial. S. Bahadur and
Sariya Bano [25] in 2017 read modified Hermite interpolation problem on
the nodes, which are obtained by projecting vertically the zeros of Legendre
polynomial onto the unit circle.
Recently, E. Berriochoa et al. [37] studied the Hermite interpolation on
the nodal arrays that meet certain separation properties. The novelty of
this article was to develop a interpolation theory that make a practical use
of these nodal systems linked to certain mechanical models that fit these
distribution.

1.8.3 Hermite-Fejér Interpolation

Fejér [49] in 1916 took an important step and proved a theorem, where the
values of the derivatives in the Hermite scheme were equal to zero.
Fejér’s theorem : If f ∈ C[−1, 1], then Hn (f, x) converges to f (x) uni-
formly on [-1,1] as n tends to infinty. Interpolation polynomials Hn (f, x) is
defined by

n
!2
X Tn (x)
Hn (f, x) = f (xkn )(1 − xkn x) ,
k=1
n(x − xkn )

where xkn are the zeros of the Chebyshev polynomial of the first kind.
Hn (f, x) satisfies the below given conditions, where k = 1, 2, ..., n.

Hn (f, xkn ) = f (xkn ) and Hn0 (f, xkn ) = 0.

Mills [89] in his paper highlights Hermite and Hemite Fejér interpolation

University of Lucknow, Lucknow–226007 20

 Interpolation on non-uniform nodes of the unit circle

as important techniques in the approximation theory. Knoop and Locher

[74] modified Hermite Fej́er interpolation at the zeros of Jacobi polynomials
by introducing more boundary conditions and obtaining pointwise conver-
gence for arbitrary α, β > −1. Fejér’s theorem has been extended to more
general nodal systems. For example, in 2001, Daruis and González-Vera [43]
extended Fejér’s result to the unit circle by considering the nodal system con-
stituted by the complex nth roots of unity. They proved that the sequence
of Hermite-Fejér interpolation polynomials uniformly converge for continous
functions on the unit circle.
Berriochoa, Cachafeiro and Garcı́a-Amor [33] extended the Fejér’s second
theorem to the unit circle. Then Berriochoa, Cachafeiro, Dı́az, and Martı́nez
Brey [34] obtained the supremum norm of the error of interpolation for an-
alytic functions and computed the order of convergence of Hermite-Fejér
interpolation on the unit circle considering the same set of nodes as of [43].
Apart from the uniform nodal system, Hermite-Fejér interpolation on the
unit circle have been also studied on some non-uniformly distributed nodes
on the unit circle (see [8] and [41]).
Higher order Hermite-Fejér Interpolation: It is a process of finding
a polynomial, which coincides with a continous function at the nodes of the
interpolation and the derivatives upto rth order (r > 1) are null at the nodal
points.
A considerable number of papers on higher order Hermite-Fejér interpolation
processes on real nodes have been published (see [111] and [135]).

1.8.4 Lacunary Interpolation

Lacunary interpolation is a method of finding a interpolatory polynomial,

which coincides with a function as well as the derivatives of the higher order

University of Lucknow, Lucknow–226007 21

 Interpolation on non-uniform nodes of the unit circle

at the nodes of the interpolation. .“Lacunary interpolation on the unit circle”

is the suggestve term that is used to describe the situation about which Szegö
remarked, “a novel feature of this investigation is to study the interpolation
polynomial for which f and f 00 are prescribed”. We define here “Lacunary
interpolation on the unit circle” as the interpolation process in which the
values of the function and its derivatives of higher order are prescribed at
given set of points on the unit circle.
The study of (0, 2) - interpolation was initiated by P. Turán [126] get an
approximate solution of differential equation of differential equation

y 00 + f y = 0.

The study of Lacunary interpolation was initiated by Surányi and Turán

[112] and Balázs and Turán [28] in the special case, when the values and
second derivatives were prescribed on the zeros of

0
πn (x) = (1 − x2 )Pn−1 (x),

where Pn−1 (x) is the (n − 1)th Legendre polynomial. At the outset the exis-
tence and uniqueness was discovered on the abscissas as the zeros of Ultras-
pherical polynomial

1
Pn(λ) (x), λ≥− , n ≥ 4,
2

whereas the explicit representation and convergence theorem were proved for
πn (x) abscissas only.
Also, K. K. Mathur and A. Sharma extended the study of P. Turán on
the zeros of Hermite polynomials, although they confined themselves to the

University of Lucknow, Lucknow–226007 22

 Interpolation on non-uniform nodes of the unit circle

existence, uniqueness and explicit representation of the (0, 2) and (0, 1, 3)

interpolants on R.
Saxena and Sharma {[100],[101]} extended the study to case of (0, 1, 3) -
interpolation. Mathur and Sharma [81] proved existence, uniqueness and
explicit representation of (0, 2) - interpolation on infinite interval. The con-
vergence theorem of Balázs and Turán [28] was further sharpened by Freud
[51]. In 1960, Kiŝ [70] extended the study of Lacunary interpolation on the
complex plane by considering (0, 2), (0, 1, 3) and (0, 1, . . . , r − 2, r)- interpo-
lation cases in the roots of unity and showed the regularity there for any
r ≥ 2. He proved the following:

Theorem 1.8.1. Let f (z) be continuous in |z| ≤ 1 and analytic for |z| < 1.
Further, let ω(δ) be the modulus of continuity of f and

1
lim ω(δ) log =0
δ→0 δ

and the given complex numbers βk ’s be such that

!
n2
βk = O , k = 1(1)n (1.6)
log n

(0,2) interpolatory polynomial

n
X n
X
Rn (z) = f (zk )Ak (z) + βk Bk (z),
k=1 k=1

converges uniformly to f (z) in |z| ≤ 1.

Abbasov [1] extended the above result to the two dimensional case. Sharma
[103] has considered the case of (0, 2, 3) - interpolation on the abscissas con-
sidered by Kiŝ. Later on he considered (0, M )- interpolation at equidistant
nodes on unit circle and proved that if a given function satisfies Dini - Lip-

University of Lucknow, Lucknow–226007 23

 Interpolation on non-uniform nodes of the unit circle

schitz condition, then interpolatory polynomial converges to it. David Fer-

guson [50] considered these interpolation problems, which could have been
uniquely solved for every choice of distinct nodes, real or complex.
Sharma and Leeming [104] again considered (0, n − 1, n) - interpolation prob-
lem and proved
“Let f (z) be continuous in |z| ≤ 1 and analytic for |z| < 1; and ω(δ) be the
modulus of continuity of f (eiφ ), 0 ≤ φ < 2π and

1
lim ω(δ) log =0
δ→0 δ

If πn (z) is a polynomial of degree ≤ 3n − 1, which interpolates to f (z) in the

roots of unity {zk }n1 together with

π (n−1) (zk ) = βkn and π (n) (zk ) = vkn ,

where ! !
n! n!
βkn =O and vkn = O ,
n3 log n n2 log n

then πn (x) converges to f (z) uniformly in |z| ≤ 1”. In 1961, A. K. Varma

[128] considered (0, 1, 3) interpolation problem on the Tchebychev nodes with
the end points. In another paper, A. K. Varma and A. Sharma [127] con-
sidered (0, 2) - interpolation problem for the same set of nodes. Also, A. K.
Varma [129] considered another case of Birkhoff interpolation and considered
the problem of existence, uniqueness, explicit representation and the problem
of convergence of the polynomials Rn (x, f ), which satisfy the properties:

Rn (xi ) = f (xi ), Rn0 (xi ) = αi ; i = 1(1)n + 2

Rn00 (xi ) = βi , Rn(iv) (xi ) = γi ; i = 2(1)n + 1,

University of Lucknow, Lucknow–226007 24

 Interpolation on non-uniform nodes of the unit circle

where xi ’s are (n + 2) zeros of (1 − x2 )Pn (x) given by

1 = x1 > x2 > . . . , xn+2 = −1 (1.7)

and Pn (x) denotes the nth Legendre polynomials.

In 1972, J. Prasad [94] studied the existence and uniqueness of (0,2)- inter-
(α,−α)
polation on the zeros of (1 − x2 )Pn (x) and obtained the explicit form
and convergence, when α = 1
3
and f ∈ C 2 [−1, 1], f 00 ∈ Lip ν, ν > 61 .
In 1981, S. D. Riemenscheider and A. Sharma [105] considered an interesting
special case of the Hermite Birkhoff interpolation involving the nth roots of
unity and derived the convergence.
In 1989, H. Gonska [57] discussed a modified general technique to derive
quantitative results on the degree of approximation by Lacunary interpola-
tion, where the values of the (R − 2)th derivatives of a function are prescribed
on some special knots. In 1990, A. K. Varma and J. Prasad [130] considered
Rn (f, z) to be the polynomial of degree ≤ 3n − 1 that interpolates the (on
|z| < 1 analytic and on |z| ≤ 1 continuous) function f (z) at the nth roots of
unity zk,n , k = 1(1)n and satisfies:






 Rn (f, zk,n ) = αk,n ,


0
 Rn (f, zk,n ) = βk,n ,



 000
Rn (f, zk,n ) = γk,n ,


and obtained convergence. In 1991, J. G. Jiang [64] studied (0,2)- type

0
interpolation based on the zeros of (1 − x2 )Pn−1 (x) and studied the degree of
approximation of the interpolatory polynomial and proved that the obtained
estimate was essentially best possible. In 1993, T. F. Xie [140] taken a
problem on the same set of zeros and considered for f ∈ C[−1, 1], denoted

University of Lucknow, Lucknow–226007 25

 Interpolation on non-uniform nodes of the unit circle

by Rn (f, 0, x) be the polynomial of degree ≤ 2n − 1, which satisfies the

conditions:

Rn (f, 0, x) = f (xk,n ) and Rn00 (f, 0, x) = 0; (k = 1(1)n)

and proved the theorem that

max |Rn (f, 0, x) − f (x)| = O(nω2 (f, n−1 )),

|z|≤1

where ω2 (f, n−1 ) is the modulus of continuity of f (x) of second order.

Siqing Xie [136] studied the problem of (0,2) - interpolation taking the nodes
Wn (x)
as the zeros of Wn (x) for n even or for n odd, where
x

0
Wn (x) = (1 − x2 )Pn−1 (x)

In 1995, Siqing Xie [137] considered an interesting interpolatory problem as-

(α,β)
suming the nodes as the vertical projection of the zeros of (1−x2 )Pn/2 (x), (n−
(α,β)
even) or (1±x)P(n−1)/2 (x), (n−odd); on unit circle and established the regu-
larity of (0, 1, . . . , r − 2, r) - interpolation on unit circle under certain restric-
tions on parameters α and β. In another paper, Siqing Xie [138] considered
a particular case for r = 3 and assuming the nodes as

z0 = 1, z2n+1 = −1, zk = cos θk + sin θk , zn+k = −zk , , k = 1(1)n.

to be the vertical projections on the unit circle of the zeros of (1 − x2 )Pn (x),
where Pn (x) stands for nth Legendre polynomial having the zeros xk =
cos θk , k = 1(1)n such that

1 > x1 > x2 > · · · > xn > −1,

University of Lucknow, Lucknow–226007 26

 Interpolation on non-uniform nodes of the unit circle

claimed the regularity, explicit forms and established a convergence theorem

of the (0, 1, 3) - interpolation. In 1998, M. R. Akhlaghi, A. M. Chak and A.
Sharma [4] considered (0,2,3) - interpolation on the zeros of πn (x). In 2000,
M. Lénard [75] obtained the convergence of the modified (0,2)- interpolation,
if the inner nodal points are the roots of the Ultraspherical polynomials with
odd integer parameter. In 2001, H. P. Dikshit [45] examined the regularity
of (0, m) and (0, 1, . . . , r − 2, r) - interpolation on the zeros of (z n + 1)(z − τ ),
where τ is any complex number. He also examined the regularity of (0,m)
interpolation on the zeros of (z 2n + 1)(z − w1 )(z − w2 ), where w1 , w2 are any
two distinct zeros of (z 2n − 1).
In 2004, W. Chen and A. Sharma [41] studied the regularity of the problem
of (0,m) - interpolation on the zeros of (z 2n + 1)(z 2 − 1) and also on the zeros
of (z 2n + 1)(z n − 1). It is clear that the nodes are not uniformly distributed
any more in these cases, because of the additional nodes at ±1. Also, P.
Mathur [85] considered (0,1,3) - interpolation on the projected zeros of an
Ultraspherical polynomial on the unit circle. After that S. Bahadur (with
M. Shukla) {[16],[22]} considered (0,1,3)* and (0,2) - interpolation on the
vertically projected zeros of Jacobi polynomial on the unit circle.

1.8.5 Pál-type Interpolation

In 1975, L. G. Pál [92] introduced the following interpolation process:

Let
−∞ < xn,n < xn−1,n < · · · < x1,n < +∞

be a system of distinct real points and put

n
Y
Wn (x) = (x − xi,n )
1=1

University of Lucknow, Lucknow–226007 27

 Interpolation on non-uniform nodes of the unit circle

The roots of yi,n (i = 1(1)n) of Wn0 (x) are interscaled between the roots of
Wn (x) i.e.

−∞ < xn,n < yn−1,n < · · · < x2,n < y1,n < +∞.

He proved that for given arbitrary numbers {a∗i,n }ni=1 and {b∗i,n }n−1
i=1 , there

exists a unique polynomial of degree ≤ 2n − 1 satisfying the conditions:


Tn (xi,n ) = a∗i,n ;

 i = 1(1)n

Tn0 (yi,n ) = b∗i,n ;


 i = 1(1)n − 1

with initial condition Tn (a) = 0, where a is a given point, different from the
nodal points xi,n (i = 1(1)n, n = 1, 2, . . . ).
In 1985, L. Szilli [120], was the first to apply this method on an infinite in-
terval for even values of n and obtained a uniquely determined polynomial
of degree ≤ 2n − 1. She also established a convergence theorem and showed
that the uniqueness fails to hold for the odd values of n. In 1990, M. R.
Akhlaghi and A. Sharma [2] studied the (0;2) - interpolation problem and
proved regularity and explicit representation of the interpolatory polynomial
when the function values were prescribed at the zeros of πn (x) and the second
derivative at the zeros of πn0 (x). They also considered the analogous prob-
lem by interchanging the nodes. Further in 1991, T. Kilgore and J. Prestin
[69] investigates the Hermite interpolant Hn (f ) of a function f ∈ C 2 [−1, 1],
where the derivatives at ±1 are interpolated upto the order r − 1. They
obtain estimates in the maximum norm over [-1,1] for (f − Hn (f ))(i) in all
possible cases, 0 ≤ i, r − 1 ≤ q. In 1995, K. K. Mathur and R. Srivastava
[83] studied the Pál - type Hermite interpolation on infinite interval. In the
same year, J. Szabados and A. K. Varma [117] studied the Pál - type (0,2)

University of Lucknow, Lucknow–226007 28

 Interpolation on non-uniform nodes of the unit circle

Rx
interpolation on the mixed zeros of Pn−1 (x) and −1
Pn−1 (t)dt and proved
that the interpolatory polynomial approximates the continuous function f
on [-1,1] in Telyakovskii - Gopengauz order. Also, T. F. Xie [141] considered
(0; 1; 2) - interpolation polynomials for approximation.
Further, in 1996, I. Joó and L. G. Pál [68] obtained Lacunary (0;0,1) - inter-
polation on the roots of Jacobi polynomials and its derivatives respectively.
Also, I. Joó and V. E. S. Szabó [66] gave a generalization of Pál - type in-
terpolation process. Also, L. G. Pál [93] gave a general Lacunary (0;0,1)-
interpolation process. Later on, R. Srivastava and K. K. Mathur [84] consid-
ered the weighted (0;0,2) - interpolation on the roots of Hermite polynomials.
In 2003, H. P. Dikshit [46] also considered the Pál - type interpolation on non
- uniformly distributed nodes on the unit circle involving the Möbius trans-
forms of the zeros of z n +1 and z n −1 along with one additional point  or two
additional points  and 1. Marcel G.de Brüin and A. Sharma [39] considered
Lacunary Pál - type interpolation, where they considered (0, m1 , m2 , . . . , mq )
- interpolation problem on the zeros of

z n − α0n , z n − α1n , . . . , z n − αqn

with
0 < α0 < · · · < αq

and showed regularity fornsufficiently large values of n. Moreover, he showed

that for a power series with radius of convergence > αq the phenomenon of
over convergence of the difference between the interpolatory polynomial of
the full power series and that of the partial sums of length (q + 1)nl will be
exhibited as n → ∞, where l ≥ 1 is an arbitrary integer. In 2006, P. Mathur
[86] considered (0,1;0) - interpolation on infinite interval. Further in 2011, S.

University of Lucknow, Lucknow–226007 29

 Interpolation on non-uniform nodes of the unit circle

Bahadur [10] considered Pál – type (0,1’0) - interpolation on two disjoint set
of nodes obtained by projecting vertically the zeros of Legendre polynomial
(1 − x2 )Pn (x) and Pn0 (x) respectively on the unit circle. In another paper,
she [8] also considered (0;0,1) - interpolation by interchanging the nodes. In
2013, M. Lénard [76] considered two set of nodes (one consists of the zeros
of a polynomial pn of degree n, while the elements of the other one are the
zeros of p0n ) different interpolation conditions are prescribed simultaneously.
Weighted (0, 1, . . . , r − 2, r) - interpolation conditions (r ≥ 2) are given on
one of the sets of the nodes with Hermite conditions on the other set and she
studied the regularity of the problem and obtained the explicit representation
of the fundamental polynomials of interpolation. Also, V. Srivastava et. al.
[109] considered trigonometric polynomials

cos (2n+1)θ
2
sin (2n+1)θ
2
rn (x) = θ
and sn (x) =
cos 2 sin 2θ

such that their zeros are interscaled and neither of the polynomials is the
extrema of the other polynomial. On these interscaled set of nodes, they
studied an interpolation process, when Lagrange data is prescribed on one
set of nodes and Hermite data is prescribed on the other set of nodes. In an-
other similar paper they [110] considered the converse of the above problem
by interchanging the conditions on the function and its derivative. Later on
S. Bahadur {[14],[15]} considered (0;0,2) and (0,2;0) - interpolation on the
vertically projected zeros of (1 − x2 )Pn (x) and Pn0 (x) onto the unit circle. Af-
ter that, she (with M. Shukla) {[19],[20]} considered the Hermite - Lagrange
and Lagrange - Hermite interpolation on the vertically projected zeros of
(α,β) (α,β)0
(1 − x2 )Pn (x) and Pn (x) onto the unit circle, obtained the regularity,
explicit forms and established a convergence theorem for that interpolatory

University of Lucknow, Lucknow–226007 30

 Interpolation on non-uniform nodes of the unit circle

polynomial.

1.9 What does notation (0,1) and (0,1;2) type

Interpolation means?

Sometimes readers will come across the notation (0, 1) and (0, 1; 2) type in-
terpolation. The comma in the notation tells us that the function and their
derivatives have been prescribed on a particular set of nodes, whereas semi-
colon in the notation tells us that the different nodal systems have been
considered. For e.g.
(0,1) interpolation: The absence of semicolon tells that only one set of
nodes is considered, whereas comma tells us that function and its first deriva-
tive are prescribed on that same set of nodes.
(0,1;2) interpolation: Single semicolon tells us that two different nodal sys-
tems have been considered for the interpolation. Also, the first half before
the semicolon tells that conditions on the function and its first derivatives
are prescribed on the one set of nodal system, whereas the second half after
the semicolon tells that conditions on the second derivative of the function
are prescribed on the other set of nodal system.

1.10 Preliminaries
(α,β)
Throughout the thesis, Jacobi polynomial of degree n is denoted by Pn (x),
where α, β > −1.
Let us define two sets T = {z ∈ C : |z| = 1} and D = {z ∈ C : |z| < 1}.

University of Lucknow, Lucknow–226007 31

 Interpolation on non-uniform nodes of the unit circle

(α,β)
The differential equation satisfied by Pn (x) is

00 0
(1 − x2 )Pn(α,β) (x) + [β − α − (α + β + 2)x]Pn(α,β) (x)

+ n(n + α + β + 1)Pn(α,β) (x) = 0. (1.8)

Figure 1.4: Szegő transformation

1 + z2
Using the Szegő transformation x = , we have
2z

00
(z 2 − 1)4 Pn(α,β) (x) + 4z(z 2 − 1) {(α + β + 2)z 2 + 1}(z 2 − 1)


0
−2z 3 (β − α) Pn(α,β) (x) − 16z 6 n(n + α + β + 1)Pn(α,β) (x) = 0.

