Linear Transformation 1 PDF
Linear Transformation 1 PDF
Linear Transformation 1 PDF
Neal, WKU
MATH 307
n
Linear Transformations
from Rn to Rm
m
n
(i) T(c u) = c T( u) , for all scalars c and all vectors u in R ;
n
(ii) T( u + v ) = T( u) + T(v ) , for all u , v in R .
It is equivalent to say that T(c u + v ) = c T( u + v ) for
all scalars c and all vectors u , v
n
in R .
n
( x1, x 2 ,. . ., x n ) in R . That is, we must define T ((x1, x 2 ,. . ., x n )) which must be a vector
in R
having m coordinates.
m
T ((x1, x 2 ,. . ., x n )) =
3
4
Example 1. Let T : R R be defined by
T (( x1 , x2 , x3 )) = (2 x1 2 x2 + 3 x3 , x1 + x2 x3 , x1 + 4 x2 , 4 x1 6 x2 + 3 x3 ).
Is T linear? What is T ((1, 2, 3)) ?
Each of the four coordinates in the range is a linear combination of the three
variables x1 , x2 , x3 from the domain; thus, T is a linear transformation. Also,
T ((1, 2, 3)) = (2 4 + 9, 1 + 2 3, 1 + 8, 4 12 + 9) = (7, 0, 9, 7).
3
2
Example 2. Let T : R R be defined by
T (( x1 , x2 , x3 )) = ( sin x1 + cos x 2 + 4 x 3, x1 + x2 + x3 + 6 ).
Neither coordinate in the range is a linear combination of the domain variables x1 ,
x2 , x3 ; thus, T is not a linear
transformation (although T is still a function).
Note: Throughout we shall let 0 n be the zero vector in R n and let 0 m be the zero vector
sides, we see that 0 m = T(0 n ) . So it is always the case that T(0 n ) = 0 m with a linear
transformation.
T ((x1, x 2 ,. . ., x n )) =
( a11 x1 + a12 x 2 + . . . + a1n x n , a21 x1 + a22 x 2 + . . . + a2n x n , . . . , am1 x1 + am2 x 2 + . . . + amn x n ),
a11
a21
then A = .
.
a
m1
a12
a22
.
.
am2
.
.
.
.
.
. . a1n
. . a2n
. .
. .
. .
.
. . am n
x2
.
can write X = as an n 1 matrix.
.
.
xn
The function value T ((x1, x 2 ,. . ., x n )) then can be computed by the matrix product
m
interchangeably write T( x ) = A X .
3
4
Example 3. Let T : R R be defined by
T (( x1 , x2 , x3 )) = (2 x1 2 x2 + 3 x3 , x1 + x2 x3 , x1 + 4 x2 , 4 x1 6 x2 + 3 x3 ).
2 2 3
1 1 1
The matrix representation is the 4 3 matrix A =
. How do we evaluate
4 0
1
4 6 3
2 2 3
7
1
1
1 1 1 0
T ((1, 2, 3)) ? Let X = 2 , then T ((1, 2, 3)) = A X =
2 = =
4 0 9
1
3
3
4 6 3
7
4
(7, 0, 9, 7) as coordinates in R .
3
2 2 3
2
1
1 1 1 1
We see that
0 = , and
4 0 1
1
4 6 3 0 4
2 2 3
3
0
1 1 1 1
0 = .
4 0 0
1
4 6 3 1 3
2 2 3
2
0
1 1 1 1
1 = , and
4 0 4
1
4 6 3 0 6
.
T A (x) = A X , where x = ( x1, x 2 ,. . ., x n ) and X = .
.
.
xn
3 2 8
, then T A : R3 R2 is defined by
For example, given the 2 3 matrix A =
4 6 3
x1
3 2 8
x
T (( x1 , x2 , x3 )) =
4 6 3 2
x3
3x1 2x 2 + 8x3
=
= ( 3x1 2x2 + 8x3 , 4x1 + 6x2 3x3 ).
4x1 + 6x2 3x3
Defining T by Action on the Standard Basis
n
Also, once we know the action on the standard basis, then we also know the
columns of the matrix representation A ; thus, we can easily evaluate T at any point.
