Linear Transformations: 4.1 Definitions and Properties

Chapter 4
Linear Transformations
4.1
Definitions and properties
Recall that a real function of one variable, y = f(x), transforms each number x in
the domain of f (a subset of R, or R itself) into exactly one number f(x) R. In
other words we can also say that function f maps or transforms a vector x (since R
is a vector space) into another vector y = f(x) in the range of f, and this meaning is
reflected by the notation:
f : x f(x)
Similarly a real function of two variables y = f(x, y) maps or transforms vectors (x, y)
in its domain (R2 , or a subset of R2 ) into vectors of its range set (R or a subset of R)
and we write:
f : (x, y) f(x, y)
In general, a function T from one vector space V to another vector space W is a
rule which assigns to every vector v V a unique vector w W, and if this function
satisfies the following two properties:
i) T (v1 + v2 ) = T (v1 ) + T (v2 ), for every v1 , v2 V,
1
ii) T (v1 ) = f(v1 ), for every v1 V and every R,

then it is called a linear transformation (or linear function, or linear mapping, or
homomorphism). Properties i) and ii) are the linearity properties.
The set of all values T (v) is called the range of T . We will denote this set by R(T ).
The vector space V is called the domain of T . A linear transformation T is sometimes
called a linear operator, especially when V = W.
If T : V W is a linear transformation then T is said to be one-to-one if for every

pair of distinct vectors v1 , v2 (v1 6= v2 ) in V the vectors T (v1 ), T (v2 ) are also distinct
vectors in W. If the range of T equals W then we say that T is onto.
Theorem: The range of the linear transformation T : V W is a vector space.

Proof: Recall that when a set S of vectors is already in a vector space, then to show that
S is a subspace (i.e., a vector space) it is only necessary to show that it is closed under
addition and scalar multiplication. Effectively, we have:
w1 , w2 R(T ) there are some v1 , v2 V : w1 + w2 = T (v1 ) + T (v2 )
T (v1 + v2 )
lin.prop
and since V is a vector space, v1 + v2 V, therefore

T (v1 + v2 ) R(T )
w1 + w2 R(T ) (closed under addition)
On the other hand, w R(T ), there is some v V such that T (v) = w, and R
we have:
w = T (v)
T ( v)
lin.prop
and since v V we have that

T ( v) R(T )
w R(T ) (closed under scalar multiplication)

2
Therefore, R(T ) is a vector space.
This theorem warrants the following definition.

The dimension of the vector space that is the range of a linear transformation T is
called the rank of T and is denoted by rank(T ). The range of T is sometimes called
rangespace.
Theorem: If T is a linear transformation T : V W then:
a) If 0 V is the zero vector of V then T (0) is the zero vector of W.
b) If v V and v is the negative of v then T (v) is the negative of T (v) in W.
c) If v1 , v2 V then T (v1 v2 ) = T (v1 ) T (v2 ).
Proof:
a) Take v V. Then
T (0)
T (0 v)
V vector space
0 T (v)
T linear
0.
R(T ) vector space
b) Take v V. We have
T (v) + T (v)
T (v + (v))
T (0)
V vector space
T linear
0.
prop. a)
c) Let v1 , v2 V, then
T (v1 v2 )
T (v1 + (v2 ))
V vector space
T linear
T (v1 ) + T (v2 )
T (v1 ) T (v2 ).
T linear

Example: Recall Pn is the vector space of all polynomials of degree n or less. Consider
the following transformations:
T1 : P2 P2 , defined by T1 (a0 + a1 x + a2 x2 ) = a2 x2 . T1 is linear. Consider

polynomials p(x), q(x) P2 , and scalar :
x + b2 x2}) = T1 (a0 +b0 +(a1 +b1 )x+(a2 +b2 )x2 ) =
T1 (a
x + a2 x2} +b
| 0 + b1{z
| 0 + a1{z
p(x)
q(x)
= (a2 + b2 )x2 = a2 x2 + b2 x2 = T1 (p(x)) + T1 (q(x)).

T1 ( p(x)) = T1 ((a0 +a1 x+a2 x2 )) = T1 ( a0 + a1 x+ a2 x2 ) = a2 x2 = T1 (p(x)).
T2 : P2 R, defined by T2 (a0 + a1 x + a2 x2 ) = |ak |, where ak is the largest size
coefficient, max{a0 , a1 , a2 }. T2 is not linear. For instance, consider polynomials
p(x) = x2 and q(x) = x2 + 2x; we have:
T2 (p(x) + q(x)) = T2 (2x2 + 2x) = 2,
whereas
T2 (p(x)) + T2 (q(x)) = 1 + 2 = 3.

