MATH 110, Spring 2023, midterm test solutions.

1. (10pp.) Let V = P2 (IR). Let I denote the identity map on V , D the differentiation map,
D2 = D ◦ D the second differentiation map, and T the map f (x) 7→ f (x − 1). (You do NOT
need to prove that these are linear maps on V ; they are.) Is the list (I, D, D2 , T ) linearly
dependent or independent in L(V )? Justify your answer.

Solution: By the Taylor-MacLauren formula,

1 00 1
f (x − 1) = f (x) − f 0 (x) + f (x) − f 000 (ξ) (1)
2 6
for some point ξ between x − 1 and x whenever f is thrice differentiable on IR. Every
polynomial is inifinitely differentiable and, moreover, p000 (x) ≡ 0 if deg p ≤ 2. So, equality (1),
when specialized to a polynomial p in V = P2 (IR), becomes
1 00
p(x − 1) = p(x) − p0 (x) + p (x) ∀p ∈ V. (2)
2
The latter can be rewritten as
 
1 2
T (p) = I − D + D (p) ∀p ∈ V
2
or, equivalently, as
1 2
T =I −D+ D in L(V ).
2
This shows that the list (I, D, D2 , T ) is linearly dependent in L(V ).
Answer: linearly dependent.

Another solution: Compute matrix representations of I, D, D2 , T with respect to, say,

the standard monomial basis used both on the domain V and the codomain V . We get
       
1 0 0 0 1 0 0 0 2 1 −1 1
M(I) =  0 1 0  , M(D) =  0 0 2  , M(D2 ) =  0 0 0  , M(T ) =  0 1 −2  .
0 0 1 0 0 0 0 0 0 0 0 1

From the zero/nonzero patterns in these matrices, we can see that

1
M(I) − M(D) + M(D2 ) = M(T ).
2
Since M is an isomorphism between L(V ) and IR3×3 , this implies
1 2
I −D+ D = T.
2

1
A third solution: Suppose a0 I + a1 D + a2 D2 + a3 T = 0 (the zero operator on V ). Plugging
in the function f1 (x) = x2 into this relation, we get (a0 I + a1 D + a2 D2 + a3 T )(x2 ) = 0 (the
zero polynomial), that is, a0 x2 + a1 2x + a2 2 + a3 (x − 1)2 = 0, i.e.,
a0 x2 + 2a1 x + 2a2 + a3 x2 − 2a3 x + a3 = 0.

Since the polynomial in the left-hand side must have all zero coefficients, we conclude
a0 + a3 = 0
a1 − a3 = 0
2a2 + a3 = 0.
This linear system has a nontrivial solution, say, a0 = 1, a1 = −1, a2 = 1/2, a3 = −1.
(In fact, this solution is unique up to scaling.) Now check that the found nontrivial linear
combination of I, D, D2 , and T is actually the zero operator. (So far, we only know that
this operator sends one function, namely f1 (x) = x2 , to zero.) Plugging in the functions
f2 (x) = x and f3 (x) = 1 into the combination I − D + D2 /2 − T , we obtain
(I − D + D2 /2 − T )(x) = x − 1 + 0 − (x − 1) = 0
(I − D + D2 /2 − T )(1) = 1 − 0 + 0 − 1 = 0.
Since the functions f1 , f2 , f3 form a basis of V , and the linear operator I −D+D2 /2−T sends
every vector of this basis to the zero vector, the operator itself must be the zero operator
by 3.4. Thus, the original operators are linearly dependent.
2. (10pp.) Let V = IR4 , let W1 = {(x1 , x2 , x3 , x4 ) : x2 + x4 = 0, xj ∈ IR for all j}, and let
W2 = {(x1 , x2 , x3 , x4 ) : x1 + x2 + x3 = 0, xj ∈ IR for all j}.
(a) Prove that W1 and W2 are subspaces of V .

Proof: The sets Wj , j = 1, 2, are contained in V because the vectors in either Wj are
real and have length 4. Next, if (x1 , x2 , x3 , x4 ) and (y1 , y2 , y3 , y4 ) ∈ W1 and a, b ∈ IR,
then (ax2 + by2 ) + (ax4 + by4 ) = a(x2 + x4 ) + b(y2 + y4 ) = 0, hence a(x1 , x2 , x3 , x4 ) +
b(y1 , y2 , y3 , y4 ) ∈ W1 . Likewise, if (x1 , x2 , x3 , x4 ) and (y1 , y2 , y3 , y4 ) ∈ W1 and a, b ∈ IR, then
(ax1 + by1 ) + (ax2 + by2 ) + (ax3 + by3 ) = a(x1 + x2 + x3 ) + b(y1 + y2 + y3 ) = 0, hence
a(x1 , x2 , x3 , x4 ) + b(y1 , y2 , y3 , y4 ) ∈ W1 . So, both W1 and W2 are closed under addition and
scalar multipliction, and are therefore subspaces of V .

(b) Determine, with proof, dim W1 ∩ W2 .

