You are on page 1of 6

Math 206, Spring 2016

Fake Assignment 9 Solutions

Due: Never

Part A.

(1) Suppose that T : W V , and furthermore that {w1 , , wn } W is given. Prove that if {w1 , , wn }
is linearly dependent if and only if {T (w1 ), , T (wn )} is linearly dependent.
Solution. Note that since T is an isomorphism, it must be injective. We saw in Problem A2 from
assignment 7 that if {w1 , , wn } is linearly independent, then so too is {T (w1 ), , T (wn )}.
For the opposite direction, recall that we have argued in class that T 1 : V W is an isomorphism.
In particular we have that T 1 is injective. Again using Problem A2 from assignment 7, the independence of {T (w1 ), , T (wn )} implies the independence of {T 1 (T (w1 )), T 1 (T (wn ))} = {w1 , , wn }.


(2) Suppose that T : W V , and let U be any subspace. Prove that the function : L(V, U ) L(W, U )
given by (S) = S T is a bijection.
Solution. First, observe that if S L(V, U ) is given, then the function S T has domain W and
codomain U . Furthermore, since T and S are both linear, so too is their composition. Hence S T
L(W, U ), and so the function does indeed take inputs from L(V, U ) to outputs from L(W, U ).
Now we show the map is injective. Suppose S1 , S2 L(V, U ) are given with (S1 ) = (S2 ). Hence
S1 T = S2 T.

()

Now observe that T


: V W is an isomorphism, and (by definition of function inverse) we have
T T 1 = idV . We then have
S1 = S1 idV
= S1 (T T

(composing with the identity does nothing)


1

(our computation above)

(composition is associative)

= (S2 T ) T 1

((S1 ) = (S2 ))

= (S1 T ) T
= S2 (T T

(composition is associative)

= S2 idV
= S2

(our computation above)


(composing with the identity does nothing).

Now well show the map is surjective. Let R L(W, U ) be given. Notice that R T 1 is a function
with domain V and codomain U . Furthermore since both R and T 1 are linear, so is their composition.
Hence R T 1 L(V, U ). Observe that
(R T 1 ) = (R T 1 ) T = R (T 1 T ) = R idW = R.
We conclude is surjective.
[Note: the proof of injectivity might seem to have come out of left field. One might have proceeded
more directly by composing with T 1 on the right side of equation () and then manipulated the
resulting (valid) equality until it became the desired equality. Alternatively, one could have argued that
S1 and S2 were the same by proving they had the same effect on a given input v V . For this, one
uses the fact that T is bijective to find some w W with T (w) = v. Then we have
S1 (v) = S1 (T (w)) = (S1 T )(w) = (S2 T )(w) = S2 (T (w)) = S2 (v).
This would then give the desired result.]

http://palmer.wellesley.edu/~aschultz/w16/math206

Page 1 of 6

Math 206, Spring 2016

Fake Assignment 9 Solutions

Due: Never

(3) Suppose that dim(V ) = 3 and T L(V, V ). Suppose further that there exists some v V so that
(T T )(v) 6= 0 and yet (T T T )(v) = 0.
(a) Prove that B = {v, T (v), T (T (v))} is a basis for V .
Solution. First we check the collection is independent. Suppose we have a relation
c1 v + c2 T (v) + c2 T (T (v)) = 0.
Applying T T to this expression (and using the linearity of T T ) gives
c1 T (T (v)) + c2 T (T (T (v))) + c3 T (T (T (T (v)))) = T (T (0)).
Now we know that T (0) = 0 (since as weve seen earlier this semester any linear transformation must take 0 to 0); hence the right side of this expression becomes
T (T (0)) = T (0) = 0.
In a similar way, using our hypothesis that T (T (T (0))) = 0 we can compute the left side to be
c1 T (T (v)) + c2 T (T (T (v))) + c3 T (T (T (T (v)))) = c1 (T (T (v))) + c2 0 + c3 T (0) = c1 T (T (v)).
We conclude that c1 T (T (v)) = 0. Now since were told T (T (v)) 6= 0, the only way this equation
can hold is if c1 = 0. Hence our original relation is really a relation between T (v) and T (T (v)):
c2 T (v) + c3 T (T (v)) = 0.
If we apply T to this expression and use linearity, we get
c2 T (T (v)) + c3 T (T (T (v))) = 0.
As before, this allows us to conclude that c2 T (T (v)) = 0, and that forces c2 = 0. So our original
relation is really a relation of the form
c3 T (T (v)) = 0.
Using the same argument as above, this then implies c3 = 0. Hence we have c1 = c2 = c3 = 0 as
desired, and so the set is independent.
To see that the set is a basis, note that its a set of 3 linearly independent vectors in a 3 dimensional
space. By our result in class, it must be a basis for V .