(1.9)

Let Z2n and T2n−2 be two distinct sets of nodes, which are obtained by
(α,β) (α,β)0
projecting vertically the zeros of Pn (x) and Pn (x) respectively on the
unit circle.

Z2n = {zk = xk + iyk = cos θk + i sin θk ; zn+k = zk ;

k = 1, 2, ..., n ; xk , yk ∈ R} (1.10)

T2n−2 = {tk = x∗k + iyk∗ = cos φk + i sin φk ; tn+k = tk ;

k = 1, 2, ..., (n − 1) ; x∗k , yk∗ ∈ R}, (1.11)

University of Lucknow, Lucknow–226007 32

 Interpolation on non-uniform nodes of the unit circle

(α,β)
where, xk = cos θk , k = 1(1)n are the zeros of Pn (x) and x∗k = cos φk ,
(α,β)0
k = 1(1)n − 1 are the zeros of Pn (x) .
The nodal polynomials W (z) and W ∗ (z) defined on Z2n and T2n−2 are given
by (1.12) and (1.13) respectively.

2n
!
Y 1 + z2 n
W (z) = (z − zk ) = Kn Pn(α,β) z , (1.12)
k=1
2z

where
Γ(α + β + n + 1)
Kn = 22n n! .
Γ(α + β + 2n + 1)

and !
2n−2
Y 0 1 + z 2 n−1
W (z) =
∗
(z − tk ) = Kn∗ Pn(α,β) z , (1.13)
k=1
2z

where
Γ(α + β + n + 1)
Kn∗ = 22n−1 (n − 1)! .
Γ(α + β + 2n + 1)

Let
R(z) = (z 2 − 1)W (z). (1.14)

The fundamental polynomials of Lagrange interpolation on the zeros of W (z),

W ∗ (z) and R(z) are respectively given by (1.15) (1.16) and (1.17).

W (z)
Lk (z) = , k = 1, 2, ..., 2n. (1.15)
(z − zk )W 0 (zk )

W ∗ (z)
Lk∗ (z) = , k = 1, 2, ..., (2n − 2). (1.16)
(z − tk )W ∗0 (tk )
R(z)
Lk (z) = , k = 0, 1, 2, ..., 2n + 1, (1.17)
(z − zk )R 0 (zk )

where z0 = 1 and z2n+1 = −1.

University of Lucknow, Lucknow–226007 33

 Interpolation on non-uniform nodes of the unit circle

We can write z = x + iy, where x, y ∈ R. If | z |= 1, then

√
| z 2 − 1 |= 2 1 − x2 . (1.18)

and r
√ √ q
| z − zk |= 2 1 − xxk − 1− x2 1 − x2k . (1.19)

To evaluate the estimates of the fundamental polynomials, we will use the

following (refer to pg.164-166 of [118])
For −1 ≤ x ≤ 1 and α ≥ β, we have

(1 − x2 )1/2 | Pn(α,β) (x) |= O(nα−1 ). (1.20)

| Pn(α,β) (x) |= O(nα ). (1.21)

0
(1 − x2 ) | Pn(α,β) (x) |= O(nα+1 ). (1.22)

α 1 1
(1 − x2 ) 2 + 4 | Pn(α,β) (x) |= O(n− 2 ). (1.23)

0
| Pn(α,β) (x) |= O(nα+2 ). (1.24)

00
| Pn(α,β) (x) |= O(nα+4 ). (1.25)

Considering set of nodes Z2n and T2n−2 such that for each k , xk , x∗k ∈ (−1, 1),
we have !−2
k
(1 − x2k )−1 ∼ . (1.26)
n
1
| Pn(α,β) (x∗k ) |∼ k −α− 2 nα . (1.27)

0 3
| Pn(α,β) (xk ) |∼ k −α− 2 nα+2 . (1.28)

00 5
| Pn(α,β) (xk ) |∼ k −α− 2 nα+4 . (1.29)

University of Lucknow, Lucknow–226007 34

 Interpolation on non-uniform nodes of the unit circle

For Legendre polynomials, above results are used after substi-

tuting α = β = 0.

University of Lucknow, Lucknow–226007 35

Chapter 2

Convergence of
Lagrange-Hermite
Interpolation on unit circle

In this chapter, we study a Lagrange-Hermite interpolation on the nodes,

(0,0)
which are obtained by projecting vertically the zeroes of the (1 − x2 )Pn (x)
(0,0)
on the unit circle, where Pn (x) stands for Legendre polynomial of degree
n. We prove the regularity of the problem, give explicit forms and establish
a convergence theorem for the same.

2.1 The Problem and the Regularity

We are interested in determining the interpolatory polynomial LHn (z) of

degree ≤ 2n + 3 satisfying the following conditions.



LHn (f, zk ) = f (zk ),
 k = 0(1)2n + 1,
(2.1)
LH0n (f, ±1) = α±1 ,



where f (zk ) and α±1 are arbitrary complex constants.

Theorem 2.1.1. LHn (z) is regular on Z2n

S
{−1, 1}.

Proof. It is sufficient, if we show the unique solution of (2.1) is LHn (z) ≡ 0.

36
 Interpolation on non-uniform nodes of the unit circle

Let us consider
LHn (z) = R(z) q(z), (2.2)

where q(z) is a linear polynomial.

Since, R(zk ) = 0 from (1.14)
=⇒ LHn (zk ) = 0 for k = 0(1)2n + 1.
On differentiating (2.2) with respect to z, we have

LH0n (z) = R(z) q 0 (z) + R 0 (z) q(z). (2.3)

Since , LH0n (±1) = 0, we get q(±1) = 0.

Therefore, we have
q(z) = az + b, (2.4)

where a and b are the arbitrary constants independent of n and z. Now

substituting z = 1 and z = −1 in (2.4), we get a = b = 0.
Hence the theorem follows.

2.2 Explicit Representation of Interpolatory

Polynomials

We shall write

2n+1
X X
LHn (z) = f (zk )Ak (z) + α±1 Bk (z), (2.5)
k=0 k=0,2n+1

where Ak (z) and Bk (z) are fundamental polynomials, each of degree at-
most 2n + 3 satisfying the conditions (2.6) and (2.7) respectively.

University of Lucknow, Lucknow–226007 37

 Interpolation on non-uniform nodes of the unit circle



Ak (zj ) = δkj
 ; j, k = 0(1)2n + 1,
(2.6)
A0k (zj ) = 0

 ; k = 0(1)2n + 1, j = 0, 2n + 1.


Bk (zj ) = 0
 ; j = 0(1)2n + 1, k = 0, 2n + 1
(2.7)
Bk0 (zj ) = δkj

 ; j, k = 0, 2n + 1.

Theorem 2.2.1. Fundamental polynomial Bk (z) is given by

R(z)(z + zk )
Bk (z) = , (2.8)
4Kn

where k = 0, 2n + 1.

Proof. Let us consider

Bk (z) = (z 2 − 1)W (z)t(z), (2.9)

where t(z) is a linear polynomial. From the first condition in (2.7), one can
verify that Bk (zj ) = 0 for j = 0(1)2n + 1.
Differentiating (2.9) with respect to z , we get

Bk0 (z) = 2zW (z)t(z) + (z 2 − 1)W 0 (z)t(z) + (z 2 − 1)W (z)t0 (z). (2.10)

For j = 0, 2n + 1, we have

Bk0 (zj ) = 2zj W (zj )t(zj ). (2.11)

Using the second set of condition of (2.7), (1.15) and (1.17), we get t(1) =
1 −1
and t(−1) = .
2Kn 2Kn

University of Lucknow, Lucknow–226007 38

 Interpolation on non-uniform nodes of the unit circle

Therefore, we have
z + zk
t(z) = . (2.12)
4Kn

Hence the theroem follows.

Theorem 2.2.2. Fundamental polynomial Ak (z) is defined differently for

various values of k
For k = 1(1)2n
(z + zk )R(z)
Ak (z) = Lk (z) + , (2.13)
(zk2 − 1)R 0 (zk )

and for k = 0, 2n + 1

h 1  1 L0k (zk )  i
Ak (z) = (z + zk )Lk (z) − + (z − zk ) . (2.14)
2zk 4zk2 2zk

Proof. Case I: For k = 1(1)2n

Let us consider
Ak (z) = Lk (z) + ck (z + zk )R(z), (2.15)

where ck is a constant independent of n and z. Using (1.14) and (1.17), we

can verify the first set of condition in (2.6)
On differentiating (2.15) with respect to z, we get

A0k (z) = L0k (z) + ck (z + zk )R 0 (z) + ck R(z).

At z = zj , we get

A0k (zj ) = L0k (zj ) + ck (zj + zk )R 0 (zj ).

To satisfy the second set of conditions in (2.6), A0k (zj ) = 0. Thus, we get

L0k (zj ) + ck (zj + zk )R 0 (zj ) = 0, (2.16)

University of Lucknow, Lucknow–226007 39

 Interpolation on non-uniform nodes of the unit circle

L0k (zj )
ck = − . (2.17)
(zj + zk )R 0 (zj )

Let us simplify further by substituting the value of L0k (zj )

Using (1.17), we have

0
Lk (z)(z − zk )R (zk ) = R(z). (2.18)

On differentiating with respect to z , we get

0 0
L0k (z)(z − zk )R (zk ) + Lk (z)R (zk ) = R 0 (z). (2.19)

Substituting z = zj in (2.19), we get

0
L0k (zj )(zj − zk )R (zk ) = R 0 (zj ). (2.20)

Putting this value of R 0 (zj ) in (2.16), we get

1
ck = . (2.21)
R 0 (zk )(zk2 − zj2 )

Since, j = 0, 2n + 1 for the second set of conditions in (2.7) implies zj2 =

(±1)2 = 1, we can rewrite (2.21) as

1
ck = . (2.22)
R 0 (zk )(zk2 − 1)

Putting value of ck from (2.22) in (2.15), we have (2.13).

Case II: For k = 0, 2n + 1, let us consider

Ak (z) = (z + zk )Lk (z)tk (z), (2.23)

University of Lucknow, Lucknow–226007 40

 Interpolation on non-uniform nodes of the unit circle

where tk (z) is a linear polynomial.

On satisfying the first set of conditions in (2.6), we get

1 

t0 (z0 ) = 
2 (2.24)
1
t2n+1 (z2n+1 ) = − 
2
Similarly, satisfying the second set of conditions in (2.6), we get

11  
t00 (z0 ) =− 0
+ L0 (z0 ) 

2 2 (2.25)
11 
t02n+1 (z2n+1 ) = − L02n+1 (z2n+1 ) 
2 2

Combining the results of (2.24) and (2.25), we get


1
tk (zk ) = 

2zk

(2.26)
0 tk (zk )  0

tk (zk ) = − 1 + 2zk Lk (zk ) 

2zk

Since, tk (z) is a polynomial of one degree, let us consider

tk (z) = a1 + b1 (z − zk ), (2.27)

where a1 and b1 are the constants independent of n and z.

Using (2.26), we have

1

tk (zk ) = a1 = 

2zk

(2.28)
 1 L0k (zk ) 
t0k (zk ) = b1 = − + 
4zk2

2zk

Substituting these values of constants a1 and b1 in (2.27), we get

h 1  1 L0k (zk )  i
tk (z) = − + (z − zk .
)
2zk 4zk2 2zk

University of Lucknow, Lucknow–226007 41

 Interpolation on non-uniform nodes of the unit circle

We get the desired result (2.14) by putting the value of tk (z) in (2.23). Hence
the theorem follows.

2.3 Estimation of Fundamental Polynomials

Lemma 2.3.1. [17] Let Lk (z) be given by (1.17), then

2n+1
X
max | Lk (z) | ≤ c log n, (2.29)
|z|=1
k=0

where c is a constant independent of n and z.

Lemma 2.3.2. Let Bk (z) be given by (2.8), then for k = 0, 2n + 1

r
2
| Bk (z) | ≤ . (2.30)
πn

Proof. From (2.8), we have

R(z)(z + zk )
| Bk (z) |= .
4Kn

Using (1.12) and (1.14), we get

(0,0)
|z 2 − 1||Pn (x)||z n ||z + zk |
| Bk (z) |= .
4

Since, max |z| = 1. Using (1.20), we have

√
| Bk (z) |≤ 1 − x2 |Pn(0,0) (x)|.

Thus, using (1.23), we have our lemma.

University of Lucknow, Lucknow–226007 42

 Interpolation on non-uniform nodes of the unit circle

Lemma 2.3.3. Let Ak (z) be given in theorem (2.13), then

2n+1
X
| Ak (z) |≤ c1 n1/2 log n,
k=0

where c1 is a constant independent of n and z.

Proof. From (2.13), the fundamental polynomial Ak (z) for k = 1(1)2n

(z + zk )R(z)
Ak (z) = Lk (z) + .
(zk2 − 1)R 0 (zk )

Then,
|z + zk ||R(z)|
|Ak (z)| ≤ |Lk (z)| + .
|zk2 − 1||R 0 (zk )|

Using (1.12) and (1.14), we get

(0,0)
|z + zk ||z 2 − 1||Pn (x)||z n |
|Ak (z)| ≤ |Lk (z)| + (0,0)0
.
|zk2 − 1|2 |Pn (xk )||zkn |

Since, |z| = 1 and |zk | = 1. Using (1.20), we have

√ (0,0)
1 − x2 |Pn (x)|
|Ak (z)| ≤ |Lk (z)| + (0,0)0
.
(1 − x2k )|Pn (xk )|

Using (1.23),(1.26) and (1.28), we get

n1/2
|Ak (z)| ≤ |Lk (z)| + .
k

Taking summation both the sides, we obtain

2n 2n 2n
X X X n1/2
|Ak (z)| ≤ |Lk (z)| + ,
k=1 k=1 k=1
k
X2n X2n
|Ak (z)| ≤ |Lk (z)| + n1/2 log n. (2.31)
k=1 k=1

University of Lucknow, Lucknow–226007 43

 Interpolation on non-uniform nodes of the unit circle

Now for k = 0, 2n + 1, from (2.14), we have

" ! #
1 1 L0k (zk )
Ak (z) = (z + zk )Lk (z) − + (z − zk ) .
2zk 4zk2 2zk

Taking modulus on both the sides , we get

!
1 L0k (zk )
|Ak (z)| ≤ |Lk (z)| + 2
+ (z 2 − zk2 )Lk (z) .
4zk 2zk

Taking summation , we have

!
X X X 1 L0k (zk )
|Ak (z)| ≤ |Lk (z)| + + (z 2 − zk2 )Lk (z) .
k=0,2n+1 k=0,2n+1 k=0,2n+1
4zk2 2zk

X X X (z 2 − zk2 )Lk (z)

|Ak (z)| ≤ |Lk (z)| + .
k=0,2n+1 k=0,2n+1 k=0,2n+1
4zk2
X (z 2 − zk2 )Lk (z)L0k (zk )
+ .
k=0,2n+1
2zk

Using (1.17), we get

X X X (z 2 − zk2 )Lk (z)

|Ak (z)| ≤ |Lk (z)| + .
k=0,2n+1 k=0,2n+1 k=0,2n+1
4zk2
X (z 2 − zk2 )R(z)R 00 (zk )
+ .
k=0,2n+1
4zk (z − zk )(R 0 (zk ))2

Using (1.12) and (1.14), we have

X X X (z 2 − zk2 )Lk (z)

|Ak (z)| ≤ |Lk (z)| + .
k=0,2n+1 k=0,2n+1 k=0,2n+1
4zk2
(0,0)
X (z + zk )(z 2 − 1)Kn Pn ((1 + z 2 )/2z)z n R 00 (zk )
+ .
k=0,2n+1
4zk (R 0 (zk ))2

University of Lucknow, Lucknow–226007 44

 Interpolation on non-uniform nodes of the unit circle

Using (1.18) and (1.23), we get

X X
|Ak (z)| ≤ |Lk (z)| + c2 + c3 n1/2 , (2.32)
k=0,2n+1 k=0,2n+1

where c2 and c3 are constants independent of z and n.

Combining (2.31) and (2.32), we obtain

2n+1
X 2n+1
X
|Ak (z)| ≤ |Lk (z)| + c2 + c3 n1/2 + n1/2 log n. (2.33)
k=0 k=0

Using lemma 2.3.1, we get

2n+1
X
|Ak (z)| ≤ c log n + c2 + c3 n1/2 + n1/2 log n. (2.34)
k=0

Hence, we have our desired lemma 2.3.3 .

2.4 Convergence

Theorem 2.4.1. Let f (z) be a function continuous on T ∪ D and analytic

on D. Let the arbitary numbers α±1 ’s be such that

| α±1 |= O(n ω2 (f, n−1 )). (2.35)

Then, the sequence of interpolatory polynomial {LHn (z)} defined by

2n+1
X X
LHn (z) = f (zk )Ak (z) + α±1 Bk (z), (2.36)
k=0 k=0,2n+1

University of Lucknow, Lucknow–226007 45

 Interpolation on non-uniform nodes of the unit circle

satisfies the relation

| LHn (z) − f (z) | = O(ω2 (f, n−1 )n1/2 log n), (2.37)

where ω2 (f, n−1 ) denotes the second modulus of continuity of f (z).

Remark 2.1 : Let f (z) be a function continuous on T ∪ D and analytic

on D and f 0  Lip α, α > 0, then the sequence {LHn (z)} converges uniformly
to f (z) on T ∪ D, which follows from (2.37) as

ω2 (f, n−1 ) ≤ n−1 ω1 (f 0 , n−1 ) = O(n−1−α ). (2.38)

To prove the Theorem 2.4.1, we shall need following.

Let f (z) be a function continuous on T ∪ D and analytic on D. Then, there
exists a polynomial Fn (z) of degree ≤ 2n+3 satisfying Jackson’s inequality
[62].

| f (z) − Fn (z) | ≤ C ω2 (f, n−1 ), z = eiθ (0 ≤ θ < 2π) (2.39)

and also an inequality by O.Kiŝ[70]

| Fn(m) (z) | ≤ C nm ω2 (f, n−1 ), m ε Z+ , (2.40)

where C is a constant independent of n and z.

Proof. Since, LHn (z) be the uniquely determined polynomial of degree ≤

2n + 3 and the polynomial Fn (z) satisfying equation (2.39) and (2.40) can

University of Lucknow, Lucknow–226007 46

 Interpolation on non-uniform nodes of the unit circle

be expressed as

2n+1
0
X X
Fn (z) = Fn (zk )Ak (z) + Fn (zk )Bk (z). (2.41)
k=0 k=0,2n+1

Then, we can write

| LHn (z) − f (z) | ≤ | LHn (z) − Fn (z) | + | Fn (z) − f (z) | .

Using (2.36) and (2.41), we have

2n+1
X
| LHn (z) − f (z) | ≤ | f (zk ) − Fn (zk ) || Ak (z) |
k=0
X
+ | α±1 − Fn0 (zk ) || Bk (z) | + | Fn (z) − f (z) |,
k=0,2n+1
2n+1
X X
| LHn (z) − f (z) | ≤ | f (zk ) − Fn (zk ) || Ak (z) | + | α±1 || Bk (z) |
k=0 k=0,2n+1
| {z } | {z }
N1 N2
X
+ | Fn0 (zk ) || Bk (z) | + | Fn (z) − f (z) | . (2.42)
| {z }
k=0,2n+1 N4
| {z }
N3

Using (2.39) and lemma 2.3.3 , we get

N1 = O(ω2 (f, n−1 )n1/2 log n). (2.43)

Using (2.35) and lemma 2.3.2, we get

N2 = O(n1/2 ω2 (f, n−1 )). (2.44)

Using (2.40) and lemma 2.3.2, we get

N3 = O(n1/2 ω2 (f, n−1 )). (2.45)

University of Lucknow, Lucknow–226007 47

 Interpolation on non-uniform nodes of the unit circle

Using (2.39), we have

N4 = O(ω2 (f, n−1 )). (2.46)

Using (2.43), (2.44), (2.45), (2.46) in (2.42), we get

| LHn (z) − f (z) | = O(ω2 (f, n−1 )n1/2 log n). (2.47)

Hence, theorem 2.4.1 follows.

