You are on page 1of 6

Mathematics 3410 - Homework 6 Solutions

Only selected problems will be graded.

1. (Nicholson 7.2.1(a)) For the matrix


 
1 2 −1 1
A= 3 1 0 2 ,
1 −3 2 0

find a basis for the kernel and image of TA , and find the rank and nullity of TA .
Solution. Using SymPy, we find that a row reduced echelon form of A is
 
1 0 1/5 3/5
R =  0 1 −3/5 1/5  .
0 0 0 0

Notice that there are 2 pivot variables, so that rank(A) = 2. The pivot variables
are x1 , x2 and the non-pivot (free) variables are x3 and x4 . Note that solving
Ax = ~0 is the same as solving Rx = ~0 where x = (x1 x2 x3 x4 )T . We see that
1 3
x1 = − x3 − x4
5 5
3 1
x2 = x3 − x4 .
5 5
It follows that
− 15 x3 − 53 x4
 1
− 35
    
−5
 3 x3 − 1 x4   3  1 
 + x4  − 5 

x= 5 5  = x3  5
 x3   1   0 
x4 0 1

and thus
− 51 − 53
   
3   1 
5  ,  − 5  .

ker(TA ) = null(A) = span 
 1   0 
0 1
Also, since R has pivots in columns 1 and 2, it follows that a basis for col(A)
comes from columns 1 and 2 of A. Therefore,
   
1 2
im(TA ) = col(A) = span  3  ,  1  .
1 −3

2. (Nicholson 7.2.2.(b)) (i) Find a basis of ker T , and (ii) find a basis of im T . You
may assume that T is linear.
(b) T : P2 → R2 where
T (p(x)) = (p(0), p(1)).
Solution. (i) Note that

ker(T ) = {p ∈ P2 | T (p(x)) = (0, 0)}


= {p ∈ P2 | (p(0), p(1)) = (0, 0)}
= {p ∈ P2 | p(0) = p(1) = 0}.

1
Note that p(0) = 0 means that x divides p(x) and p(1) = 0 means that x − 1
divides p(x). It follows that

p(x) = x(x − 1)q(x)

where q(x) is another polynomial. Since the degree of p is at most 2, it follows


that q(x) is a constant. Therefore, p(x) = ax(x − 1) with a ∈ R. Therefore

ker(T ) = {ax(x − 1) | a ∈ R} = span(x(x − 1)).

Since any non-zero vector (polynomial in this case) is linearly independent,

{x(x − 1)}

is a basis for ker(T ).


(ii) Note that

im(T ) = {(p(0), p(1)) | p ∈ P2 }


= {(a, a + b + c) | a + bx + cx2 ∈ P2 }
= {(a, a + b + c) | a, b, c ∈ R}.

We claim that this set equals all of R2 . Note that given (x, y) ∈ R2 Then we
would need to solve
x = a and y = a + b + c.
There are many solutions of this. Just take a = x, b = y, and c = 0. It follows
that R2 ⊆ im(T ). Since we have by definition im(T ) ⊆ R2 , we obtain

im(T ) = R2 .

We can choose any basis of R2 . For instance, the standard basis e1 = (1, 0) and
e2 = (0, 1) works. Thus a basis of im(T ) is {e1 , e2 }.
3. (Nicholson 7.2.2(e)) (i) Find a basis of ker T , and (ii) find a basis of im T . You
may assume that T is linear.
(e) T : M22 → M22 where
   
a b a+b b+c
T =
c d c+d d+a

Solution. (i) Note that


n a b  o
ker(T ) = ∈ M22 | a + b = b + c = c + d = d + a = 0 .
c d

Note that the last set of equations can be written in matrix form as Ax = ~0 where
x = (a b c d)T and  
1 1 0 0
 0 1 1 0 
A=  0 0 1 1 

1 0 0 1
Row reducing using SymPy, we find that a row reduced echelon form of R is
 
