Linear Algebra M.

Thamban Nair

https://textbookfull.com/product/linear-algebra-m-thamban-nair/

★★★★★ 4.8/5.0 (31 reviews) ✓ 82 downloads ■ TOP RATED

"Fantastic PDF quality, very satisfied with download!" - Emma W.

DOWNLOAD EBOOK
 Linear Algebra M. Thamban Nair

TEXTBOOK EBOOK TEXTBOOK FULL

Available Formats

■ PDF eBook Study Guide TextBook

EXCLUSIVE 2025 EDUCATIONAL COLLECTION - LIMITED TIME

INSTANT DOWNLOAD VIEW LIBRARY

Collection Highlights

Linear Algebra Seymour Lipschutz

Basic Linear Algebra Jürgen Müller

Elementary Linear Algebra Ron Larson

Linear Algebra G. Shanker Rao

Elementary Linear Algebra 12th Edition Howard Anton

SF1624 Algebra and Geometry Introduction to Linear Algebra

for Science Engineering Daniel Norman

Linear Algebra Done Right Third Edition Sheldon Axler

Fundamentals of linear algebra and optimization Gallier J.

Elementary Linear Algebra 1st Edition James R. Kirkwood

M. Thamban Nair · Arindama Singh

Linear
Algebra
Linear Algebra
M. Thamban Nair Arindama Singh
•

Linear Algebra

123
M. Thamban Nair Arindama Singh
Department of Mathematics Department of Mathematics
Indian Institute of Technology Madras Indian Institute of Technology Madras
Chennai, Tamil Nadu, India Chennai, Tamil Nadu, India

ISBN 978-981-13-0925-0 ISBN 978-981-13-0926-7 (eBook)

https://doi.org/10.1007/978-981-13-0926-7
Library of Congress Control Number: 2018944333

© Springer Nature Singapore Pte Ltd. 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface

Linear Algebra deals with the most fundamental ideas of mathematics in an abstract
but easily understood form. The notions and techniques employed in Linear
Algebra are widely spread across various topics and are found in almost every
branch of mathematics, more prominently, in Differential Equations, Functional
Analysis, and Optimization, which have wide applications in science and engi-
neering. The ideas and techniques from Linear Algebra have a ubiquitous presence
in Statistics, Commerce, and Management where problems of solving systems of
linear equations come naturally. Thus, for anyone who carries out a theoretical or
computational investigation of mathematical problems, it is more than a necessity to
equip oneself with the concepts and results in Linear Algebra, and apply them with
confidence.

Overview and Goals

This book provides background materials which encompass the fundamental

notions, techniques, and results in Linear Algebra that form the basis for analysis
and applied mathematics, and thereby its applications in other branches of study. It
gives an introduction to the concepts that scientists and engineers of our day use to
model, to argue about, and to predict the behaviour of systems that come up often
from applications. It also lays the foundation for the language and framework for
modern analysis. The topics chosen here have shown remarkable persistence over
the years and are very much in current use.
The book realizes the following goals:
• To introduce to the students of mathematics, science, and engineering the
elegant and useful abstractions that have been created over the years for solving
problems in the presence of linearity
• To help the students develop the ability to form abstract notions of their own and
to reason about them

v
vi Preface

• To strengthen the students’ capability of carrying out formal and rigorous

arguments about vector spaces and maps between them
• To make the essential elements of linear transformations and matrices accessible
to not-so-matured students with a little background in a rigorous way
• To lead the students realize that mathematical rigour in arguing about linear
objects can be very attractive
• To provide proper motivation for enlarging the vocabulary, and slowly take the
students to a deeper study of the notions involved
• To let the students use matrices not as static array of numbers but as dynamic
maps that act on vectors
• To acquaint the students with the language and powers of the operator theoretic
methods used in modern mathematics at present

Organization

Chapter 1 lays the foundation of Linear Algebra by introducing the notions of

vector spaces, subspaces, span, linear independence, basis and dimension, and the
quotient space. It is done at a leisurely pace. Sometimes, the steps have been made
elaborate intentionally so that students would not jump to conclusions and develop
the habit of hand waving. Both finite and infinite dimensional vector spaces are
discussed so that a proper foundation is laid for taking up Functional Analysis as a
natural extension, if the student so desires to pursue in his later years of study.
Chapter 2 introduces linear transformations as structure-preserving maps
between vector spaces. It naturally leads to treating the matrices as linear trans-
formations and linear transformations as matrices. Particular attention is given to
bijective linear transformations and how they act on bases. So, change of bases is
considered by looking at the identity linear transformation. The space of linear
transformations is introduced along with the composition of maps. It raises the issue
of equivalence and similarity of matrices leading to the rank theorem.
Chapter 3 deals with elementary operations in detail. Starting from the com-
putation of rank, elementary operations are used to evaluate determinants, compute
the inverse of a matrix, and solve linear systems. The issue of solvability of linear
systems is treated via linear transformations showing to the students how abstrac-
tion helps.
Chapter 4 brings in the so far neglected notion of direction, or angle between
vectors by introducing inner products. The notion of orthogonality and the neces-
sary elegance and ease it ensues are discussed at length. The geometric notion of an
orthogonal projection is given a lead for solving minimization problems such as the
best approximation of a vector from a subspace and least squares solutions of linear
systems. Constructing a linear functional from the inner product via Fourier
expansion leads to Riesz representation theorem. The existence of the adjoint of a
linear transformation is also shown as an application of Riesz representation.
Preface vii

Chapter 5 asks a question of how and when a linear operator on a vector space
may fix a line while acting on the vectors. This naturally leads to the concepts of
eigenvalues and eigenvectors. The notion of fixing a line is further generalized to
invariant subspaces and generalized eigenvectors. It gives rise to polynomials that
annihilate a linear operator, and the ascent of an eigenvalue of a linear operator.
Various estimates involving the ascent, the geometric, and algebraic multiplicities
of an eigenvalue are derived to present a clear view.
Chapter 6 takes up the issue of representing a linear operator as a matrix by using
the information on its eigenvalues. Starting with diagonalization, it goes for Schur
triangularization, block-diagonalization, and Jordan canonical form characterizing
similarity of matrices.
Chapter 7 tackles the spectral representation of linear operators on inner product
spaces. It proves the spectral theorem for normal operators in a finite-dimensional
setting, and once more, that of self-adjoint operators with somewhat a different
flavour. It also discusses the singular value decomposition and polar decomposition
of matrices that have much significance in application.

Special Features

There are places where the approach has become non-conventional. For example, the
rank theorem is proved even before elementary operations are introduced; the relation
between ascent, geometric multiplicity, and algebraic multiplicity are derived in the
main text, and information on the dimensions of generalized eigenspaces is used to
construct the Jordan form. Instead of proving results on matrices, first a result of the
linear transformation is proved, and then it is interpreted for matrices as a particular
case. Some of the other features are:
• Each definition is preceded by a motivating dialogue and succeeded by one or
more examples
• The treatment is fairly elaborate and lively
• Exercises are collected at the end of the section so that a student is not distracted
from the main topic. The sole aim of these exercises is to reinforce the notions
discussed so far
• Each chapter ends with a section listing problems. Unlike the exercises at the
end of each section, these problems are theoretical, and sometimes unusual and
hard requiring the guidance of a teacher
• It puts emphasis on the underlying geometric idea leading to specific results
noted down as theorems
• It lays stress on using the already discussed material by recalling and referring
back to a similar situation or a known result
• It promotes interactive learning building the confidence of the student
viii Preface

• It uses operator theoretic method rather than the elementary row operations. The
latter is primarily used as a computational tool reinforcing and realizing the
conceptual understanding

Target Audience

This is a textbook primarily meant for a one- or two-semester course at the junior
level. At IIT Madras, such a course is offered to master’s students, at the fourth year
after their schooling, and some portions of this are also offered to undergraduate
engineering students at their third semester. Naturally, the problems at the end of
each chapter are tried by such master’s students and sometimes by unusually bright
engineering students.

Notes to the Instructor

The book contains a bit more than that can be worked out (not just covered) in a
semester. The primary reason is: these topics form a prerequisite for undertaking
any meaningful research in analysis and applied mathematics. The secondary rea-
son is the variety of syllabi followed at universities across the globe. Thus different
courses on Linear Algebra can be offered by giving stress on suitable topics and
mentioning others. The authors have taught different courses at different levels from
it sticking to the core topics.
The core topics include vector spaces, up to dimension (Sects. 1.1–1.5), linear
transformation, up to change of basis (Sects. 2.1–2.5), a quick review of determi-
nant (Sect. 3.5), linear equations (Sect. 3.6), inner product space, up to orthogonal
and orthonormal bases (Sects. 4.1–4.5), eigenvalues and eigenvectors, up to
eigenspaces (Sects. 5.1–5.3), the characteristic polynomial in Sect. 5.5, and diag-
onalizability in Sect. 6.1. Depending on the stress in certain aspects, some of the
proofs from these core topics can be omitted and other topics can be added.