4
3
Example 4. Evaluate T ((2, 3, 8, 10)) for T : R R defined by
1 2 1 2
Or we can write A = 2 1 2 0 , where the columns of A are the action on
3 4 4 6
2
1 2 1 2
36
3
the st. basis. Then T ((2, 3, 8, 10)) = 2 1 2 0
= 9 = (36, 9, 10) .
8
10
3 4 4 6
10
One-to-One Transformations
n
in the domain
To prove that a function is one-to-one, we often assume that T( u) = T(v ) and then
n
m
A.
Theorem 2.1. Let T : R R be a lineartransformation with matrix representation
Proof. We know that T(0 n ) = 0 m for any linear transformation. Suppose first that T is
one-to-one. If T( z ) = 0 m also, then z = 0 n because T is one-to-one.
Now suppose that z = 0 n is the onlysolution of T( z ) = 0 m . To prove that T is one
to-one, assume
that T( x ) = T( y ) . Then 0 m = T( x ) T( y ) = T( x y ) which means that
T is one-to-one.
QED
x y =0 n . Hence, x = y and
1 0 . . .
0 1 . . .
. . 1 . .
. . . 1 .
0 . . . 1
. .
. .
. .
. .
. cn
.
.
0
1 . . . 0
. 1 . . .
. . 1 . .
. . . 1 .
0 . . . 1
0 . . . 0
the n n identity with
additional rows that are all 0
1 . . . c1
. 1 . . .
. . 1 . cn
0 . . . 0
. . . . .
0 . . . 0
free variables remain in the
last non-zero row
In the first case, A X = 0 has only the trivial solution; thus, T is one-to-one. In the
second case, A X = 0 has infinite solutions; thus, T is not one-to-one.
n
3
4
Example 5. Let T : R R be defined by
T (( x1 , x2 , x3 )) = (2 x1 2 x2 + 3 x3 , x1 + x2 x3 , x1 + 4 x2 , 4 x1 6 x2 + 3 x3 ).
Is T one-to-one?
Solution. We let A be the matrix representation and look at the reduced row echelon
form of homogeneous system A 0 :
2 2 3 0
1
1
1 1 0
0
rref
4 0 0
1
0
4 6 3 0
0
0
1
0
0
0
0
1
0
0
0
Is T is one-to-one?
Solution. We look at the reduced row echelon form of homogeneous system A 0 :
1 2 1 2 0
1 0
2 1 2 6 0
0 1
3 4 4 3 0 rref 0 0
0 1 0 2 0
0 0
1 0 2 5 0
0 0
0
0
1
0
0
3
2
1
0
0
0
0
0
2
2
Example 7. Let T : R R be the reflection about the x -axis. Is T one-to-one?
1 0
Onto Transformations
n
that T(v ) = b . (The vector v is a pre-image of b .) That is, the equation T( x ) = b always
n
m
Theorem 2.5. Let T : R R be a linear transformation with matrix representation A .
Then T is onto if and only if the matrix equation A X = B always has a solution for
every m 1 matrix B .
1 .
0 1
. .
. .
0 0
. . c1
. . c2
1 . .
. 1 .
0 0 cm
n
m
Theorem 2.7. Let T : R R be a linear transformation with matrix representation A .
Assume m < n . Then T is onto if and only if the reduced row echelon form of A does
not contain a row of all 0's.
1 0 . . . . .
0 1 . . . . .
. . 1 . . . .
. . . 1 . . .
0 . . . 1 . .
no rows become all 0
1 0 . . . . .
0 1 . . . . .
. . 1 . . . .
. . . 1 . . .
0 0 . . . . 0
a row becomes all 0
Theorem 2.9. Let T : R R be a linear transformation. Then T is either both one-toone and onto (i.e., a bijection), or T is neither one-to-one nor onto.
Proof. Let A be the matrix representation of T . By Theorems 4 and 8, T is one-to-one if
and only if det (A) 0 if and only if T is onto. Thus T is one-to-one if and only if T is
onto. That is, T is either both one-to-one and onto, or T is neither one-to-one nor onto.
4
3
Example 8. Let T A : R R be defined by the matrix representation
1 0 2
A = 1 2 1
2 1
0
where T A (x) = A X . Is T A one-to-one and/or onto? Is the point (4, 2, 6) in the range of
T A ? If so, find a pre-image.
1 0 0 1.2
0 1 0 0.4
0 0 1 2.6
Because no rows became all 0, there will never be an inconsistency. So every system
A X = B will have a solution; therefore, T A is onto.
