Scribd
Upload
206 views

Linear Transformations and Matrices

This document discusses linear transformations and matrices. It begins by explaining that a matrix A defines a linear transformation T from Rn to Rm by the rule T(v) = Av. This transformation maps vectors to other vectors by a linear operation. The document then proves several theorems about the relationship between matrices and linear transformations, including that every linear transformation T can be represented by a matrix A such that T(v) = Av. It provides examples of linear transformations and discusses how to find the kernel, range, and rank of a linear transformation based on its matrix. Finally, it explains how change of basis transformations are also linear transformations, and how to represent them using matrices.

Uploaded by

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

Linear Transformations and Matrices

This document discusses linear transformations and matrices. It begins by explaining that a matrix A defines a linear transformation T from Rn to Rm by the rule T(v) = Av. This transformation maps vectors to other vectors by a linear operation. The document then proves several theorems about the relationship between matrices and linear transformations, including that every linear transformation T can be represented by a matrix A such that T(v) = Av. It provides examples of linear transformations and discusses how to find the kernel, range, and rank of a linear transformation based on its matrix. Finally, it explains how change of basis transformations are also linear transformations, and how to represent them using matrices.

Uploaded by

Raul
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Chapter 5

Linear Transformations and Matrices


5.1

Linear transformations from Rn to Rm

When A is an m n matrix and v is a vector of dimension n then Av is a vector of


dimension m. Since matrix multiplication of A by a vector is a linear operation it
follows that the matrix product Av produces a natural linear transformation from Rn
to Rm .
Theorem:. If A is an m n matrix (m rows and n columns), and T : Rn Rm is
defined by T (v) = Av, for every v V; then T is a linear transformation.
Proof. Since A has size m n, for any vector v Rn we have that the product Av
exists and is an m-dimensional vector, that is, T (v) = Av Rm . We have to prove that
T is a linear transformation. Let v1 , v2 Rn , and be a scalar. We have:
T (v1 + v2 ) = A(v1 + v2 ) = A(v1 ) + A(v2 ) T (v1 + v2 ) = T (v1 ) + T (v2 ),
and:
T ( v1 ) = A( v1 ) = Av1 T ( v1 ) = T (v1 ).

The matrix-based linear transformation T , defined by T (v) = Av is often denoted as
TA .
1

There is a reverse connection: for every linear transformation T : Rn Rm , there is a


matrix A such that T (v) = Av.
Theorem: Suppose that T : Rn Rm is a linear transformation. Let {e1 , . . . , en }
denote the standard basis of Rn and let A be the matrix whose columns are the vectors
T (e1 ), . . . , T (en ). Then for every u Rn ,
T (u) = Au.
Proof: Let u = (u1 , . . . , un ) 0 Rn . Then
u = u1 e1 + + un en .
By the linearity properties of T we have
T (u) = u1 T (e1 ) + + un T (en ),
which is a linear combination of the columns of A, so we have
T (u) = Au,
as we wanted.

Thus, to each matrix A there corresponds a linear transformation TA , and to each


linear transformation T there corresponds a matrix, that we often denote as AT . Note
that it is enough to know the image of the vectors in the basis to compute the matrix
of the transformation and thus to allow determining the image of any vector.
Example: Let T : R3 R3 be the linear transformation given by
T ((u1 , u2 , u3 ) 0 ) = (u1 + u2 + u3 , u1 u3 , u1 + 2u2 3u3 ) 0 .
The image of any vector is easily computed by substituting its components into the
definition of T . The matrix of this transformation is found by computing the image of
the standard basis vectors:
T (e1 ) = (1, 1, 1) 0 ;

T (e2 ) = (1, 1, 2) 0 ;
2

T (e3 ) = (1, 0, 3) 0 ,

and thus

A= 1

1
0
2

1 .

Notice that the entries of the matrix A are just the coefficients of u1 , u2 , u3 in the
definition of T .

Example: If A is the m n zero matrix then TA is the zero linear transformation that
maps every v Rn into the zero vector in Rm .

Example: If n = m and In is the n n identity matrix, then TIn is the linear transformation, Rn Rn that maps every vector into itself and is called the identity linear
transformation.

Example: Consider the matrix

A=

and the linear transformation associated to it, TA : R3 R2 .