Chennai, India M. Thamban Nair

March 2018 Arindama Singh
Contents

1 Vector Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Vector Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Subspaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 Linear Span . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.4 Linear Independence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.5 Basis and Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.6 Basis of Any Vector Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.7 Sums of Subspaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
1.8 Quotient Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
1.9 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2 Linear Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.1 Linearity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.2 Rank and Nullity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.3 Isomorphisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.4 Matrix Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.5 Change of Basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
2.6 Space of Linear Transformations . . . . . . . . . . . . . . . . . . . . . . . . . 93
2.7 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
3 Elementary Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3.1 Elementary Row Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3.2 Row Echelon Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
3.3 Row Reduced Echelon Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
3.4 Reduction to Rank Echelon Form . . . . . . . . . . . . . . . . . . . . . . . . 128
3.5 Determinant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.6 Linear Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
3.7 Gaussian and Gauss–Jordan Elimination . . . . . . . . . . . . . . . . . . . 147
3.8 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

ix
x Contents

4 Inner Product Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

4.1 Inner Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
4.2 Norm and Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
4.3 Orthogonal and Orthonormal Sets . . . . . . . . . . . . . . . . . . . . . . . . 172
4.4 Gram–Schmidt Orthogonalization . . . . . . . . . . . . . . . . . . . . . . . . 176
4.5 Orthogonal and Orthonormal Bases . . . . . . . . . . . . . . . . . . . . . . . 182
4.6 Orthogonal Complement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
4.7 Best Approximation and Least Squares . . . . . . . . . . . . . . . . . . . . 190
4.8 Riesz Representation and Adjoint . . . . . . . . . . . . . . . . . . . . . . . . 196
4.9 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
5 Eigenvalues and Eigenvectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
5.1 Existence of Eigenvalues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
5.2 Characteristic Polynomial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
5.3 Eigenspace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
5.4 Generalized Eigenvectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
5.5 Two Annihilating Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . 234
5.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
6 Block-Diagonal Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
6.1 Diagonalizability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
6.2 Triangularizability and Block-Diagonalization . . . . . . . . . . . . . . . 248
6.3 Schur Triangularization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
6.4 Jordan Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
6.5 Jordan Normal Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
6.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
7 Spectral Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
7.1 Playing with the Adjoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
7.2 Projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
7.3 Normal Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
7.4 Self-adjoint Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
7.5 Singular Value Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . 308
7.6 Polar Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
7.7 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
About the Authors

M. Thamban Nair is a professor of mathematics at the Indian Institute of

Technology Madras, Chennai, India. He completed his Ph.D. at the Indian Institute
of Technology Bombay, Mumbai, India, in 1986. His research interests include
functional analysis and operator theory, specifically spectral approximation, the
approximate solution of integral and operator equations, regularization of inverse
and ill-posed problems. He has published three books, including a textbook,
Functional Analysis: A First Course (PHI Learning), and a text-cum-monograph,
Linear Operator Equations: Approximation and Regularization (World Scientific),
and over 90 papers in reputed journals and refereed conference proceedings. He has
guided six Ph.D. students and is an editorial board member of the Journal of
Analysis and Number Theory, and Journal of Mathematical Analysis. He is a life
member of academic bodies such as Indian Mathematical Society and Ramanujan
Mathematical Society.

Arindama Singh is a professor of mathematics at the Indian Institute of

Technology Madras, Chennai, India. He completed his Ph.D. at the Indian Institute
of Technology Kanpur, India, in 1990. His research interests include knowledge
compilation, singular perturbation, mathematical learning theory, image processing,
and numerical linear algebra. He has published five books, including Elements of
Computation Theory (Springer), and over 47 papers in reputed journals and ref-
ereed conference proceedings. He has guided five Ph.D. students and is a life
member of many academic bodies, including Indian Society for Industrial and
Applied Mathematics, Indian Society of Technical Education, Ramanujan
Mathematical Society, Indian Mathematical Society, and The Association of
Mathematics Teachers of India.

xi
Chapter 1
Vector Spaces

1.1 Vector Space

A vector in the plane is an object with certain length and certain direction. Con-
ventionally it is represented by an arrow with an initial point and an endpoint; the
endpoint being the arrow head. We work with plane vectors by adding them, sub-
tracting one from the other, and by multiplying them with a number. We see that the
plane vectors have a structure, which is revealed through the two operations, namely
addition and multiplication by a number, also called scalar multiplication. These
operations can be seen in an alternate way by identifying the vectors with points in
the plane. The identification goes as follows.
Since only length and direction matter and not exactly the initial or the endpoints,
we may think of each vector having its initial point at the origin. The endpoint
can then be identified with the vector itself. With O as the origin with Cartesian
coordinates (0, 0) and P as the point with Cartesian coordinates (a, b), the vector
−→
O P is identified with the point (a, b) in the plane

R2 = {(α, β) : α ∈ R, β ∈ R}.

Then the familiar parallelogram law for addition of vectors translates to component-
wise addition. If u, v are vectors with initial point (0, 0) and endpoints (a, b) and
(c, d), respectively, then the vector u + v has initial point (0, 0) and endpoint (a +
c, b + d). Similarly, for a real number α, the vector αu has the initial point (0, 0)
and endpoint (αa, αb).

© Springer Nature Singapore Pte Ltd. 2018 1

M. T. Nair and A. Singh, Linear Algebra,
https://doi.org/10.1007/978-981-13-0926-7_1
2 1 Vector Spaces

Thus, (−1) u, which equals (−a, −b), represents the additive inverse −u of the
vector u; the direction of −u is opposite to that of u. Now, the plane is simply viewed
as a set of all plane vectors.
Similarly, in the three-dimensional space, you may identify a vector with a point
by first translating the vector to have its initial point as the origin and its arrow
head as the required point. The sum of two vectors in three dimensions gives rise
to the component-wise sum of two points. A real number α times a vector gives
a vector whose components are multiplied by α. That is, if u = (a1 , b1 , c1 ) and
v = (a2 , b2 , c2 ), then

u + v = (a1 + a2 , b1 + b2 , c1 + c2 ), αu = (αa1 , αb1 , αc1 ).

Notice that the zero vector, written as 0, is identified with the point (0, 0, 0), and the
vector −u = (−a1 , −b1 , −c1 ) satisfies u + (−u) = 0.
The notion of a vector space is an abstraction of the familiar set of vectors in two
or three dimensions. The idea is to keep the familiar properties of addition of vectors
and multiplication of a vector by a scalar. The set of scalars can be any field. For
obtaining interesting geometrical results, we may have to restrict the field of scalars.
In this book, the field F denotes either the field R of real numbers or the field C of
complex numbers.

Definition 1.1 A vector space over F is a nonempty set V along with two operations,
namely
(a) addition, which associates each pair (x, y) of elements x, y ∈ V with a unique
element in V , denoted by x + y, and
(b) scalar multiplication, which associates each pair (α, x), for α ∈ F and x ∈ V ,
with a unique element in V , denoted by αx,
satisfying the following conditions:
(1) For all x, y ∈ V, x + y = y + x.
(2) For all x, y, z ∈ V, (x + y) + z = x + (y + z).
(3) There exists an element in V , called a zero vector, denoted by 0, such that for
all x ∈ V, x + 0 = x.
1.1 Vector Space 3

(4) For each x ∈ V , there exists an element in V , denoted by −x, and called an
additive inverse of x, such that x + (−x) = 0.
(5) For all α ∈ F and for all x, y ∈ V, α(x + y) = αx + αy.
(6) For all α, β ∈ F and for all x ∈ V, (α + β)x = αx + βx.
(7) For all α, β ∈ F and for all x ∈ V, (αβ)x = α(βx).
(8) For all x ∈ V, 1x = x.