3
b
R
v
R 4 such that
Now
we
can
say
that
for
any
vector
,
there
is
a
pre-image
T(v ) = b . (In fact we can see from rref( A ) that there will be infinite pre-images, which
also verifies that T A is not one-to-one.)
1
0 2
1 2 1
2 1
0
4 4
3 2 rref
2 6
1 0
0 1
0 0
x1 = 4.8 + 1.2 t
0 1.2 4.8
x2 = 3.6 + 0.4 t
0 0.4 3.6
x3 = 4.4 + 2.6 t
1 2.6 4.4
x4 = t
If t = 0, then (4.8, 3.6, 4.4, 0) is one particular pre-image of (4, 2, 6). That is,
T ((4.8, 3.6, 4.4, 0)) = (4, 2, 6).
3
4
3
Example 9. Let T : R R be the embedding map which places a point from R into
R 4 by adding a 0 in the 4th coordinate. Is T one-to-one and/or onto?
Solution. Here we can define T by T ((x, y, z)) = (x, y, z, 0) . The matrix representation is
1 0 0
0 1 0
the 4 3 matrix A =
. We see that A is already in rref form and has a row of
0 0 1
0 0 0
all 0's. By Theorem 7, T is not onto. (In particular, every element in R with a non-zero
4th coordinate will cause an inconsistency; thus, these elements will not be in the
range.) However, the homogeneous system A X = 0 has only the trivial solution of X =
0; thus, T is one-to-one.
n
n
n
When a linear transformation T : R R is both one-to-one and onto, then we can
1
n
n
find the inverse transformation T : R R such that if T ((x1, x 2 ,. . ., x n )) =
If A is the matrix representation of the one-to-one and onto T , then we know that
1
det (A) 0. Therefore A exists. Because T ((x1, x 2 ,. . ., x n )) = (y1, y2 ,. .., yn ) is equivalent
to A X = Y , we can apply the matrix inverse and say that X = A 1 Y . Thus, A
1
be the matrix representation of T because ( x1, x 2 ,. . ., x n ) = T 1((y1, y2 ,. . ., yn )) .
must
1 2 1
Solution. The matrix representation of T is A = 0 1 1 and det (A) = 1 0; thus T
1 2 4
is both one-to-one and onto (i.e., a bijection). The matrix representation of T
2 6 1
1 5 1 . Thus,
1 4 1
is A
Note: T 1((1, 0, 1)) = (1, 0, 0) , T 1((2, 1, 2)) = (0, 1, 0), and T 1((1, 1, 4)) = (0, 0, 1).
2
2
1
Example 11. Let T : R R be the reflection about the x -axis. Find T .
1 0
1 0
. Then A 1 =
also. So
Solution. Here T ((x, y)) = (x, y) and A =
0 1
0 1
T 1((x, y)) = (x, y) = T ((x, y)) . In other words, T is its own inverse. If you reflect
about the x -axis, then you undo the process by reflecting back about the x -axis.
null space (or kernel of T ) is the set of all vectors x R n such that T( x ) = 0 m . We know
T( x ) = 0 m . The set of all solutions to AX = 0 is called the null space. The nullity is the
dimension of this null space and is given by the number of independent parameters ( r ,
s , t , etc.) used
in writing the form of all solutions to AX = 0 . And if T is one-to-one,
then the nullity is simply 0.
The range of T is the set of all vectors y R m for which T( x ) = y for some x R n . All
of the columns in the matrix representation A are actually elements in the range of T .
The first column in A is the value T ((1, 0, 0, . . ., 0)) , the second column in A is the
value T ((0, 1, 0, . . ., 0)) , etc. But some of these columns may be linear combinations of
independent of
T(e1 ) = (1, 1, 2, 0, 1)
T(e2 ) = (0, 1, 1, 1, 1)
T(e4 ) = (1, 1, 2, 0, 0)
(a)
(b)
(c)
(d)
(e)
T(e3 ) = (0, 1, 1, 1, 1)
T(e5 ) = (1, 0, 1, 1, 1)
A = 2
0
0
0
1 1
1 1
1 1
1 1
1 1
1 0
2 1
0 1
0 1
T (( x1 , x2 , x3 , x4 , x5 )) =
( x1 + x4 + x5 , x1 x2 + x3 + x4 , 2 x1 x2 + x3 + 2 x4 + x5 , x2 + x3 x5 , x1 + x2 x3 + x5 )
(b) det (A) = 0 ; thus, A is neither one-to-one nor onto.
