Math 113 Homework 5 Solutions (Starred problems)

Solutions by Guanyang Wang, with edits by Tom Church.

Exercise 5.C.1 Suppose T ∈ L(V ) is diagonalizable. Prove that V = null T ⊕

range T .
Proof. Let v1 , ..., vn be a basis of V with respect to which T has a diagonal ma-
trix. So for every j ∈ 1, 2, ...n, we have some λj ∈ F such that T vj = λj vj . By
renumbering, we can choose m ∈ {1, 2, ..., n} such that
λj = 0 for j = 1, 2...m
and
λj 6= 0 for j = m + 1, ...n
So we have V = span{v1 , ..., vm } ⊕ span{vm+1 , ...vn }. Now we claim that null T =
span{v1 , ..., vm } and range T = span{vm+1 , ...vn }
First we prove null T = span{v1 , ..., vm }. Notice that span{v1 , ..., vm } is the
eigenspace of T corresponding to 0. Hence every element v ∈ span{v1 , ..., vm }
satisfies T v = 0, so v ∈ null T . Meanwhile any u ∈ null T , we have T u = 0,
therefore u is an eigenvector of T corresponding to 0. So u ∈ span{v1 , ..., vm }.
Hence we have null T = span{v1 , ..., vm }.
Then we prove range T = span{vm+1 , ...vn }. For any j ∈ m + 1, ...n, we have
T (λ−1
j vj ) = vj , therefore range T ⊃ span{vm+1 , ...vn }. Meanwhile for every y ∈
range T , we have y = T x for some x ∈ V . Since v1 , ..., vn is a basis of V , we have
x = a1 v1 + ... + an vn for some a1 , · · · , an ∈ F. Therefore
y = T x = T (a1 v1 + ... + an vn )
= λm+1 am+1 vm+1 + · · · + λn an vn ∈ span{vm+1 , ..., vn }

Hence we have range T = span{vm+1 , ...vn }. Therefore we have proved our claim,
so we can conclude that V = null T ⊕ range T . 

Exercise 5.C.2 Prove the converse of the statement in the exercise above or
give a counterexample to the converse.
Proof. The converse of the statement in the exercise above is false. As an example,
define T ∈ L(F2 ) by
T (w, z) = (w + z, z)
The eigenvector-eigenvalue equation T (w, z) = λ(w, z) is equivalent to the sys-
tem of equations
w + z = λw and z = λz
After solving the equations, we have 1 is the only eigenvalue of T and that
E(1, T ) = {(w, 0) : w ∈ F}.
Since 1 is the only eigenvalue of T , 5.41 shows that T is not diagonalizable.
Because 0 is not an eigenvalue of T , we know that T is invertible. Thus null T =
{0} and range T = F2 . Hence F2 = null T ⊕ range T , providing a counterexample
to the converse of the previous exercise.

1
2

Exercise 5.C.8. Suppose T ∈ F5 and dim E(8, T ) = 4. Prove that T − 2I and

T − 6I is invertible.
Proof. From 5.38, we know that
dim E(8, T ) + dim E(2, T ) + dim E(6, T ) ≤ dim F5 .
Since dim E(8, T ) = 4 and dim F5 = 5, the inequality above can be written as
dim E(2, T ) + dim E(6, T ) ≤ 1
Thus we have dim E(2, T ) = 0 or dim E(6, T ) = 0. In other words, 2 is not
an eigenvalue of T or 6 is not an eigenvalue of T . Hence T − 2I or T − 6I is
invertible. 
Exercise 5.C.14 Find T ∈ L(C3 ) such that 6 and 7 are eigenvalues of T and
such that T does not have a diagonal matrix with respect to any basis of C3 .
Proof. Define T ∈ L(C3 ) by
T (z1 , z2 , z3 ) = (6z1 + z2 , 6z2 , 7z3 )
The eigenvector-eigenvalue equation T (z1 , z2 , z3 ) = λ(z1 , z2 , z3 ) is equivalent to the
system of equations
6z1 + z2 = λz1
6z2 = λz2
7z3 = λz3

After solving the equations, we have 6 and 7 are the only eigenvalues of T and from
our definition we have z2 is the eigenvector of T corresponding to 6 and z3 is the
eigenvector of T corresponding to 7. We also have
E(6, T ) = span((1, 0, 0)) and E(7, T ) = span((0, 0, 1))
Thus
dim E(6, T ) = dim E(7, T ) = 1.
Now 5.41 shows that T is not diagonalizable since dim E(6, T ) + dim E(7, T ) =
2 < 3 = dim(C3 ). 