University of Lucknow, Lucknow–226007 48

Chapter 3

Convergence of Interpolatory
polynomial between Lagrange
and Hermite

In this chapter, we study an interpolation problem ,which is an intermediate

problem between Lagrange and Hermite. We considered this problem on the
(0,0)
nodes obtained by projecting vertically the zeroes of the (1−x2 )Pn (x) onto
(0,0)
the unit circle, where Pn (x) stands for Legendre polynomial of degree n.
We prove the regularity of the problem, give explicit forms and establish a
convergence theorem for the same.

3.1 The Problem and the Regularity

We are interested in determining the interpolatory polynomial HLn (z) of

degree ≤ 2n + 1 satisfying the following conditions.



HLn (f, zk ) = τk
 ; k = 0, 2n + 1,
(3.1)
HL0n (f, zk ) = βk

 ; k = 1(1)2n,

where τk and βk are arbitrary complex constants.

49
 Interpolation on non-uniform nodes of the unit circle

Theorem 3.1.1. HLn (z) is regular on Z2n

S
{−1, 1}.

Proof. It is sufficient if we show the unique solution of (3.1) is HLn (z) ≡ 0.

For k = 1(1)2n, let us consider

HL0n (zk ) = 0. (3.2)

Since zk ’s are the zeros of W (z), so we can write

HL0n (z) = a1 W (z), (3.3)

where a1 is a constant independent of n and z. On integrating (3.3), we get

Z z
HLn (z) = a1 W (t)dt. (3.4)
−1

Now, since HLn (±1), we have

Z 1
HLn (1) = a1 W (t)dt. (3.5)
−1

R1
Since, −1
W (t)dt 6= 0. This implies a1 = 0.
Hence, the theorem follows.

3.2 Explicit Representation of Interpolatory

Polynomials

We shall write

X 2n
X
HLn (z) = τk Uk (z) + βk Vk (z), (3.6)
k=0,2n+1 k=1

where Uk (z) and Vk (z) are unique polynomials, each of degree atmost

University of Lucknow, Lucknow–226007 50

 Interpolation on non-uniform nodes of the unit circle

2n + 1 satisfying the conditions



Uk (zj ) = δkj
 ; j, k = 0, 2n + 1
(3.7)
Uk0 (zj ) = 0

 ; k = 0, 2n + 1, j = 1(1)2n,

and



Vk (zj ) = 0
 ; j = 0, 2n + 1, k = 1(1)2n
(3.8)
Vk0 (zj ) = δkj

 ; j, k = 1(1)2n.

Theorem 3.2.1. Fundamental polynomial Vk (z) for k = 1(1)2n is given as

Z z
2
Vk (z) = (z − 1)Lk (z) − (t2 − 1)Lk0 (t)dt
−1
Z z Z z
Sk (1)
+ (1 − 2zk ) Lk (t)dt − R 1 W (t)dt. (3.9)
−1 −1
W (t)dt −1

Proof. Let

Z z
2
Vk (z) = (z − 1)Lk (z) + Sk (z) + a W (t)dt, (3.10)
−1

where Vk (z) is atmost of degree 2n+1 satisfying the conditions given in (3.8),
then

Z 1
Vk (1) = Sk (1) + a W (t)dt,
−1
Z 1
0 = Sk (1) + a W (t)dt,
−1

University of Lucknow, Lucknow–226007 51

 Interpolation on non-uniform nodes of the unit circle

Sk (1)
a = −R 1 . (3.11)
−1
W (t)dt

On differentiating (3.10) with respect to z, we get

Vk0 (z) = 2zLk (z) + (z 2 − 1)Lk0 (z) + Sk0 (z) + aW (z). (3.12)

At z = zj , we have

Vk0 (zj ) = 2zj Lk (zj ) + (zj2 − 1)Lk0 (zj ) + Sk0 (zj ). (3.13)

On satisfying the second set of conditions given in (3.8) for j 6= k, we have

0 = (zj2 − 1)Lk0 (zj ) + Sk0 (zj ),

Sk0 (zj ) = −(zj2 − 1)Lk0 (zj ).

Since, Lk (zj ) = 0 for j 6= k, we can write

Sk0 (z) = −(z 2 − 1)Lk0 (z) + ak Lk (z), (3.14)

where ak is a constant independent of n and z. Integrating above equation,

we get

Z z Z z
Sk (z) = − 2
(t − 1)Lk0 (t)dt + ak Lk (t)dt. (3.15)
−1 −1

Now, to determine the constant ak , we use the second set of conditions given
in (3.8) for j = k,

Vk0 (zk ) = 2zk Lk (zk ) + (zk2 − 1)Lk0 (zk ) + Sk0 (zk ). (3.16)

University of Lucknow, Lucknow–226007 52

 Interpolation on non-uniform nodes of the unit circle

Substituting the value of Sk0 (z) from (3.14) at z = zk in (3.16), we get

ak = 1 − 2zk . (3.17)

Hence, from (3.10), (3.11), (3.15) and (3.17), we have theorem 3.2.1 .

Theorem 3.2.2. For k = 0, 2n + 1

Rz
−z
W (t)dt
Uk (z) = (−1)k R 1 k . (3.18)
−1
W (t)dt

Proof. Let
Z z
Uk (z) = (−1) bkk
W (t)dt, (3.19)
−zk

where Uk (z) is atmost of degree 2n + 1. At z = zj , we get

Z zj
k
Uk (zj ) = (−1) bk W (t)dt. (3.20)
−zk

On satisfying the conditions in (3.7), we get

1
bk = R 1 .
−1
W (t)dt

Hence, the theorem follows.

3.3 Estimation of Fundamental Polynomials

Lemma 3.3.1. Let Vk (z) be given by theorem 3.2.1, then

2n
X
| Vk (z) | = O(log n).
k=1

Proof. From (3.9), we have

University of Lucknow, Lucknow–226007 53

 Interpolation on non-uniform nodes of the unit circle

Z z
2
Vk (z) = (z − 1)Lk (z) − (t2 − 1)Lk0 (t)dt
−1
Z z Z z
Sk (1)
+ (1 − 2zk ) Lk (t)dt − R 1 W (t)dt. (3.21)
−1 −1
W (t)dt −1

On taking modulus both the sides, we get

Z z Z z
2
|Vk (z)| ≤ |(z − 1)Lk (z)| + (1 − t 2
)Lk0 (t)dt + (1 − 2zk ) Lk (t)dt
| {z } −1 −1
I1 | {z }
I2
Z z
Sk (1)
+ R1 W (t)dt . (3.22)
W (t)dt −1
| −1 {z }
I3

Taking summation on both the sides, we get

2n
X 2n
X 2n
X 2n
X
|Vk (z)| ≤ I1 + I2 + I3 . (3.23)
k=1 k=1 k=1 k=1

From (3.22), we have

I1 = |(z 2 − 1)Lk (z)|. (3.24)

Since, |zk | = 1 and using (1.12) and (1.15), we get

(0,0)
|(z 2 − 1)||Pn (x)||z n |
I1 = (0,0)0
. (3.25)
|(z − zk )||Pn (xk )|

Using (1.18) and (1.19), we get

√ (0,0)
2 1 − x2 |Pn (x)|
I1 = √ q √ ,
(0,0)0
p
2 2
2 1 − xxk − 1 − x 1 − xk |Pn (xk )|

University of Lucknow, Lucknow–226007 54

 Interpolation on non-uniform nodes of the unit circle

√ √ (0,0)
q √ p
2 1 − x |Pn (x)| 1 − xxk + 1 − x2 1 − x2k
2
= p (0,0)0
,
(1 − xxk )2 − (1 − x2 )(1 − x2k ) |Pn (xk )|
√ √ (0,0)
q p
2 1 − x |Pn (x)| 1 − xxk + (1 − xxk )2 − (x − xk )2
2
= (0,0)0
,
|x − xk | |Pn (xk )|
√ √ (0,0) √
2 1 − x2 |Pn (x)| 1 − xxk
≤ (0,0)0
.
|x − xk | |Pn (xk )|

1
For | x − xk | ≥ | 1 − x2k |, we have
2
√ √ (0,0) √
2 2 1 − x2 |Pn (x)| 1 − xxk
I1 ≤ (0,0)0
, (3.26)
(1 − x2k )|Pn (xk )|
√ (0,0)
4 1 − x2 |Pn (x)|
≤ (0,0)0
.
(1 − x2k )|Pn (xk )|

Using (1.20), (1.26) and (1.28), we get

!
n−1
I1 = O .
(k 2 n−2 )(k −3/2 n2 )

1
I1 = O(n−1 k − 2 ). (3.27)

1
The estimate remains the same in the case, when | x − x∗k | < | 1 − x∗2
k |.
2
On taking summation , we get

2n
!
X log n
I1 = O 1 . (3.28)
k=1 n2

Now, to determine I2 , let us first determine Lk0 (z). From (1.15), we have

W (z)
Lk (z) = .
(z − zk )W 0 (zk )

0
Lk (z)(z − zk )W (zk ) = W (z).

University of Lucknow, Lucknow–226007 55

 Interpolation on non-uniform nodes of the unit circle

Differentiating above equation with respect to z, we get

0 0
Lk0 (z)(z − zk )W (zk ) + Lk (z)W (zk ) = W 0 (z).

0
W 0 (z) − Lk (z)W (zk )
Lk0 (z) = .
(z − zk )W 0 (zk )

Taking Modulus on both the sides, we get

W 0 (z) Lk (z)
|Lk0 (z)| ≤ 0 + . (3.29)
(z − zk )W (zk ) (z − zk )

From (3.22), we have

Z z Z z
I2 = 2
(1 − t )Lk0 (t)dt + (1 − 2zk ) Lk (t)dt .
−1 −1

Z z Z z
I2 ≤ 2
| (t − 1) || Lk0 (t) | dt + (1 + 2 | zk |) | Lk (t) | dt.
−1 −1

Using (3.29), (1.12), (1.18), (1.19), (1.22), (1.23), (1.26), (1.28) and following
the similar scheme as used to evaluate I1 , we get

2n
X
I2 = O(log n). (3.30)
k=1

Also Rz
W (t)dt
I3 = Sk (1) R−1
1 .
−1
W (t)dt

Using (1.12), (1.15) and (3.15), we have

2n
X
I3 = O(1). (3.31)
k=1

Using (3.28), (3.30) and (3.31) in (3.23), lemma follows.

University of Lucknow, Lucknow–226007 56

 Interpolation on non-uniform nodes of the unit circle

Lemma 3.3.2. Let Uk (z) be given by theorem 3.2.2, then

X
| Uk (z) |= O(1).
k=0,2n+1

Proof.
X
| Uk (z) |=| U0 (z) | + | U2n+1 (z) | .
k=0,2n+1

Z z
1
| U0 (z) |≤ R 1 | W (t)dt | .
| −1 W (t)dt | −1

Using (1.12), we have

| U0 (z) = O(1). (3.32)

Similarly, we have
| U2n+1 (z) |= O(1). (3.33)

Combining (3.32) and (3.34), we get lemma 3.3.2

3.4 Convergence

Theorem 3.4.1. Let f (z) be a function continuous on T ∪ D and analytic

on D. Let the arbitrary numbers βk ’s be such that

| βk |= O(n ω2 (f, n−1 )) ; k = 1(1)2n. (3.34)

Then, the sequence of interpolatory polynomial {HLn (z)} defined by

X 2n
X
HLn (z) = f (zk )Uk (z) + βk Vk (z), (3.35)
k=0,2n+1 k=1

University of Lucknow, Lucknow–226007 57

 Interpolation on non-uniform nodes of the unit circle

satisfies the relation

| HLn (z) − f (z) | = O(nω2 (f, n−1 )logn), (3.36)

where ω2 (f, n−1 ) be the second modulus of continuity of f (z).

Remark 3.1: Let f (z) be a function continuous on T ∪ D and analytic

on D and f 0  Lip α, α > 0, then the sequence {HLn (z)} converges uniformly
to f (z) on T ∪ D, which follows from (3.36) as

ω2 (f, n−1 ) ≤ n−1 ω2 (f 0 , n−1 ) = O(n−1−α ), (3.37)

where ω2 (f, n−1 ) be the second modulus of continuity of f (z).

To prove the Theorem 3.4.1, we shall need following.
Let f (z) be a function continuous on T ∪ D and analytic on D. Then, there
exists a polynomial Fn (z) of degree ≤ 2n+1 satisfying Jackson’s inequality
[62].

| f (z) − Fn (z) | ≤ C ω2 (f, n−1 ), z = eiθ (0 ≤ θ < 2π) (3.38)

and also an inequality by O.Kiŝ[70]

| Fn(m) (z) | ≤ C nm ω2 (f, n−1 ) , m ε Z+ (3.39)

where C is a constant independent of n and z.

Proof. Since, HLn (z) be the uniquely determined polynomial of degree ≤

2n+1 and the polynomial Fn (z) satisfying (3.38) and (3.39) can be expressed
as
2n
0
X X
Fn (z) = Fn (zk )Uk (z) + Fn (zk )Vk (z). (3.40)
k=0,2n+1 k=1

University of Lucknow, Lucknow–226007 58

 Interpolation on non-uniform nodes of the unit circle

Then, we can write

| HLn (z) − f (z) | ≤ | HLn (z) − Fn (z) | + | Fn (z) − f (z) | .

Using (3.35) and (3.40), we have

X
| HLn (z) − f (z) | ≤ | f (zk ) − Fn (zk ) || Uk (z) |
k=0,2n+1
2n
X
+ | βk − Fn0 (zk ) || Vk (z) | + | Fn (z) − f (z) |,
k=1
X 2n
X
| HLn (z) − f (z) | ≤ | f (zk ) − Fn (zk ) || Uk (z) | + | βk || Vk (z) |
k=0,2n+1 k=1
| {z } | {z }
M1 M2
2n
X
+ | Fn0 (zk ) || Vk (z) | + | Fn (z) − f (z) | . (3.41)
| {z }
k=1 M4
| {z }
M3

Using (3.38) and lemma 3.3.2 , we get

M1 = O(ω2 (f, n−1 )). (3.42)

Using (3.34) and lemma 3.3.1, we get

M2 = O(nω2 (f, n−1 ) log n). (3.43)

Using (3.39) and lemma 3.3.1., we get

M3 = O(nω2 (f, n−1 ) log n). (3.44)

Using (3.39), we have

M4 = O(ω2 (f, n−1 )). (3.45)

University of Lucknow, Lucknow–226007 59

 Interpolation on non-uniform nodes of the unit circle

Using (3.42), (3.43), (3.44), (3.45) in (3.41), we get

| HLn (z) − f (z) | = O(nω2 (f, n−1 ) log n). (3.46)

Hence, theorem 3.4.1 follows.

University of Lucknow, Lucknow–226007 60

Chapter 4

A note on Hermite-Fejér
Interpolation on the
non-uniform nodes of the unit
circle

This chapter deals with the Hermite-Fejér interpolation problem on the unit
circle with the nodal system containing the vertically projected zeros of Ja-
cobi’s polynomial with boundary points on the unit circle. Three cases with
different nodal structures is worked upon in this chapter and rate of conver-
gence is obtained for each case. Moreover, a comparison of all the three cases
and provided some important conclusions.

4.1 The Problem

The three cases to study Hermite-Fejér interpolation problem on the unit

circle are as under
Case I: Interpolatory polynomial Jn (z) of degree ≤ 4n − 1 satisfying follow-
ing conditions. 

Jn (f, zk ) = αk , k = 1(1)2n,

(4.1)
Jn0 (f, zk ) = 0,

 k = 1(1)2n,

61
 Interpolation on non-uniform nodes of the unit circle

where αk ’s are the arbitrary complex constants.

Case II: Interpolatory polynomial Gn (z) of degree ≤ 4n + 1 satisfying fol-
lowing conditions.


Gn (f, zk ) = βk , k = 0(1)2n + 1,

(4.2)
Gn0 (f, zk ) = 0,

 k = 1(1)2n,

where βk ’s are the arbitrary complex constants.

Case III: Interpolatory polynomial Hn (z) of degree ≤ 4n + 3 satisfying
following conditions.


Hn (f, zk ) = γk , k = 0(1)2n + 1,

(4.3)
Hn0 (f, zk ) = 0,

 k = 0(1)2n + 1,

where γk ’s are the arbitrary complex constants.

4.2 Explicit Representation of Interpolatory

Polynomial

Referring to the Case I, II and III mentioned in (4.1) - (4.3), we give following
theorems.

Theorem 4.2.1. Let Jn (z) be a polynomial satisfying (4.1), given by

2n
X
Jn (z) = αk Mk (z), (4.4)
k=1

University of Lucknow, Lucknow–226007 62

 Interpolation on non-uniform nodes of the unit circle

where Mk (z) is a unique polynomial of degree ≤ 4n − 1, such that

n o
Mk (z) = Lk2 (z) 1 − 2(z − zk )Lk0 (zk ) , (4.5)

satisfying the conditions



Mk (zj ) = δkj ,
 j, k = 1(1)2n,
(4.6)
M0k (zj ) = 0,

 j, k = 1(1)2n.

Proof. Let
Mk (z) = mk W (z)Lk (z) + Lk2 (z), (4.7)

where Mk (z) is atmost of degree 4n − 1 and mk are complex constants.

At z = zj
Mk (zj ) = mk W (zj )Lk (zj ) + Lk2 (zj ).

Since zj ’s [j = 1(1)2n] are the zeros of W (z), we get

Mk (zj ) = Lk2 (zj ).

Using (1.15), we get

Mk (zj ) = δkj .

Now, differentiating (4.7) with respect to z, we get

M0k (z) = mk W 0 (z)Lk (z) + mk W (z)Lk0 (z) + 2Lk (z)Lk0 (z).

At z = zk , we have

M0k (zk ) = mk W 0 (zk )Lk (zk ) + mk W (zk )Lk0 (zk ) + 2Lk (zk )Lk0 (zk ).

University of Lucknow, Lucknow–226007 63

 Interpolation on non-uniform nodes of the unit circle

On satisfying the second set of conditions given in (4.6), we get

0 = mk W 0 (zk ).1 + mk .0.Lk0 (zk ) + 2.1.Lk0 (zk ),

2Lk0 (zk )
mk = − . (4.8)
W 0 (zk )

Substituting (1.12), (1.15) and (4.8) in (4.7), the theorem follows.

Theorem 4.2.2. Let Gn (z) be a polynomial satisfying (4.2), given by

2n+1
X
Gn (z) = βk Nk (z), (4.9)
k=0

where Nk (z) is a unique polynomial of degree ≤ 4n + 1, such that

For k = 1(1)2n

!
Lk0 (zk ) + L0k (zk )
Nk (z) = − W (z)Lk (z) + Lk (z)Lk (z). (4.10)
W 0 (z )
k

For k = 0, 2n + 1
W (z)Lk (z)
Nk (z) = , (4.11)
W (zk )

satisfying the conditions



Nk (zj ) = δkj ,
 j, k = 0(1)2n + 1,
(4.12)
Nk0 (zj ) = 0,

 j = 1(1)2n, k = 0(1)2n + 1.

Proof. For k = 1(1)2n, Let

Nk (z) = nk W (z)Lk (z) + Lk (z)Lk (z), (4.13)

where nk are complex constants.

University of Lucknow, Lucknow–226007 64

 Interpolation on non-uniform nodes of the unit circle

Differentiating (4.13) with respect to z, we get

Nk0 (z) = nk W 0 (z)Lk (z) + nk W (z)L0k (z) + Lk0 (z)Lk (z) + Lk (z)L0k (z).

At z = zk , we have

Nk0 (zk ) = nk W 0 (zk )Lk (zk ) + nk W (zk )L0k (zk ) + Lk0 (zk )Lk (zk ) + Lk (zk )L0k (zk ).

Using second set of conditions given in (4.12), we get

0 = nk W 0 (zk ).1 + 0.L0k (zk ) + Lk0 (zk ).1 + 1.L0k (zk ),

 L 0 (z ) + L0 (z ) 
k k k k
nk = − . (4.14)
W (zk )
0

For k = 0, 2n + 1’
let
Nk (z) = n1k W (z)Lk (z), (4.15)

where n1k are complex constants. For z = zk , we have

Nk (zk ) = n1k W (zk )Lk (zk ).

Using the first set of condition of (4.12), we have

1
n1k = , as W (zk ) 6= 0. (4.16)
W (zk )

From (4.13), (4.14), (4.15) and (4.16), the theorem follows.

