Solutions to Homework 11

Olena Bormashenko
December 11, 2011

1. Check the 10 properties of vector spaces to see whether the following sets
with the operations given are vector spaces. If they are vector spaces, give
an argument for each property showing that it works; if not, provide an
example (with numbers) showing a property that does not work.
(a) V is the set of 2 × 2 matrices of the form
 
1 a
A=
0 1
with the operations defined as
A  B = AB (normal matrix multiplication)
c  A = cA
Note: Don’t forget that these definitions only apply to the 2 × 2
matrices in the form I wrote above!

Solution:

This is NOT a vector space. It fails many of the properties, in

particular, property (B) which states that c · ~x must be in V for any
~x ∈ V . Here, letting  
1 1
A=
0 1
and c = 2, we see that
 
2 2
c · A = cA = ∈
/V
0 2
Note: I actually wanted this to be a vector space! I just goofed and
defined scalar multiplication wrong. . .
(b) V is R2 , with addition and scalar multiplication defined as follows:
     
x1 y x + y1
 1 = 1
x2 y2 x2 + y2
   
x 2cx1
c 1 =
x2 2cx2

1
 Solution:

This is NOT a vector space. It fails property (7), which states

that a  (b  ~u) = (ab)  ~u. For example, if
 
1
~u =
1

and a = b = 2, then we see that

      
1 4 16
a  (b  ~u) = 2  2  =2 =
1 4 16

while    
1 8
(ab)  ~u = 4  =
1 8
so we see that a  (b  ~u) 6= (ab)  ~u.
(c) V is the set of functions from R to the positive real numbers: that is,
V is the set of functions f with domain R such that f (x) > 0 for all
x ∈ R. Let f and g be in V . We define f  g and c  f by defining
the values these functions take on at any x ∈ R:

(f  g)(x) = f (x)g(x)
(c  f )(x) = f (x)c

Note: In questions like this, it’s good to be strict about your nota-
tion! Here, f and g are functions, while f (x) and g(x) are values that
those functions take at x. So for example, notation like f (x)  g(x)
doesn’t make any sense, because the operations  and  are only
defined on the elements in the vector space, and not on the values
the functions take.

Solution:

This IS a vector space. Let’s check the properties. Since here, our
‘vectors’ are functions, I’ll be calling them f, g and h when checking
the properties. The scalars will be c and d.
(A) f  g ∈ V : Clearly, f  g is a function defined on all of R. Thus,
the only thing to check is that (f  g)(x) > 0 for all x. But we
have that
(f  g)(x) = f (x)g(x)
and since f (x) > 0 and g(x) > 0, we have that (f  g)(x) > 0
for all x, as required.

2
(B) c  f ∈ V : Since f (x) > 0, we have that (c  f )(x) = f (x)c is
defined for all x ∈ R, and furthermore, this is clearly positive.
Therefore, c  f is in V .
(1) f  g = g  f : To check that two functions are equal, we need
to check that they are the same at all x ∈ R. Thus, we need to
check that (f  g)(x) = (g  f )(x) for all x ∈ R. We have that
(f  g)(x) = f (x)g(x) = g(x)f (x) = (g  f )(x)
so the property holds.
(2) f  (g  h) = (f  g)  h: Again, we need to check that the
functions are the same at all x ∈ R. If x ∈ R, then using the
definition,
(f  (g  h))(x) = f (x)(g  h)(x) = f (x)g(x)h(x)
= (f  g)(x)h(x) = ((f  g)  h)(x)
as required.
(3) There exists an element ~0 ∈ V such that ~0  f = f = f  ~0 for
all f : Here, we need to define the zero vector. Since this is a
function, we need to state its value for all x ∈ R. Let z be the
function such that z(x) = 1 for all x. Let us show that z is the
zero vector for this vector space: for all x ∈ R, we have that
(f  z)(x) = f (x)z(x) = f (x) = (z  f )(x)
which shows that (f  z) = f = (z  f ), as required.
(4) There exists an element −f ∈ V such that f  (−f ) = ~0 =
(−f )  f for all f ∈ V : Since our ~0 is the function z from above,
we need to define −f to fit it. We define
1
(−f )(x) =
f (x)
It is easy to check that this is a function in V , as its domain is R,
and it only takes on positive values. Now, note that for x ∈ R,
1
(f  (−f ))(x) = f (x)(−f )(x) = f (x) =1
f (x)
= z(x) = (−f )(x)f (x) = ((−f )  f )(x)
Thus, for this definition of −f , f  (−f ) = z = (−f )  f , as
required.
(5) c  (f  g) = (c  f )  (c  g): Let x ∈ R. Then, using the rules
from above,
(c  (f  g))(x) = (f  g)(x)c = (f (x)g(x))c = f (x)c g(x)c
= (c  f )(x)(c  g)(x) = ((c  f )  (c  g))(x)
so the functions are equal for all x ∈ R, and the property holds.