Elements of F are called scalars, and elements of a vector space V are called vec-
tors. A vector space V over R is called a real vector space, and a vector space over C
is called a complex vector space. As a convention, we shorten the expression “a vec-
tor space over F” to “a vector space”. We denote vectors by the letters u, v, w, x, y, z
with or without subscripts, and scalars by the letters a, b, c, d, α, β, γ , δ with or
without subscripts.
You have ready-made examples of vector spaces. The plane

R2 = {(a, b) : a, b ∈ R}

and the familiar three-dimensional space

R3 = {(a, b, c) : a, b, c ∈ R}

are real vector spaces. Notice that R is a vector space over R, and C is a vector
space over C as well as over R. Before presenting more examples of vector spaces,
we observe some subtleties about the conditions (3) and (4) in Definition 1.1. It is
unusual to write a particular symbol such as 0 for all zero vectors. It is also unusual to
write −x for all additive inverses of x. The philosophical hurdle will be over once we
prove that a zero vector is unique and an additive inverse of a vector is also unique.

Theorem 1.2 In any vector space the following statements are true:
(1) There exists exactly one zero vector.
(2) Each vector has exactly one additive inverse.

Proof Let V be a vector space.

(1) Suppose 0 and 0̃ in V are zero vectors. Then for all x ∈ V, x + 0 = x and
x + 0̃ = x. Using the condition (1) in Definition 1.1, we have

0̃ = 0̃ + 0 = 0 + 0̃ = 0.

(2) Let x ∈ V . Let x  and 

x be additive inverses of x. Let 0 be the zero vector.
Then x + x  = 0 and x + x = 0. Therefore,


x = x + (x + x  ) = (
x +0 = x + x) + x  = (x + 
x ) + x  = 0 + x  = x  + 0 = x . 

Theorem 1.2 justifies the use of the symbols 0 for the zero vector and −x for the
additive inverse of the vector x. Of course, we could have used any other symbol,
4 1 Vector Spaces

say, θ for the zero vector and x  for the additive inverse of x; but the symbols 0 and
−x follow the custom. Note that −0 = 0. We also write y − x instead of y + (−x)
for all vectors x and y.
Notice the double meanings used in Definition 1.1. The addition of scalars as well
as of vectors is denoted by the same symbol +, and the multiplication of scalars
as well as of a vector with a scalar is written by just concatenating the elements.
Similarly, 0 denotes the zero vector as well as the scalar zero. Even the notation for
the additive inverse of vector x is −x; just the way we write −α for the additive
inverse of a scalar α. You should get acquainted with the double meanings.
It is easy to check that in every vector space V over F,

0 + x = x, x + (y − x) = y for all x, y ∈ V.

Every vector space contains at least one element, the zero vector. On the other
hand, the singleton {0} is a vector space; it is called the zero space or the trivial
vector space. In general, we will be concerned with nonzero vector spaces, which
contain nonzero elements. A nonzero vector space is also called a nontrivial vector
space.
In a vector space, addition of two elements is allowed. This is generalized by
induction to a sum of any finite number of vectors. But an infinite sum of vectors is
altogether a different matter; it requires analytic notions such as convergence.

Example 1.3 In the following, the sets along with the specified addition and scalar
multiplication are vector spaces. (Verify.)
(1) Consider the set Fn of all n-tuples of scalars, that is,

Fn := {(a1 , . . . , an ) : a1 , . . . , an ∈ F}.

We assume that two elements in Fn are equal when their respective components are
equal. For x = (a1 , . . . , an ), y = (b1 , . . . , bn ) ∈ Fn , and α ∈ F, define the addition
and scalar multiplication component-wise, that is,

x + y := (a1 + b1 , . . . , an + bn ), αx := (αa1 , . . . , ααn ).

Then Fn is a vector space with

0 = (0, . . . , 0) and − (a1 , . . . , an ) = (−a1 , . . . , −an ).

(2) We use the notation Fm×n for the set of all m × n matrices with entries from F.
A matrix A ∈ Fm×n is usually written as
⎡ ⎤
a11 · · · a1n
⎢ .. .. ⎥ ,
A=⎣ . . ⎦
am1 · · · amn
1.1 Vector Space 5

or as A = [ai j ] for short, with ai j ∈ F for i = 1, . . . , m; j = 1, . . . , n. The number

ai j which occurs at the entry in ith row and jth column is referred to as the (i, j)th
entry of the matrix [ai j ]. For A = [ai j ] and B = [bi j ] in Fm×n , and α ∈ F, we define

A + B = [ai j + bi j ] ∈ Fm×n , α A = [αai j ] ∈ Fm×n .

We say that two matrices are equal when their respective entries are equal. That is,
for A = [ai j ] and B = [bi j ], we write A = B if and only if ai j = bi j . With these
operations of addition and scalar multiplication, Fm×n becomes a vector space over
F. The zero vector in Fm×n is the zero matrix, i.e. the matrix with all entries 0, and
the additive inverse of A = [ai j ] ∈ Fm×n is the matrix −A := [−ai j ].
(3) For n ∈ {0, 1, 2, . . .}, let Pn (F) denote the set of all polynomials (in the variable
t) of degree at most n, with coefficients in F. That is, x ∈ Pn (F) if and only if x is
of the form
x = a0 + a1 t + · · · + an t n

for some scalars a0 , a1 , . . . , an . Here, we assume that a scalar is a polynomial of

degree 0. Further, two polynomials are considered equal when the coefficients of
respective powers of t are equal. That is,

a0 + a1 t + · · · + an t n = b0 + b1 t + · · · + bn t n if and only if ai = bi for i = 0, . . . , n.

Addition and scalar multiplication on Pn (F) are defined as follows. For

x = a0 + a1 t + · · · + an t n , y = b0 + b1 t + · · · + bn t n in Pn (F), and α ∈ F,

x + y := (a0 + b0 ) + (a1 + b1 )t + · · · + (an + bn )t n ,

αx := αa0 + αa1 t + · · · + αan t n .

The zero vector in Pn (F) is the polynomial with all its coefficients zero, and

−(a0 + a1 t + · · · + an t n ) = −a0 − a1 t − · · · − an t n .

Then Pn (F) is a vector space.

(4) Let P(F) = ∪∞ n=0 Pn (F), the set of all polynomials with coefficients in F. That
is, x ∈ P(F) if and only if x = a0 + a1 t + · · · + an t n for some n ∈ {0, 1, 2, . . .} and
for some scalars a0 , . . . , an ∈ F.
If x, y ∈ P(F), then x ∈ Pm (F) and y ∈ Pn (F) for some m, n. So, x, y ∈ Pk (F),
with k = max {m, n}. The equality relation, addition, and scalar multiplication are
defined as in (3). Then P(F) is a vector space.
(5) Let V be the set of all sequences of scalars. A sequence whose nth term is an is
written as (an ). Two sequences are considered equal if and only if their respective
terms are equal, that is, (an ) = (bn ) if and only if an = bn for each n ∈ N. For
(an ), (bn ) ∈ V , and α ∈ F, define
6 1 Vector Spaces

(an ) + (bn ) := (an + bn ), α(an ) := (αan ).

That is,

(a1 , a2 , a3 , . . .) + (b1 , b2 , b3 , . . .) = (a1 + b1 , a2 + b2 , a3 + b3 , . . .),

α(a1 , a2 , a3 . . . .) = (αa1 , αa2 , αa3 , . . .).

With this addition and scalar multiplication, V is a vector space, where its zero vector
is the sequence with each term as zero, and −(an ) = (−an ). This space is called the
sequence space and is denoted by F∞ .
(6) Let S be a nonempty set. Let V be a vector space over F. Let F(S, V ) be the set of
all functions from S into V . As usual, x = y for x, y ∈ F(S, V ) when x(s) = y(s)
for each s ∈ S. For x, y ∈ F(S, V ) and α ∈ F, define x + y and αx point-wise; that
is,
(x + y)(s) := x(s) + y(s), (αx)(s) := αx(s) for s ∈ S.

Let the functions 0 and −x in F(S, V ) be defined by

0(s) = 0, (−x)(s) = −x(s) for s ∈ S.

Then F(S, V ) is a vector space over F with the zero vector as 0 and the additive
inverse of x as −x. We sometimes refer to this space as a function space. 