University of Lucknow, Lucknow–226007 65

 Interpolation on non-uniform nodes of the unit circle

Theorem 4.2.3. Let Hn (z) be a polynomial satisfying (4.3), given by

2n+1
X
Hn (z) = γk Pk (z), (4.17)
k=0

where Ck (z) is a unique polynomial of degree ≤ 4n + 3, such that

n o
Pk (z) = L2k (z) 1 − 2(z − zk )L0k (zk ) , (4.18)

satisfying the conditions



Pk (zj ) = δkj ,
 j, k = 0(1)2n + 1,
(4.19)
Pk0 (zj ) = 0,

 j, k = 0(1)2n + 1.

Proof. Proceed similar to the proof of the therorem 4.2.1

4.3 Estimates of Fundamental Polynomials

Lemma 4.3.1. Let Mk (z) be given by (4.5), then


 1

 O(log n), −1 < α ≤ − ,
2


2n 

1 1
X
1
| Mk (z) | = O(n 2 +α log n), − <α≤ , (4.20)
k=1


 2 2

O(n 12 +α ),
 1
 α> .
2

Proof. Taking modulus on both sides of (4.5), we get

2Lk0 (zk )Lk (z)W (z)

| Mk (z) |= Lk2 (z) − ,
W 0 (zk )

2Lk0 (zk )Lk (z)W (z)

| Mk (z) | ≤ |Lk2 (z)| + .
W 0 (zk )

University of Lucknow, Lucknow–226007 66

 Interpolation on non-uniform nodes of the unit circle

Taking summations on both the sides, we get

2n
X 2n
X
| Mk (z) | ≤ max | Lk (z) | | Lk (z) |
k=1 k=1
2n
X | L 0 (zk ) || Lk (z) || W (z) |
k
+ 2 . (4.21)
k=1
| W 0 (zk ) |

Using (1.12), (1.15), (1.20), (1.26) and (1.28), we have

2n 2n 2n
X X 1 X 1
| Mk (z) |≤ C1 −α+ 12
+ C2 , (4.22)
k=1 k=1 k k=1
k −2α+1

where C1 and C2 are constants independent of n and z. Further calculations

give us (4.20).

Lemma 4.3.2. Let Nk (z) be given by (4.10) and (4.11), then






 O(log n), −1 < α ≤ 0,
2n+1 

1
X
| Nk (z) | = 2α
O(n log n), 0<α≤ , (4.23)
k=0


 2

O(n2α ),
 1
 α> .
2

Proof. Since Nk (z) is defined in a different way for k = 1(1)2n and k =

0, 2n + 1. We can write

2n+1
X 2n
X X
Nk (z) = Nk (z) + Nk (z).
k=0 k=1 k=0,2n+1

University of Lucknow, Lucknow–226007 67

 Interpolation on non-uniform nodes of the unit circle

Taking modulus on both the sides,

2n+1
X 2n
X X
| Nk (z)| ≤ | Nk (z)| + | Nk (z) | . (4.24)
k=0 k=1 k=0,2n+1
| {z } | {z }
I1 I2

Now,

2n 2n
X | L 0 (zk ) || Lk (z) || W (z) |
k
X | L0 (zk ) || Lk (z) || W (z) |
k
I1 ≤ +
k=1
|W 0 (z )
k | k=1
| W 0 (zk ) |
X2n
+ | Lk (z) || Lk (z) | . (4.25)
k=1

Using (1.12), (1.15), (1.17) (1.20), (1.26) and (1.28), we have

I1 = O(log n), (−1 < α ≤ 0). (4.26)

Now, we have
X | Lk (z) || W (z) |
I2 ≤ . (4.27)
k=0,2n+1
| W (zk ) |

Using (1.12), (1.17) and (1.21) , we have

I2 = O(1), (−1 < α ≤ 0). (4.28)

Combining (4.26) and (4.28), after a little computations for different ranges
of α, the lemma follows.

Lemma 4.3.3. Let Pk (z) be given by (4.18), then


 1

O(log n), −1 < α ≤ ,
2


2n+1 

1
X
| Pk (z) | ≤ O(n2α−1 log n), < α ≤ 1, (4.29)
k=0


 2

O(n2α−1 ),

 α > 1.

University of Lucknow, Lucknow–226007 68

 Interpolation on non-uniform nodes of the unit circle

Proof. Proceed similar to the proof of Lemma 4.3.1

4.4 Convergence

Remark 4.1: Let f (z) be a function continuous on T ∪ D and analytic on

D and f (r)  Lip q, q > 0, then the sequence {Jn (z)}, {Gn (z)} and {Hn (z)}
converges uniformly to f (z) in T ∪ D, which follows from (4.34) as

3
ωr (f, n−1 ) = O(n−r+1−q ), {q > + α − r}, (4.30)
2

where ωr (f, n−1 ) be the rth modulus of continuity of f (z), such that

h 3 1i
r− ≤α<r− , (4.31)
2 2

where r = 2, 3, ... .
To prove the theorems , we shall need following.
Let f (z) be a function continuous on T ∪ D and analytic on D. Then, there
exists a polynomial Fn (z) of degree ≤ 4n − 1 satisfying Jackson’s inequality
[62]. 
1
O(ω(f, n−1 )),


 −1 < α ≤ − ,
2




1
| f (z) − Fn (z) |= O(ω2 (f, n−1 )), − < α ≤ ,1 (4.32)


 2 2
 1
O(ωr (f, n−1 )), α > ,


2
where C is a constant independent of n and z.

Theorem 4.4.1. Let f (z) be a function continuous on T ∪ D and analytic

on D. Then, Jn (z) defined by

2n
X
Jn (z) = f (zk )Mk (z), (4.33)
k=1

University of Lucknow, Lucknow–226007 69

 Interpolation on non-uniform nodes of the unit circle

satisfies the relation


1
O(ω(f, n−1 ) log n);


 −1 < α ≤ − ,
2




| Jn (z) − f (z) |= O(ω2 (f, n−1 ) log n); 1 1 (4.34)
− <α≤ ,


 2 2

O(ωr (f, n−1 )n 21 +α );
 1
 α> ,
2

where ωr (f, n−1 ) be the rth modulus of continuity of f (z).

Proof. Since Jn (z) be the uniquely determined polynomial of degree ≤ 4n−

1 and the polynomial Fn (z) satisfying (4.32) can be expressed as

2n
X
Fn (z) = Fn (zk )Mk (z), (4.35)
k=1

Then,

| Jn (z) − f (z) | ≤ | Jn (z) − Fn (z) | + | Fn (z) − f (z) |, (4.36)

Using (4.35), we get

2n
X
| Jn (z) − f (z) | ≤ | f (zk ) − Fn (zk ) || Mk (z) | + | Fn (z) − f (z) | . (4.37)
k=1

Using (4.32) and Lemma 4.3.1, we have Theorem 4.4.1

Theorem 4.4.2. Let f (z) be a function continuous on T ∪ D and analytic

on D. Then, Gn (z) defined by

2n+1
X
Gn (z) = f (zk )Nk (z), (4.38)
k=0

University of Lucknow, Lucknow–226007 70

 Interpolation on non-uniform nodes of the unit circle

satisfies the relation


O(ω(f, n−1 ) log n),




 −1 < α ≤ 0,


| Gn (z) − f (z) |= O(ω2 (f, n−1 ) log n), 1 (4.39)
0<α≤ ,


 2

O(ωr (f, n−1 )n2α ),
 1
 α> ,
2

where ωr (f, n−1 ) be the rth modulus of continuity of f (z).

Theorem 4.4.3. Let f (z) be a function continuous on T ∪ D and analytic

on D. Then, Hn (z) defined by

2n+1
X
Hn (z) = f (zk )Pk (z), (4.40)
k=0

satisfies the relation


1
O(ω(f, n−1 ) log n),


 −1 < α ≤ ,
2




| Hn (z) − f (z) |= −1 1 (4.41)
O(ω2 (f, n ) log n), < α ≤ 1,


 2

O(ωr (f, n−1 )n2α−1 ),

 α > 1,

where ωr (f, n−1 ) be the rth modulus of continuity of f (z).

Proof of Theorem 4.4.2 and Theorem 4.4.3 can be obtained proceeding

similarly as in Theorem 4.4.1

4.5 Conclusion

This chapter contains a problem involving three nodal structures different

from each other in a perspective of inclusion of the end points (±1) on the
real line. We can conclude from all the calculations done in this chapter that

University of Lucknow, Lucknow–226007 71

 Interpolation on non-uniform nodes of the unit circle

increase in nodal points gradually increase the degree of the interpolatory

polynomial formed, but the convergence came out to be more stronger in
Case I. Thus, for certain modulus of continuity, the interpolatory polynomial
in the Case I, where function values and the derivative values are prescribed
on Z2n excluding end points ±1, can survive through high range of α for
convergence purpose as compared to interpolatory polynomials obtained for
Case II and Case III.

University of Lucknow, Lucknow–226007 72

Chapter 5

Extension of Pál-type
Hermite-Fejér Interpolation
onto the unit circle

This chapter is devoted to the study of a Pál-type Hermite Fejér interpolation

problem on the unit circle considering two disjoint sets of nodes. The nodal
points are obtained by projecting vertically the zeros of the Jacobi polynomial
(α,β) (α,β)0
Pn (x) and its derivative Pn (x), together with ±1 onto the unit circle.
The Lagrange data are prescribed on Z2n , the Hermite data are prescribed
on T2n−2 and generalized Hermite-Fejér boundary conditions are prescribed
at ±1. An explicit representation of the interpolatory polynomial is given
and the convergence is studied for analytic functions on the unit disk. The
results are of interest to approximation theory.

5.1 The Problem & Explicit Representation

of Interpolatory Polynomial

Here, we are interested in determining the convergence of interpolatory poly-

nomial Rn (z) of degree ≤ 4n + 2r − 1 on the set of nodes Z2n and T2n−2
with Hermite-Fejér boundary conditions at ±1 satisfying the conditions.

73
 Interpolation on non-uniform nodes of the unit circle






Rn (zk ) = αk ; k = 1, 2, ..., 2n,


R0n (tk ) = βk ; k = 1, 2, ..., (2n − 2), (5.1)




R(m)

n (±1) = 0 ; m = 0, 1, ..., r ,


where αk and βk are complex constants and r < ∞.

Rn (z) can be written in the form given below

2n
X 2n−2
X
Rn (z) = αk Ak (z) + βk Bk (z). (5.2)
k=1 k=1

Here, Ak (z) and Bk (z) are the first and second fundamental polynomials each
of degree ≤ 4n + 2r − 1 satisfying (5.3) and (5.4) respectively.

For k = 1, 2, ..., 2n





 Ak (zj ) = δkj ; j = 1, 2, ..., 2n,


0
Ak (tj ) = 0 ; j = 1, 2, ..., 2n − 2, (5.3)




A(m)

k (±1) = 0 ; m = 0, 1, ..., r ,


and for k = 1, 2, ..., 2n − 2






Bk (zj ) = 0 ; j = 1, 2, ..., 2n,


0
Bk (tj ) = δkj ; j = 1, 2, ..., 2n − 2, (5.4)




B (m) (±1) = 0


k ; m = 0, 1, ..., r .

Explicit expressions of the polynomials Bk (z) and Ak (z) are given in

Theorems 5.1.1 and 5.1.2 respectively.
Remark: The equations (5.5) and (5.6) have been developed while deriving

University of Lucknow, Lucknow–226007 74

 Interpolation on non-uniform nodes of the unit circle

out explicit representation of the interpolatory polynomial. Readers can get

the motivation to form such expression from the idea to maintain the degree
of the polynomial as well as simultaneously satisfy the conditions required
to form the fundamental polynomial.

Z z
Jk (z) = z n+1 (z 2 − 1)r Lk∗ (z) dz (5.5)
0

and
Z z
J1j (z) = z n−j (z 2 − 1)r W ∗ (z) dz ; j = 0, 1 , (5.6)
0

where J1j (−1) = (−1)j+1 J1j (1).

Theorem 5.1.1. For k = 1, 2, ..., (2n − 2), second fundamental polynomial

is given by (5.7)

 
Bk (z) = z −n
W (z) bk Jk (z) + b0k J10 (z) + b1k J11 (z) , (5.7)

where
1
bk = , (5.8)
W (tk )tk (t2k − 1)r

−bk (Jk (1) + Jk (−1))

b1k = (5.9)
2J11 (1)

and
bk (Jk (−1) − Jk (1))
b0k = . (5.10)
2J10 (1)

Proof. Consider (5.7), where Bk (z) is atmost of the degree (4n + 2r − 1)

satisfying the conditions given in (5.4).

University of Lucknow, Lucknow–226007 75

 Interpolation on non-uniform nodes of the unit circle

At z = zj , j = 1, 2, ..., n

 
Bk (zj ) = zj−n W (zj ) bk Jk (zj ) + b0k J10 (zj ) + b1k J11 (zj ) .

Since zj ’s are the zeros of the polynomial W (z), so Bk (zj ) = 0.

Differentiating Bk (z) with respect to z gives us

Bk0 (z) = − nz −n−1 W (z) + z −n W 0 (z) bk Jk (z) + b0k J10 (z) + b1k J11 (z)
 

+z −n W (z) bk Jk0 (z) + b0k J100 0

(z) + b1k J11 (z)
h n  1 + z2  o n 0
 1 + z2 
= − nz −n−1 Kn Pn(α,β) z n + Kn z −n Pn(α,β) zn
2z 2z
2
(α,β) 1 + z
 oi  
n−1
+Pn nz × bk Jk (z) + b0k J10 (z) + b1k J11 (z)
2z
+z −n W (z) bk Jk0 (z) + b0k J100 0


(z) + b1k J11 (z) ,
h 0
 1 + z 2 i 
Bk0 (z) = Kn Pn(α,β) bk Jk (z) + b0k J10 (z) + b1k J11 (z)
2z
+z −n W (z) bk Jk0 (z) + b0k J100 0


(z) + b1k J11 (z) .

Since tj ’s are the zeros of the polynomial W ∗ (z), so Bk0 (tj ) = t−n
j W (tj )bk Jk (tj ).
0

Using (1.16) and (5.5), we have

j W (tj )bk tj
Bk0 (tj ) = t−n (tj − 1)r Lk∗ (tj )
n+1 2

= tj W (tj )bk (t2j − 1)r δkj .

0
Using condition Bk (tj ) = δkj given in (5.4), at j = k, we get (5.8).
One can verify the results for j 6= k.
(m)
Also, from Bk (±1) = 0 for m = 0, 1, ..., r, we get (5.9) and (5.10).
Hence, the theorem 5.1.1 follows.

Theorem 5.1.2. For k = 1, 2, ..., 2n, first fundamental polynomial is given

University of Lucknow, Lucknow–226007 76

 Interpolation on non-uniform nodes of the unit circle

by (5.11)

(z 2 − 1)r+1 Lk (z)W ∗ (z)

 
Ak (z) = + z W (z) Sk (z) + a0k J10 (z) + a1k J11 (z) ,
−n
(zk2 − 1)r+1 W ∗ (zk )
(5.11)
where

0
z
z n (z 2 − 1)r (z − 1)W ∗ (z) + ck W ∗ (z)
Z  2 
Sk (z) = − dz,
0 W 0 (zk )(zk2 − 1)r+1 W ∗ (zk ) (z − zk )
(5.12)

Sk (1) − Sk (−1)
a0k = , (5.13)
2J10 (1)

−(Sk (1) + Sk (−1))

a1k = (5.14)
2J11 (1)

and
0
(1 − zk 2 )W ∗ (zk )
ck = . (5.15)
W ∗ (zk )

Proof. Consider (5.11), where Ak (z) is atmost of the degree (4n + 2r − 1)

satisfying the conditions given in (5.3).
At z = zj , j = 1, 2, ..., 2n, we have

(zj2 − 1)r+1 lk (zj )W ∗ (zj ) −n

 
Ak (zj ) = +zj W (zj ) Sk (zj )+a0k J10 (zj )+a1k J11 (zj ) .
(zk2 − 1)r+1 W ∗ (zk )

Since zj ’s are the zeros of the polynomial W (z)

(zj2 − 1)r+1 Lk (zj )W ∗ (zj )

Ak (zj ) =
(zk2 − 1)r+1 W ∗ (zk )
(zj2 − 1)r+1 δkj W ∗ (zj )
=
(zk2 − 1)r+1 W ∗ (zk )
= δkj .

University of Lucknow, Lucknow–226007 77

 Interpolation on non-uniform nodes of the unit circle

Differentiating Ak (z) with respect to z

1 h
A0k (z) = W ∗
2z(r + 1)(z 2 − 1)r Lk (z) +

(z)
(zk2 − 1)r+1 W ∗ (zk )
0
i
(z 2 − 1)r+1 Lk0 (z) + (z 2 − 1)r+1 Lk (z)W ∗ (z)
n o 
+ − nz −n−1 W (z) + z −n W 0 (z) Sk (z) + a0k J10 (z) + a1k J11 (z)
 
+z W (z) Sk (z) + a0k J10 (z) + a1k J11 (z) .
−n 0 0 0

Since tj ’s are the zeros of W ∗ (z), we have

0
(t2j − 1)r+1 Lk (tj )W ∗ (tj ) n o
A0k (tj ) = + − nt−n−1
W (t j ) + t−n
W 0
(t j )
(zk2 − 1)r+1 W ∗ (zk ) j j
 
j W (tj )Sk (tj ).
Sk (tj ) + a0k J10 (tj ) + a1k J11 (tj ) + t−n 0

From the second condition given in (5.3), we have

0
(t2j − 1)r+1 Lk (tj )W ∗ (tj ) n o
0 = + − nt−n−1
W (t j ) + t−n
W 0
(t j )
(zk2 − 1)r+1 W ∗ (zk ) j j
 
j W (tj )Sk (tj ),
Sk (tj ) + a0k J10 (tj ) + a1k J11 (tj ) + t−n 0

0
(t2j − 1)r+1 Lk (tj )W1∗ (tj ) n −n−1 o
j W
t−n W W
−n
(tj )Sk0 (tj ) = − + nt (tj ) − t 0
(t j )
(zk2 − 1)r+1 W ∗ (zk ) j j
 
× Sk (tj ) + a0k J10 (tj ) + a1k J11 (tj )
0
( !
(t2j − 1)r+1 Lk (tj )W ∗ (tj ) 1 + t2
j
= − + nt−n−1 Kn Pn(α,β) tnj
(zk2 − 1)r+1 W ∗ (zk ) j
2tj
( ! ! ))
2 2
0 1 + tj 1 + tj
− t−n
j Kn Pn(α,β) tnj + nKn Pn(α,β) tn−1
j
2tj 2tj
!
× Sk (tj ) + a0k J10 (tj ) + a1k J11 (tj ) .

University of Lucknow, Lucknow–226007 78

 Interpolation on non-uniform nodes of the unit circle

After a little computation, we get

0
(t2j − 1)r+1 Lk (tj )W ∗ (tj )
Sk0 (tj ) = − −n .
tj W (tj )(zk2 − 1)r+1 W ∗ (zk )

We can write above equation as

0
z n (z 2 − 1)r (z − 1)W ∗ (z) + ck W ∗ (z)
 2 
Sk0 (z) =− 0 . (5.16)
W (zk )(zk2 − 1)r+1 W ∗ (zk ) (z − zk )

Integrating (5.16) provides us with a polynomial Sk (z) of degree (3n+2r −1)

given by (5.12).
To establish the validity of Sk (z), we must have

0
[(z 2 − 1)W ∗ (z) + ck W ∗ (z)]|z=zk = 0,

which in turn gives (5.15). Similarly, the constants a0k and a1k can be found
(m)
out by satisfying the condition Ak (±1) = 0 for m = 0, 1, ..., r.
Hence, the theorem 5.1.2 follows.

5.2 Estimates of Fundamental Polynomials

We need to calculate estimates in order to obtain the rate of convergence of

interpolatory polynomials.

Lemma 5.2.1. Let Ak (z) be given by (5.11), then for z ∈ T ∪ D.