3
 (6) (c + d)  f = (c  f )  (d  f ): Let x ∈ R. Then,
((c + d)  f )(x) = f (x)c+d = f (x)c f (x)d
= (c  f )(x)(d  f )(x) = ((c  f )  (d  f ))(x)
so the functions are equal for all x ∈ R, and the property holds.
(7) (cd)  f = c  (d  f ): Let x ∈ R. Then,
c
((cd)  f )(x) = f (x)cd = (f (x)d )c = [(d  f )(x)]
= (c  (d  f ))(x)
as required.
(8) 1  f = f : Let x ∈ R. Then,
(1  f )(x) = f (x)1 = f (x)
so the property holds, as required.
2. Check whether the following sets W are non-empty subsets of V , closed
under vector addition, and closed under scalar multiplication to check
whether they are subspaces of the provided vector spaces V . If they are,
show that everything works; if they are not, give an example that doesn’t
work. (You do NOT need to check whether V is a vector space!)
Note: If V = Rn , we assume that it has the standard vector addition
and scalar multiplication, and similarly for V = Mmn (the set of m ×
n matrices.) Feel free to use the normal plus and scalar multiplication
notation in those cases, instead of  and .

(a) Is W = {c1 [1, 2] + c2 [1, −1] c1 , c2 ∈ R} a subspace of R2 ? (Here,

we’re allowing vectors in R2 to be row vectors.)

Solution:

This IS is a subspace. It’s easy to check that it is a non-empty

subset of R2 (clearly, all the vectors in it have two coordinates, and
for example, the vector [1, 2] is in W .) Thus, we need to check that
it’s closed under addition and closed under scalar multiplication.
(1) ~x + ~y ∈ W for all ~x, ~y ∈ W : Let ~x and ~y be in W . Then, by
definition of W , that means there exist scalars c1 , c2 , d1 , and d2
such that
~x = c1 [1, 2] + c2 [1, −1]
~y = d1 [1, 2] + d2 [1, −1]
Then, we see that
~x + ~y = (c1 + d1 )[1, 2] + (c2 + d2 )[1, −1]
This is clearly in W since c1 + d1 and c2 + d2 are scalars.

4
 (2) a~x ∈ W for all a ∈ R, ~x ∈ W : Let ~x be in W . We can write it
just like in part (1) above. Then,
a~x = a(c1 [1, 2] + c2 [1, −1]) = (ac1 )[1, 2] + (ac2 )[1, −1]
which is clearly in W since ac1 and ac2 are scalars.
(b) Is W = {the set of 2 × 2 diagonal matrices} a subspace of M22 ?

Solution:

This IS is a subspace. It’s clearly a subset of M22 , and it’s clearly

non-empty. Let us now check the other two subspace properties.
Since the ‘vectors’ in this vector space are matrices, we will use the
letters A and B for them.
(1) A + B ∈ W for all A, B ∈ W : Let A and B be in W . Then, we
must have that
   
a 0 b 0
A= 1 ,B = 1
0 a2 0 b2
for some a1 , a2 , b1 and b2 . Then, we see that
 
a1 + b2 0
A+B =
0 a2 + b2
which is clearly a diagonal matrix, and hence is in W .
(2) cA ∈ W for all c ∈ R, A ∈ W : Let A be in W , and let it be
written like it is in part (1) above. Then,
   
a 0 ca1 0
cA = c 1 =
0 a2 0ca2
which is clearly a diagonal matrix, and hence is in W .
(c) Let W be the set of functions from [0, 1] to R such that f (0) = 1,
and let V be the vector space of all functions from [0, 1] to R with
the vector addition and scalar multiplication defined normally, that
is:
(f  g)(x) = f (x) + g(x)
(c  f )(x) = cf (x)
Is W a subspace of V ?

Solution:

This is NOT a vector space. It’s easy to check that the function f
such that f (x) = x+ 1 is in W , while 2f is not in W , since 2f (0) = 2.
Thus, property (2) doesn’t hold. (Property (1) can be easily demon-
strate not to hold, either.)

5
 ~ m
(d) Let A be an m n × n matrix,
o and let b be a vector in R . For which
values of ~b is ~x A~x = ~b a subspace of Rn ?