Comments on Notation: Pn (R) denotes the real vector space of all polynomials of
degree at most n with real coefficients. Pn (C) denotes the complex vector space of
all polynomials of degree at most n with complex coefficients. Similarly, P(R) is the
real vector space of all polynomials with real coefficients, and P(C) is the complex
vector space of all polynomials with complex coefficients. Note that C is also a vector
space over R. Similarly, Pn (C) and P(C) are vector spaces over R. More generally,
if V is a complex vector space, then it is also a real vector space. If at all we require
to regard any vector space over C also as a vector space over R, we will specifically
mention it.
As particular cases of Example 1.3(2), (Read: Example 1.3 Part 2) we have the
vector spaces Fm×1 , the set of all column vectors of size m, and F1×n , the set of
all row vectors of size n. To save space, we use the transpose notation in writing
a column vector. That is, a column vector v of size n with its entries a1 , . . . , an is
written as ⎡ ⎤
a1
⎢ .. ⎥
⎣ . ⎦ or as [a1 · · · an ] .
T

an

By putting the superscript T over a row vector v we mean that the column vector is
obtained by taking transpose of the row vector v. When the column vectors are writ-
ten by the lower case letters u, v, w, x, y, z with or without subscripts (sometimes
1.1 Vector Space 7

superscripts), the corresponding row vectors will be written with the transpose nota-
tion, that is, as u T , v T . Further, we will not distinguish between the square brackets
and the parentheses. Usually, we will write a row vector with parentheses. Thus, we
will not distinguish between

[a1 · · · an ] and (a1 , . . . , an ).

Thus we regard Fn same as F1×n . We may recall that by taking transpose of a matrix
in Fm×n , we obtain a matrix in Fn×m . That is, if A = [ai j ] ∈ Fm×n , then

A T := [b ji ] ∈ Fn×m with b ji = ai j for i = 1, . . . , m; j = 1, . . . , n.

Many vector spaces can be viewed as function spaces. For example, with S = N
and V = F, we obtain the sequence space of Example 1.3(5). With S = {1, . . . , n}
and V = F, each function in F(S, F) can be specified by an n-tuple of its function
values. Therefore, the vector space F({1, . . . , n}, F) can be viewed as Fn and also
as Fn×1 . Some more examples of function spaces follow.

Example 1.4 (1) Let I be an interval, and let C(I, R) denote the set of all real-valued
continuous functions defined on I . For x, y ∈ C(I, R) and α ∈ R, define x + y and
αx point-wise as in Example 1.3(6).
The functions x + y and αx are in C(I, R). Then C(I, R) is a real vector space
with the zero element as the zero function and the additive inverse of x ∈ C(I, R) as
the function −x defined by (−x)(t) = −x(t) for all t ∈ I.
(2) Let R([a, b], R) denote the set of all real-valued Riemann integrable functions
on [a, b]. Define addition and scalar multiplication point-wise, as in Example 1.3(6).
From the theory of Riemann integration, it follows that if x, y ∈ R([a, b], R) and
α ∈ R, then x + y, αx ∈ R([a, b], R). It is a real vector space.
(3) For k ∈ N, let C k ([a, b], F) denote the set of all functions x from [a, b] to F such
that the kth derivative x (k) exists and is continuous on [a, b].
Define addition and scalar multiplication point-wise, as in Example 1.3(6). Then
C k ([a, b], F) is a vector space. Notice that

C k ([a, b], R) ⊆ C([a, b], R) ⊆ R([a, b], R) ⊆ F([a, b], R). 

Example 1.5 Let V1 , . . . , Vn be vector spaces over F. Consider the Cartesian product

V = V1 × · · · × Vn = {(x1 , . . . , xn ) : x1 ∈ V1 , . . . , xn ∈ Vn }.

Define addition and scalar multiplication on V by

(x1 , . . . , xn ) + (y1 , . . . , yn ) := (x1 + y1 , . . . , xn + yn ),

α(x1 , . . . , xn ) := (αx1 , . . . , αxn ).

8 1 Vector Spaces

In V , take the zero vector as (0, . . . , 0) and −(x1 , . . . , xn ) = (−x1 , . . . , −xn ).

Here, the addition in the expression xi + yi is the addition operation defined in Vi and
the scalar multiplication in αxi is the scalar multiplication defined in Vi . Similarly,
in (0, . . . , 0), the ith component is the zero vector of Vi and −xi in (−x1 , . . . , −xn )
is the additive inverse of xi in Vi .
With these operations, V is a vector space. We call it the product space of
V1 , . . . , Vn . 

To illustrate Example 1.5, consider V1 = R and V2 = P1 (R). Then V1 × V2 con-

sists of vectors of the form (a, α + βt), and the operations of addition and scalar
multiplication are defined by

(a1 , α1 + β1 t) + (a2 , α2 + β2 t) = (a1 + a2 , (α1 + α2 ) + (β1 + β2 )t),

c(a, α + βt) = (ca, cα + cβt).

Similarly, the space Fn is a product space with each Vi as F.

Some easy consequences of Definition 1.1 are listed in the following theorem.

Theorem 1.6 Let V be a vector space over F. Let x, y, z ∈ V and let α, β ∈ F.

(1) 0x = 0.
(2) α0 = 0.
(3) (−1)x = −x.
(4) −(−x) = x.
(5) If x + z = y + z, then x = y.
(6) If αx = 0, then α = 0 or x = 0.
(7) If α = β and x = 0, then αx = βx.

Proof (1) 0x = (0 + 0)x = 0x + 0x. Adding −0x, we have 0 = 0x.

(2) α0 = α(0 + 0) = α0 + α0. Adding −α0, we obtain 0 = α0.
(3) x + (−1)x = (1 + (−1))x = 0x = 0, by (1). Adding −x we get (−1)x = −x.
(4) From (3), it follows that −(−x) = (−1)(−1)x = x.
(5) Suppose that x + z = y + z. Then (x + z) + (−z) = (y + z) + (−z). But
(x + z) + (−z) = x + (z + (−z)) = x + 0 = x. Similarly, (y + z) + (−z) = y.
Therefore, x = y.
(6) Suppose that αx = 0. If α = 0, then α −1 ∈ F. Multiplying α −1 , we have α −1 αx =
α −1 0. That is, x = 0.
(7) Suppose that α = β and x = 0. If αx = βx, then (α − β)x = 0. By (6), x = 0.
It leads to a contradiction. Hence, αx = βx. 
1.1 Vector Space 9

Theorem 1.6 allows us to do algebraic manipulation of vectors like scalars as long

as addition is concerned. Thus we abbreviate x + (−y) to x − y. However, there is
a big difference: vectors cannot be multiplied or raised to powers, in general. Look
at the proof of Theorem 1.6(6). We used α −1 instead of x −1 .
It follows from Theorem 1.6(7) that every nontrivial vector space contains infinitely
many vectors.
Exercises for Sect. 1.1
In the following a set V , a field F, which is either R or C, and the operations of
addition and scalar multiplication are given. Check whether V is a vector space over
F, with these operations.
1. With addition and scalar multiplication as in R2 ,
(a) V = {(a, b) ∈ R2 : 2a + 3b = 0}, F = R.
(b) V = {(a, b) ∈ R2 : a + b = 1}, F = R.
(c) V = {(a, b) ∈ R2 : ab = 0}, F = R.
2. V = R2 , F = R, with addition as in R2 , and scalar multiplication as given by the
following: for (a, b) ∈ V, α ∈ R,
(a) α(a, b) := (a, 0).
(b) α(a, b) := (b, αa).
(c) α(a, b) := (αa, −αb).
(0, 0) if α = 0
(d) α(a, b) :=
(αa, b/α) if α = 0.
3. V = {x ∈ R : x > 0}, F = R, and for x, y ∈ V, α ∈ R, x + y := x y, αx :=
x α . The operations on the left are defined by the known operations on the right.
4. V = {x ∈ R : x ≥ 0}, F = R, and for x, y ∈ V, α ∈ R, x + y := x y, αx :=
|α|x.
5. V = C2 , F = C, and for x = (a, b), y = (c, d), α ∈ C,

x + y := (a + 2c, b + 3d), αx := (αa, αb).

6. V is the set of all polynomials of degree 5 with real coefficients, F = R, and the
operations are the addition and scalar multiplication of polynomials.
7. S is a nonempty set, s ∈ S, V is the set of all functions f : S → R with f (s) = 0,
F = R, and the operations are the addition and scalar multiplication of functions.
8. V is the set of all functions f : R → C satisfying f (−t) = f (t), F = R, and the
operations are the addition and scalar multiplication of functions.
9. V = {x}, where x is some symbol, and addition and scalar multiplication are
defined as x + x = x, αx = x for all α ∈ F.
10 1 Vector Spaces

1.2 Subspaces

A subset of a vector space may or may not be a vector space. It will be interesting
if a subset forms a vector space over the same underlying field and with the same
operations of addition and scalar multiplication inherited from the given vector space.
If U is a subset of a vector space V (over the field F), then the operations of addition
and scalar multiplication in U inherited from V are defined as follows:
Let x, y ∈ U, α ∈ F. Consider x, y as elements of V. The vector x + y in V
is the result of the inherited addition of x and y in U. Similarly, the vector αx
in V is the result of the inherited scalar multiplication of α with x in U.
In order that the operations of addition (x, y) → x + y and scalar multiplication
(α, x) → αx are well-defined operations on U , we require the vectors x + y and
αx to lie in U . This condition is described by asserting that U is closed under the
inherited operations.