2n
X  
2 r/2 r+1
| Ak (z) | = O (1 − x ) n log n , (5.17)
k=1

r
where −1 < α ≤ .
2

University of Lucknow, Lucknow–226007 79

 Interpolation on non-uniform nodes of the unit circle

Proof. From (5.11), we have

2n 2n 2n
X X (z 2 − 1)r+1 Lk (z)W ∗ (z) X
| Ak (z) | ≤ 2 r+1 W ∗ (z )
+ z −n W (z)Sk (z)
k=1 k=1
(zk − 1) k
k=1
| {z } | {z }
I1 I2
2n
X
+ z −n W (z)(a0k J10 (z) + a1k J11 (z)) .
k=1
| {z }
I3

We can write as
2n
X
| Ak (z) | ≤ I1 + I2 + I3 (5.18)
k=1

Using (1.15) we have

2n
X (z 2 − 1)r+1 W (z)W ∗ (z)
I1 =
k=1
(zk2 − 1)r+1 (z − zk )W 0 (zk )W ∗ (zk )
n   o
2 r+1 (α,β) 1+z 2
X2n (z − 1) K n Pn 2z
zn
= n   o0 ×
2 (α,β) 1+z 2
k=1 (zk − 1)r+1 (z − zk ) Kn Pn 2z
z n
|z=zk
n 0
 2
 o
(α,β) 1+z
Kn∗ Pn 2z
z n−1
n   o
(α,β)0 1+zk2 n−1
Kn∗ Pn 2zk
z k

(α,β)
Since Pn (xk ) = 0 and | zk |= 1, we get

2n (α,β) (α,β)0
X | (z 2 − 1)r+1 || Pn (x) || Pn (x) |
I1 = 2 (α,β)0
.
k=1 | (zk2 − 1)r+2 || z − zk || Pn (xk ) |2

Using (1.18) and (1.19), we get

2n r+1 (α,β) (α,β)0

X 2r+1 (1 − x2 ) 2 | Pn (x) || Pn (x) |
I1 = 2
r+2 √ √
q
(α,β)0
p
k=1 2r+2 (1 − x2k ) 2 2 1 − xxk − 1 − x2 1 − x2k | Pn (xk ) |2

University of Lucknow, Lucknow–226007 80

 Interpolation on non-uniform nodes of the unit circle

2n r+1 (α,β) (α,β)0 √

X (1 − x2 ) 2 | Pn (x) || Pn (x) | 1 − xxk
≤ r+2 (α,β)0
.
k=1 (1 − x2k ) 2 | x − xk || Pn (xk ) |2

1
For | x − xk | ≥ 2
| 1 − x2k |, we have

2n r+1 (α,β) (α,β)0 √

X 2(1 − x2 ) 2 | Pn (x) || Pn (x) | 1 − xxk
I1 ≤ r+2 (α,β)0
k=1 (1 − x2k )
(1 − x2k ) | Pn
2 (xk ) |2
2n √ r p (α,β) (α,β)0
X 2 2(1 − x2 ) 2 {( (1 − x2 ) | Pn (x) |)} | Pn (x) |
≤ r+4 (α,β)0
.
k=1 (1 − x2k ) 2 | Pn (xk ) |2

Using (1.20), (1.24), (1.26) and (1.28), we get

2n
!
X 1
I1 = O (1 − x2 )r/2 nr+1 .
k=1
k r−2α+1

From r − 2α + 1 ≥ 1, we get

  n ro
2 r/2 r+1
I1 = O (1 − x ) n log n −1<α≤ . (5.19)
2

The reader can verify that estimate remains the same in the case, where
1
| x − xk | < 2
| 1 − x2k |.
Following similar scheme as above gives

  n ro
2 r/2 r
I2 = O (1 − x ) n log n −1<α≤ (5.20)
2

and

  n ro
2 r/2 r
I3 = O (1 − x ) n log n −1<α≤ . (5.21)
2

University of Lucknow, Lucknow–226007 81

 Interpolation on non-uniform nodes of the unit circle

Combining (5.19), (5.20) and (5.21) gives lemma 5.2.1.

Lemma 5.2.2. Let Bk (z) be given by (5.7), then for z ∈ T ∪ D

2n−2
X  
2 r/2 r
| Bk (z) | = O (1 − x ) n log n , (5.22)
k=1

r−1
where −1 < α ≤ .
2

Proof. Consider (5.7), we have

| Bk (z) |= z −n W (z) bk Jk (z) + b0k J10 (z) + b1k J11 (z) .

 
(5.23)

2n−2
X 2n−2
X 2n−2
X
| Bk (z) |≤ z −n
W (z)bk Jk (z) + z −n W (z)((b0k J10 (z) + b1k J11 (z)) .
k=1 k=1 k=1
| {z } | {z }
M1 M2
(5.24)
2n−2
X
| Bk (z) | ≤ M1 + M2 . (5.25)
k=1

Using (5.8) and (5.5), we have

2n−2
( )Z
z
X 1
M1 = z −n W (z) z n+1 (z 2 − 1)r Lk∗ (z)dz .
k=1
W (tk )tk (t2k − 1)r 0

Using (1.16), we get

2n−2
( )
X 1
M1 = z −n W (z) ×
k=1
W (tk )tk (t2k − 1)r
( )
W
Z z ∗
(z)
z n+1 (z 2 − 1)r dz
0 (z − tk )W ∗0 (tk )

University of Lucknow, Lucknow–226007 82

 Interpolation on non-uniform nodes of the unit circle

2n−2
( )
z −n W (z) z
W ∗ (z)
X Z
≤ max z n+1 (z 2 − 1)r dz
k=1
W (tk )tk (t2k − 1)r |z|=1 0 (z − tk )W ∗0 (tk )
2n−2
z −n W (z)(z 2 − 1)r W ∗ (z) z
X Z
≤ z n+1 dz .
k=1
W (tk )tk (t2k − 1)r (z − tk )W ∗0 (tk ) 0

(α,β)0
Since Pn (x∗k ) = 0 and | tk |= 1, using (1.12) and (1.13) we get

2n−2 (α,β) (α,β)0

2 X | Pn (x) || (z 2 − 1)r || Pn (x) |
M1 ≤ .
n + 2 k=1 | Pn(α,β) (x∗k ) || (t2k − 1)r+1 || z − tk || Pn(α,β)00 (x∗k ) |

Owing to (1.18) and (1.19), we have

0
X (1 − x2 )r/2 | Pn(α,β) (x) || Pn(α,β) (x) | 1 − xx∗k
2n−2 p
1
M1 ≤ r+1 (α,β) ∗ (α,β)00
.
(n + 2) k=1 (1 − x∗2
k ) 2 | P n (x k ) | | x − x ∗
k || P n (x∗
k ) |

For | x − x∗k | ≥ 1
2
| 1 − x∗2
k |, we have

√ 2n−2 (α,β) (α,β)0

2 2 X (1 − x2 )r/2 | Pn (x) || Pn (x) |
M1 ≤ r+3 (α,β) ∗ (α,β)00
. (5.26)
(n + 2) k=1 (1 − x∗2
k ) 2 | P n (xk ) | | P n (x∗
k ) |

Using (1.21),(1.24),(1.26),(1.27) and (1.29), we get

2n
!
X 1
M1 = O (1 − x2 )r/2 nr .
k=1
k r−2α

From r − 2α ≥ 1, we get

  n r − 1o
2 r/2 r
M1 = O (1 − x ) n log n −1<α≤ . (5.27)
2

University of Lucknow, Lucknow–226007 83

 Interpolation on non-uniform nodes of the unit circle

The estimate remains the same in the case, where | x − x∗k | < 1
2
| 1 − x∗2
k |.

Similarly, we have

  n r − 1o
2 r/2 r
M2 = O (1 − x ) n log n −1<α≤ . (5.28)
2

Combining (5.27) and (5.28) give us desired lemma 5.2.2 .

5.3 Convergence

Remark 5.1 :To prove the Theorem 5.3.1, we shall need following.
Let f (z) be a function continuous on T ∪ D and analytic on D. Then, there
exists a polynomial Fn (z) of degree ≤ 2n+1 satisfying Jackson’s inequality
[62].
| f (z) − Fn (z) | ≤ C ωr+1 (f, n−1 ), r≥0 (5.29)

and also an inequality by O.Kiŝ [70]

| Fn(m) (z) | ≤ C nm ωr+1 (f, n−1 ) , m ε Z+ , (5.30)

where ωr+1 (f, n−1 ) denotes the (r + 1)th modulus of continuity of f (z) and
C is a constant independent of n and z.

Theorem 5.3.1. Let f (z) be a function continuous on T ∪ D and analytic

on D and f (r)  Lip ν, ν ≥ 1 + δ, δ > 0. Let the arbitrary numbers βk ’s is such
that
| βk |= O n ωr+1 (f, n−1 ) ,


k = 1, 2, ..., (2n − 2). (5.31)

University of Lucknow, Lucknow–226007 84

 Interpolation on non-uniform nodes of the unit circle

Then sequence {Rn (z)} is defined by

2n
X 2n−2
X
Rn (z) = f (zk )Ak (z) + βk Bk (z), (5.32)
k=1 k=1

satisfies the following relation for | z |≤ 1

 
2 r/2 r+1 −1
| Rn (z) − f (z) | = O (1 − x ) n ωr+1 (f, n ) log n , (5.33)

where ωr+1 (f, n−1 ) be the (r + 1)th modulus of continuity of f (z).

Proof. Since Rn (z) be the uniquely determined polynomial of degree ≤

4n + 2r − 1 and the polynomial Fn (z) satisfying equation (5.29) and (5.30)
can be expressed as

2n 2n−2
0
X X
Fn (z) = Fn (zk )Ak (z) + Fn (zk )Bk (z). (5.34)
k=1 k=1

Then, we can write

| Rn (z) − f (z) | ≤ | Rn (z) − Fn (z) | + | Fn (z) − f (z) | . (5.35)

Using (5.32) and (5.34), we have

2n
X
| Rn (z) − f (z) | ≤ | f (zk ) − Fn (zk ) || Ak (z) |
k=1
2n−2
X
+ | βk − Fn0 (zk ) || Bk (z) | + | Fn (z) − f (z) |
k=1
X2n 2n−2
X
≤ | f (zk ) − Fn (zk ) || Ak (z) | + | βk || Bk (z) |
k=1 k=1
| {z } | {z }
N1 N2

University of Lucknow, Lucknow–226007 85

 Interpolation on non-uniform nodes of the unit circle

2n−2
X
+ | Fn0 (zk ) || Bk (z) | + | Fn (z) − f (z) | .
| {z }
k=1 N4
| {z }
N3

| Rn (z) − f (z) | ≤ N1 + N2 + N3 + N4 , (5.36)

where

2n
X
N1 = | f (zk ) − Fn (zk ) || Ak (z) | .
k=1

From (5.29) and (5.17), we have

 
2 r/2 −1 r+1
N1 = O (1 − x ) ωr+1 (f, n )n log n . (5.37)

2n−2
X
N2 = | βk || Bk (z) | .
k=1

From (5.31) and (5.22), we have

 
2 r/2 −1 r
N2 = O (1 − x ) n ωr+1 (f, n ))n log n . (5.38)

2n−2
X
N3 = | Fn0 (zk ) || Bk (z) | .
k=1

From (5.30) and (5.22), we have

 
2 r/2 −1 r
N3 = O (1 − x ) n ωr+1 (f, n )n log n . (5.39)

N4 = | Fn (z) − f (z) | .

University of Lucknow, Lucknow–226007 86

 Interpolation on non-uniform nodes of the unit circle

From (5.29), we have

 
−1
N4 = O ωr+1 (f, n ) . (5.40)

Using (5.37),(5.38),(5.39) and (5.40) in (5.36), we get

 
2 r/2 r+1 −1
| Rn (z) − f (z) | = O (1 − x ) n ωr+1 (f, n ) log n .

Hence, Theorem 5.3.1 follows.

5.4 Conclusion

This chapter poses a completely new problem by introducing the generalized

Hermite-Fejér boundary conditions at the points ±1. Since these additional
nodes gradually increase the degree of the interpolatory polynomial. So,
the order of convergence must also depend on that increment, which can
be seen in (5.33) as we require the (r + 1)th modulus of continuity for the
convergence purpose. Since this problem is posed considering generalized
Hermite-Fejér boundary conditions only at ±1, a subtle open problem is to
consider the generalized Hermite-Fejér boundary conditions at ±1 as well as
on all the nodal points, where Lagrange and Hermite data are prescribed (i.e
±1 ∪ Z2n ∪ T2n−2 ). This will provide a much broader aspect of convergence
and comparisons to the problem in this chapter.

University of Lucknow, Lucknow–226007 87

Chapter 6

Revisiting Pál-type Hermite-

Fejér Interpolation onto the
unit circle

In this chapter, authors brought into consideration the set of non-uniformly

distributed nodes on the unit circle to investigate a Pál-type (1;0) interpo-
lation problem in account with Hermite-Fejér boundary condition. These
nodes are obtained by projecting vertically the zeros of Jacobi polynomial
as well as zeros of its derivative onto the unit circle. Explicitly representing
the interpolatory polynomial as well as establishment of convergence theo-
rem are the key highlights of this chapter. The field of approximation theory
entertains the results proved.

6.1 The Problem and the Explicit Represen-

tation of Interpolatory Polynomial

Here, we are interested in determining the convergence of interpolatory poly-

nomial Fn (z) on the two distinct set of nodes Z2n and T2n−2 , with Hermite
boundary conditions at ±1 satisfying the conditions.

88
 Interpolation on non-uniform nodes of the unit circle


Fn (tk ) = ηk




 ; k = 1(1)2n − 2,


(1)
 Fn (zk ) = ϕk ; k = 1(1)2n, (6.1)



Fn(m) (±1) = 0

 ; m = 0(1)r,

(m)
where, ηk and ϕk are complex constants. Also, Fn (z) denotes mth
derivative of Fn (z). We shall write Fn (z) satisfying (6.1)

2n−2
X 2n
X
Fn (z) = ηk Ck (z) + ϕk Dk (z), (6.2)
k=1 k=1

where Ck (z) and Dk (z) are unique polynomials each of degree ≤ 4n + 2r − 1.

For k = 1(1)2n − 2

Ck (tj ) = δkj




 ; j = 1(1)2n − 2,


0

 Ck (zj ) = 0 ; j = 1(1)2n, (6.3)




 (m)
Ck (±1) = 0
 ; m = 0(1)r.

For k = 1(1)2n

Dk (tj ) = 0




 ; j = 1(1)2n − 2,


0

 Dk (zj ) = δkj ; j = 1(1)2n, (6.4)




Dk(m) (±1) = 0

 ; m = 0(1)r.

Remark: The equations (6.5) and (6.6) have been developed while de-
riving out explicit representation of the interpolatory polynomial. Readers
can get the motivation to form such expression from the idea to maintain
the degree of the polynomial as well as simultaneously satisfy the conditions
required to form the fundamental polynomial.

University of Lucknow, Lucknow–226007 89

 Interpolation on non-uniform nodes of the unit circle

Z z
Mk (z) = z n+2α (z 2 − 1)r−2α Lk∗ (z)dz. (6.5)
0

Z z
M1j (z) = z n+2α+j (z 2 − 1)r−2α−2 W (z)dz ; j = 0, 1 , (6.6)
0

where M1j (−1) = (−1)j+1 M1j (1).

Theorem 6.1.1. For k = 1(1)2n,

 
Dk (z) = (z − 1)
2 2+2α −n−2α−1
z W (z) gk Mk (z) + g0k M10 (z) + g1k M11 (z)
∗

(6.7)

zk
gk = , (6.8)
(zk2 − 1)2+r W ∗ (zk )

−gk [Mk (1) + Mk (−1)]

g1k = (6.9)
2M11 (1)

and
gk [Mk (−1) − Mk (1)]
g0k = . (6.10)
2M10 (1)

Proof. Consider (6.7), where Dk (z) is atmost of degree (4n+2r−1) satisfying

the conditions given in (6.4).
0
Clearly, Dk (tj ) = 0 for j = 1(1)2n − 2 and from Dk (zj ) = δkj for j = 1(1)2n.
Using (6.5), at j = k, we get (6.8).
(m)
One can verify the results for j 6= k. Also, from Dk (±1) = 0 for m = 0(1)r,
we get (6.9) and (6.10) using (6.6).
Hence, the theorem follows.

University of Lucknow, Lucknow–226007 90

 Interpolation on non-uniform nodes of the unit circle

Theorem 6.1.2. For k = 1(1)2n − 2,

(z 2 − 1)r+1 Lk∗ (z)W (z)

Ck (z) = (6.11)
(t2k − 1)r+1 W (tk )
 
2
+ (z − 1)2+2α −n−2α−1
z W (z) Nk (z) + h0k M10 (z) + h1k M11 (z) ,
∗

where

0
z
z n+2α+1 (z 2 − 1)r−2−2α (z 2 − 1)W (z) + dk W (z)
Z  
Nk (z) = − dz,
0 W ∗0 (tk )(t2k − 1)r+1 W (tk ) (z − tk )
(6.12)
Nk (1) − Nk (−1)
h0k = , (6.13)
2M10 (1)

−[Nk (1) + Nk (−1)]

h1k = , (6.14)
2M11 (1)

(1 − tk 2 )W 0 (tk )
dk = . (6.15)
W (tk )

Proof. Consider (6.11), where Ck (z) is atmost of degree (4n+2r−1) satisfying

the conditions given in (6.3).
0
Clearly, Ck (tj ) = δkj for j = 1(1)2n−2 and Ck (zj ) = 0 for j = 1(1)2n (j 6= k),
provides us with a polynomial Nk (z) of degree (3n + 2r − 1) given by (6.12).
0
One can verify [Ck (z)]z=zk = 0.
0
[(z 2 − 1)W (z) + dk W (z)]z=zk = 0 provides us (6.15).
(m)
Also, from Ck (±1) = 0 for m = 0(1)r, we get (6.13) and (6.14).
Hence, the theorem follows.

University of Lucknow, Lucknow–226007 91

 Interpolation on non-uniform nodes of the unit circle

6.2 Estimates of Fundamental Polynomials

We need to calculate estimates in order to obtain the rate of convergence of

interpolatory polynomials.

Lemma 6.2.1. Let Ck (z) be given by (6.11), then

2n−2  
2 r−2
X
| Ck (z) | = O (1 − x ) n log n ,
2
r
(6.16)
k=1

r−1
where −1 < α ≤ .
2

Proof. Consider (6.11)

2n−2 2n−2
X X (z 2 − 1)r+1 Lk∗ (z)W (z)
| Ck (z) | ≤ (6.17)
k=1 k=1
(t2k − 1)r+1 W (tk )
| {z }
I1
2n−2
X
+ (z 2 − 1)2+2α z −n−2α−1 W ∗ (z)Nk (z)
k=1
| {z }
I2
2n−2
X
+ (z 2 − 1)2+2α z −n−2α−1 W ∗ (z)(h0k M10 (z) + h1k M11 (z)) ,
k=1
| {z }
I3

Now, we can write above as

2n−2
X
| Ck (z) | ≤ I1 + I2 + I3 . (6.18)
k=1

We have

University of Lucknow, Lucknow–226007 92

 Interpolation on non-uniform nodes of the unit circle

2n−2
X (z 2 − 1)r+1 Lk∗ (z)W (z)
I1 = .
k=1
(t2k − 1)r+1 W (tk )

Using (1.16), we get

2n−2
X (z 2 − 1)r+1 W ∗ (z)W (z)
I1 = .
k=1
(t2k − 1)r+1 (z − tk )W ∗0 (tk )W (tk )

Using (1.12), (1.13), (1.18) and (1.19),we get

2n−2 r+1 (α,β) (α,β)0

1 − xx∗k
p
X (1 − x2 ) 2 | Pn (x) || Pn (x) |
I1 ≤ c r+2 (α,β)00 (α,β)
,
k=1 (1 − x∗2
k )
2 | x − x∗k || Pn (x∗k ) || Pn (x∗k ) |

where c is a constant independent of n and z.

For | x − x∗k | ≥ 1
2
| 1 − x∗2
k |, we have

r+1 (α,β) (α,β) 0

2n
1 − xx∗k
p
X 2(1 − x2 ) 2 | Pn (x) || Pn (x) |
I1 ≤ c r+2 (α,β)00 (α,β)
.
k=1 (1 − x∗2
k )
2 (1 − x∗2
k ) | Pn (x∗k ) || Pn (x∗k ) |

Using (1.20), (1.22), (1.26), (1.27) and (1.29), we get

2n
!
X 1
I1 = O (1 − x2 )r−2/2 nr−1 .
k=1
k r−2α

From r − 2α ≥ 1, we get

  n r − 1o
2 r−2 r−1
I1 = O (1 − x ) 2 n log n −1<α≤ . (6.19)
2

The reader can verify that estimate remains the same in the case, where
| x − x∗k | < 1
2
| 1 − x∗2
k |.