Solution: The intersection W1 ∩W2 consists of all vectors satisfying both conditions x2 +x4 =
0 and x1 +x2 +x3 = 0. Let us show that the vectors (−1, 0, 1, 0) and (1, −1, 0, 1) form a basis
for W1 ∩ W2 . Indeed, these vectors are linearly independent because the first of them has a
nonzero in its third coordinate while the the second has zero in that position, and the first
has zero in the fourth position while the second has a nonzero in that position. Moreover,
these vectors span W1 ∩ W2 because any vector satisfying the condition x2 + x4 = 0 and
x1 + x2 + x3 = 0 can be written as

(x1 , xx , x3 , x4 ) = x3 (−1, 0, 1, 0) + x4 (1, −1, 0, 1).

This shows that dim W1 ∩ W2 = 2.

Answer: 2.

(c) Is the sum W1 + W2 direct? Determine, with proof, dim(W1 + W2 ).

Solution: We just saw that W1 ∩ W2 is non-trivial, so the sum W1 + W2 cannot be direct.

Note that W1 = null ϕ1 where ϕ1 : (x1 , x2 , x3 , x4 ) 7→ x2 + x4 , and W2 = null ϕ2 where

ϕ2 : (x1 , x2 , x3 , x4 ) 7→ x1 + x2 + x3 . Since both are nonzero linear functionals, their ranges
are both equal to IR, and hence their nullspaces have dimension dim IR4 − dim IR = 4 − 1 = 3
by the Fundamental Theorem of Linear Maps.

So, dim(W1 + W2 ) = dim W1 + dim W2 − dim W1 ∩ W2 = 3 + 3 − 2 = 4.

Answers: not direct; 4.

3. (10pp.) Let V be the vector space of all trigonometric polynomials (in x) with real
coefficients of degree at most 2 , i.e. V := span{1, sin x, cos x, sin(2x), cos(2x)}. The list
(1, sin x, cos x, sin(2x), cos(2x)) is a basis of V . You do NOT need to prove it. Consider the
linear operator (do NOT check linearity)
T ∈ L(V ) : (T f )(x) = f 00 (x) + f (x).
(a) Find the matrix representation of T in this basis used for the domain and the codomain.

Solution: We calculate

T (1) = 0 + 1 = 1,
T (sin x) = − sin x + sin x = 0,
T (cos x) = − cos x + cos x = 0,
T (sin(2x)) = −4 sin(2x) + sin(2x) = −3 sin(2x),
T (cos(2x)) = −4 cos(2x) + cos(2x) = −3 cos(2x),
so the matrix representation of T with respect to the basis (1, sin x, cos x, sin(2x), cos(2x))
used on both sides is  
1 0 0 0 0
 0 0 0 0 0 
 
M(T ) =  0 0 0 0 0 
 0 0 0 −3 0 
0 0 0 0 −3

(b) What are dim null T and dim range T ? Justify your answers.

Solution: By 3.117, dim range T equals the column rank of M(T ) (or its row rank by
3.118). Notice that columns 1, 4 and 5 are linearly independent because they each have
a nonzero component at different slots. They also span the other columns since all others
are zero. Therefore dim range T = 3. Now, by the Fundamental Theorem of Linear Maps,
dim null T = dim V − dim range T = 5 − 3 = 2.
Answers: dim null T = 2 and dim range T = 3.
4. (10pp.) Consider the linear map T : P2 (IR) → P4 (IR) : f (x) 7→ f (x2 ) and the linear
functional ϕ : P4 (IR) → IR : f (x) 7→ f 00 (0). (NO need to prove they are linear.)
(a) Write down the domain and co-domain of the linear functional T 0 (ϕ).

Solution: T 0 (ϕ) = ϕ ◦ T ∈ (P2 (IR))0 , so the domain of T 0 (ϕ) is P2 (IR), the codomain is IR.

Answers: P2 (IR), IR.

(b) Write down the action of T 0 (ϕ). (E.g., if your functional were from IR4 to IR and added
up all coordinates, your formula would be (x1 , x2 , x3 , x4 ) 7→ x1 + x2 + x3 + x4 .)

Solution: T 0 (ϕ)(p(x)) = ϕ(T (p(x))) = ϕ(p(x2 )) = (p(x2 ))00 (0) = (2xp0 (x2 ))0 (0) = 2p0 (0).

Answer: p(x) 7→ 2p0 (0).

(c) What is dim null T 0 ? Is T 0 an isomorphism? Justify your answers.

Solution: First observe that range T = span(1, x2 , x4 ) because T (1) = 1, T (x) = x2 and
T (x2 ) = x4 . Since the list 1, x2 , x4 is linearly independent in P4 (IR), we conclude that these
functions/vectors form a basis for range T , and therefore dim range T = 3.

Hence dim(range T )0 = dim P4 (IR) − dim range T = 5 − 3 = 2 by 3.104. So, by 3.107 (a),
dim null T 0 = dim(ran T )0 = 2.

This shows that T 0 has a nontrivial nullspace, so is not injective, so is not an isomorphism.

Answers: 2; not an isomorphism.