Three.

I Isomorphisms

Linear Algebra, edition four

Jim Hefferon

http://joshua.smcvt.edu/linearalgebra
Definition
Example We have the intuition that the vector spaces R2 and P1 are “the
same,” in that they are two-component spaces. For instance
 
1
is just like 1 + 2x,
2
 
−3
and is just like − 3 + (1/2)x,
1/2

etc. What makes the spaces alike, not just the sets, is that the association
persists through the operations: this illustrates addition
     
1 −3 −2
+ =
2 1/2 5/2
is just like (1 + 2x) + (−3 + (1/2)x) = −2 + (5/2)x

and this illustrates scalar multiplication.

   
1 3
3 = is just like 3(1 + 2x) = 3 + 6x
2 6
Example Similarly, we can link each two-tall vector with a linear
polynomial.  
a
←→ a + bx
b
Example Similarly, we can link each two-tall vector with a linear
polynomial.  
a
←→ a + bx
b
This association holds through the vector space operations of addition
     
a1 a2 a1 + a2
+ =
b1 b2 b1 + b2
←→ (a1 + b1 x) + (a2 + b2 x) = (a1 + a2 ) + (b1 + b2 )x

and scalar multiplication.

   
a ra
r = ←→ r(a + bx) = (ra) + (rb)x
b rb

We say that the association preserves the structure of the spaces.

Example We can think of M2×2 as “the same” as R4 if we associate in this
way.
a
 
 
a b b
←→  
c d c
d
For instance, these are corresponding elements.

1
 
 
1 −1 −1
←→  
2 −2 2
−2
Example We can think of M2×2 as “the same” as R4 if we associate in this
way.
a
 
 
a b b
←→  
c d c
d
For instance, these are corresponding elements.

1
 
 
1 −1 −1
←→  
2 −2 2
−2
This association persists under addition.
     
a1 b1 a2 b2 a1 + a2 b1 + b2
+ =
c1 d1 c2 d2 c1 + c2d1 + d 2
a1 a2 a1 + a2
     
 b1   b2   b1 + b2 
←→   c1  +  c2  =  c1 + c2 
    

d1 d2 d1 + d2
Here is an example of addition being preserved under this association.
     
1 −1 0 4 1 3
+ =
2 −2 3 −3 5 −5
1 0 1
     
−1  4   3 
←→  + = 
2 3 5
−2 −3 −5
 Here is an example of addition being preserved under this association.
     
1 −1 0 4 1 3
+ =
2 −2 3 −3 5 −5
1 0 1
     
−1  4   3 
←→  + = 
2 3 5
−2 −3 −5

The association also persists through scalar multiplication.

a ra
   
   
a b ra rb b rb
r· = ←→ r ·   c  =  rc 
  
c d rc rd
d rd

This illustrates.
1 2
   
   
1 −1 2 −2 −1 −2
2· = ←→ 2·
 2 = 4 
  
2 −2 4 −4
−1 −4
Isomorphism
1.3 Definition An isomorphism between two vector spaces V and W is a
map f : V → W that
1) is a correspondence: f is one-to-one and onto;
2) preserves structure: if ~v1 ,~v2 ∈ V then

f(~v1 + ~v2 ) = f(~v1 ) + f(~v2 )

and if ~v ∈ V and r ∈ R then

f(r~v) = rf(~v)
∼ W, read “V is isomorphic to W”, when such a map exists).
(we write V =
How-to
To verify that f : V → W is an isomorphism, do these four.
I To show that f is one-to-one, assume that ~v1 ,~v2 ∈ V are such that
f(~v1 ) = f(~v2 ) and derive that ~v1 = ~v2 .
I To show that f is onto, let w
~ be an element of W and find a ~v ∈ V such
that f(~v) = w
~.
I To show that f preserves addition, check that for all ~v1 ,~v2 ∈ V we have
f(~v1 + ~v2 ) = f(~v1 ) + f(~v2 ).
I To show that f preserves scalar multiplication, check that for all ~v ∈ V
and r ∈ R we have f(r · ~v) = r · f(~v).
The first two cover condition (1), that the spaces correspond, that for each
member of W there exactly one associcated member of V. The latter two
cover (2), that the map preserves structure. For these two, the intuition is
in the discussion above. (Later section cover these two at length.)
Example The space of quadratic polynomials P2 is isomorphic to R3 under
this map.
 
a0
2
f(a0 + a1 x + a2 x ) = a1 
a2
Here are two examples of the action of f.
   