University of Lucknow, Lucknow–226007 93

 Interpolation on non-uniform nodes of the unit circle

Following similar scheme as above gives

 
2 r−2 r
I2 = O (1 − x ) n log n .
2 (6.20)

 
2 r−2 r
I3 = O (1 − x ) 2 n log n . (6.21)

Combining (6.19), (6.20) and (6.21) give our desired lemma.

Lemma 6.2.2. Let Dk (z) be given by (6.7), then

2n  
2 r−2
X
| Dk (z) | = O (1 − x ) 2 n log n ,
r−1
(6.22)
k=1

r−1
where, −1 < α ≤ .
2

Proof. Consider (6.7)

| Dk (z) |= (z 2 − 1)2+2α z −n−2α−1 W ∗ (z) [gk Mk (z) + g0k M10 (z) + g1k M11 (z)] .
(6.23)

2n
X 2n
X
| Dk (z) | ≤ (z 2 − 1)2+2α z −n−2α−1 W ∗ (z)gk Mk (z) (6.24)
k=1 k=1
X2n
+ (z 2 − 1)2+2α z −n−2α−1 W ∗ (z)[g0k M10 (z) + g1k M11 (z)] ,
k=1

2n
X
| Dk (z) | ≤ I1 + I2 , (6.25)
k=1

where
 
2 r−2 r−1
I1 = O (1 − x ) n log n .
2 (6.26)

University of Lucknow, Lucknow–226007 94

 Interpolation on non-uniform nodes of the unit circle

 
r−2
2 r−1
I2 = O (1 − x ) 2 n log n . (6.27)

Combining (6.26) and (6.27) gives (6.22).

6.3 Convergence

In this section, we shall need following.

Remark 6.3.1. Let f (z) be a function continuous on T ∪ D and analytic on

D and f (r)  Lip ν, ν > 0, then the sequence {Fn (z)} converges uniformly to
f (z) in z ∈ T ∪ D, which follows as

ωr (f, n−1 ) = O(n−r+1−ν ) , 0 < ν < 1, (6.28)

where ωr (f, n−1 ) be the rth modulus of continuity of f (z).

Remark 6.3.2. Let f (z) be a function continuous on T ∪ D and analytic on

D. Then, there exists a polynomial Fn (z) of degree ≤ 4n + 2r − 1 satisfying
Jackson’s inequality [62].

| f (z) − Fn (z) | ≤ C ωr+1 (f, n−1 ) (6.29)

and also an inequality by O.Kiŝ. [70]

| Fn(m) (z) | ≤ C nm ωr+1 (f, n−1 ), m ε Z+ , (6.30)

where C is a constant independent of n and z.

Theorem 6.3.1. Let f (z) be a function continuous on T ∪ D and analytic

University of Lucknow, Lucknow–226007 95

 Interpolation on non-uniform nodes of the unit circle

on D. Let the arbitary numbers ϕk ’s be such that

| ϕk |= O(n ωr+1 (f, n−1 )) , k = 1(1)2n. (6.31)

Then, sequence {Fn (z)} defined by

2n−2
X 2n
X
Fn (z) = f (zk )Ck (z) + ϕk Dk (z), (6.32)
k=1 k=1

satisfies the relation,

r−2
| Fn (z) − f (z) | = O((1 − x2 ) 2 nr ωr+1 (f, n−1 ) log n), (6.33)

where ωr+1 (f, n−1 ) be the (r + 1)th modulus of continuity of f (z).

Proof. Since Fn (z) be the uniquely determined polynomial of degree ≤

4n + 2r − 1 and the polynomial Fn (z) satisfying equation (6.29) can be
expressed as

2n−2 2n
0
X X
Fn (z) = Fn (zk )Ck (z) + Fn (zk )Dk (z) (6.34)
k=1 k=1

Then

| Fn (z) − f (z) | ≤ | Fn (z) − Fn (z) | + | Fn (z) − f (z) |, (6.35)

2n−2
X
| Fn (z) − f (z) | ≤ | f (zk ) − Fn (zk ) || Ck (z) |
k=1
2n
X
+ {| ϕk | + | Fn0 (zk ) |} | Dk (z) | + | Fn (z) − f (z) |
k=1

University of Lucknow, Lucknow–226007 96

 Interpolation on non-uniform nodes of the unit circle

Using equations (6.29), (6.30), (6.31), Lemma 6.2.1 and Lemma 6.2.2, we get
(6.33).

University of Lucknow, Lucknow–226007 97

Chapter 7

Higher order Hermite-Fejér

Interpolation on the unit circle.

In this chapter, we study the approximation of functions using a higher-

order Hermite-Fejér interpolation process on the unit circle. The system of
nodes is composed of vertically projected zeros of Jacobi polynomials onto
the unit circle with boundary points at ±1. Values of the polynomial and its
first four derivatives are fixed by the interpolation conditions at the nodes.
Convergence is obtained and the rate of convergence is estimated for analytic
functions on a suitable domain,

7.1 The Problem

Here, we are interested in determining the convergence of interpolatory poly-

nomial Qn (z) of degree less than (2n + 2)(r + 1) on the distinct set of nodes
{±1} ∪ Z2n with Hermite conditions at all points satisfying


Qn (zk ) = αk ,
 k = 0, 1, ..., 2n + 1
(7.1)
Q(r)

n (zk ) = 0, k = 0, 1, ..., 2n + 1, r = 1, 2, 3, 4 ,


98
 Interpolation on non-uniform nodes of the unit circle

(r)
where αk0 s are complex constants and Qn (zk ) denotes rth derivative of Qn (z)
at z = zk .

7.2 Explicit Representation of Interpolatory

Polynomial

Theorem 7.2.1. We shall write Qn (z) satisfying (7.1)

2n+1
X
Qn (z) = f (zk )A0k (z), (7.2)
k=0

where A0k (z) is a polynomial of degree less than (2n + 2)(r + 1) satisfying
the conditions given in (7.3).
For j, k = 0, 1, ..., 2n + 1 ,


A0k (zj ) = δkj ,

(7.3)
A(r)

0k (zj ) = 0 ; r = 1, 2, 3, 4 ,


such that
4
X
5
A0k (z) = [Lk (z)] + cpk Apk (z), (7.4)
p=1

Apk (z) = [R(z)]p (Lk (z))5−p , (7.5)

5 L0k (zk )
c1k = − , (7.6)
R 0 (zk )
5  00 0 2

c2k = − Lk (zk ) + 10 (Lk (zk )) , (7.7)
2! [R 0 (zk )]2
5  00 0 000 0 3

c3k = − − 18 Lk (zk ) Lk (z k ) + Lk (zk ) − 198(Lk (zk )) , (7.8)
3! [R 0 (zk )]3

University of Lucknow, Lucknow–226007 99

 Interpolation on non-uniform nodes of the unit circle

5
c4k =− 0 4
L0000 000 0
k (zk ) − 24Lk (zk ) Lk (zk )
4! [R (zk )]
 (7.9)
+ (L00k (zk ))2 − 156(L0k (zk ))2 L00k (zk ) + 2544(L0k (zk ))4 .

Proof. Let A0k (z) be written as

4
X
5
A0k (z) = [Lk (z)] + cpk [R(z)]p (Lk (z))5−p . (7.10)
p=1

At z = zj , where j = 0, 1, ..., 2n + 1 ,

4
X
5
A0k (zj ) = [Lk (zj )] + cpk [R(zj )]p (Lk (zj ))5−p .
p=1

Using (1.14), we have R(zj ) = 0 and from (1.17), we have

A0k (zj ) = δkj . (7.11)

Clearly, the first set of condition in (7.3) is satisfied.

In order to determine the cpk ’s, we use the second set of conditions of (7.3).
On differentiating A0k (z) in (7.10) one time with respect to z, we get

4
hX i0
A00k (z) = 5L0k (z)[Lk (z)]4 + c1k [R(z)(Lk (z)) ] + 4 0
cpk [R(z)]p (Lk (z))5−p .
p=2
(7.12)
Clearly, at z = zj (j 6= k), we have A00k (z) = 0.
At z = zk , A00k (z) must be equal to zero. We have

5 L0k (zk ) + c1k R 0 (zk ) = 0,

which provides (7.6). In a similar manner, differentiating (7.10) two, three

and four times with respect to z gives (7.7), (7.8) and (7.9) respectively, by
using conditions given in (7.3).

University of Lucknow, Lucknow–226007 100

 Interpolation on non-uniform nodes of the unit circle

7.3 Estimation of the fundamental polynomi-

als

In order to find the estimates, we intend to represent the constants cpk in

general form as given under (p=1,2,3,4)

p
[ 2 ] p−s
5 X X (s) (p−sr)
cpk = epsr [Lk (zk )]r Lk (zk ), (7.13)
p![R 0 (zk )]p s=0 r=s

 
p
where epsr are the constants independent of n and z and denotes
2
(s)
greatest integer function Also, Lk (zk ) denotes sth derivative of Lk (z) at
z = zk .

Lemma 7.3.1. Let Lk (z) be given by (1.17), then for z ∈ T ∪ D, we have


1
| Lk (z) |= O 3 (7.14)
k −α+ 2

Proof. For k = 1, 2, ..., 2n,

R(z)
Lk (z) = . (7.15)
(z − zk )R 0 (zk )

Taking modulus on the both sides,

| R(z) |
| Lk (z) |= .
| z − zk || R 0 (zk ) |

Using (1.14), we have

| (z 2 − 1)W (z) |
| Lk (z) |= .
| z − zk || {2zW (z) + (z 2 − 1)W 0 (z))}z=zk |

University of Lucknow, Lucknow–226007 101

 Interpolation on non-uniform nodes of the unit circle

Since zk0 s are the zeros of W (z), using (1.12), we get

 
(α,β) 1+z 2
(z 2 − 1)Kn Pn 2z
zn
| Lk (z) | = n   o0
(α,β) 1+z 2
| z − zk | (zk2 − 1) Kn Pn 2z
zn
 2 z=zk
(α,β) 1+z
(z 2 − 1)Pn 2z
zn
= n  2 o
(α,β) (α,β)0 zk −1
| z − zk | (zk2 − 1) nzkn−1 Pn (xk ) + zkn Pn (xk ) 2z 2
k
(α,β)
2|z − 1||Pn (x)||z|n
2
= (α,β)0
.
|z− zk | |zk |n−2 |zk2 − 1|2 |Pn (xk )|

Using (1.18) and (1.19), we get

√ (α,β)
2.2 1 − x2 |Pn (x)|
| Lk (z) | = √ q √ (α,β)0
p
4(1 − x2k ) 2 1 − xxk − 1 − x2 1 − x2k |Pn (xk )|
√ (α,β)
q √ p
1 − x2 |Pn (x)| 1 − xxk + 1 − x2 1 − x2k
= √ p (α,β)0
2(1 − x2k ) (1 − xxk )2 − (1 − x2 )(1 − x2k )|Pn (xk )|
√ (α,β)
q p
1 − x2 |Pn (x)| 1 − xxk + 1 − x2 − x2k + x2 x2k
= √ (α,β)0
2(1 − x2k ) | x − xk | |Pn (xk )|
√ (α,β)
q p
1 − x2 |Pn (x)| 1 − xxk + (1 − xxk )2 − (x − xk )2
= √ (α,β)0
2(1 − x2k ) | x − xk | |Pn (xk )|
√ (α,β) √
1 − x2 |Pn (x)| 1 − xxk
≤ (α,β)0
.
(1 − x2k ) | x − xk | |Pn (xk )|

1
For | x − xk |≥ 2
| 1 − x2k |, we get

√ (α,β)
1 − x2 |Pn (x)|
| Lk (z) | ≤ C (α,β)0
,
(1 − x2k )3/2 |Pn (xk )|

where C is constant independent of n and z. Using (1.20), (1.26) and (1.28),

we have
!
1
| Lk (z) | = O 3 . (7.16)
k −α+ 2

University of Lucknow, Lucknow–226007 102

 Interpolation on non-uniform nodes of the unit circle

1
Similarly, for | x − xk |≤ 2
| 1 − x2k |, we get the same result as (7.16).
For k = 0 and k = 2n + 1, we have

| L0 (z) |=| L2n+1 (z) |= O(1). (7.17)

From (7.16) and (7.17), we have Lemma 7.3.1.

Lemma 7.3.2. Let cpk be given by (7.13), then

!
1
| cpk |= O . (7.18)
Knp np(α−1) k p/2−pα

Proof. From (1.14) and (1.12), we have

!
Kn n−2 2 0
R (zk ) =
0
zk (zk − 1)2 Pn(α,β) (xk ).
2

Taking modulus on both the sides, we get

!
Kn 0
| R 0 (zk ) |= | zk |n−2 | (zk2 − 1) |2 | Pn(α,β) (xk ) | .
2

From (1.18), (1.26) and (1.28), we have

1
| R 0 (zk ) |= O(Kn k −α+ 2 nα ). (7.19)

Similarly, from (1.9), (1.17) and (1.18), we have

(s)
| Lk (zk ) |= O(ns ). (7.20)

Using (7.19) and (7.20) in (7.13), we have Lemma 7.3.2.

University of Lucknow, Lucknow–226007 103

 Interpolation on non-uniform nodes of the unit circle

Lemma 7.3.3. Let A0k (z) be given by (7.4) and cpk given by (7.13), then
for z ∈ T ∪ D,
2n+1
X
| A0k (z) | = O(log n), (7.21)
k=0

1
where −1 < α ≤ .
2

Proof. From (1.14) and (1.12), we have

!
1 + z2 n
R(z) = Kn (z 2 − 1)Pn(α,β) z . (7.22)
2z

Taking modulus on both the sides and using (1.18) and (1.20), we have

| R(z) |= O(Kn nα−1 ). (7.23)

1
For | xk − x | ≥ | 1 − x2k | and from (7.4) and (7.5), we have
2

2n+1
X 2n+1
X 2n+1
XX 4
| A0k (z) | = 5
| Lk (z) | + | cpk | | R(z) |p | Lk (z) |5−p .
k=0 k=0 k=0 p=1

Using (7.23), Lemma 7.3.1 and Lemma 7.3.2, we get

2n+1 2n+1 2n+1 4

!
X X 1 XX 1
| A0k (z) | = O 15 + 15 ,
k=0 k=0 k −5α+ 2 k=0 p=1 k −5α−p+ 2
2n+1 4
!
XX 1
= O 15 .
k=0 p=0 k −5α−p+ 2

Now, we get

2n+1
( )
X 13 p
| A0k (z) |= O(log n), −1<α≤ − . (7.24)
k=0
10 5

1
Similarly, for | xk − x |≤ 2
| 1 − x2k |, we get the same result.

University of Lucknow, Lucknow–226007 104

 Interpolation on non-uniform nodes of the unit circle

Since range of α with p = 4 lies in the intersection of all the cases. Hence,
the lemma follows.

7.4 Convergence

Theorem 7.4.1. Let f (z) is a function continous on T ∪ D and analytic

on D. The sequence of interpolatory polynomial {Qn (z)} satisfies the below
relation
| Qn (z) − f (z) | = O ω(f, n−1 ) log n


(7.25)

where ω(f, n−1 ) represents the modulus of continuity of the function f (z).

Remark 7.1 : To prove the Theorem 7.4.1, we shall need following.

Let f (z) be a function continuous on T ∪ D and analytic on D. Then, there
exists a polynomial Fn (z) of degree less than (2n + 2)(r + 1) satisfying Jack-
son’s inequality [62].

| f (z) − Fn (z) | ≤ C ω(f, n−1 ), r ≥ 0, (7.26)

where ω(f, n−1 ) denotes the first modulus of continuity of f (z) as well as C
is a constant independent of n and z.

Proof. Let f (z) be a function that is continous on T ∪ D and analytic on

D. Since Qn (z) is the uniquely determined polynomial of degree less than
(2n + 2)(r + 1) and the polynomial Fn (z) satisfying equation (7.26) can be
expressed as
2n+1
X
Fn (z) = Fn (zk )Ak (z). (7.27)
k=0

Then,

| Qn (z) − f (z) | ≤ | Qn (z) − Fn (z) | + | Fn (z) − f (z) | . (7.28)

University of Lucknow, Lucknow–226007 105

 Interpolation on non-uniform nodes of the unit circle

Using (7.2) and (7.27), we have

2n+1
X
| Qn (z) − f (z) | ≤ | f (zk ) − Fn (zk ) || Ak (z) | + | Fn (z) − f (z) | .
| {z }
k=0 N2
| {z }
N1

From (7.26) and (7.21), we have

N1 = O ω(f, n−1 ) log n .

(7.29)

From (7.26), we have

N2 = O ω(f, n−1 ) .


(7.30)

Using (7.29) and (7.30) in (??), we get

| Qn (z) − f (z) | = O ω(f, n−1 ) log n .

Hence, Theorem 7.4.1 follows.

University of Lucknow, Lucknow–226007 106

Chapter 8

Convergence of Lagrange
Interpolation on the
non-uniform nodes on the unit
circle

In this chapter, we staunchly study the approximation using Lagrange inter-

polation on the unit circle. Nodal system constitutes the vertically projected
zeros of Jacobi polynomial onto the unit circle with boundary points at ±1.
Moreover, convergence is obtained by considering analytic functions on a
suitable domain accompanied by some numerical experiments.

8.1 The Problem and Explicit Representa-

tion of Interpolatory Polynomial

Here, we are interested in determining the convergence of Lagrange inter-

polatory polynomial Ln (z) of degree atmost (2n + 1) on the distinct set of
nodes {−1, 1} ∪ Z2n satisfying

Ln (zk ) = αk , k = 0, 1, ..., 2n + 1 (8.1)

107
 Interpolation on non-uniform nodes of the unit circle

where αk0 s are complex constants.

Theorem 8.1.1. We shall write Ln (z) satisfying (8.1) as

2n+1
X
Ln (z) = f (zk )Lk (z) (8.2)
k=0

where Lk (z) is a polynomial of degree atmost 2n+1 satisfying (8.3)

Lk (zj ) = δkj , k, j = 0, 1, ..., 2n + 1. (8.3)

8.2 Estimation of the Fundamental Polyno-

mial

Lemma 8.2.1. Let Lk (z) be given by (1.17), then for z ∈ T ∪ D, we have

2n+1
X
|Lk (z)| = O(log n), (8.4)
k=0

1
where −1 < α ≤ .
2

Proof. For k = 1, 2, ..., 2n.