1 0 0 1
 0 1 0 −1 
R=  0 0

1 1 
0 0 0 0

2
Solving Rx = ~0, we see that

a = −d, b = d, c = −d
 
a b
Thus ∈ ker(T ) if
c d
     
a b −d d −1 1
= =d .
c d −d d −1 1

and  
−1 1
ker(T ) = span .
−1 1
Therefore n  −1 1  o
B=
−1 1
is a basis since it is a single non-zero matrix. (Any non-zero vector is linearly
independent).
(ii) We now compute a basis of im(T ). Note that
n  a b  o
im(T ) = T | a, b, c, c ∈ R
c d
n a + b b + c  o
= | a, b, c, c ∈ R .
c+d d+a

Observe that
         
a+b b+c 1 0 1 1 0 1 0 0
=a +b +c +d .
c+d d+a 0 1 0 0 1 0 1 1

Thus im(T ) lies in the span of the four matrices appearing in the last equation.
However, notice that the last matrix is a linear combination of the first three
since        
0 0 1 0 0 1 1 1
= + − .
1 1 0 1 1 0 0 0
It follows that
 1 0   0 1   1 1 
im(T ) = span , , . (1)
0 1 1 0 0 0

Moreover, we notice that these three matrices are linearly independent. Let
c1 , c2 , c3 ∈ R and suppose
       
1 0 1 1 0 1 0 0
c1 + c2 + c3 = .
0 1 0 0 1 0 0 0

This simplifies to    
c1 + c2 c2 + c3 0 0
= .
c3 c1 0 0
From the bottom row of the matrices, we see that c1 = c3 = 0. Using these values
in the top row, it follows that c2 = 0. It follows that the set of matrices in (1) is
a basis. Thus a basis of im(T ) is given by
(     )
1 0 0 1 1 1
, , .
0 1 1 0 0 0

3
4. Nicholson 7.2.3. Let P : V → R and Q : V → R be linear transformations, where
V is a vector space. Define T : V → R2 by T (v) = (P (v), Q(v)).
(a) Show that T is a linear transformation.
(b) Show that ker T = ker P ∩ ker Q, the set of vectors in both ker P and ker Q.
Solution. (a) (i) Note that

T (~0) = (P (~0), Q(~0)) = (~0, ~0)

since P and Q are linear transformations.


(ii) Let v, w ∈ V . Then

T (v + w) = (P (v + w), Q(v + w)) by definition of T


= (P (v) + P (w), Q(v) + Qw)) since P and Q are linear transformations
= (P (v) + Q(v)) + (P (w) + Q(w)) by addition in R2
= T (v) + T (w) by definition of T.

(iii) Let c ∈ R and v ∈ V . Then

T (cv) = (P (cv), Q(cv)) by definition of T


= (cP (v), cQ(v))) since P and Q are linear transformations
= c(P (v), Q(v)) by scalar mutliplication in R2
= cT (v) by definition of T.

It follows that T is a linear transformation.


(b) Note that

ker(T ) = {v ∈ V | T (v) = (0, 0)} by definition of kernel


= {v ∈ V | (P (v), Q(v)) = (0, 0)} by definition of T
= {v ∈ V | P (v) = 0 and Q(v) = 0}
= {v ∈ V | P (v) = 0} ∩ {v ∈ V | Q(v) = 0}
= ker(P ) ∩ ker(Q).

5. Nicholson 7.2.6 (a),(b),(c),(j),(k) In each case either prove the statement or give
an example in which it is false. Throughout, let T : V → W be a linear transfor-
mation where V and W are finite dimensional.
(a) If V = W , then ker T ⊆ im T .
(b) If dim V = 5, dim W = 3, and dim(ker T ) = 2, then T is onto.
(c) If dim V = 5 and dim W = 4, then ker T 6= {0}.
(j) If dim V ≤ dim W , then T is one-to-one.
(k) If T is onto, then dim V ≥ dim W .
Solution.
(a) This statement is false. Let T : R2 → R2 satisfy

T ((x, y)) = (x + y, 0).

Note that

ker(T ) = {(x, y) ∈ R2 | (x + y, 0) = (0, 0)}


= {(x, y) ∈ R2 | y = −x}
= {(x, −x) ∈ R2 | x ∈ R}
= span((1, −1)).

4
On the other hand,

im(T ) = {(x + y, 0) | x, y ∈ R}
= {(x, 0) | x ∈ R}
= span((1, 0)).