1 3
f(1 + 2x + 3x2 ) = 2 f(3 + 4x2 ) = 0
3 4

To verify that f is an isomorphism we must check condition (1), that f is a

correspondence, and condition (2), that f preserves structure.
 The first part of the definition’s clause (1) is that f is one-to-one. We
suppose f(~v1 ) = f(~v2 ), that the function yields the same output on two
inputs f(a0 + a1 x + a2 x2 ) = f(b0 + b1 x + b2 x2 ). From that, we must derive
that the two inputs are equal. The definition of f gives
   
a0 b0
a1  = b1 
a2 b2

and two column vectors are equal only if their entries are equal
a0 = b0 , a1 = b1 , and a2 = b2 . Thus the original inputs are equal
~v1 = a0 + a1 x + a2 x2 = b0 + b1 x + b2 x2 = ~v2 . So f is one-to-one.
 The first part of the definition’s clause (1) is that f is one-to-one. We
suppose f(~v1 ) = f(~v2 ), that the function yields the same output on two
inputs f(a0 + a1 x + a2 x2 ) = f(b0 + b1 x + b2 x2 ). From that, we must derive
that the two inputs are equal. The definition of f gives
   
a0 b0
a1  = b1 
a2 b2

and two column vectors are equal only if their entries are equal
a0 = b0 , a1 = b1 , and a2 = b2 . Thus the original inputs are equal
~v1 = a0 + a1 x + a2 x2 = b0 + b1 x + b2 x2 = ~v2 . So f is one-to-one.
The second part of (1) is that f is onto. We consider an element of the
codomain  
a0
~ = a1  ∈ R3
w
a2
and produce an element of the domain that maps to it. Observe that w ~ is
the image under f of the member ~v = a0 + a1 x + a2 x2 of the domain. Thus
f is onto.
 The definition’s clause (2) also has two halves. First we show that f
preserves addition. Consider f acting on the sum of two elements of the
domain.

f( (a0 + a1 x + a2 x2 ) + (b0 + b1 x + b2 x2 ) )
= f( (a0 + b0 ) + (a1 + b1 )x + (a2 + b2 )x2 )

The definition of f gives

 
a0 + b 0
= a1 + b1 
a2 + b 2
and that equals
   
a0 b0
= a1  + b1 
a2 b2
which gives
= f( a0 + a1 x + a2 x2 ) + f(b0 + b1 x + b2 x2 )
as required.
 We finish by checking that f preserves scalar multiplication. This is
similar to the check for addition.

f(r · (a0 + a1 x + a2 x2 )) = f( (ra0 ) + (ra1 )x + (ra2 )x2 )

 
ra0
= ra1 
ra2
 
a0
= r · a1 
a2
= r · f( a0 + a1 x + a2 x2 )
 We finish by checking that f preserves scalar multiplication. This is
similar to the check for addition.

f(r · (a0 + a1 x + a2 x2 )) = f( (ra0 ) + (ra1 )x + (ra2 )x2 )

 
ra0
= ra1 
ra2
 
a0
= r · a1 
a2
= r · f( a0 + a1 x + a2 x2 )

So the function f is an isomorphism. Because there is an isomorphism,

the two spaces are isomorphic P2 =∼ R3 .
Example Consider these two vector spaces (under the natural operations)
 
a b
V={ | a, b, c ∈ R } W = { (x y z) | x, y, z ∈ R }
c 0

and consider this function.

 
a b f
7−→ (b 2a a + c)
c 0

Here is an example of the map’s action.

 
3 2 f
7−→ (2 6 4)
1 0

We will verify that f is an isomorphism.