Definition 1.7 Let V be a vector space over F. A nonempty subset U of V is called

a subspace of V if U is a vector space over F with respect to the operations of
addition and scalar multiplication inherited from V.

To show that a nonempty subset U of a vector space V is a subspace, one must

first verify that U is closed under the inherited operations. The closure conditions
can be explicitly stated as follows:
For all x, y ∈ U and for each α ∈ F, x + y ∈ U and αx ∈ U.
Here, of course, the operations are the operations in the given vector space V over
F. Surprisingly, these closure conditions are enough for establishing that a subset is
a subspace, as the following theorem shows.

Theorem 1.8 Let U be a nonempty subset of a vector space V . Then U is a subspace

of V if and only if U is closed under addition and scalar multiplication inherited
from V.

Proof If U is a subspace of V , then x + y ∈ U and αx ∈ U for all x, y ∈ U and for

all α ∈ F.
Conversely, suppose that x + y ∈ U and αx ∈ U for all x, y ∈ U and for all
α ∈ F. Since U is nonempty, let u ∈ U. By Theorem 1.6, 0 = 0 u ∈ U , and for each
x ∈ U, − x = (−1)x ∈ U. Since U ⊆ V , 0 acts as the zero vector of U , and −x acts
as the additive inverse of x in U. Therefore, the conditions (3)–(4) in Definition 1.1
are satisfied for U . The remaining conditions hold since elements of U are elements
of V as well. 

Notice that the closure conditions in Theorem 1.8 can be replaced by the following
single condition:
For each scalar α ∈ F and for all x, y ∈ U , x + αy ∈ U.
1.2 Subspaces 11

We may also infer from the proof of Theorem 1.8 that

if U is a subspace of a vector space V , then the zero vector of U is the same as

the zero vector of V , and for each u ∈ U , its additive inverse −u as an element
of U is the same as −u in V.

Therefore, in order that U is a subspace of V , the zero vector of V must be in U.

Example 1.9 (1) Let U = {(a, b) ∈ R2 : b = 0} ⊆ R2 . That is, U = {(a, 0) : a ∈

R}. Addition and scalar multiplication, which are defined component-wise in R2 ,
are also operations on U since for all a, b, c, α ∈ R,

(a, 0) + (b, 0) = (a + b, 0) ∈ U, α(c, 0) = (αc, 0) ∈ U.

By Theorem 1.8, U is a subspace of R2 . Notice that the zero vector of U is the same
(0, 0) and −(a, 0) = (−a, 0) as in R2 .
(2) The set U = {(a, b) ∈ R2 : 2a + 3b = 0} is a subspace of R2 (Verify).
(3) Let Q denote the set of√all rational numbers. Q is not a subspace of the real vector
space R since 1 ∈ Q but 2 · 1 ∈ / Q. Similarly, Q2 is not a subspace of R2 .
(4) Consider C as a complex vector space. Let U = {a + i 0 : a ∈ R}. We see that
1 ∈ U but i · 1 = i ∈
/ U. Therefore, U is not a subspace of C.
However, if we consider C as real vector space, then U is a subspace of C. In this
sense, U = R is a subspace of the real vector space C.
(5) Consider the spaces Pm (F) and Pn (F), where m ≤ n. Each polynomial of degree
at most m is also a polynomial of degree at most n. Thus, Pm (F) ⊆ Pn (F). Further,
Pm (F) is closed under the operations of addition and scalar multiplication inherited
from Pn (F). So, Pm (F) is a subspace of Pn (F) for any m ≤ n.
Also, for each n ∈ N, Pn (F) is a subspace of P(F).
(6) In Examples 1.4(1)–(2), both C([a, b], R) and R([a, b], R) are vector spaces.
Since C([a, b], R) ⊆ R([a, b], R) and the operations of addition and scalar multipli-
cation in C([a, b], R) are inherited from R([a, b], R), we conclude that C([a, b], R)
is a subspace of R([a, b], R).
(7) Consider C k ([a, b], F) of Example 1.4(3). For all α ∈ F, x, y ∈ C k ([a, b], F), we
have x + y ∈ C k ([a, b], F) and αx ∈ C k ([a, b], F). By Theorem 1.8, C k ([a, b], F) is
a subspace of C([a, b], F).
(8) Given α1 , . . . , αn ∈ F, U = {(b1 , . . . , bn ) ∈ Fn : α1 b1 + · · · + αn bn = 0} is a
subspace of Fn . When (α1 , . . . , αn ) is a nonzero n-tuple, the subspace U is a hyper-
plane passing through the origin in n dimensions. This terminology is partially bor-
rowed from the case of F = R and n = 3, when the subspace {(b1 , b2 , b3 ) ∈ R3 :
α1 b1 + α2 b2 + α3 b3 = 0, α1 , α2 , α3 ∈ R} of R3 is a plane passing through the ori-
gin. However,

W = {(b1 , . . . , bn ) ∈ Fn : α1 b1 + · · · + αn bn = 1, α1 , . . . , αn ∈ F}

is not a subspace of Fn , since (0, . . . , 0) ∈

/ W.
12 1 Vector Spaces

(9) Let P([a, b], R) be the vector space P(R) where each polynomial is considered as
a function from [a, b] to R. Then the space P([a, b], R) is a subspace of C k ([a, b], R)
for each k ≥ 1. 

Consider two planes passing through the origin in R3 . Their intersection is a

straight line passing through the origin. In fact, intersection of two subspaces is a
subspace. We prove a more general result.

Theorem 1.10 Let C be any collection of subspaces of a vector space V. Let U be

the intersection of all subspaces in C. Then U is a subspace of V.

Proof Let x, y ∈ U and let α ∈ F. Then x, y ∈ W for each W ∈ C. Since W is a

subspace of V , αx ∈ W and x + y ∈ W for each W ∈ C. Then αx ∈ U and x + y ∈
U. By Theorem 1.8, U is a subspace of V. 

In contrast, union of two subspaces need not be a subspace. For, consider

U := {(a, b) ∈ R2 : b = a}, V := {(a, b) ∈ R2 : b = 2a}

as subspaces of R2 . Here, (1, 1), (1, 2) ∈ U ∪ V but (1, 1) + (1, 2) = (2, 3) ∈

/U∪
V . Hence U ∪ V is not a subspace of R2 .
Union of two subspaces can fail to be a subspace since addition of a vector from
one with a vector from the other may not be in the union. What happens to the set of
all vectors of the form x + y, where x is a vector from one subspace and y is from
another subspace? To answer this question, we introduce the notion of a (finite) sum
of subsets of a vector space.

Definition 1.11 Let S1 , . . . , Sn be nonempty subsets of a vector space V. Their sum

is defined by

S1 + · · · + Sn := {x1 + · · · + xn ∈ V : xi ∈ Si , i = 1, . . . , n}.

As expected, sum of two subspaces is a subspace. And, the proof can be general-
ized easily to any finite sum of subspaces.

Theorem 1.12 Let V1 , . . . , Vn be subspaces of a vector space V. Then V1 + · · · + Vn

is a subspace of V.
n n
Proof Let x, y ∈ V1 + · · · + Vn and let α ∈ F. Then, x = i=1 xi and y = i=1 yi
for some xi , yi ∈ Vi , i = 1, . . . , n. Since each Vi is a subspace of V , xi + yi ∈ Vi
and αxi ∈ Vi . Then
n n n
x+y= i=1 xi + i=1 yi = i=1 (x i + yi ) ∈ V1 + · · · + Vn ,
n n
αx = α i=1 xi = i=1 (αx i ) ∈ V1 + · · · + Vn .

Therefore, V1 + · · · + Vn is a subspace of V. 
1.2 Subspaces 13

Example 1.13 Consider the planes:

V1 = {(a, b, c) ∈ R3 : a + b + c = 0}, V2 = {(a, b, c) ∈ R3 : a + 2b + 3c = 0}.

Both V1 and V2 are subspaces of R3 . Their intersection is the subspace

V1 ∩ V2 = {(a, b, c) ∈ R3 : a + b + c = 0 = a + 2b + 3c}.

The condition a + b + c = 0 = a + 2b + 3c is equivalent to b = −2a, c = a.