Note that the sets


ker(T ) 6= im(T )
and in fact neither set can be contained in the either. There only intersection
point is (0, 0).
(b) This statement is true. Note that by the dimension formula

dim(ker(T )) + dim(im(T )) = 5

Since dim(ker(T )) = 2 it follows that dim(im(T )) = 5 − 2 = 3. Since


im(T ) ⊆ W is a subspace and t dim(im(T )) = dim(W ) = 3 it follows that
im(T ) = W .
(c) This statement is true. Note that by the dimension formula

dim(ker(T )) + dim(im(T )) = 5

Since dim(im(T )) ≤ dim(W ) = 4 it follows that

dim(ker(T )) ≥ 5 − 4 = 1.

Therefore ker(T ) is at least a one-dimensional subspace of V and thus cannot


be the zero vector.
(j) This is false. Just let V = W and find a T where ker(T ) 6= {~0}. For
instance, take the example from question (a) above. That is, V = W = R2
and T ((x, y)) = (x + y, 0). In this case, ker(T ) = span((1, −1)) and thus T
is not one-to-one (Since T is one-to-one if and only if ker(T ) = {~0}).
(k) This is true. Note that if T is onto, then im(T ) = W and thus

dim(im(T )) = dim(W ). (2)

By the dimension theorem,

dim(ker(T )) + dim(im(T )) = dim(V ).

Subbing in, equation (2) we find

dim(ker(T )) + dim(W ) = dim(V ).

Since dim(ker(T )) ≥ 0, we obtain dim(W ) ≤ dim(V ).


6. (Nicholson 7.2.7) Show that linear independence is preserved by one-to-one trans-
formations and that spanning sets are preserved by onto transformations. More
precisely, if T : V → W is a linear transformation, show that:
(a) If T is one-to-one and {v1 , . . . , vn } is independent in V , then {T (v1 ) , . . . , T (vn )}
is independent in W .
(b) If T is onto and V = span {v1 , . . . , vn }, then W = span {T (v1 ) , . . . , T (vn )}.
Solution. (a) Let c1 , . . . , cn ∈ R and assume

c1 T (v1 ) + · · · cn T (vn ) = ~0.

5
Since T is a linear transformation c1 T (v1 ) + · · · cn T (vn ) = T (c1 v1 + · · · + cn vn )
and T (~0) = ~0. Thus,
T (c1 v1 + · · · + cn vn ) = T (~0).
As T is one-to-one, it follows that

c1 v1 + · · · + cn vn = ~0.

Since {v1 , . . . , vn } is linearly independent, it follows that c1 = c2 = · · · = cn = 0.


Therefore {T (v1 ) , . . . , T (vn )} is independent.
(b) Let w ∈ W . Since T is onto, there exists v ∈ V such that

T (v) = w.

Since V = span {v1 , . . . , vn }, it follows that there exists c1 , . . . , cn ∈ R such that

v = c1 v1 + · · · + cn vn .

Therefore
T (c1 v1 + · · · + cn vn ) = w
and as T is a linear transformation

c1 T (v1 ) + · · · + cn T (vn ) = w.

This implies that w ∈ span {T (v1 ) , . . . , T (vn )} and therefore

W = span {T (v1 ) , . . . , T (vn )} .

7. (Nicholson 7.2.16) Use the dimension theorem to prove Theorem 1.3.1: If A is an


m × n matrix with m < n, the system Ax = 0 of m homogeneous equations in n
variables always has a nontrivial solution.
Solution. Recall that

null(A) = {x ∈ Rn | Ax = 0}.

Note that Ax = 0 has a non-trivial solution if and only if null(A) has a non-zero
solution. This is if and only if

dim(null(A)) ≥ 1.

Let TA : Rn → Rm given by
TA (x) = Ax.
The dimension theorem tells us that

dim(ker(TA )) + dim(im(TA )) = n

or
dim(null(A)) + dim(col(A)) = n
since null(A) = ker(TA ) and col(A) = im(TA ). Since dim(col(A)) = rank(A) ≤ m.
It follows that
dim(null(A)) = n − rank(A) ≥ n − m ≥ 1,
since m < n.

You might also like