Math 115A

Homework 6
Due Sunday, February 19, 2023
Remember: If a problem deals with an arbitrary vector space V or W , do not assume that
the vector space is finite-dimensional unless the problem explicitly says so!

I Invertible functions and vector space isomorphisms

Recall that for any sets X and Y and any function f : X → Y , to say that g: Y → X is an
inverse of f means that both

g ◦ f = 1X and f ◦ g = 1Y .

We will now weaken this definition and explore the consequences. To say that g is a left
inverse of f means that g ◦ f = 1X , and to say that g is a right inverse of f means that
f ◦ g = 1Y . Thus an inverse of f is merely a function g that is both a left inverse and a right
inverse simultaneously. (For this reason, an inverse is sometimes called a two-sided inverse.)

1. Let X and Y be any nonempty sets, and let f : X → Y be a function.

(a) Prove that f has a left inverse if and only if f is injective (one-to-one).
(b) Prove that f has a right inverse if and only if f is surjective (onto).
(Note: The (⇐=) direction of this relies on the Axiom of Choice. Try to explicitly
point out where in your proof you used the Axiom of Choice.)
(c) Prove that f has a (two-sided) inverse if and only if f is bijective (one-to-one and
onto).
(Hint on last page.)
(d) Prove that if f has a (two-sided) inverse, that inverse is unique.

2. Let V and W be finite-dimensional vector spaces with dim(V ) = dim(W ), and let
T : V → W be linear.
(a) Prove that if S: W → V is a right inverse of T , then T is invertible and S is the
inverse of T .
(b) Prove that if S: W → V is a left inverse of T , then T is invertible and S is the
inverse of T .

3. (a) Give an example of vector spaces V and W and a linear function T : V → W that
has a right inverse, but does not have a left inverse.
(Hint on last page.)
 (b) For the function T you chose in part (a), give two different linear functions S1
and S2 that are right inverses of T . This shows that, in general, right inverses are
not unique.
(c) Give an example of vector spaces V and W and a linear function T : V → W that
has a left inverse, but does not have a right inverse.
(Hint on last page.)
(d) For the function T you chose in part (c), give two different linear functions S1
and S2 that are left inverses of T . This shows that, in general, left inverses are
not unique.

II Invariant subspaces, and projections revisited

The following definitions are used in several of the exercises in this section, and will show
up later in the course.
Definition. Let V be a vector space, and let T be a linear operator on V (that is, let
T : V → V be linear). For a subspace W of V , to say that W is T -invariant (or invariant
under T ) means that
T (W ) ⊆ W
(Or, equivalently, for all x ∈ W , T (x) ∈ W as well.)
If W is a T -invariant subspace of V , we define the restriction of T to W to be the function
TW : W → W defined by
TW (x) = T (x) for all x ∈ W
In other words, TW gives the same outputs as T , but is only defined on W instead of on all
of V .

4. Let V be a vector space, and let T : V → V be linear. Prove that the subspaces {0},
V , im(T ), and ker(T ) are all T -invariant.

5. Let V be a vector space and W1 a subspace of V , and suppose that T : V → V is

the projection onto W1 along some subspace W2 . Prove that both W1 and W2 are
T -invariant, and that TW1 = 1W1 (the identity map on W1 ). What is TW2 ?

6. Let V be a vector space, and let T : V → V be linear. Suppose that W is a T -invariant

subspace of V such that V = im(T ) ⊕ W .
(a) Prove that W ⊆ ker(T ).
(b) Show that if V is finite-dimensional, then W = ker(T ).
(c) Show by example that if V is not finite-dimensional, then the conclusion of (b) is
not necessarily true (that is, W may be a proper subset of ker(T )).
(Hint on last page.)
 7. Let V be a vector space, and let W1 and W2 be subspaces such that V = W1 ⊕ W2 .
Let T : V → V be the projection onto W1 along W2 .
(Note: Do not assume in this problem that V is finite-dimensional!)
(a) Prove that im(T ) = W1 and ker(T ) = W2 . (Thus V = im(T ) ⊕ ker(T ).)
(b) Prove that T 2 = T .
(Note: Here, T 2 means T ◦ T .)

8. Let V be a vector space, and let T : V → V be a linear operator on V such that T 2 = T .

(Note: Do not assume in this problem that V is finite-dimensional!)
(a) Prove that V = im(T ) ⊕ ker(T ).
(Hint on last page.)
(b) Prove that T is the projection onto im(T ) along ker(T ).

Note that, as a result of the previous two problems, you have proved the following theorem:
Theorem. Let V be a vector space, and let T be a linear operator on V . Then T is a
projection if and only if T 2 = T .
Congratulations! You’ve solved your first problem from the UCLA Math Department’s
“Basic Qualifying Exam”!