If T is one-to-one then the inverse transformation T 1 is defined from W 0 V,
where W 0 is the range of T (a subspace of W) as follows. For each pair v V and
w W 0 with w = T (v), define T 1 (w) = v.
Theorem: If T : V W is a one-to-one linear transformation and the range of T is W 0

(a subspace of W), then T 1 is a one-to-one linear transformation from W 0 to V.
Proof: Consider w1 , w2 W 0 and R. Then, there exist v1 , v2 V such that
T (v1 ) = w1 , T 1 (w1 ) = v1 and T (v2 ) = w2 , T 1 (w2 ) = v2 . First, notice that
T 1 (T (v1 )) = T 1 (w1 ) = v1 ,
and
T (T 1 (w1 )) = T (v1 ) = w1 .
4
We have:
w1 ,w2 W 0
(w1 + w2 )
T linear
= v1 + v2
(T (v1 ) + T (v(2)))
T 1 (T (v1 + v2 )) =
T 1 (w1 ) + T 1 (w2 ),
def. of T 1
and
T 1 (w1 ) = T 1 (T (v1 ))
T 1 (T (v1 )) = v1 = T 1 (w1 ),
T linear
and therefore, T 1 is linear.
Two vector spaces V and W are said to be isomorphic if there is a one-to-one and
onto linear transformation T : V W; T is then called an isomorphism between V
and W. An automorphism is an isomorphism of a space with itself.
Theorem: If T : V W is an isomorphism between V and W and {v1 , v2 , . . . , vn } is a

basis of the vector space V, then {T (v1 ), T (v2 ), . . . , T (vn )} is a basis of the vector space
W. Note: It follows that isomorphic vector spaces have the same dimension.
Proof: Consider the linear combination
0 = 1 T (v1 ) + + n T (vn )
T (1 v1 + + n vn ),
T linear
Since T (0) = 0 and T is an isomorphism:
1 v1 + + n vn = 0
1 = = n = 0,
v1 ,...,vn lin.ind.
and T (v1 ), . . . , T (vn ) are linearly independent.

In addition, consider w W. We have, for some v V:
w = T (v)
T (1 v1 + + n vn )
v1 ,...,vn
basis
T linear
1 T (v1 ) + + n T (vn ),
and w arbitrary can be expressed as a linear combination of the vectors T (v1 ), . . . , T (vn );
thus, {T (v1 ), T (v2 ), . . . , T (vn )} is a basis for W.
If T : V W is a linear transformation and 0 is the zero vector of W, then the

set of vectors in v V for which T (v) = 0 is called the kernel of T and is sometimes
denoted by ker(T ).
Theorem: The kernel of a linear transformation T : V W is a vector space.

Proof: Since ker(T ) is a subset of the vector space V it is only necessary to prove that
it is closed under addition and scalar multiplication in order to confirm that it is a
vector space (a subspace of V). Consider u, v ker(T ), and R. We have to prove
that u + v ker(T ), and that u ker(T ). In effect,
T (u + v)
T (u) + T (v)
T linear
T ( u)
0+0=0
u + v ker(T ),
u,vker(T )
T (u)
T linear
.0 = 0
u ker(T ),
uker(T )
and this completes the proof.
The nullity of a linear transformation T is the dimension of the vector space that
is the kernel of T . The kernel of T is sometimes called the nullspace of T .
The following theorem is given without proof.

Theorem: If T : V W is a linear transformation and the vector space V has dimension n, then:
rank (T) + nullity (T) = n
6
Note: rank(T ) is the dimension of the range of T (a vector space in W), whereas
nullity(T ) is the dimension of the kernel of T (a vector space in V).
Example: Let T : R3 R2 be defined by:
T ( (v1 , v2 , v3 ) ) = (v1 , v2 ).
Show that T is a linear transformation; find the rank of T and a basis of the rangespace;
find the nullity of T and a basis of the kernel. Finally, decide whether or not T is oneto-one, onto, or an isomorphism.
* Consider u, v R3 , and R:
T (u+v) = T ((u1 +v1 , u2 +v2 , u3 +v3 )) = (u1 +v1 , u2 +v2 ) = (u1 , u2 )+(v1 , v2 ) =
= T (u) + T (v).
Also
T (u) = T ((u1 , u2 , u3 )) = T ((u1 , u2 , u3 )) = (u1 , u2 ) = (u1 , u2 ) = T (u)
and T is linear.
* Obviously, R(T ) = R2 , rank(T ) = 2 and B0 = {e1 = (1, 0), e2 = (0, 1)} is a basis
for the rangespace.
* We can compute the nullity of T by

n rank(T ) = nullity(T ) = 3 2 = 1.
To find the kernel, we have to find the vectors (v1 , v2 , v3 ) R3 such that T ((v1 , v2 , v3 )) =
(v1 , v2 ) = (0, 0), that is, the nullspace is the set of vectors of the form (0, 0, v3 ), v3
R. A basis for this subspace is given by B1 = {e3 = (0, 0, 1)}.
* Finally, T cannot be an isomorphism, because dim(R3 ) 6= rank(T ). Since (1, 1, 1) 6=