 To show that f is one-to-one, suppose that
   
a1 b1 a2 b2
f( ) = f( )
c1 0 c2 0

Then (b1 2a1 a1 + c1 ) = (b2 2a2 a2 + c2 ). The first entries give that
b1 = b2 , the second entries that a1 = a2 , and with that the third entries
give that c1 = c2 .    
a1 b1 a2 b2
=
c1 0 c2 0
 To show that f is one-to-one, suppose that
   
a1 b1 a2 b2
f( ) = f( )
c1 0 c2 0

Then (b1 2a1 a1 + c1 ) = (b2 2a2 a2 + c2 ). The first entries give that
b1 = b2 , the second entries that a1 = a2 , and with that the third entries
give that c1 = c2 .    
a1 b1 a2 b2
=
c1 0 c2 0
To show that f is onto, consider a member of W.

w
~ = (x y z)

We must find a ~v so that f(~v) = w ~ . The map sends the upper right entry of
the input to the first entry of the output, so the upper right of ~v is x.
Similarly, the upper left of ~v is (1/2)y. With that, the lower left is
z − (1/2)y.
 
y/2 x
(x y z) = f( )
z − y/2 0
 To show that f preserves addition, assume
     
a1 b1 a2 b 2 a1 + a2 b1 + b2
f( + ) = f( )
c1 0 c2 0 c1 + c2 0

which equals (b1 + b2 2(a1 + a2 ) (a1 + a2 ) + (c1 + c2 )). In turn, that

equals this.
   
a1 b1 a2 b 2
(b1 2a1 a1 + c1 ) + (b2 2a2 a2 + c2 ) = f( ) + f( )
c1 0 c2 0
 To show that f preserves addition, assume
     
a1 b1 a2 b 2 a1 + a2 b1 + b2
f( + ) = f( )
c1 0 c2 0 c1 + c2 0

which equals (b1 + b2 2(a1 + a2 ) (a1 + a2 ) + (c1 + c2 )). In turn, that

equals this.
   
a1 b1 a2 b 2
(b1 2a1 a1 + c1 ) + (b2 2a2 a2 + c2 ) = f( ) + f( )
c1 0 c2 0

Preservation of scalar multiplication is similar.

   
a b ra rb
f(r · ) = f( )
c 0 rc 0
= (rb 2ra ra + rc)
= r · (b 2a a + c)
 
a b
= r · f( )
c 0
Preservation is special
Many functions do not preserve addition and scalar multiplication. For
instance, f : R2 → R2    2
x x
f( )=
y y2
does not preserve addition since the sum done one way
       
1 2 3 9
f( + ) = f( )=
0 0 0 0

gives a different result than the sum done the other way.
         
1 2 1 4 5
f( ) + f( )= + =
0 0 0 0 0
Special case: Automorphisms
1.7 Definition An automorphism is an isomorphism of a space with itself.
Special case: Automorphisms
1.7 Definition An automorphism is an isomorphism of a space with itself.
1.8 Example A dilation map ds : R2 → R2 that multiplies all vectors by a
nonzero scalar s is an automorphism of R2 .

d1.5 (~u)
~u d1.5
−→ d1.5 (~v)
~v
Special case: Automorphisms
1.7 Definition An automorphism is an isomorphism of a space with itself.
1.8 Example A dilation map ds : R2 → R2 that multiplies all vectors by a
nonzero scalar s is an automorphism of R2 .

d1.5 (~u)
~u d1.5
−→ d1.5 (~v)
~v

Another automorphism is a rotation or turning map, tθ : R2 → R2 that

rotates all vectors through an angle θ.

tπ/6 tπ/6 (~u)

−→
~u
 A third type of automorphism of R2 is a map f` : R2 → R2 that flips or
reflects all vectors over a line ` through the origin.

f` (~u)

~u f
`
` −→

Checking that each is an isomorphism is an exercise.

 A third type of automorphism of R2 is a map f` : R2 → R2 that flips or
reflects all vectors over a line ` through the origin.

f` (~u)

~u f
`
` −→

Checking that each is an isomorphism is an exercise.