Solution:
n o
Assume that W = ~x A~x = ~b is a subspace of Rn . Then, if ~x0
is in W , so is 2~x0 . Since ~x0 is in W , we have that

A~x0 = ~b

This means that

A(2~x0 ) = 2A~x0 = 2~b
But for 2~x0 to be in W , we must by definition have A(2~x0 ) = ~b.
Therefore, we must have that
~b = 2~b

which implies that our only possibility is ~b = ~0.

Now, let us show that if ~b = ~0, then we indeed have a subspace. Let
n o
W = ~x A~x = ~0

Since A is m × n, this is clearly a subset of Rn , and since it contains

~0 in Rn , it’s non-empty. Let us now check the other two properties
of subspaces.
(1) ~u + ~v ∈ W for all ~u, ~v ∈ W : Let ~u and ~v be in W . Then, by
definition of W , we have that

A~u = ~0
A~v = ~0

Therefore, using the properties of matrix-vector multiplication,

A(~u + ~v ) = A~u + A~v = ~0 + ~0 = ~0

Since a vector is in W precisely if A times that vector is ~0, this

shows that ~u + ~v is in W , as required.
(2) c~u ∈ W for all c ∈ R, ~u ∈ W : Similarly to above, we have that

A~u = ~0

Therefore,
A(c~u) = cA~u = c~0 = ~0
which shows that c~u is in W , as required.

6
 (e) Is W = {A ∈ M22 |A| = 1} a subspace of M22 ?

Solution:

This is NOT a subspace. For example,

 
1 0
I2 =
0 1

satisfies |I2 | = 1 and hence is in W , but the determinant of 2I2 can

be easily checked to be 4, and therefore, 2I2 is not in W . This means
that property (2) doesn’t hold.

3. Prove the following statements using the properties of vector spaces – at

each step, name which property you’re using to justify it. Be careful not
to use ‘intuitively obvious’ statements which aren’t in the properties!
(Yes, these are the statements that you were asked to show in class – I’m
not sure most people got there, so I’m giving them again.)

(a) a~0 = ~0 for all scalars a

Proof:
By property (3), ~0 + ~0 = ~0. Therefore, using property (5),

a~0 = a(~0 + ~0) = a~0 + a~0

By property (4), there exists an element (−a~0) (which we don’t know

to be the same as (−a)~0 – it’s merely notation for the additive inverse
of a~0) such that a~0 + (−a~0) = ~0. Thus, adding it to the right of the
above equation,

a~0 + (−a~0) = (a~0 + a~0) + (−a~0)

Note that we have to put in the brackets to make sure that we’re
doing the exact same thing to both sides. Now, using property (2)
to rearrange the right-hand side,

a~0 + (−a~0) = a~0 + (a~0 + (−a~0))

and applying property (4),

~0 = a~0 + ~0

Finally, by property (3), a~0 + ~0 = a~0, and hence

~0 = a~0

as required. PHEW!

7
(b) 0~v = ~0 for all ~v ∈ V .

Proof:
This proof is pretty similar to the last one. Clearly, since 0 is just a
scalar, 0 + 0 = 0. Therefore, using property (6),
0~v = (0 + 0)~v = 0~v + 0~v
By property (4), there exists a vector −0~v such that 0~v + (−0~v ) = ~0.
Adding it to both sides on the right,
0~v + (−0~v ) = (0~v + 0~v ) + (−0~v )
Rearranging using property (2), and simplifying the left-hand side
with property (4),
~0 = 0~v + (0~v + (−0~v ))
Finally, using property (4), then property (3),
~0 = 0~v + ~0 = 0~v

as required.

(c) (−1)~v = −~v for all ~v ∈ V .

Proof:
By property (8), 1~v = ~v . Therefore,
~v + (−1)~v = 1~v + (−1)~v
Using property (6), 1~v + (−1)~v = (1 + (−1))~v = 0~v . From part (b)
above, we know that 0~v = ~0. Thus,
~v + (−1)~v = ~0
From property (4), there exists a vector −~v (not necessarily equal
to (−1)~v – that’s what we have to prove!) such that (−~v ) + ~v = ~0.
Adding it to both sides on the left, we get
(−~v ) + (~v + (−1)~v ) = (−~v ) + ~0
Using property (2) on the left, and property (3) on the right,
((−~v ) + ~v ) + (−1)~v = −~v
Using property (4),
~0 + (−1)~v = −~v
and finally, using property (3),
(−1)~v = −~v
as required.