9. Let V be a vector space, and let T : V → V be linear. For any positive integer k, T k
denotes the “k-fold composition” of T with itself:

Tk = T
| ◦ T ◦{z· · · ◦ T} .
k times

In particular, note that for any k ≥ 0, T k+1 = T ◦ T k , or in other words T k+1 (x) =
T (T k (x)) for all x ∈ V . You will use this fact many times in this problem.
(a) Prove that for any k > 0, im(T k+1 ) ⊆ im(T k ) and ker(T k+1 ) ⊇ ker(T k ).
Note: By applying this repeatedly, this means that for any j > k, im(T j ) ⊆
im(T k ) and ker(T j ) ⊇ ker(T k ).
(b) Now assume that V is finite-dimensional. Prove that there exists some k > 0
for which rank(T k+1 ) = rank(T k ). Then prove that, for the same value of k,
nullity(T k+1 ) = nullity(T k ) as well.
(Hint on last page.)
(c) Conclude from the previous parts that, with the same value of k as in part (b),
im(T k+1 ) = im(T k ) and ker(T k+1 ) = ker(T k ).
(d) With the same k as in parts (b) and (c), prove that V = im(T k ) ⊕ ker(T k ).
(Hint on last page.)
 You have proved that if V is finite-dimensional and T : V → V is linear, there is a
positive integer k such that V = im(T k ) ⊕ ker(T k ).

III Coordinates and matrices

10. Label each of the following statements as true or false. If true, explain why briefly (no
need to give a rigorous proof). If false, either give a counterexample or say how the
statement should be modified to make it true. In all of the following, V and W are
finite-dimensional vector spaces over a field F , with ordered bases B and C, respectively,
and T : V → W .
(a) If S: V → W also, then if C [T ]B = C [S]B , then T = S.
(b) If m = dim(V ) and n = dim(W ), then C [T ]B is an m × n matrix.
(c) For any x ∈ V , C [T (x)] = C [T ]B B [x].
(d) If T is invertible, then C [T ]−1
B = C [T
−1
]B .

11. Let B = [X 3 , X 2 , X, 1] and C = [1, X, X 2 ], which are ordered bases for P3 (R) and
P2 (R), respectively. (Note the orders of these bases!) Define T : P3 (R) → P2 (R) by
T (f ) = f 0 .
(a) Compute C [T ]B .
(Note: Part (a) is the only part of this problem where you’re allowed to take
derivatives.)
(b) Let f (X) = 4 − 5X 2 + 2X 3 . Compute B [f ]. Then, without taking a derivative,
compute T (f ).
(c) Define S: P2 (R) → R2 by S(f ) = (f (−1), f (1)). Compute D [S]C , where D is
the standard ordered basis of R2 . Then, without taking any derivatives, compute
D [ST ]B , the matrix of the composition ST with respect to the bases B and D.

12. Let V be an n-dimensional vector space, and let T : V → V be a linear transformation.

Suppose that W is a T -invariant subspace of V having dimension k. Show that there
is an ordered basis B for V such that B [T ]B has the block-upper-triangular form
 
A B
0 C

where A is a k × k matrix, and 0 is the (n − k) × k zero matrix.

(Hint on last page.)

13. Let V be a finite-dimensional vector space, and let W1 and W2 be two subspaces of
V such that V = W1 ⊕ W2 . Let T be the projection onto W1 along W2 . Find an
ordered basis B for V such that B [T ]B is a diagonal matrix. What are the entries on
 the diagonal?
(Hint on last page.)

14. Let V and W be finite-dimensional vector spaces, with dim(V ) = dim(W ), and let
T : V → W be a linear transformation. Show that there exist ordered bases B and C
for V and W , respectively, such that C [T ]B is a diagonal matrix. What are the entries
on the diagonal?
(Hint on last page.)
Hints for a few exercises:
Hint for problem 1(c): For one direction of this, you can save a bit of writing by just using
parts (a) and (b).
Hint for problem 3: For parts (a) and (c), the previous problem implies that this is only
possible if the dimensions of V and W are different, or are the same but infinite. For all
parts of the problem, a couple nice examples like this appeared several times in Homework
5.
Hint for problem 6(c): One possible example appeared several times in Homework 5.
Hint for problem 8: For one part of this, it will be helpful to write an arbitrary x ∈ V as
T (x) + (x − T (x)).
Hints for problem 9:
(i) For part (b): Try a proof by contradiction.
(ii) For part (d): Problem 10 from Homework 5 (or just Rank–Nullity) will save you a lot
of effort.
(iii) Also for part (d): There are several ways to prove this, but you’ll almost certainly need
to use induction somewhere. One way starts by proving that, with the same k as in
parts (b) and (c), for all j > k, ker(T j ) = ker(T k ) and/or im(T j ) = im(T k ). Another
way is to prove (by induction) that for all j ∈ {1, 2, . . . , k}, im(T k ) ∩ ker(T j ) = {0}.
Hint for problem 12: Use the basis extension theorem for subspaces.
Hint for problem 13: Use the basis extension theorem again, or better yet problem 16 from
Homework 3.
Hint for problem 14: Use the basis extension theorem for subspaces again, but this time, do
it twice; start by following the exact same steps as in the proof of the Rank–Nullity Theorem.