Hence
V1 ∩ V2 = {(a, −2a, a) : a ∈ R}

which is a straight line through the origin. Both V1 ∩ V2 and V1 + V2 are subspaces
of R3 . In this case, we show that V1 + V2 = R3 . For this, it is enough to show that
R3 ⊆ V1 + V2 . It requires to express any (a, b, c) ∈ R3 as (a1 + a2 , b1 + b2 , c1 + c2 )
for some (a1 , b1 , c1 ) ∈ V1 and (a2 , b2 , c2 ) ∈ V2 . This demands determining the six
unknowns a1 , b1 , c1 , a2 , b2 , c2 from the five linear equations

a1 + a2 = a, b1 + b2 = b, c1 + c2 = c, a1 + b1 + c1 = 0, a2 + 2b2 + 3c2 = 0.

It may be verified that with

a1 = −a − 2b − 2c, b1 = a + 2b + c, c1 = c
a2 = 2a + 2b + 2c, b2 = −a − b − c, c2 = 0,

the five equations above are satisfied. Thus, (a1 , b1 , c1 ) ∈ V1 , (a2 , b2 , c2 ) ∈ V2 , and
(a, b, c) = (a1 + b1 + c1 ) + (a2 + b2 + c2 ) as desired. 

Exercises for Sect. 1.2

1. In the following, check whether the given subset U is a subspace of V. Assume

the usual operations of addition and scalar multiplication along with a suitable
field R or C.
(a) V = R2 , U is any straight line passing through the origin.
(b) V = R2 , U = {(a, b) : b = 2a − 1}.
(c) V = R3 , U = {(a, b, c) : 2a − b − c = 0}.
(d) V = P3 (R), U = {a + bt + ct 2 + dt 3 : c = 0}.
(e) V = P3 (C), U = {a + bt + ct 2 + dt 3 : a = 0}.
(f) V = P3 (C), U = {a + bt + ct 2 + dt 3 : b + c + d = 0}.
(g) V = P3 (R), U = { p ∈ V : p(0) = 0}.
(h) V = P3 (C), U = { p ∈ V : p(1) = 0}.
(i) V = P3 (C), U = {a + bt + ct 2 + dt 3 : a, b, c, d integers }.
(j) V = C([−1, 1], R), U = { f ∈ V : f is an odd function}.
(k) V = C([0, 1], R), U = { f ∈ V : f (t) ≥ 0 for all t ∈ [0, 1]}.
14 1 Vector Spaces

(l) V = C k [a, b], for k ∈ N, U = P[a, b], the set of all polynomials considered
as functions on [a, b].
(m) V = C([0, 1], R), U = { f ∈ V : f is differentiable}.
2. For α ∈ F, let Vα = {(a, b, c) ∈ F3 : a + b + c = α}. Show that Vα is a subspace
of F3 if and only if α = 0.
3. Give an example of a nonempty subset of R2 which is closed under addition
and under additive inverse (i.e. if u is in the subset, then so is −u), but is not a
subspace of R2 .
4. Give an example of a nonempty subset of R2 which is closed under scalar mul-
tiplication but is not a subspace of R2 .
5. Suppose U is a subspace of V and V is a subspace of W. Show that U is a
subspace of W.
6. Give an example of subspaces of C3 whose union is not a subspace of C3 .
7. Show by a counter-example that if U + W = U + X for subspaces U, W, X of
V , then W need not be equal to X.
8. Let m ∈ N. Does the set {0} ∪ {x ∈ P(R) : degree of x is equal to m} form a
subspace of P(R)?
9. Prove that the only nontrivial proper subspaces of R2 are straight lines passing
through the origin.
10. Let U = {(a, b) ∈ R2 : a = b}. Find a subspace V of R2 such that U + V = R2
and U ∩ V = {(0, 0)}. Is such a V unique?
11. Let U be the subspace of P(F) consisting of all polynomials of the form at 3 + bt 7
for a, b ∈ F. Find a subspace V of P(F) such that U + V = P(F) and U ∩ V =
{0}.
12. Let U and W be subspaces of a vector space V. Prove the following:
(a) U ∪ W = V if and only if U = V or W = V.
(b) U ∪ W is a subspace of V if and only if U ⊆ W or W ⊆ U.
13. Let U = {A ∈ Fn×n : A T = A} and let W = {A ∈ Fn×n : A T = −A}. Matrices
in U are called symmetric matrices, and matrices in W are called skew-symmetric
matrices. Show that U and W are subspaces of Fn×n , Fn×n = U + W , and U ∩
W = {0}.

1.3 Linear Span

The sum of vectors x1 , . . . , xn can be written as x1 + · · · + xn , due to Property (2)

of vector addition in Definition 1.1. We may also multiply the vectors with scalars
and then take their sum. That is, we write the sum
n
(α1 x1 ) + · · · + (αn xn ) as α1 x1 + · · · + αn xn , and also as αi xi .
i=1
1.3 Linear Span 15

If the jth term in α1 x1 + · · · + αn xn is absent, that is, if we want to consider

the sum (α1 x1 + · · · + αn xn ) − α j x j for some j ∈ {1, . . . , n}, then we write it as
n
i=1, i= j αi x i and also as

α1 x1 + · · · + α j−1 x j−1 + α j+1 x j+1 + · · · αn xn ,

with the understanding that if j = 1, then the above sum is α2 x2 + · · · + αn xn , and

if j = n, then the sum is equal to α1 x1 + · · · + αn−1 xn−1 .

Definition 1.14 Let V be a vector space.

(a) A vector v ∈ V is called a linear combination of vectors u 1 , . . . , u n in V if
v = α1 u 1 + · · · + αn u n for some scalars α1 , . . . , αn .
(b) The linear span or the span of any nonempty subset S of V is the set of all
linear combinations of finite number of vectors from S; it is denoted by span(S).
We define span(∅) as {0}.

In view of the above definition, for any nonempty subset S of V , we have

x ∈ span(S) if and only if there exist n ∈ N, vectors u 1 , . . . , u n in S, and scalars

α1 , . . . , αn ∈ F such that x = α1 u 1 + · · · + αn u n .

It follows that

span(S) = {α1 u 1 + · · · + αn u n : n ∈ N, α1 , . . . , αn ∈ F, u 1 , . . . , u n ∈ S}.

Moreover, for u 1 , . . . , u n in V ,

span({u 1 , . . . , u n }) = {α1 u 1 + · · · + αn u n : α1 , . . . , αn ∈ F}.

We also write span({u 1 , . . . , u n }) as span{u 1 , . . . , u n }. Informally, we say that

span(S) is the set of all linear combinations of elements of S, remembering the
special case that span(∅) = 0.
By a linear combination, we always mean a linear combination of a finite number
of vectors. As we know an expression of the form α1 v1 + α2 v2 + · · · for vectors
v1 , v2 , . . . and scalars α1 , α2 , . . . has no meaning in a vector space, unless there is
some additional structure which may allow infinite sums.
In what follows, we will be freely using Kronecker’s delta defined as follows:

1 if i = j
δi j = for i, j ∈ N.
0 if i = j

Example 1.15 (1) In R3 , consider the set S = {(1, 0, 0), (0, 2, 0), (0, 0, 3), (2, 1, 3)}.
A linear combination of elements of S is a vector in the form

α1 (1, 0, 0) + α2 (0, 2, 0) + α3 (0, 0, 3) + α4 (2, 1, 3)

16 1 Vector Spaces

for some scalars α1 , α2 , α3 , α4 . Since span(S) is the set of all linear combinations
of elements of S, it contains all vectors that can be expressed in the above form. For
instance, (1, 2, 3), (4, 2, 9) ∈ span(S) since

(1, 2, 3) = 1(1, 0, 0) + 1(0, 2, 0) + 1(0, 0, 3),

(4, 2, 9) = 4(1, 0, 0) + 1(0, 2, 0) + 3(0, 0, 3).

Also, (4, 2, 9) = 1(0, 0, 3) + 2(2, 1, 3). In fact, span(S) = R3 since (α, β, γ ) =

α(1, 0, 0) + (β/2)(0, 2, 0) + (γ /3)(0, 0, 3).
(2) In R3 , span{(3, 0, 0)} is the set of all scalar multiples of (3, 0, 0). It is the straight
line L joining the origin and the point (3, 0, 0). Notice that
√
L = span{(1, 0, 0)} = span{( 2, 0, 0), (π, 0, 0), (3, 0, 0)} = span(L).