R(z)
Lk (z) = . (8.5)
(z − zk )R 0 (zk )

Talking modulus on both the sides, we get

| R(z) |
| Lk (z) | = ,
| z − zk || R 0 (zk ) |
| (z 2 − 1)W (z) |
= .
| z − zk || {2zW (z) + (z 2 − 1)W 0 (z)}z=zk |

University of Lucknow, Lucknow–226007 108

 Interpolation on non-uniform nodes of the unit circle

Since zk0 s are the zeros of W (z), using (1.12), we get

 
(α,β) 1+z 2
(z 2 − 1)Kn Pn 2z
zn
| Lk (z) | = n   o0
(α,β) 1+z 2
| z − zk | (zk2 − 1) Kn Pn 2z
zn
 2 z=zk
(α,β) 1+z
(z 2 − 1)Pn 2z
zn
= n  2 o
(α,β) (α,β)0 zk −1
| z − zk | (zk2 − 1) nzkn−1 Pn (xk ) + zkn Pn (xk ) 2z 2
k
(α,β)
2|z − 1||Pn (x)||z|n
2
= (α,β)0
.
|z− zk | |zk |n−2 |zk2 − 1|2 |Pn (xk )|

Using (1.18) and (1.19), we get

√ (α,β)
2.2 1 − x2 |Pn (x)|
| Lk (z) | = √ q √ (α,β)0
p
4(1 − x2k ) 2 1 − xxk − 1 − x2 1 − x2k |Pn (xk )|
√ (α,β)
q √ p
1 − x2 |Pn (x)| 1 − xxk + 1 − x2 1 − x2k
= √ p (α,β)0
2(1 − x2k ) (1 − xxk )2 − (1 − x2 )(1 − x2k )|Pn (xk )|
√ (α,β)
q p
1 − x2 |Pn (x)| 1 − xxk + 1 − x2 − x2k + x2 x2k
= √ (α,β)0
2(1 − x2k ) | x − xk | |Pn (xk )|
√ (α,β)
q p
1 − x2 |Pn (x)| 1 − xxk + 1 + x2 x2k − 2xxk + 2xxk − x2 − x2k
= √ (α,β)0
2(1 − x2k ) | x − xk | |Pn (xk )|
√ (α,β)
q p
1 − x2 |Pn (x)| 1 − xxk + (1 − xxk )2 − (x − xk )2
= √ (α,β)0
2(1 − x2k ) | x − xk | |Pn (xk )|
√ (α,β) √
1 − x2 |Pn (x)| 1 − xxk
≤ (α,β)0
.
(1 − x2k ) | x − xk | |Pn (xk )|

1
For | x − xk |≥ 2
| 1 − x2k |,

√ (α,β)
1 − x2 |Pn (x)|
| Lk (z) | ≤ C (α,β)0
,
(1 − x2k )3/2 |Pn (xk )|

University of Lucknow, Lucknow–226007 109

 Interpolation on non-uniform nodes of the unit circle

where C is constant independent of n and z. Using (1.20), (1.26) and (1.28),

we have
!
1
| Lk (z) | = O 3 . (8.6)
k −α+ 2

1
Similarly, for | x − xk |≤| 1 − x2k |, we get the same result as (8.6).
2
For k = 0 and k = 2n + 1, we have

| L0 (z) |=| L2n+1 (z) |= O(1) (8.7)

Now, on taking summmation we get

2n+1 2n+1
!
X X 1
|Lk (z)| = O 3 , (8.8)
k=0 k=0 k −α+ 2
2n+1
X 1
|Lk (z)| = O(log n), −1 < α ≤ .
k=0
2

Hence, we have Lemma 8.2.1.

8.3 Convergence

Theorem 8.3.1. Let f (z) be a function continuous on T∪D and analytic on

D. The sequence of interpolatory polynomial {Ln (z)} satisfies the following
relation
|Ln (z) − f (z)| = O(ω(f, n−1 ) log n), (8.9)

where ω(f, n−1 ) represents the modulus of continuity of the function f (z).

Remark 5.1 : To prove the Theorem 8.3.1, we shall need following.

Let f (z) be a function continuous on T ∪ D and analytic on D. Then, there
exists a polynomial Fn (z) of degree less than (2n + 2) satisfying Jackson’s

University of Lucknow, Lucknow–226007 110

 Interpolation on non-uniform nodes of the unit circle

inequality[62].
| f (z) − Fn (z) | ≤ C ω(f, n−1 ) (8.10)

where ω(f, n−1 ) denotes the modulus of continuity of f (z) as well as C is a

constant independent of n and z.

Proof. Let f (z) be a function that is continuous on T ∪ D and analytic on D.

Since, Ln (z) be the uniquely determined polynomial of degree atmost 2n + 1
and the polynomial Fn (z) satisfying (8.10) can be expressed as

2n+1
X
Fn (z) = Fn (zk )Lk (z). (8.11)
k=0

Therefore,

|Ln (z) − f (z)| = |Ln (z) − Fn (z) + Fn (z) − f (z)|

≤ |Ln (z) − Fn (z)| + |Fn (z) − f (z)|.

Using (8.2) and (8.11), we have

2n+1
X
|Ln (z) − f (z)| ≤ |f (zk ) − Fn (zk )||Lk (z)| + |Fn (z) − f (z)| .
| {z }
k=0
| {z }

We have,
|Ln (z) − f (z)| ≤ N1 + N2 . (8.12)

Using Lemma 8.2.1 and (8.10), we have

2n+1 2n+1
!
X X
N1 = |f (zk ) − Fn (zk )||Lk (z)| = O ω(f, n−1 ) |Lk (z)|
k=0 k=0
1
= O(ω(f, n−1 ) log n), −1 < α ≤ . (8.13)
2

University of Lucknow, Lucknow–226007 111

 Interpolation on non-uniform nodes of the unit circle

and
N2 = |Fn (z) − f (z)| = O(ω(f, n−1 )). (8.14)

Using (8.13) and (8.14) in (8.12), we get

|Ln (z) − f (z)| = O(ω(f, n−1 ) log n)

Hence, Theorem 8.3.1 follows.

8.4 Numerical Experiments

To visualize the contribution of this chapter, we carried out some numerical

experiments.
We work in the following way for both the examples we included.

• Nodal system {−1, 1} ∪ Z2n is used with varying values of n.

• We detail a test function f (z) in each example and the Lagrange inter-
polating polynomial Ln (z) that interpolates the test function is always
plot in black.

• To showcase the interpolation behavior on the boundary and within

the unit circle, we considered two cases in each example, where z ∈ T
and z ∈ D.

Example 1. We consider a rational function f continuous on z ∈ T ∪ D

and analytic in D satisfying the hypothesis of theorem 8.3.1 defined by

2z
f (z) = .
z2+2

Case 1: Here z ∈ T i.e z = eiθ , θ ∈ [0, 2π) and we choose Jacobi’s

polynomial parameter as α = β = 0. We have plotted the graphic of absolute

University of Lucknow, Lucknow–226007 112

 Interpolation on non-uniform nodes of the unit circle

values of f and its interpolant Ln (z) constructed with the 2n + 2 roots of T as

nodes. The results can be seen in the following Fig. 8.1, Fig. 8.2 and Fig.
8.3 showing for n = 2, n = 4 and n = 8 respectively. In right of each of the
figures the error term |Ln (z) − f (z)| has been shown to get a good grasp of
the errors involved in interpolating the test function with the varying number
of nodes.
It can be clearly seen as the value of n increases, we get a good fit of the
function via our interpolatory polynomial and for n = 8 i.e with number of
nodes=18, maximum error can be seen of order 10−3 in right of Fig. 8.3.
So, a rational function f (z) analytic in T can be very well approximated via
Lagrange interpolatory polynomial on the projected nodes of the zeros of
Jacobi polynomial.
0.5
2.0

1.8 0.4

1.6
0.3
1.4

1.2 0.2

1.0
0.1
0.8

1 2 3 4 5 6 0 1 2 3 4 5 6

Figure 8.1: |f (eiθ )| and |L2 (eiθ )|; |L2 (z) − f (z)| for θ ∈ [0, 2π) and n = 2.

0.14
2.0

0.12
1.8

0.10
1.6

0.08
1.4

0.06
1.2

0.04
1.0

0.02
0.8

0.6 1 2 3 4 5 6 0 1 2 3 4 5 6

Figure 8.2: |f (eiθ )| and |L4 (eiθ )|; |L4 (z) − f (z)| for θ ∈ [0, 2π) and n = 4.

University of Lucknow, Lucknow–226007 113

 Interpolation on non-uniform nodes of the unit circle

0.008

1.8
0.006
1.6

1.4
0.004

1.2

1.0 0.002

0.8

0.6 1 2 3 4 5 6 0 1 2 3 4 5 6

Figure 8.3: |f (eiθ )| and |L8 (eiθ )|; |L8 (z) − f (z)| for θ ∈ [0, 2π) and n = 8.

Case 2: Here z ∈ D and n = 16. We took largest

! rectangular region of
−1 1
D. We express z = x + iy, where x, y ∈ √ , √ and we choose Jacobi’s
2 2
polynomial parameter as α = β = 0. We have plotted the f and its inter-
polant L16 (z) constructed with the 34 roots of T as nodes. The results can
be seen in the following Fig. 8.4. In right of the Fig. 8.4, maximum error
|Ln (z) − f (z)| can be observed less than or equal to 6 × 10−6 .

!
−1 1
Figure 8.4: |f (x+iy)| and |L16 (x+iy)|; |L16 (z)−f (z)| for x, y ∈ √ ,√
2 2
and n = 16.

Example 2. We consider a exponential function f continuous on z ∈

T ∪ D and analytic in D satisfying the hypothesis of theorem 8.3.1 defined by

f (z) = ez .

Case 1: Here z ∈ T i.e z = eiθ , θ ∈ [0, 2π) and we choose Jacobi’s poly-

University of Lucknow, Lucknow–226007 114

 Interpolation on non-uniform nodes of the unit circle

1
nomial parameter as α = β = . We have plotted the graphic of absolute
2
values of f and its interpolant L8 (z) constructed with the 18 roots of T as
nodes. The results can be seen in the following Fig. 8.5. In right of the Fig.
8.5, graphic of the error term |L8 (z) − f (z)| can be seen with error less than
8 × 10−10

8. × 10-10
2.5

6. × 10-10
2.0

4. × 10-10
1.5

1.0 2. × 10-10

0.5
1 2 3 4 5 6
1 2 3 4 5 6

Figure 8.5: |f (eiθ )| and |L8 (eiθ )|; |L8 (z) − f (z)| for θ ∈ [0, 2π) and n = 8.

Case 2: Here z ∈ D and n = 8. We took largest ! rectangular region of

−1 1
D. We express z = x + iy, where x, y ∈ √ , √ and we choose Jacobi’s
2 2
1
polynomial parameters as α = β = . We have plotted the f and its inter-
2
polant L8 (z) constructed with the 18 roots of T as nodes. The results can
be seen in the following Fig. 8.6. In right of the Fig. 8.6, the error term
|L8 (z) − f (z)| can be observed less than or equal to 2 × 10−11 .

!
−1 1
Figure 8.6: |f (x + iy)| and |L8 (x + iy)|; |L8 (z) − f (z)| for x, y ∈ √ ,√
2 2
and n = 8.

University of Lucknow, Lucknow–226007 115

 Interpolation on non-uniform nodes of the unit circle

We also took random values except this rectangular domain within the
unit circle and results were found to be same indicating Lagrange interpola-
tory polynomials on the projected nodes delivers a very good approximation
of the analytic functions within unit circle.

8.5 Discussions

Taking into account the Theorem 8.3.1, results obtained via numerical exam-
ples are similar to the expected ones as per the theory developed throughout
the chapter. Polynomials Ln (z) converges to the test function f (z), which
clearly indicates that Lagrange interpolatory polynomial formed using the
projected nodes of the zeros of Jacobi polynomials does a pretty good job in
approximating suitable class functions on the unit circle.

8.6 Materials and Methods

The experiments given in the section 8.4 were obtained by using personal
codes elaborated with software Wolfram Mathematica [134] without using
any extra precision. All computations were performed on the Macbook Pro
13 inch 2019 edition (1.4 GHz Quad-Core Intel Core i5 and 8 GB 2133 MHz
LPDDR3).

University of Lucknow, Lucknow–226007 116

Bibliography

[1] Abbasov E.A., (1972), The convergence of (0,2)- Interpolation polyno-

mial constructed for the functions of variable (Russian), Azerbaidzan
Gos. Univ. Ucen, Zap. Ser. Fiz. Mat. Nauk., 1, pp. 45-51.
[2] Akhlaghi M.R. and Sharma A., (1990), Some Pál - type Interpolation
problems, Approx. Optimization and Computing Theory and Applica-
tions; Elsevier Science Publisher B.V.(North Holland), IMACS, pp. 37-
40.
[3] Akhlaghi M.R., Jakjmovski A. and Sharma A., (1990), Equiconvergence
of some complex interpolatory polynomials, Numer. Math., 6, pp. 635 –
649. https://link.springer.com/article/10.1007/BF01386433
[4] Akhlaghi M.R., (1998), Interpolation on the zeros of πn (x), Approx.
Theory and its Applications, 42, pp. 55- 74.
[5] Bahadur S. and Mathur K.K., (1998), Quasi – Hermite Interpolation on
the unit circle, Ganita, 49(1), pp. 1-6.
[6] Bahadur S., (2010), (0,1,2)-Interpolation on the unit circle, South East
Asian J. Math. & Math Sc., 9(3), pp. 23-31.
[7] Bahadur S. and Mathur K.K., (2011), Weighted (0,2)*- Interpolation on
unit circle, International Journal of Applied Mathematics and Statis-
tics, 20(M11), pp. 73-78. http://www.ceser.in/ceserp/index.php/
ijamas/article/view/593
[8] Bahadur, S., (2011), (0;0,1) interpolation on the unit circle, Inter-
national Journal of Mathematical Analysis, 5(29), pp. 1429-1434.
https://www.researchgate.net/publication/267015902_001_
interpolation_on_the_unit_circle
[9] Bahadur, S., (2011), Convergence of Hermite Interpolation, Global Jour-
nal of Theoretical and Applied Mathematics Sciences, 1(2), pp. 115-119.
https://www.ripublication.com/Volume/gjtamsv1n3.htm
[10] Bahadur, S., (2011), Pál - type (0,1;0)-Interpolation on unit circle, Ad-
vances in Theoretical Mathematics and Applications, 6(1), pp. 35-39.
https://www.ripublication.com/Volume/atamv6n1.htm

117
 Interpolation on non-uniform nodes of the unit circle

[11] Bahadur, S., (2012), A study of certain Interpolatory processes on unit

circle, Lap Lambert pub. Co., Germany. https://www.amazon.com/
Study-Certain-Interpolatory-Processes-Circle/dp/3847378546

[12] Bahadur S. and Mathur K.K., (2013), Weighted (0,1,3)- Interpolation

on unit circle, American J. Math. Stat., 3(6), pp. 337–347. http://
article.sapub.org/10.5923.j.ajms.20130306.06.html

[13] Bahadur S., (2012), Convergence of Lagrange Interpolation on

the unit circle, International Mathematical Forum, 7(4), pp.
189 - 192. http://www.m-hikari.com/imf/imf-2012/1-4-2012/
bahadurIMF1-4-2012.pdf

[14] Bahadur S., (2012), Pál - type (0;0,2)-Interpolation on the unit circle;
Int. J. Math. Sci., 11(1-2), pp. 103 - 111.

[15] Bahadur S., (2014), (0,2;0)-Interpolation on the unit cir-

cle, Italian Journal of Pure and App. Math., 32, pp. 57-66.
https://www.researchgate.net/publication/290542102_0_2_
0-Interpolation_on_the_unit_circle

[16] Bahadur S. and Shukla M., (2013), (0,1,3)*- Interpolation on the

Projected Nodes On the Unit Circle, International Journal Of Pure
and Applied Mathematical Sciences, 7(1), pp. 91-102. https://www.
ripublication.com/Volume/ijpamsv7n1.htm

[17] Bahadur S. and Shukla M., (2014), A New Kind of Hermite In-
terpolation, Adv. Inequal. Appl., http://scik.org/index.php/aia/
article/view/1265

[18] Bahadur S. and Shukla M., (2014), On the Derivative of Jacobi’s Poly-
nomial, Asian Journal Mathematics and Applications, Article ID ama
0163, pp. 1-6. https://scienceasia.asia/files/163.pdf

[19] Bahadur S. and Shukla M., (2014), Hermite-Lagrange Interpolation

on a Unit Circle, Journal of Advances in Mathematics, 9(1). https:
//rajpub.com/index.php/jam/article/view/6891

[20] Bahadur S. and Shukla M., (2014), Convergences of Lagrange -

Hemite Interpolation, Italian Journal of Pure and Applied Math-
ematics, 33, pp.. 255-262 https://ijpam.uniud.it/online_issue/
201433/21-BahadurShukla.pdf

[21] Bahadur S. and Shukla M., (2015), Weighted Pál - type Interpolation
On The Unit Circle, Int. J. Emer. Tech. Comp. App. Sci., 12(2), pp.
135-139 http://iasir.net/IJETCASpapers/IJETCAS15-350.pdf

University of Lucknow, Lucknow–226007 118

 Interpolation on non-uniform nodes of the unit circle

[22] Bahadur S. and Shukla M., (2014), (0,2)-Interpolation On The Unit Cir-
cle, International Journal of Advancements in Research and Technology,
3(1).

[23] Bahadur S. and Shukla M., (2016), On the Convergence of Derivative

of Hermite Interpolating Polynomial, International Journal of Research
Studies in Sciences, Engineering and Technology, 3(5), pp. 9-13. http:
//ijrsset.org/pdfs/v3-i5/2.pdf

[24] Bahadur S., (2017), On Lagrange – Hermite Interpolating Polynomial,

J. Emer. Tech. and Innovative Research, 99(11). https://www.jetir.
org/papers/JETIR1711004.pdf

[25] Bahadur S., (2017) and Bano S., Modified Hermite Interpolation on
The Unit Circle, Annals of Pure and Applied Mathematics, 14(3), pp.
531-537.

[26] Burkett, J.; Varma, A.K., (1995), On Birkhoff interpolation (0;2) case.,
11(2), pp. 59-66. https://zbmath.org/?q=an%3A0930.41004

[27] Bajpai P., Weighted (0,2)-Interpolation on the extended Tchebychev

nodes of second kind, Approx. Theory appl., 11(2), pp. 59-66.

[28] Balázs J., Turan P., (1958), Notes On Interpolation III, Acta. Math.
Acad. Sci. Hung., 9, pp. 195-214. https://link.springer.com/
article/10.1007/BF02023872

[29] Balázs J., Sulyozött, (1961), (0,2)-Interpolácio Ultraszherikus Poli-

nomok Gyökein, MTA, Math. Fiz. Ozot. Közl, 11(3), pp. 305-338.

[30] Balázs J., (1998), Modified weighted (0,2)-Interpolation, Approx. The-

ory, Marcel Dekker Inc., New York, pp. 61-73.

[31] Balázs K., Kilgore T., (1993), Interpolation and Simultaneous mean
convergence of derivatives, Acta. Math. Hungar, 61, pp. 63-72. https:
//link.springer.com/article/10.1007/BF01872098

[32] Berriochoa E., Cachafeiro A., Martinez Brey E., (2011), Some im-
provements to the Hermite - Féjer interpolation on the unit cir-
cle and bounded interval, Computers and Mathematics with Ap-
plications, 61(4), pp. 1228–1240. https://www.sciencedirect.com/
science/article/pii/S0898122110009909

[33] Berriochoa E., Cachafeiro A., Garcı́a-amor J.M., (2012), An extension of

Fejér’s condition for Hermite interpolation. Complex Analysis and Op-
erator Theory, 6, pp. 651–664.https://link.springer.com/article/
10.1007/s11785-011-0161-2

University of Lucknow, Lucknow–226007 119

 Interpolation on non-uniform nodes of the unit circle

[34] Berriochoa E., Cachafeiro A., Dı́az J., Martı́nez Brey E., (2013), Rate of
convergence of Hermite- Fejér interpolation on the unit circle, Journal
of Applied Mathematics. Article ID 407128, 8 pages https://doi.org/
10.1155/2013/407128.
[35] Berriochoa E., Cachafeiro A., Dı́az J., (2014), Hermite interpolation on
the Unit Circle Considering up to the Second Derivative, ISRN Mathe-
matical Analysis, http://dx.doi.org/10.1155/2014/808519.
[36] Berriochoa E., Cachafeiro A., Garcı́a-Amor J.M., (2020),
Gibbs–Wilbraham phenomenon on Lagrange interpolation based
on analytic weights on the unit circle, J. Comput. Appl. Math., 365.
https://doi.org/10.1016/j.cam.2019.112376.
[37] Berriochoa E., Cachafeiro A., Rabade H.G. and Garcı́a-Amor J.M.,
(2022), Mechanical Models for Hermite Interpolation on the Unit Circle,
Mathematics, , 9(9). https://www.mdpi.com/2227-7390/9/9/1043
[38] Brück R., (1996), Lagrange interpolation in non–uniformly distributed
zeros on unit circle, Analysis, 16(3), pp. 273–282 https://www.
degruyter.com/document/doi/10.1524/anly.1996.16.3.273/html.
[39] Bruin Marcel G. de and Sharma A., (2003), Lacunary Pál - type
Interpolation and over – convergence, Computational and Function
Theory, 3(1), pp. 305-323. https://sci-hub.hkvisa.net/10.1007/
BF03321040.
[40] Cavaretta A.S., Sharma A. and Varga R. S., (1980), Interpolation in
the roots of unity : An extension of a theorem of J. L. Walsh, Re-
sults in Mathematics, 3(2), pp. 155-191. https://link.springer.com/
article/10.1007/BF03323355.
[41] Chen W. and Sharma A., (2004), Lacunary interpolation on some non-
uniformly distributed nodes on the unit circle, Annales Universitatis
Scientiarum Budapestinensis, 16, pp. 69-82. http://ac.inf.elte.hu/
Vol_016_1996/069.pdf
[42] Cheney E.W., (1966), Introduction to Approximation Theory, McGraw-
Hill Book Co., New York. https://www.worldcat.org/title/
introduction-to-approximation-theory/oclc/243547
[43] Daruis L. and González-vera P., (2001), A Note on Hermite -Fejér
Interpolation for the Unit Circle, Applied Mathematic Letters, 14,
pp. 997-1003. https://www.sciencedirect.com/science/article/
pii/S0893965901000787
[44] Davis P.J., (1963), Interpolation and Approximation, Blaise-
lell Pub. Co., Waltham Mass. https://www.worldcat.org/title/
interpolation-and-approximation/oclc/228351

University of Lucknow, Lucknow–226007 120

 Interpolation on non-uniform nodes of the unit circle

[45] Dikshit H.P., (2001), Birkhoff Interpolation on some perturbed roots of

unity, Nonlinear Analysis Forum, 6(1), pp. 97-102.