(b) Compute RepB,B (T ).
Solution. We know that

RepB,B (T ) = RepB (T (v))

RepB (T (T (v)))

RepB (T (T (T (v))) .

Observe that
T (v) = 0v + 1T (v) + 0T (T (v))
T (T (v)) = 0v + 0T (v) + 1T (T (v))
T (T (T (v))) = 0 = 0v + 0T (v) + 0T (T (v)),
and so we get

0
RepB,B (T ) = 1
0

0
0
1

0
0 .
0


[Note: This is problem 58 from section 3.4 of our textbook.]


http://palmer.wellesley.edu/~aschultz/w16/math206

Page 2 of 6

Math 206, Spring 2016

Fake Assignment 9 Solutions

Due: Never

Part B.
(1) Suppose that T L(W, V ) and that ker(T ) = {0}. Prove that W ' im(T ).
Solution. By rank nullity we know that
dim(W ) = dim(ker(T )) + dim(im(T )).
Since ker(T ) = {0} we have dim(ker(T )) = 0, and so our previous equation gives dim(W ) = dim(im(T )).
Since dimension characterizes isomorphism, we have W ' im(T ).

(2) Let {w0 , , w5 } be linearly independent vectors in Rn , and define
W3 = span{w0 , w1 , w2 , w3 }
W4 = span{w0 , w1 , w2 , w3 , w4 }
W5 = span{w0 , w1 , w2 , w3 , w3 , w5 }.
Define functions T L(W5 , W4 ), S L(W4 , W3 ) and R L(W5 , W3 ) by
T (c0 w0 + c1 w1 + c2 w2 + c3 w3 + c4 w4 + c5 w5 ) = c1 w0 + 2c2 w1 + 3c3 w2 + 4c4 w3 + 5c5 w4
S(c0 w0 + c1 w1 + c2 w2 + c3 w3 + c4 w4 ) = c1 w0 + 2c2 w1 + 3c3 w2 + 4c4 w3
R(c0 w0 + c1 w1 + c2 w2 + c3 w3 + c4 w4 + c5 w5 ) = 2c2 w0 + 6c3 w1 + 12c4 w2 + 20c5 w3 .
(a) It is a fact (that you dont have to prove) that the following sets are bases for W5 and W3 (respectively):
B = {w0 + w2 , w1 + w3 , w2 + w4 , w3 + w5 , w4 + w0 , w5 + w1 }
D = {w0 , w0 + w1 , w0 + w1 + w2 , w0 + w1 + w2 + w3 }.
Compute RepD,B (R).
Solution. The first column of RepD,B (R) is

2
0

RepD (R(w0 + w2 )) = RepD (2w0 ) =


0 .
0
The second column of this matrix is
RepD (R(w1 + w3 )) = RepD (6w1 ) .
For this latter expression we need to solve for c1 , c2 , c3 , c4 R with
6w1 = c1 (w0 ) + c2 (w0 + w1 ) + c3 (w0 + w1 + w2 ) + c4 (w0 + w1 + w2 + w3 ).
Redistributing on the right side and substracting 6w1 from both sides leaves the equation
0 = (c1 + c2 + c3 + c4 )w0 + (c2 + c3 + c4 6)w1 + (c3 + c4 )w2 + c4 w3 .
Linear independence of the collection {w0 ,
if they exist at all correspond precisely to
augmented matrix

1 1 1 1
0 1 1 1

0 0 1 1
0 0 0 1

, w3 } then tells us that the desired coefficients


solutions to the linear system represented by the

http://palmer.wellesley.edu/~aschultz/w16/math206

Page 3 of 6

0
6
.
0
0

Math 206, Spring 2016

Fake Assignment 9 Solutions

Due: Never

The relevant computation will then yield

6
6

RepD (6w1 ) =
0 .
0 D
The rest of the columns are computed in a similar way; we are left with

2 6
2 6
0
0
0
6 12
6 12
0
.
RepD,B (R) =
0
0
12 20
12 20
0
0
0
20
0
20

(b) Use your answer to calculate RepD (R((w0 + w2 ) + (w1 + w3 ) (w3 + w5 ))) via a matrix product.
Solution. We can see that the input vector has

RepB ((w0 + w2 ) + (w1 + w3 ) (w3 + w5 )) =

1
1
0
1
0
0

Hence we have
RepD (R((w0 + w2 ) + (w1 + w3 ) (w3 + w5 ))) = RepD,B (R) RepB ((w0 + w2 ) + (w1 + w3 ) (w3 + w5 ))