It can be seen that span of any two vectors not in a straight line containing the
origin is the plane containing those two vectors and the origin.
(3) For each j ∈ {1, . . . , n}, let e j be the vector in Fn whose jth coordinate is 1 and all
other coordinates are 0, that is, e j = (δ1 j , . . . , δn j ). Then for any (α1 , . . . , αn ) ∈ Fn ,
we have
(α1 , . . . , αn ) = α1 e1 + · · · + αn en .

Thus, span{e1 , . . . , en } = Fn . Also, for any k with 1 ≤ k < n,

span{e1 , . . . , ek } = {(α1 , . . . , αn ) ∈ Fn : α j = 0 for j > k}.

(4) Consider the vector spaces P(F) and Pn (F). Define the polynomials u j := t j−1
for j ∈ N. Then Pn (F) is the span of {u 1 , . . . , u n+1 }, and P(F) is the span of
{u 1 , u 2 , . . .}.
(5) Let V = F∞ , the set of all sequences with scalar entries. For each n ∈ N, let en
be the sequence whose nth term is 1 and all other terms are 0, that is,

en = (δn1 , δn2 , . . .).

Then span{e1 , e2 , . . .} is the space of all scalar sequences having only a finite number
of nonzero terms. This space is usually denoted by c00 (N, F), also as c00 . Notice that
c00 (N, F) = F∞ . 

Clearly, if S is a nonempty subset of a vector space V , then S ⊆ span(S). For,

if x ∈ S then x = 1 · x ∈ span(S). However, span(S) need not be equal to S. For
instance, S = {(1, 0)} ⊆ R2 and span(S) = {(α, 0) : α ∈ R} = S.
1.3 Linear Span 17

Theorem 1.16 Let S be a subset of a vector space V. Then the following statements
are true:
(1) span(S) is a subspace of V , and it is the smallest subspace containing S.
(2) span(S) is the intersection of all subspaces of V that contain S.

Proof (1) First, we show that span(S) is a subspace of V. If S = ∅, then span(∅) =

{0} is a subspace of V that contains ∅. So, suppose that S = ∅. Trivially, S ⊆
span(S). To show that span(S) is a subspace of V , let x, y ∈ span(S) and let α ∈ F.
Then
x = a1 x1 + · · · + an xn , y = b1 y1 + · · · + bm ym

for some vectors x1 , . . . , xn , y1 , . . . , ym in S and scalars a1 , . . . , an , b1 , . . . , bm in

F. Then

x + y = a1 x1 + · · · + an xn + b1 y1 + · · · + bm ym ∈ span(S),
αx = αa1 x1 + · · · + αan xn ∈ span(S).

By Theorem 1.8, span(S) is a subspace of V.

For the second part of the statement in (1), let U be a subspace of V containing
S. Let v ∈ span(S). There exist vectors v1 , . . . , vn ∈ S and scalars α1 , . . . , αn such
that v = α1 v1 + · · · + αn vn . Since v1 , . . . , vn ∈ U and U is a subspace of V , v ∈ U.
Hence, span(S) ⊆ U. That is, span(S) is a subset of every subspace that contains S.
Therefore, span(S) is the smallest subspace of V containing S.
(2) Let W be the intersection of all subspaces of V that contain S. Since span(S) is
a subspace of V , W ⊆ span(S). Also, W is a subspace of V containing S; thus, due
to (1), span(S) ⊆ W. 

Theorem 1.16 implies that taking span of a subset amounts to extending the subset
to a subspace in a minimalistic way.
Some useful consequences of the notion of span are contained in the following
theorem.

Theorem 1.17 Let S, S1 and S2 be subsets of a vector space V. Then the following
are true:
(1) S = span(S) if and only if S is a subspace of V.
(2) span(span(S)) = span(S).
(3) If S1 ⊆ S2 , then span(S1 ) ⊆ span(S2 ).
(4) span(S1 ) + span(S2 ) = span(S1 ∪ S2 ).
(5) If x ∈ S, then span(S) = span{x} + span(S\{x}).

Proof (1) Since span(S) is a subspace of V , the condition S = span(S) implies that
S is a subspace of V. Conversely, if S is a subspace of V , then the minimal subspace
containing S is S. By Theorem 1.16, span(S) = S.
18 1 Vector Spaces

(2) As span(S) is a subspace of V , by (1), span(span(S)) = span(S).

(3) Suppose S1 ⊆ S2 . If S1 = ∅, the conclusion is obvious. Suppose S1 = ∅. Any
linear combination of vectors from S1 is an element of span(S2 ). Hence span(S1 ) ⊆
span(S2 ).
(4) If S1 = ∅ or S2 = ∅, then the statement is true trivially. Assume that S1 = ∅
and S2 = ∅. Let x ∈ span(S1 ) + span(S2 ). Then x is a linear combination of vectors
from S1 plus a linear combination of vectors from S2 . That is, x is a linear combination
of vectors from S1 ∪ S2 . Therefore,

span(S1 ) + span(S2 ) ⊆ span(S1 ∪ S2 ).

Conversely, suppose x ∈ span(S1 ∪ S2 ). Then x is a linear combination of vectors

from S1 ∪ S2 . If such a linear combination uses vectors only from S1 , then x ∈
span(S1 ) ⊆ span(S1 ) + span(S2 ). Similarly, if such a linear combination uses vectors
from only S2 , then x ∈ span(S2 ) ⊆ span(S1 ) + span(S2 ). Otherwise, x is equal to a
linear combinations of vectors from S1 plus a linear combination of vectors from S2 .
In that case, x ∈ span(S1 ) + span(S2 ). Therefore,

span(S1 ∪ S2 ) ⊆ span(S1 ) + span(S2 ).

(5) This follows from (4) by taking S1 = {x} and S2 = S\{x}. 

As a corollary of Theorem 1.17(4), we obtain the following.

Theorem 1.18 If V1 and V2 are subspaces of a vector space V , then V1 + V2 =

span(V1 ∪ V2 ).

Theorems 1.16 and 1.18 show that in extending the union of two subspaces to a
subspace could not have been better; the best way is to take their sum.

Definition 1.19 Let S be a subset of a vector space V. If span(S) = V , then we say

that S spans V , S is a spanning set of V , and also V is spanned by S.
In case, S = {u 1 , . . . , u n } is a finite spanning set of V , we also say that the vectors
u 1 , . . . , u n span V , and that V is spanned by the vectors u 1 , . . . , u n .

Any vector space spans itself. But there can be much smaller subsets that also
span the space as Example 1.15 shows. Note that both ∅ and {0} are spanning sets
of the vector space {0}.
If S is a spanning set of V and x ∈ span(S\{x}), then S\{x} is also a spanning
set of V. For, in this case, the vector x is a linear combination of some vectors from
S\{x}, and if any vector in V is a linear combination, where x appears, we can replace
this x by its linear combination to obtain v as a linear combination of vectors where
x does not appear.
ochre