[46] Dikshit H.P., (2003), Pál - type Interpolation on non - uniformly

distributed nodes on the unit circle Journal of Comp. and Ap-
plied Math., 155(2), pp. 253-261. https://www.researchgate.
net/publication/253558985_Pal-type_interpolation_on_
nonuniformly_distributed_nodes_on_the_unit_circle

[47] Eneduanya S.A.N., (1984), On Interpolation polynomials us-

ing the roots of Ultraspherical polynomials, Demonstratio Math.,
17(4), pp. 1043-1050. https://www.degruyter.com/document/doi/
10.1515/dema-1984-0423/html

[48] Erdös P. and Vértesi P., (1980), On almost everywhere divergence of

Lagrange Interpolatory polynomials for arbitrary system of nodes, Acta.
Math. Acad. Sci. Hung., 36, pp. 71-89. https://link.springer.com/
article/10.1007/BF01897094

[49] Fejér L., (1916), Über Interpolation, Gött. Nachr, 6, pp. 66–91. https:
//www.math.technion.ac.il/hat/people/fejer.html

[50] Ferguson D., (1969), The question of Uniqueness for G. D. Birkhoff

Interpolation problem, J. Approx. Theory, 2, pp. 1-28. https://www.
sciencedirect.com/science/article/pii/0021904569900288

[51] Freud G., (1958), Bemerkungen Uber die Kanverg enzeins interpolation
sverfahrens Von. P. Turan; Acta. Math. Acad. Sci. Hung., 9, pp. 337-341.
https://link.springer.com/article/10.1007/BF02020263

[52] Freud G., (1964), On Jacksonian Interpolation process in Approximation

Theory, Proc. of Conference in Overfach Birkhouser Basel and Stuttgart,
pp. 227-232.

[53] Freud G. and Vértesi P., (1967), A new proof of A. F. Timan Ap-
proximation Theorem; Studia Sci. Math. Hung., 2, pp. 403-414. https:
//mathscinet.ams.org/mathscinet-getitem?mr=221165

[54] Freud G., (1973), On polynomial approximation with weight func-

tion exp(− 21 x2k ), Ibid, 24, pp. 363-371. https://link.springer.com/
article/10.1007/BF01958048

[55] Freud G. and Sharma A., (1974), Some good sequences of Interpolatory
polynomials, Cand. J. of Maths. 26(1), pp. 233-246. https://www.
cambridge.org/core/services/aop-cambridge-core/content/
view/BCE1E28A193914D480419ED3F1C53A5C/S0008414X0004877Xa.
pdf/some-good-sequences-of-interpolatory-polynomials.pdf

University of Lucknow, Lucknow–226007 121

 Interpolation on non-uniform nodes of the unit circle

[56] German A.H., (1980), Interpolation in the complex domain, Anal.

Math., 6(2), pp. 121-135.
[57] Gonska H., (1989), Degree of Approximation by Lacunary In-
terpolators (0, 1, . . . , R − 2, R)-Interpolation, Rocky Mountain J.
Math., 19(1), pp. 157-171. https://sci-hub.hkvisa.net/10.1216/
RMJ-1989-19-1-157
[58] Goodman T.N.T, Ivanov K.G. and Sharma A., (1996), Hermite Interpo-
lation in the roots of unity; J. Approx. Theory., 84, pp. 41-60. https://
www.sciencedirect.com/science/article/pii/S0021904596900040
[59] Gopengauz I.E., (1967), On the theorem of A. F. Timan on Approxima-
tion of continuous functions on a line segment, Math. Zametski, 2, pp.
163-172.
[60] Grünwald G., (1943), On the theory of interpolation, Acta Mathematica,
75, pp. 219-245.
[61] Ivanov K.G. and Saff E.B., (1990), Behaviour of the Lagrange Inter-
polants in the roots of unity, comp. methods and function theory (Val-
paraiso, 1989); Lecture notes in Math., 1435, Springer, Berline, 2, pp.
81-87. https://link.springer.com/chapter/10.1007/BFb0087899
[62] Jackson D., (1911), Ueber die Genauigkeit der Annäherung stetiger
Funktionen durch ganze rationale Funktionen gegebenen Grades und
trigonometrische Summen gegebener Ordnung, Göttingen.
[63] Jackson D., (1930), The Theory of Approximation; Amer. Math. Soc.,
Coll. Publ. Vol. II, New York.
[64] Jiang Gong Jian, (1991), Approximation by type (0,2)- Interpolation
polynomials, Neimenggu Shida Xuebao Ziran Kexue Ban, 1, pp. 7-14.
[65] Jiang Tian Zi, (1993), Asymptotic expansion of the Lebesgue constants
associated with Lagrange polynomial Interpolation on unit circle, Soo-
chow J. Math., 19(4), pp. 429-434.
[66] Joó I. and Szabo V.E.S., (1995), A generalization of Pál
Interpolation process; Acta. Sci. Math., 60, pp. 429-438.
http://pub.acta.hu/acta/showCustomerArticle.action?id=9452&
dataObjectType=article&returnAction=showCustomerVolume&
sessionDataSetId=53f9b579aad83d96&style=
[67] Joó I., (1995), On weighted (0,2)- Interpolation, Annals
Univ. Sci. Budapest, 38, pp. 185-222. http://pub.acta.hu/
acta/showCustomerArticle.action?id=9452&dataObjectType=
article&returnAction=showCustomerVolume&sessionDataSetId=
53f9b579aad83d96&style=

University of Lucknow, Lucknow–226007 122

 Interpolation on non-uniform nodes of the unit circle

[68] Joó I. and Pál L.G., (1996), Lacunary (0;0,1)-Interpolation on the roots
of Jacobi polynomials and their derivatives respectively I (Existence,
Explicit Formulae, Unicity); Analysis Math., 22, pp. 289-298.

[69] Kilgore T. and Prestin J., (1991), On the order of simultaneous ap-
proximation by Lagrange - Hermite Interpolation; Math. Nachr., 150,
pp. 143-150. https://onlinelibrary.wiley.com/doi/abs/10.1002/
mana.19911500111

[70] Kiš O., (1960), Remarks on Interpolation (Russian), Acta Math. Acad.
Sci. Hungar, 11, pp. 49-64.

[71] Kiš O. and Vértesi P., (1967), On a new Interpolation process; Ann.
Univ. Sci. Budapest Etövos Sect. Maths, 10, pp. 117-128.

[72] Korenin, L.Ya., (1982), Interpolation on Standard Closed Sets On the

Unit Circle, Application of Function Analysis in Approx. Theory, pp.
98-103.

[73] Krebsz A., (2004), Weighted (0,1,3)- Interpolation on the roots of the
classical orthogonal polynomials, Mathematica Pannonica, 15(1), pp.
21-35.

[74] Knoop H., Locher F., (1990), Hermite-Fejér type interpolation and
Korovkin’s theorem, Bulletin of the Australian Mathematical Society,
42(3), pp. 383-390. https://doi:10.1017/S0004972700028549.

[75] Lénard M., (2000), Modified (0,2)-Interpolation on the roots of

Jacobi polynomials II (convergence), Acta Math. Hung., 89(1-
2), pp. 161-177. https://www.researchgate.net/publication/
246546226_Modified_0_2-Interpolation_on_the_Roots_of_
Jacobi_Polynomials_II_Convergence

[76] Lénard M., (2013), A (0,2)-Pál-type Interpolation problem;

Annales Univ. Sci. Budapest., Sect. Comp., 41, pp. 97-102.
https://www.researchgate.net/publication/267659191_A_
02-type_Pal_interpolation_problem

[77] Li, Zhong Kai, (1991), A note on convergent polynomials of Interpola-

tion, Approx. Theory Appl. 7(3), pp. 19-26.

[78] Mastroianni G. and Vértesi P., (1993), Mean convergence of Lagrange

Interpolation on arbitrary system of nodes, Acta. Math. Hung. Szeged,
57, pp. 429-441.

[79] Mastroianni G., Milovanovic G.V. and Notarangelo I., (2012), An In-
terpolation Process of Lagrange - Hermite Type, De L’ Instinct Math.,
Nouvelleserie Tome, 91(105), pp. 163-175.

University of Lucknow, Lucknow–226007 123

 Interpolation on non-uniform nodes of the unit circle

[80] Milovanovic G.V., Notarangelo I. and Paster P., (2016), Lagrange - Her-
mite Interpolation on the real axis, Calcolo, 53(2), pp. 235-261.

[81] Mathur K.K. and Sharma A., (1961), Some interpolatory properties of
Hermite Polynomials, Acta. Math. Acad. Sci. Hung., 12, pp. 193-207.

[82] Mathur K.K. and Saxena R. B., (1993), On weighted (0,1,3)-

Interpolation on the abscissas as the zeros of Hermite polynomials; Acta.
Math. Hungary, 62(1-2), pp. 31-47.

[83] Mathur K.K. and Srivastava R., (1995), Pál – type interpolation on
infinite interval, J. of Math. Anal. and Appl., 192, pp. 346-359.

[84] Mathur K.K. and Srivastava R., (1996), Weighted (0;0,2)-Interpolation

on the roots of Hermite polynomials; Acta. Math. Hung., 70(1-2), pp.
57-73.

[85] Mathur P., (2004), (0,1,3)-Interpolation on the projected zeros of an

Ultraspherical polynomial on the unit circle, Bull. of Allahabad Math.
Soc., 19, pp. 49-72.

[86] Mathur P., (2006), (0,1;0)-Interpolation on infinite interval (−∞, +∞ ),

Analysis in Theory and Application, 22(2), pp. 105-113.

[87] Mathur P., (2006), Weighted (0,1,3)-Interpolation on an arbitrary set of

nodes, Acta Math. Acad. Paed. Nyireg., 22(2), pp. 265-273.

[88] Mathur P. and Srivastava A. K., (2010), Hermite Interpolation on unit

circle, J. Ultra. Phys. Sci., 21.

[89] Mills T.M., (1980), Some techniques in Approximation theory, Math.

Scientist, 5, pp. 105-120.

[90] Min Guo Huo, (1992), Mean convergence of derivatives of the Hermite
Interpolation operator, Adv. in Math (China), 21(3), pp. 329-333.

[91] Nevái P., (1979), Orthogonal Polynomials (Memoirs of the American

Mathematical Society), No. 213, Providence Rhode Island.

[92] Pál L.G., (1975), A new modification of H-F Interpolation; Anals. Math.,
1, pp. 197-205.

[93] Pál L.G., (1996), A general Lacunary (0;0,1)-Interpolation process;

Anals. Univ. Budapest, Sect. Comp., pp. 291-301.

[94] Prasad J., (1972), Remarks on Lacunary Interpolation, Mathematica,

14(37) pp. 147-165.

University of Lucknow, Lucknow–226007 124

 Interpolation on non-uniform nodes of the unit circle

[95] Prasad J., (1992), On the degree of approximation of the Hermite and
Hermite- Féjer Interpolation, International J. Math. Sci., 15(1) pp. 47-
56.

[96] Prasad J., (1992), On the order of convergence and mean convergence
of interpolatory process, Ganita, 43(1-2) pp. 181-204.

[97] Prasad J., (1993), On Hermite and Hermite-Féjer Interpolation of higher

order, Demonstratio Maths., 26(2) pp. 413-425.

[98] Rainville R.D., (1963), Special Functions, Mac Millan Co., New York.

[99] Sakai R., (1991), Hermite - Féjer Interpolation, Approx. Theory, Collog.
Math. Soc. Janos Bolyai, North – Holland, Amsterdam, pp. 591-601.

[100] Saxena R.B. and Sharma A., (1958), On Some Interpolatory properties
of Legendre polynomial; Acta. Math. Acad. Sci. Hung., 9, pp. 345-358.

[101] Saxena R.B. and Sharma A., (1959), Convergence of Interpolatory

polynomials, Acta. Math. Acad. Sci. Hung., 10, pp. 157-175.

[102] Saxena R.B., (1967), On Polynomial Interpolation, Studia. Sci. Math.

Hung., 2, pp. 167-183.

[103] Sharma A., (1964), Some remarks on Lacunary Interpolation in the

roots of unity, Israel J. Maths., 2, pp. 41-69.

[104] Sharma A. and Leeming D., (1969), Lacunary Interpolation (0, n−1, n)-
case, Mathematical (cluj), 11(34), pp. 155-162.

[105] Sharma A. and Riemenschneider S.D., (1981), Birkhoff Interpolation

in the nth roots of unity: Convergence; Canad. J. Math., 33(2), pp.
362-371.

[106] Shen Xie. Chang and Shuai Bin Peny, (1991), The order of approxima-
tion of Lagrange Interpolating polynomials based on asymptotic roots
of unity; J. Math., 11(3), pp. 287-297.

[107] Shi Y.G., (1992), A comparison of Hermite - Féjer Interpolation and

Lagrange Interpolation; Chinese Ann. Math. Sec. A, 13(4), pp. 411-416.

[108] Shi Y.G., (1997), Mean convergence of Truncated Hermite Interpola-

tion on an arbitrary system of nodes, Acta. Math. Hung., 76(1-2), pp.
45-48.

[109] Srivastava V., Mathur P. and Mathur N., (2011), A new kind of Pál-
type Interpolation II, Int. J. Contemp. Math. Sciences, 6(4-5), pp. 2237-
2246.

University of Lucknow, Lucknow–226007 125

 Interpolation on non-uniform nodes of the unit circle

[110] Srivastava V., Mathur P. and Mathur N., (2013), A new kind of Pál -
type (0;0,1)-Interpolation; Int. J. of Math. and soft computing, 3, pp.
21-32.

[111] Sung H.S., Ko D.H. and Sakai R., (2017), Lp Convergence of

Higher order Hermite or Hermite-Fejér Interpolation polynomials with
exponential-type weights ( II ), Global Journal of Pure and Applied
Mathematics, 13, pp. 7401–7426.

[112] Surányi J. and Turán P., (1995), Notes On Interpolation I, Acta. Math.
Sci. Hung., 6, pp. 67-79.

[113] Szabadós J., (1974), On Some Interpolating procedure based on the

roots of unity; Acta. Math. Acad. Sci. Hung., 25, pp. 159-169.

[114] Szabadós J. and Varma A.K., (1985), On the almost everywhere diver-
gence of Lagrange Interpolation on infinite interval; Acad. Sci. Math.
Hung., 48(14), pp. 443-450.

[115] Szabadós J. and Vértesi P., (1990), Interpolation of Functions, World

Scientific Publishers.

[116] Szabadós J. and Varma A.K., (1990), On (0,1,2)-Interpolation on uni-

form metric; Proc. Amer. Math. Soc., 109(4), pp. 945-979.

[117] Szabadós J. and Varma A.K., (1995), On a convergent Pál - type (0,2)-
Interpolation process; Acta. Math. Hung., 66(4), pp. 301-326.

[118] Szegő, G., (1975), Orthogonal Polynomials, Amer. Math. Soc. Coll.,
23.

[119] Szilli L., (1984), Weighted (0,2)-Interpolation on the roots of Hermite

Polynomials, Annals Univ. Sci. Budapest, 27, pp. 155-166.

[120] Szilli L., (1985), A convergence theorem for the Pál method of Inter-
polation on the roots of Hermite polynomials; Analysis Math., 11, pp.
75-84.

[121] Szilli L., (1993), Weighted (0,2) Interpolation on the roots of Classical
Orthogonal Polynomials; Bull. of Allahabad Math. Soc., 8, pp. 1-10.

[122] Telyakovskii S.A., (1966), Two theorems on Approximation of func-

tions by Algebraic Polynomials; Mat. Sb. (N.S.), 70(112), pp. 252-265.

[123] Timan A.F., (1951), Strengthening of Jackson’s Theorem on the best

Approximation of continuous functions by polynomials on the finite seg-
ment of real axis; Dok. Aka. Nauk., SSSC, 70(112), pp. 17-70.

University of Lucknow, Lucknow–226007 126

 Interpolation on non-uniform nodes of the unit circle

[124] Trefethen L.N., (2011), Six Myths of Polynomial Interpolation and

Quadrature, Math. Today, 47, pp. 184-188.https://people.maths.ox.
ac.uk/trefethen/mythspaper.pdf

[125] Tu Tian Liang, (1990), On Hermite Interpolation in roots of unity,

Approximation, Optimization and Computing, pp. 197-200.

[126] Turan P., (1979), A Turan’s problem on the (0,2)-Interpolation based

on the zeros of Jacobi polynomials; Acta Mathematica Acad. Sci. Hung.,
33, pp. 317-321.

[127] Varma A.K. and Sharma A., (1961), Some interpolatory properties of
Tchebicheff polynomials, (0,2) case modified, Pub. Math. Debreen, 8,
pp. 336-349.

[128] Varma A.K., (1961), Some interpolatory properties of Tchebicheff poly-

nomials, ( 0,1;3)-case; Duke Math. J., 28, pp. 449-462.

[129] Varma A.K., (1982), On some open problems of P. Turán concern-

ing Birkhoff interpolation; Transactions of the American Mathematical
Society, 274(2).

[130] Varma A.K. and Prasad J., (1990), Rate of convergence for Lacunary
Interpolation processes based on the roots of unity; Acta Math. Hung.,
55(3), pp. 211-217.

[131] Vecchia B.D. and Mastroianni G., (1992), The Hermite Interpolations,
Calcolo, 30(4), pp. 371-394.

[132] Vertesi P. and Xu Yuan, (1992), Weighted Lp convergence of Hermite

Interpolation of higher order; Acta. Math. Hung., 59(3-4), pp. 423-438.

[133] Wang Zi Yu and Tian Ji Shan, (1991), On the pointwise estimations

of approximation of functions and their derivatives by Hermite Interpo-
lation polynomial; Chinese J. Math, 19(3), pp. 243-256.

[134] Wolfram Research, (2020), Inc., Mathematica, Version 12.1, Cham-

paign, IL .https://www.wolfram.com/mathematica/?source=nav

[135] Xiang, S., He, G., (2015), The Fast Implementation of Higher Order
Hermite–Fejér Interpolation, SIAM Journal on Scientific Computing, 37,
pp. A1727–A1751. https://doi.org/10.1137/140971488

[136] Xie S., (1993), On a problem of (0,2) Interpolation, Approx. Theory

Appl., 9(1), pp. 9-17. https://doi.org/10.1007/BF02836485

[137] Xie S., (1995), Regularity of (0, 1, . . . , r − 2, r) and (0, 1, . . . , r − 2, r)*-

Interpolation on some sets of the unit circle, J. Approx. Theory, 82(1),
pp. 54-59.

University of Lucknow, Lucknow–226007 127

 Interpolation on non-uniform nodes of the unit circle

[138] Xie S., (1996), (0,1,3)*- Interpolation on the unit circle, Acta Math.
Sinica, 39(5), pp. 690-700.

[139] Xie S., (2001) Weighted least square convergence of Lagrange interpo-
lation on the unit circle, Approximation Theory and its Applications,
17, pp. 60–68. https://doi.org/10.1023/A:1015502428520

[140] Xie T.F., (1993), On the approximation of (0,2)-Interpolation, A Chi-

nese summary appears in Chinese Ann. Math. Ser. A, 14(2), pp. 201-208.

[141] Xie T.F., (1995), On the Approximation of (0;1;2)-Interpolation poly-

nomials, Chinese J. Comp. Math., 16(4), pp. 429-436.

?????

University of Lucknow, Lucknow–226007 128

View publication stats