1
2 6
2 6
0
0

0
6 12
6 12
0
0
=

0
12 20
12 20
1

0
0
0
0
20
0
20
0

2
6
6
2

0 6
6
0


=
0 + 0 20 = 20 .
20
20
0
0

[Though the problem doesnt ask for it, heres a quick reality check that things are as they should
be. Note that one has
R((w0 + w2 ) + (w1 + w3 ) (w3 + w5 )) = 0 + 2w0 + 0 + 6w1 6w1 20w3 = 2w0 20w3 .
But this is the same vector as
2(w0 ) + 0(w0 + w1 ) + 20(w0 + w1 + w2 ) 20(w0 + w1 + w2 + w3 ).
Hence the output under R of the given vector really does have the stated D-coordinate representation.]
(c) It is a fact (that you dont have to check) that R = ST . Consider the basis C = {w0 , w1 , w2 , w3 , w4 }
of W4 . Calculate RepC,B (T ),RepD,C (S) and then verify (by direct calculation) that
RepD,B (R) = RepD,C (S)RepC,B (T ).
http://palmer.wellesley.edu/~aschultz/w16/math206

Page 4 of 6

Math 206, Spring 2016

Fake Assignment 9 Solutions

Due: Never

Solution. We know that

RepC,B (T ) = RepC (T (w0 + w2 ))

RepC (T (w1 + w3 ))

RepC (T (w5 + w1 )) .

Note that we have

RepC (T (w0 + w2 )) = RepC (2w1 ) =

0
2
0
0
0
0

RepC (T (w1 + w3 ))) = RepC (w0 + 3w2 ) =

RepC (T (w2 + w4 )) = RepC (2w1 + 4w3 ) =

RepC (T (w4 + w0 )) = RepC (4w3 ) =

0
0
0
4
0
0

0
2
0
4
0
0

RepC (T (w3 + w5 ) = RepC (3w2 + 5w4 ) =

RepC (T (w5 + w1 ))) = RepC (5w4 + w1 ) =

1
0
3
0
0
0

0
0
3
0
5
0

0
1
0
0
5
0

Hence we have

RepC,B (T ) =

0
2
0
0
0

1
0
3
0
0

0
2
0
4
0

0
0
3
0
5

0
0
0
4
0

1
0
0
0
5

We compute RepD,C (S) column-by-column as well; for instance, the third column is

2
2

RepD (S(w2 )) = RepD (2w1 ) =


0 .
0
Doing a few more calculations like this gives

0 1 2
0
0
0 0
2
3
0
RepD,C (S) =
0 0
0
3 4
0 0
0
0
4

0
0

0
0

Performing matrix multiplication

0 1 0
1 2
0
0
2 0 2

0
2 3
0
0 3 0

0
0
3 4
0 0 4
0
0
0
4
0 0 0

does indeed verify that

0 0 1
2 6
2 6
0
0
0 0 0
0
6 12
6 12
0

3 0 0 =
0
0
12 20
12 20
0 4 0
0
0
0
20
0
20
5 0 5

http://palmer.wellesley.edu/~aschultz/w16/math206

B,D

Page 5 of 6

Math 206, Spring 2016

Fake Assignment 9 Solutions

Due: Never


(3) Suppose that T L(V, V ) and that there exists some basis B = {v1 , , vn } of V for which there are
real numbers 1 , , n which satisfy T (vi ) = i vi . Write T k for the function which is the composition
of T with itself k times. (So, for example, T 2 = T T .) Compute (with justification)
RepB,B (T k ).
Solution. We start by observing that for all k 1, we have T k (vi ) = ki vi . This follows since
each application of T to vi yields i vi ; we have k such applications, and hence a factor of ki after k
applications of T .
With this observation in hand, recall that the ith column of RepB,B (T k ) is precisely RepB (T k (vi )).
Since we know that T k (vi ) = ki vi from above, we know RepB (T k (vi )) is a vector whose only nonzero
coordinate is in the ith position, with value ki . So we get
k

1 0
0
0
0 k2 0 0

k
0 k3
0
RepB,B (T ) = 0
.

..
..
.
.

. .
.
0

kn


This argument is very hand-wavy, so you should be suspicious. The appropriate way to justify this claim is via mathematical
induction, but we havent talked about that in class yet. Perhaps by the end of the semester well discuss induction, and hence will
be able to put this proof on a more solid footing.

http://palmer.wellesley.edu/~aschultz/w16/math206

Page 6 of 6

You might also like