(1, 1, 1) and T ((1, 1, 1)) = (1, 1) = T ((1, 1, 1)), we conclude that T is not oneto-one, but it is onto, since for every (u1 , u2 ) R2 there is at least one vector in
R3 , for instance (u1 , u2 , 0) such that T ( (u1 , u2 , 0) ) = (u1 , u2 ).
If T1 : U V and T2 : V W are linear transformations, respectively, from the

vector space U to vector space V, and from V to vector space W, then the composition,
denoted T2 oT1 is a transformation U W, defined for u U by:
(T2 oT1 )(u) = T2 (T1 (u)).
Note: T1 (u) is in V so T2 maps this to the value T2 (T1 (u)) in W.
Theorem: Given two linear transformations T1 : U V and T2 : V W:
a) the composition T2 oT1 is also a linear transformation from U W.
b) If T2 and T1 are both one-to-one then T2 oT1 is also one-to-one.
c) If T2 and T1 are both onto transformations then T2 oT1 is also onto.
Example: Consider again T : R3 R2 defined by:

T ( (v1 , v2 , v3 ) ) = (v1 , v2 ),
and let T 0 : R2 R3 be the transformation defined by
T 0 ((w1 , w2 )) = (w1 , w2 , 0).
Let us show that T 0 is linear, and describe the properties of T 0 oT (kernel and a basis,
nullity, rangespace and a basis, rank, whether or not is one-to-one, onto or isomorphism).
* Consider u, v R2 , and R:
T 0 (u + v) = T ((u1 + v1 , u2 + v2 )) = (u1 + v1 , u2 + v2 , 0) = (u1 , u2 , 0) + (v1 , v2 , 0) =
= T 0 (u) + T 0 (v).
Also
T (u) = T 0 ((u1 , u2 )) = T 0 ((u1 , u2 )) = (u1 , u2 , 0) = (u1 , u2 , 0) = T 0 (u)
and T 0 is also linear.
* The composition T 0 oT is given by

(T 0 oT )(u1 , u2 , u3 ) = (u1 , u2 , 0).
Then, rank(T 0 oT ) = 2 and B0 = {e1 = (1, 0, 0), e2 = (0, 1, 0)} is a basis for the
rangespace of the composition.
* The nullity of T 0 oT by
n rank(T 0 oT ) = nullity(T 0 oT ) = 3 2 = 1.
The kernel consists of vectors (v1 , v2 , v3 ) R3 such that (T 0 oT )((v1 , v2 , v3 )) =
(v1 , v2 , 0) = (0, 0, 0), that is, again the nullspace is the set of vectors of the form
(0, 0, v3 ), v3 R. A basis for this subspace is given by B1 = {e3 = (0, 0, 1)}.
* Finally, we have again that T 0 oT cannot be an isomorphism, because dim(R3 ) 6=
rank(T 0 oT ). If we consider again the vectors (1, 1, 1) 6= (1, 1, 1), we have
(T 0 oT )((1, 1, 1)) = (1, 1, 0) = (T 0 oT )((1, 1, 1)), and it follows that T is not oneto-one, but it is onto, since for every (u1 , u2 , 0) R3 there is at least one vector
in R3 , for instance (u1 , u2 , 0) itself, such that (T 0 oT )( (u1 , u2 , 0) ) = (u1 , u2 , 0).

Linear Transformations: 4.1 Definitions and Properties

Uploaded by

Copyright:

Available Formats

Linear Transformations: 4.1 Definitions and Properties

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Linear Transformations: 4.1 Definitions and Properties

Uploaded by

Copyright:

Available Formats

Chapter 4

Definitions and properties

ii) T (v1 ) = f(v1 ), for every v1 V and every R,

If T : V W is a linear transformation then T is said to be one-to-one if for every

Theorem: The range of the linear transformation T : V W is a vector space.

and since V is a vector space, v1 + v2 V, therefore

w1 + w2 R(T ) (closed under addition)

and since v V we have that

w R(T ) (closed under scalar multiplication)

Therefore, R(T ) is a vector space.

This theorem warrants the following definition.

R(T ) vector space

T1 : P2 P2 , defined by T1 (a0 + a1 x + a2 x2 ) = a2 x2 . T1 is linear. Consider

= (a2 + b2 )x2 = a2 x2 + b2 x2 = T1 (p(x)) + T1 (q(x)).

Theorem: If T : V W is a one-to-one linear transformation and the range of T is W 0

and therefore, T 1 is linear.

Theorem: If T : V W is an isomorphism between V and W and {v1 , v2 , . . . , vn } is a

Since T (0) = 0 and T is an isomorphism:

and T (v1 ), . . . , T (vn ) are linearly independent.

If T : V W is a linear transformation and 0 is the zero vector of W, then the

Theorem: The kernel of a linear transformation T : V W is a vector space.

and this completes the proof.

The following theorem is given without proof.

* We can compute the nullity of T by

* Finally, T cannot be an isomorphism, because dim(R3 ) 6= rank(T ). Since (1, 1, 1) 6=

If T1 : U V and T2 : V W are linear transformations, respectively, from the

Example: Consider again T : R3 R2 defined by:

* The composition T 0 oT is given by

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.