(c)
A B = 2
0
0
1
1
0
1
1
1
1
2
1 1 4
1
0 2 2
0
1 3 6 rref 0
0
1 1 0 1 4 2
1 1 0 1 5 5
0
0 0 0
1 1 0
0 0 1
0
0
0
0
0
0
0 0 3
1 0 2
1 0 1
0 1 0
0 0 0
(d) rref (A 0) = 0
0
0
1
0
0
0
0
1
0
0
0
0
0
1
0
0
0
1
1
0
0
0
0
0
(x1 , x2 , x3 , x4 , x5 ) = (0, s t , s , t , t )
where s and t are any real numbers.
Then any vector in the null space has the form s v1 + t v2 , which is a linear combination
of {v1, v 2 } . So {v1, v 2 } is then a basis for the null space.
(e) From rref (A) , we see that the original first, second, and fourth columns create the
of T will have three dimensions T .
3 3 identity in the rref. Thus, the range
Thus, rank T = 3 and the range of T is the collections of all linear combinations of the
three vectors T(e1 ) = (1, 1, 2, 0, 1) , T(e2 ) = (0, 1, 1, 1, 1) , and T(e4 ) = (1, 1, 2, 0, 0) , which
are the first, second, and fourth columns in the origianl matrix.
In other words, if we look at just the original first, second, and fourth columns as
vectors in R 5, then any vector in the range of T can be
written as a linear combination
of justthese three vectors after rref is performed. For c = (4, 2, 6, 2, 5) as in Part (c), we
Note:
Another way to label all linear combinations of a collection of vectors is to
the preceding example, the null space is
simply call
it the span of these vectors.
In
span
{v1, v2 } and the
range is span{ T(e1 ) , T(e2 ) , T(e4 ) }.
Here is a description of the preceding transformtation:
5
5
T
: R R
Domain: R
(5 dimensions)
span{ e1, e2 , e4 }
Two-dimensional
Null Space
span {v1, v 2 }
Range:
5
A Three Dimensional Subpace of R
{0,
0, 0, 0, 0}
5
T is not one-to-one: Two dimensions of R all get mapped to {0, 0, 0, 0, 0}.
T is not onto: The range is 3-d; only three dimensions of R5 have pre-images.
MATH 307
Linear Transformations: Rn to Rn
Exercises
3
3
1. Let T : R R be defined in functional form by
T (( x1 , x2 , x3 )) = ( x1 2 x2 + 2 x3 , 2 x1 + x2 + x3 , x1 + x2 ).
(a)
(b)
(c)
(d)
(e) Give both the matrix representation and the functional form of T
4
4
2. Let T : R R be defined by
MATH 307
Linear Transformations: Rn to Rm
Exercises
6
4
1. Let T : R R be defined by
T (( x1 , x2 , x3 , x4 , x5 , x6 )) =
( x1 + 2 x2 + 4 x4 + 5 x5 3 x6 , 3 x1 7 x2 + 2 x3 + x5 + 4 x6 ,
2 x1 5 x2 + 2 x3 + 4 x4 + 6 x5 + x6 , 4 x1 9 x2 + 2 x3 4 x4 4 x5 + 7 x6 )
(a) Give the matrix representation of T .
(b) Determine and explain whether or not T is onto.
(c) Describe the range of T and give the rank of T .
(d) Determine and explain whether or not T is one-to-one.
(e) Give the form of the null space and give the nullity of T . Also give a basis for the
null space.
(f) Explain how the domain is divided into two subspaces and how these subspaces are
4
mapped into R via T .
5
3
2. Let T : R R defined by matrix representation
1 4 5 0 9
A = 3 2 1 0 1
2 3 5 1 8
(a) Give the function form of T .
(b) Determine and explain whether or not T is onto.
(c) Describe the range of T and give the rank of T .
(d) Determine if (3, 2, 1) is in the range of T . If so, give a pre-image.
(e) Determine and explain whether or not T is one-to-one.
(f) Give the form of the null space and give the nullity of T . Also give a basis for the
null space.
3
4
3. Let T : R R be defined by