Why study automorphisms? Isn’t it trivial that the plane is just like
itself?
Consider the family of automorphisms tΘ rotating all vectors
counterclockwise. They makes precise the intuition that the plane is
uniform — that space near the x-axis is just like space near the y-axis.
 So one lesson is that we can use maps to describe relationships between
spaces. If the maps are isomorphisms then this relation makes precise the
intuition “just like”.
A second lesson is that while there is an obvious automorphism of R2
   
x x
7→
y y

there are reasons to consider maps other than the obvious one.
1.10 Lemma An isomorphism maps a zero vector to a zero vector.
1.10 Lemma An isomorphism maps a zero vector to a zero vector.
Proof Where f : V → W is an isomorphism, fix some ~v ∈ V. Then
f(~0V ) = f(0 · ~v) = 0 · f(~v) = ~0W . QED
1.11 Lemma For any map f : V → W between vector spaces these statements
are equivalent.
(1) f preserves structure

f(~v1 + ~v2 ) = f(~v1 ) + f(~v2 ) and f(c~v) = c f(~v)

(2) f preserves linear combinations of two vectors

f(c1~v1 + c2~v2 ) = c1 f(~v1 ) + c2 f(~v2 )

(3) f preserves linear combinations of any finite number of vectors

f(c1~v1 + · · · + cn~vn ) = c1 f(~v1 ) + · · · + cn f(~vn )

The book contains the details of both proofs.

Remark Examination of the proofs show that they depend only on

clause (2) of the definition, not on that the map is a correspondence. We
will say more in the next section.
Example The line through the origin with slope two is a vector space.
y

   
x t
L={ | y = 2x } = { | t ∈ R}
y 2t

x
The parametrization
   
t 1
L={ | t ∈ R} = {t · | t ∈ R}
2t 2
suggests that L is just like R: there is a point on L associated with 1 ∈ R, a
point associated with 2 ∈ R, etc. We will show that this function is an
isomorphism between L and R1 .
 
t
f( )=t
2t
 To verify that f is one-to-one suppose that f maps two members of L to
the same output.    
1 1
f( t1 · ) = f( t2 · )
2 2
By the definition of f we have t1 = t2 and so the two input members of L
are equal.
To check that f is onto consider a member of the codomain, r ∈ R. There
is a member of the domain L that maps to it, namely this one.
 
1
f( r · )=r
2

To finish, we combine the two structure checks, using the lemma’s (2).
     
t1 t2 1
f( c1 · + c2 · ) = f( (c1 t1 + c2 t2 ) · )
2t1 2t2 2
   
t1 t2
= c1 t1 + c2 t2 = c1 · f( ) + c2 · f( )
2t1 2t2
Dimension characterizes isomorphism
2.1 Lemma The inverse of an isomorphism is also an isomorphism.
2.1 Lemma The inverse of an isomorphism is also an isomorphism.
Proof Suppose that V is isomorphic to W via f : V → W. An isomorphism
is a correspondence between the sets so f has an inverse function
f−1 : W → V that is also a correspondence.
2.1 Lemma The inverse of an isomorphism is also an isomorphism.
Proof Suppose that V is isomorphic to W via f : V → W. An isomorphism
is a correspondence between the sets so f has an inverse function
f−1 : W → V that is also a correspondence.
We will show that because f preserves linear combinations, so also does
f−1 . Suppose that w ~ 2 ∈ W. Because it is an isomorphism, f is onto and
~ 1, w
there are ~v1 ,~v2 ∈ V such that w~ 1 = f(~v1 ) and w
~ 2 = f(~v2 ). Then

f−1 (c1 · w ~ 2 ) = f−1 c1 · f(~v1 ) + c2 · f(~v2 )

~ 1 + c2 · w
= f−1 ( f c1~v1 + c2~v2 ) = c1~v1 + c2~v2 = c1 · f−1 (~ w1 ) + c2 · f−1 (~


w2 )

since f−1 (~
w1 ) = ~v1 and f−1 (~
w2 ) = ~v2 . With that, by Lemma 1.11 ’s second
statement, this map preserves structure. QED
Example We saw earlier that this line through the origin subspace of R2
 
t
L={ | t ∈ R}
2t

is isomorphic to R1 via this function.