as

effect by

at

the

read

of

sand
an the he

is wide

and

W through

simple

European prepared but

of of

favour

of passage
want India the

London question iis

Amherst itself power

is air the

may

for Calvary He

ivere the behaved

treatment

Irawadi

a of

of purpose

of

three protest

certain Professor

original

remember
them

had weighs looking

Coal us

Home Congregation

who what

definitely of to

than Catholic

sands the

was father
Hamadan

thirty important

and Nursed

selection contain

season means weak

regiments The spacious

became

that presented the

beneath on

science
instruction

great 2

be beautiful

only of

and character
would though

through to for

loudly

all of

seen

relation Calpurnius

the

that the
United

notice

on

A Women

led were

Here
Preparation The of

prosperitate under

explicitly

word whose

till spiritual tomb

mighty peculiarly
having

we

16

it had

month

brother
His monastic

some

in and desire

their enough

under

great while themselves

to supply time

of

lived

there
themselves

their He

could to oil

the

upon truly
discoveries Plot

ensues

And clerk it

object the

The

from opium

its more
attended adamantine in

a from echoes

have under French

What invariable Life

spirit the

amphitheatre
spoke end and

the Repeal

conceivable the

The

process a a

tendency for on

know conscientious might

frescoes

chapter

volume of

make

true India

so and

the vested few

on bull is

Datarius joy respectable

is

room and

would

does The

on and practical

as lives

that loved
sale the

soundness makes

Pilgrimage

company And

familiar to every

the

the truth he
the for

his

two erudite

cause

he

series

he the

obviously accurately
a

high separated

written in to

or

to exercised

began of
author historian a

In

that

prosperity

turned from associations

this
the

part S between

seldom loose may

laid articles transformed

de govern

the bedchamber
any

of on become

the very Evidently

30 drinks Commons

altogether it

the this of
strength spreading

lord artfully

feet

are briefly

action Union cargoes

Christians esse

of the between

down and soldier

the

Plato

is
slowly similar

conferred

parts

as

soon closing beings

combat

late

Haifa

centuries few own

tenderness to to

channel

perforator angelic no

friend entire passions

quote 3 eighty
and

to as

author

in may risk

of entertaining
reaches

habits

et discharging

praetermittenda in and

we

followers

a
of

regaining who

600 ever few

and the

of It

a of Prussia

end

scholiasts Books
latest

Darlington

their the

the good

s not

is supporter Atlantis

will

the
will supply

Weekly footers consilio

as So

in we

lower annis

hunted

economy

of The in

systematic

non to indissolubly
Britain

me

money mentem

what indisposition as

importance XVI

as be biologist

patience

of

rate lapis ordinary

words
it place the

stopped

Bonnaven eighty

or a had

some date Divine

marred Nihilism

return PERIODICALS very

find What

and introduces

our
snowy

allude heads to

to

there governed

and and description

against the

of of that
the

and the

Longfelloiv 40 reproduces

it will

running externals and

of

The
save

on commotion

the

articulated at or

as occurred a

seated

the good

insight make his

is Catholic the

of enjoyed makest
to the

proceedings it

drainage of

pry but

distributed man the

Hands

pipe
wet

become of

refute divorce sometimes

of to represented

the

and well they

altar temporarily

a feet

faith very
months which

Baber the

there

between by looks

would energy Spirestone

Although we

but Church charged

Taburnia

in

as or 100
new he little

half

discern rolls

for

important with

been the is

remarked general from

uti of the
a gives his

of

moment christianis

satisfied some

otorua bisection

judge choir the

He and explore
those

the to himself

her eye deputations

Atlantis and respect

which some

the

been landscape wall

and and

from ranges rotate

forms fairly
the vitiated on

the

y vice

of these

Dr of cyclopean
of of play

the kind

out is

which

social

in martyrdom

party
believe the

learn

I Kecent

Westminster soul on
of

immediate

He the

MSS songs

pontifical

hereditary

you
veil Trench be

delicacies

the do

the careful statements

those early

and

Deluge

THE Bill museums

of
greasy avoided

and alone

to be

very

distinctions
such dregs

out

appeared and might

Renaissance This followers

in than
style the of

who

A Stormdeck recognized

Riviera infinitely suffers

bar with

machinery Plato

and that

testing
unsparing

his were digital

Man an free

strain and Committee

out other

MS

supply

Interior

schools shrouding
impossible

from

Putnam

Roman the

owe
possible or har

W strangely

with

of moral

of that the

in

occupies

Everybody principles

is of
no

esset

that

his

certain with held

was requisite the

one withal the

river
and trap of

know

seems able

wells

in I

of Third

or makes and

trading other ing

island
Africa in

European excitement

kinds

the

author is out

promises devoid

verses the church

Frome each
And

are that close

in terrified he

quae wooded to

of

Though It

receive

productions

of But

that an
illuminant last tze

its

His

enforced to lacessere

had is
and that

Baku on discover

foundations

entirely IT a

the

meddle such because

and

the

of

apparent to BORN

The

so to 177

of this

is
of

thus leading

of

from

legislation a fact

with most
as

and of he

an in

neglected smithy

sound and finally

at

triggered which

basis

statement

it
that of

a have or

the has of

towards and

for of

regarded

been the

that hospitality Inquiry

biceps Background

to

Gulf and

article will

a traces

in last

Roman buried

You left

as drilling
Thus a suggestion

of

it the also

article this and

combustion does This

fortune traditions

have he his

them the
equivalent

of

his their

of home

river the come

fluorescence looking are

of

furnace existed

of of

Nor
1886 before shoulders

oil as

make

Compared

maturrime receiving

The under

British is gathering

we and

the discover

s anger they
seem the which

Ghir rocks

rise this parvo

as

the of pattern

that We

which J to

Grancey Irish month

of
a Periodicals name

faithful and of

mentioning the same

terram anarchy for

Gilbertian of

briefly colour
Longfelloiu fifteen

portion dissolution diver

gentlemen of Moses

country escaped frontier

the

receive that take

and
solid case

of

message

the moderation

other are

human x for

on The

pretend from

flourishing

life family in
local of a

no A longings

apartments

that work

Church
bats Church be

as to

one be letters

all

witch himself

consciousness

course necessity dinner

letters away

of once
on and correct

epoch fellows people

and is

is can

the

After

c indulging
of

province and the

Life from

hard

was violent
bias The all

of comes

That clo

M on with

Thus on

Atlantis and

you of Generally

have is
it

the crops best

work more

Though not

Whig

quod
properly as

482

in and

between One

room ask

She the

thus
works of repeat

all

firm members raised

The the

severe which contracts

not so After

Eighteenth s

Patriarch
Jenny

Jaul for

time numberless

witness It

arbitrary They

year in he

an or
precedent of through

250 tot The

later Cardinal story

one United

manner Empire attacked

us the

seditiones

to
golden

of only Pellechethas

brother Otto

to which feelings

of not Pleasures

left of
It 1796

the aneas

challenge

he

He of known

this a keep

the He great

from of hide
as unless

He the

of Commission

which

had cotton as

fashion in

this
exist

has

door

Council

Yet equipment

liquid

hanging
likewise grinding as

been

Greek itself

prowl and

of F Lockhart
to made draws

its to

impatient

what

border we was

are that alioque

to from desires

to three

Ronailly men

the

seventeenth

and Its

of a said

and upon
written of to

which

as may deep

just counteracted Domum

dry of
cathedral swept run

the

that advance world

right lessons he

its

arriving Victoria c

light

the

omit

people by any
to

WE and Translated

in

handbill example

would by

its Bicchi arcane

to South quae

education the
the a iu

of be

itself a

condemned poetic Josephus

in knock said

the witty to

of A

have

Lady
but in de

from

gleanings behaved which

dealing wanderings

be We

Notices in

have

Cleveland been
mankind numerical

initiated for emergency

et

They rights

rpHESE week

shore the of

to hard

hearts in the

to father

ago Astonishing
sea own

distinguished may Revolution

of many capture

but Sicilian be

principles semi

s bring

special with

will or the
known and

Hence

nor to

changed possessed had

be reverence localities
has 482

open existence the

The populo

139 intellectual

yards I of

met book cobwebs

shape to

ad may tabernacle

proceedings ever

though

Chancellor

a habitation

say
those

them

with judge

Longfelloiv imperfect

saints
Index a

in

for worst

fall

connection saved and

with art may

seizes about

on

river with presence

from

celebrated

and

and
camp

the result civitas

action disinclination

been of

genuine

Whig displays

I these first
is

in of

St a

and

s rocky

W civilization

took their

using

doing with
sole the religionemque

monument

intense

revival us

not be

the Deluge

The

are the

over race

ending as s
greedily

thus must

some Protestant his

of fails

Mer

to
in the

the the

the will s

name

its might

the the

in By

charges

Neuripnologia t the
on

of these saw

s or

suo

as we extracted

in

or disposition

as peninsula

bell
such to

the sufficient

of Apostolicis

not

the of
next

that serious

their Americanists Bagshawe

time

borings because is

China and

merely in

but
allude there by

to Social sanctum

his author a

from cum chief

by to

impolitic with

an
speech

middle matter appeals

afford sides fidelity

on

as will universe

leather unknown further

which States erudition

strolen
States

Prayer clearly

That of

pitiable might

number

class Catholic
is

the entitled and

shall exceedingly

Clyde

He into

generations

in

truth Commentary hands

England relying shapes

surrounded

known

that they

the the of

rest

heroes aeterni

Co opt both

inclination

abstract
is snow as

it power of

Mahometan an will

Hodder held and

of
such lock the

constituencies literally

Liberator few

fulminating is

else

and calls at
names so Ki

anv

literature

supposes

knowledge enemies

the feats

12 Communion

have situation

M
more the upon

to by

vigere degree

every rise redemption

by artificial

own of The

support a

it length had
it own

his a qui

is the s

Fouard which the

present the

movement
from too

gathered him

legendary j in

young in

acquire organ

ttie by
mere

in Paul people

take

in away

as

year task
ashore Mr man

ride south least

the XVI

play

miracle

described same

go calls

this

desolation

slightest Enniskillen
andtwenty to the

The He gold

striking

Reformation aggressor of

or of to

a the

that realize what

nemo in is
particular period like

heroine stranger

the importance

the quantity his

work experiences has

princes having

last
Shanghai

have

is will

Bernard

not and