Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
2 views

mcnotes21

The document provides an overview of vectors and matrices, defining key concepts such as n-dimensional vectors, matrix dimensions, and operations like addition, scalar multiplication, and multiplication. It explains the properties of matrix operations, including the transpose, and introduces special types of matrices such as symmetric, upper triangular, and diagonal matrices. Additionally, it outlines the rules for transposes and the structure of linear vector spaces formed by matrices.

Uploaded by

eugenio
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
2 views

mcnotes21

The document provides an overview of vectors and matrices, defining key concepts such as n-dimensional vectors, matrix dimensions, and operations like addition, scalar multiplication, and multiplication. It explains the properties of matrix operations, including the transpose, and introduces special types of matrices such as symmetric, upper triangular, and diagonal matrices. Additionally, it outlines the rules for transposes and the structure of linear vector spaces formed by matrices.

Uploaded by

eugenio
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Vectors, Matrices, Matrix Operations

3.1 Vectors
An n-dimensional real vector x = (x1 , ..., xn ) ∈ Rn is an ordered n-tuple of
real numbers. As a matter of notational convention, whenever we talk of an
n-dimensional vector we mean a column vector; that is, a single column of n
rows: ⎡ ⎤
x1
⎢ x2 ⎥
x = [xi ]i=1,...,n = ⎢ ⎥
⎣ ... ⎦
xn
Letting a prime (0 ) denote the transpose, a row vector is simply the transpose
of a column vector:
x0 = [x1 x2 ... xn ]
For the set of all real vectors we may define summation and scalar multi-
plication, as well as inner product, in the ordinary way. Then, the set of all
real vectors x ∈ Rn is a linear vector space, and endowed with the Euclid-
ean distance/norm forms our well known n-dimensional Euclidean vector
space.3

3.2 Matrices
A real matrix is defined as a rectangular array of real numbers.4 In particular,
the matrix ⎡ ⎤
a11 a12 ... a1n
⎢ a21 a22 ... a2n ⎥
j=1,...,n ⎢
A = [aij ]i=1,...,m = ⎣ ⎥
... ... ... ⎦
am1 am2 ... amn
for aij ∈ R ∀i ∈ {1, 2, ..., m}, j ∈ {1, 2, ..., n} and m, n ∈ N∗ = {1, 2, ...}, is a
matrix of dimensions m × n. More compactly we write this matrix as A = [aij ].
Notice that, just as an n-dimensional (real) vector x = (x1 , ..., xn ) ∈ Rn is
an ordered set or ordered n-tuple of real numbers, an m×n matrix is an ordered
set or ordered n-tuple of m-dimensional vectors. Thus, we may write a matrix
as an array of column vectors

A = [aj ] = [a1 a2 ... an ]

with the understanding that aj = (a1j , a2j , ..., amj ) ∈ Rm is an m-dimensional


column vector for all j = 1, .., n.
3 For more details on vector spaces and Euclidean spaces, see Takayama (1993), ch. 1;

Simon and Blume (1994), chs. 10 & 27; or the MathCamp notes by Peter and Leeat. ♠
ed.note: It is so convinient to transfer responsibility to somebody else! :-)
4 In most of our discussion we restrict focus to matrices over the set of real numbers, R.

All notions, however, can be extended to the complex plane, C, or any arbitrary field. It’s a
good exercise for the reader to check what (if any) would have to change should, throughout
this text, we had set C wherever R appears.

19
3.3 Matrix Transpose
The transpose A0 of an m × n matrix A = [aij ] is an n × m matrix defined
simply as ⎡ 0 ⎤ ⎡ ⎤
a1 a11 a21 ... am1
⎢ a02 ⎥ ⎢ a12 a22 ... am2 ⎥
0 ⎢
A = [aji ] = ⎣ ⎥=⎢ ⎥
... ⎦ ⎣ ... ... ... ⎦
a0n a1n a2n ... amn

3.4 Matrix Addition and Scalar Multiplication


The set of all matrices of the same dimensions forms a vector space. This
requires only that we appropriately define matrix addition and scalar multi-
plication:
Definition 94 Take any m × n matrices A = [aij ], B = [bij ], C = [cij ], and
any real scalars λ, µ. We define matrix addition as
C = A + B ⇔ cij = aij + bij ∀i, j
and scalar multiplication as
C = λA ⇔ cij = λaij ∀i, j
We can then easily show the following properties:
Lemma 95 Matrix addition and scalar multiplication, defined as above, satisfy:
(i) Commutative Law: A + B = B + A
(ii) Associative Law: (A + B) + C = A + (B + C)
(iii) Distributive Law: (λ + µ)A = λA + µA
(iv) Distributive Law: λ(A + B) = λA + λB
(v) Distributive Law: λ(µA) = (λµ)A
It then follows that:
Proposition 96 The set of all m × n real matrices endowed with matrix addi-
tion and scalar multiplication forms a linear vector space.
Exercise 97 Persuade yourself that you can prove Lemma 95 and Proposition
96 from first principles.

3.5 Matrix Multiplication


Let A and B be matrices. The matrix product of A and B (denoted AB) is
defined when A is m×n and B is n×p. When this holds (i.e., there are as many
columns in A as there are rows in B), we say that A and B are conformable.
The i, kth element of AB is then given by
n
X
(AB)ik = aij bjk
j=1

20
and AB is an m × p matrix.

Matrix multiplication satsfies the following properties:


Lemma 98 Provided dimension conformity, matrix multiplication satisfies:
(i) Associative Law: (AB)C = A(BC)
(ii)Associative Law: (λA)B = λ(AB)
(iii) Distributive Law: A(B + C) = AB + AC
(iv) Distributive Law: (A + B)C = AC + BC
Note that if AB exists, BA need not be definted; even if it does exist, it is
not generally true that AB = BA (when this is true, we say that A and B are
commuting). This is why we need both (iii) and (iv) above; the first distributive
law deals with pre-multiplication of a matrix sum by another matrix, while the
second deals with post-multiplication of a matrix sum with another matrix.

3.6 Special Matrices


Definition 99 An m × n matrix A is called square if m = n.
A matrix A is symmetric if A0 = A, or equivalently aij = aji ∀i, j. This
imposes that A is square.
A matrix A is upper triangular if it is square and aij = 0 for j = i+1, ..., n
and i = 1, ..., n − 1. And A is lower triangular iff A0 is upper triangular.
A matrix A is diagonal if it is both upper and lower triangular; equivalently,
iff aij = 0 for i 6= j.
The null matrix (denoted 0) is defined by the relation:
A + (−1)A = 0, or equivalently, A + 0 = A
It can thus take any dimension (it must have the same dimension as
A), and is the matrix with all elements equal to zero.
The identity matrix (denoted I) is defined by the relation:
IA = AI = A
It is the diagonal matrix with ones on its diagonal, and takes the di-
mension necessary for conformability (that is, if A above is m × n, I is m × m
in the first part of the above relation and n × n in the second part).

3.7 Transpose Rules


Lemma 100 The following rules hold for matrix transposes:
(A0 )0 = A
(A + B)0 = A0 + B 0
(AB)0 = B 0 A0
A0 A = 0 ⇔ A = A0 = 0
A0 A 6 = AA 6= A0 A0 in general, but
A0 A = AA = A0 A0 for symmetric A

21
Note that AB defined ⇒ (AB)0 = B 0 A0 defined.

Exercise 101 Find examples illustrating the above facts.

22

You might also like