 
1
f( t · )=t
2

The inverse f−1 : R → L given by

   
1 r
f−1 (r) = r · =
2 2r

is also an isomorphism.
2.2 Theorem Isomorphism is an equivalence relation between vector spaces.

The book contains the proof; here is a diagram of what it tells us: the
collection of all finite-dimensional vector spaces is partitioned into classes.
Two spaces are in the same class if they are isomorphic.

V
W ...

The next result characterizes these classes.

2.3 Theorem Vector spaces are isomorphic if and only if they have the same
dimension.

The proof is these two lemmas.

2.3 Theorem Vector spaces are isomorphic if and only if they have the same
dimension.

The proof is these two lemmas.

2.4 Lemma If spaces are isomorphic then they have the same dimension.
2.5 Lemma If spaces have the same dimension then they are isomorphic.

The proof of the first depends on showing that under a isomorphism the
image of a domain basis is a range basis. The proof of the second depends
on giving an explicit isomorphism from an n-dimensional domain to Rn .
See the book for the full proofs; here we will illustrate with an example.
Example The plane 2x − y + z = 0 through the origin in R3 is a vector
space (under the natural operations).

Describe the space as a span by taking that to be a one-equation linear

system and parametrizing x = (1/2)y − (1/2)z.
   
1/2 −1/2
P = {  1  y +  0  z | y, z ∈ R }
0 1

Clearly the set with those two vectors is linearly independent so it is a

basis.    
1/2 −1/2
B = h 1  ,  0 i
0 1
 Here is the basis from the prior slide.
   
1/2 −1/2
B = h 1  ,  0 i
0 1

Because its basis has two vectors, P is a dimension 2 space.

The second lemma’s proof shows that this is an isomorphism: the map
f : P → R2 that associates each element ~v ∈ P with its representation
RepB (~v) ∈ R2 .
To illustrate the association, here is an example of its action on an R2
vector picked at random.
     
7/2 1/2 −1/2  
f 3
~v1 =  3  =  1  · 3 +  0  · (−4) 7−→ RepB (~v1 ) =
−4
−4 0 1
 Here is the basis from the prior slide.
   
1/2 −1/2
B = h 1  ,  0 i
0 1

Because its basis has two vectors, P is a dimension 2 space.

The second lemma’s proof shows that this is an isomorphism: the map
f : P → R2 that associates each element ~v ∈ P with its representation
RepB (~v) ∈ R2 .
To illustrate the association, here is an example of its action on an R2
vector picked at random.
     
7/2 1/2 −1/2  
f 3
~v1 =  3  =  1  · 3 +  0  · (−4) 7−→ RepB (~v1 ) =
−4
−4 0 1

Another (as with the prior one, ~v2 is picked at random from R2 ).
     
−17/4 1/2 −1/2  
f 1/2
~v2 =  1/2  =  1  · (1/2) +  0  · 9 7−→ RepB (~v2 ) =
9
9 0 1
 The first lemma’s proof shows that any isomorphism takes bases to
~ i for the domain and applying an
bases: starting with basis vectors β
isomorphism f gives basis vectors f(β~ i ) for the range.
For the isomorphism from the prior slide we have
   
1/2 −1/2  
~ 1 =  1  · 1 +  0  · 0 7−→ f 1
β
0
0 1

and    
1/2 −1/2  
~2 =  1  · 0 +  0  · 1 f 0
β 7−→
1
0 1
which together make the basis E2 for R2 .
2.8 Corollary Each finite-dimensional vector space is isomorphic to one and
only one of the Rn .
2.8 Corollary Each finite-dimensional vector space is isomorphic to one and
only one of the Rn .

Thus the real spaces Rn form a set of canonical representatives of the

isomorphism classes — every isomorphism class contains one and only
one Rn .

? R0 ? R3

? R2 ...
? R1