The kernel of TA is given by
ker(TA ) = {u = (u1 , u2 , u3 ) 0 R3 : Au = 0}.
This set includes all vectors satisfying 2u1 2u3 = 0 and 4u1 +u2 u3 = 0. By solving
this system, we get that the kernel is given by
ker(TA ) = {(1, 3, 1) 0 : R},
which is, obviously, a vector space of dimension 1, spanned by vector (1, 3, 1) 0 .
The range of TA is given by
R(TA ) = {v = (v1 , v2 ) 0 R2 : v1 = 2u1 2u3 , v2 = 4u1 + u2 u3 , u1 , u2 , u3 R},
3

and its dimension is easily computed as


rank(TA ) = n nullity(TA ) = 3 1 = 2.
Finally, notice that the rank of A is also 2.

Theorem: If TA : Rn Rm is the linear transformation associated to the m n matrix


A then:
1. rank(TA ) = rank(A).
2. The maximum possible value for rank(TA ) is equal to the min(m, n).

Theorem: If A is an n n matrix and TA : Rn Rn is the corresponding linear


transformation then each of the following statements is equivalent (if one is true so
are all of the others):
1. A is non-singular (it has an inverse).
2. The range of TA is Rn .
3. TA is one-to-one.
1
4. The linear transformation TA
exists and its matrix is A1 .

Notice that a one-to-one linear transformation corresponds exactly to a matrix A


that is non-singular.

5.2

Change of basis as a linear transformation

Recall from chapter 3 that if B = {v1 , . . . , vn } is a basis of Rn , the matrix (v1 | . . . | vn ),


whose columns are the basis vectors in B is the transition matrix from coordinates
4

with respect to B to the standard coordinates. Also, the inverse of this matrix is the
transition matrix from standard coordinates to coordinates with respect to the basis B.

If P is the transition matrix from coordinates in a basis B of Rn to standard coordinates, then considered as the matrix of a linear transformation TP (x) = Px, the linear
transformation actually maps the standard basis vectors ei to the new basis vectors
vi , that is, TP (ei ) = vi .

Example: Suppose we want to change the basis in R2 by a rotation of the axes

radians counterclockwise. What are the coordinates of a vector with respect to this
new basis B = {v1 , v2 }? Consider the linear transformation T that transforms the
vectors in the standard basis e1 , e2 into the new basis vectors:

0


0
1 1
v1 = cos , sin
= ,
4
4
2 2
and

0
v2 = sin , cos
=
4
4


1 1
,
2 2

0

that is, T (e1 ) = v1 and T (e2 ) = v2 . The matrix associated to the linear transformation
is the transition matrix

P=

1
2

12

1
2

1
2

which corresponds to the transition matrix from coordinates with respect to B to standard coordinates. Then the coordinates of a vector v with respect to the new basis are
given by
P1 v .
The inverse of a counterclockwise rotation is a clockwise rotation, so we have:


 1

1
cos

sin

4
4
2
2
=
P1 =



1
1
sin 4
cos 4
2 2
5

and we can compute the new coordinates of any vector v.


Consider the vector v = (1, 1) 0 ; its coordinates with respect to the basis B is given by
the product


1
2
=

P1 x =
1
1

2
1
0
2

Alternatively we could have noticed that v = 2v1 , and we get that the coordinates

of v with respect to B are ( 2, 0) 0 .




5.2.1

1
2

1
2

Change of coordinates from basis B to basis B 0

Given a basis B of Rn with transition matrix PB , and another basis B 0 with transition
matrix PB 0 , we can obtain the transition matrix to change from coordinates in the basis
B to coordinates in the basis B 0 directly from these two matrices, instead of computing
the coordinates of one set of vectors with respect to the other.
To do that, we first change from coordinates with respect to B to standard coordinates
using PB , and then change from standard coordinates to coordinates with respect to
1
1
B 0 using PB
0 . The matrix M = PB 0 PB is the transition matrix from coordinates with

respect to B to coordinates with respect to B 0 .

Example: Consider the two basis of R2 given by B = {(1, 2) 0 , (1, 1) 0 } and B 0 =


{(3, 1) 0 , (5, 2) 0 }. The transition matrices from coordinates with respect to B and B 0 to

1 1
3 5
and PB 0 =
, respecstandard coordinates are given by PB =
2 1
1 2
tively. If we want to find the coordinates of vector u with respect to B 0 and we know
that its coordinates with respect to B are given by (4, 1) 0 , we can simply compute

2 5
1 1
4
25
1

=
,
TM (u) = Mu = PB
0 PB u =
1 3
2 1
1
16
to obtain those coordinates.


6

You might also like