Exercise 8.C.8 Suppose T ∈ L(V ). Prove that T is invertible if and only if the
constant term in the minimal polynomial of T is nonzero.
Proof. For any polynomial f (x), the constant term is the value f (0) at 0 (since if
f (x) = an xn + · · · + a1 x + a0 , then f (0) = an 0n + · · · + a1 0 + a0 = a0 ). Therefore
the constant term in the minimal polynomial p(x) is nonzero if and only if p(0) is
nonzero; in other words, if and only if 0 is not a root of p(x). Since the roots of the
minimal polynomial are the eigenvalues of T , we conclude that:
constant term of p(x) is nonzero ⇐⇒ 0 is not a root of p(x)
⇐⇒ 0 is not an eigenvalue of T
⇐⇒ (T − 0I) = {0}
⇐⇒ T is injective
⇐⇒ T is invertible 
 3

Question 1.
• Give an example of an operator T on V = C3 whose minimal polynomial
is (x + 2)2 .
• Give an example of an operator S on W = C4 whose minimal polynomial
is (x2 + 1)(x − 3)2 .
• What are the eigenvalues of the operators T and S in parts a) and b)?
Proof. • Let T : C3 → C3 be defined by
T (x, y, z) = (−2x, −2y, −2(y + z))
We claim that the minimal polynomial of T is f (x) = (x + 2)2 . First we
show that f (T ) = (T + 2I)2 = 0. We have that
(T + 2I)(x, y, z) = (0, 0, −2y)
Thus, (T + 2I)2 (x, y, z) = (T + 2I)(0, 0, −2y). Since
(T + 2I)(0, 0, −2y) = (0, 0, 0)
we get that (T − I)2 (x, y, z) = (0, 0, 0) for all (x, y, z) ∈ C3 . That is,
f (T ) = 0.
We have found one polynomial f (x) ∈ UT , which has degree 2 (recall
from class that UT is the set of all polynomials with F (T ) = 0). It is
obvious that T is not a multiple of the identity, so no degree-1 polynomial
x − λ is contained in UT . Therefore f (x) has the smallest possible degree
in UT , and so the minimal polynomial of T is mT (x) = (x + 1)2 .
• Give an example of an operator S on W = C5 whose minimal polynomial
is (x2 + 1)(x − 3)2 .
Let S : C5 → C5 be defined by
S(x, y, z, w, t) = (ix, −iy, 3z, 3w, 3t + w)
We begin by noting that i −i and 3 are eigenvalues of S:
S(1, 0, 0, 0, 0) = (i, 0, 0, 0, 0) = i · (1, 0, 0, 0, 0)
S(0, 1, 0, 0, 0) = (0, −i, 0, 0, 0) = −i · (0, 1, 0, 0, 0)
S(0, 0, 1, 0, 0) = (0, 0, 3, 0, 0) = 3 · (0, 0, 1, 0, 0)
So we compute
(S − 3I)(x, y, z, w, t) = ((i − 3)x, (−i − 3)y, 0, 0, w)
and therefore.
(S − 3I)2 (x, y, z, w, t) = ((i − 3)2 x, (−i − 3)2 y, 0, 0, 0)
Meanwhile,
(S 2 + I)(x, y, z, w, t) = (0, 0, 10z, 10w, 10t + 6w)
Thus applying (S 2 + I) to the result of (S − 3I)2 (x, y, z, w, t) we get
(S 2 + I)(S − 3I)2 (x, y, z, w, t) = (0, 0, 0, 0, 0).
This shows that (S 2 + I)(S − 3I)2 = 0, so if f (x) = (x2 + 1)(x − 3)2 then
f (x) ∈ US . However, since −i, i and 3 are eigenvalues of S, we know that
−i,i and 3 are roots of the minimal polynomial. Therefore the only smaller
possibility for the minimal polynomial is (x−i)(x+i)(x−3) = (x2 +1)(x−3),
since this is the only polynomial of degree < 4 with both −i, i and 3 as
4

roots. So we just need to show that (S 2 + I)(S − 3I) 6= 0, and this we can
do by direct computation. Indeed, applying S 2 + I to the result of S − 3I
that we found above, we get
(S 2 + I)(S − 3I)(x, y, z, w, t) = (0, 0, 0, 0, 10w)
which is non-zero. Thus the minimal polynomial of S is indeed
mS (x) = (x2 + 1)(x − 3)2 .
• Since all the eigenvalues are precisely all the zeros of the minimal polyno-
mial (8.49), we just need to compute the zeros of the minimal polynomial.
In the first case, the minimal polynomial (x + 2)2 , so −2 is the only eigen-
value of operator T . In the second case, the minimal polynomial of S is
(x2 + 1)(x − 3)3 , the roots of this polynomial are −i, i and 3. Therefore
−i, i and 3 are all the eigenvalues of S.


Question 2. Let V = R4 , and let T ∈ L(V ) be the operator with matrix

 
2 0 0 0
 0 3 0 1 
 
 0 0 3 0 
0 0 0 3
Find the minimal polynomial of T .
Proof. We claim that f (x) = (x − 2)(x − 3)2 is the minimal polynomial. First we
show that (T − 2I)(T − 3I)2 = 0
     
0 0 0 0 −1 0 0 0 −1 0 0 0
0 1 0 1 0 0 0 1 0 0 0 1
(T − 2I)(T − 3I)2 = 
    
0 0 1 0 ·  0 0 0 0 ·  0 0 0
  
0
0 0 0 1 0 0 0 0 0 0 0 0
   
0 0 0 0 1 0 0 0
0 1 0 1 0 0 0 0
= 0 0 1 0 · 0 0 0 0
  

0 0 0 1 0 0 0 0
 
0 0 0 0
0 0 0 0
= 0 0 0 0


0 0 0 0

Therefore we have (T − 2I)(T − 3I)2 = 0, so f (x) = (x − 2)(x − 3)2 ∈ UT .

However, since 2, and 3 are eigenvalues of T , we know that 2 and 3 are roots
of the minimal polynomial. Therefore the only smaller possibility for the minimal
polynomial is (x − 2)(x − 3), since this is the only polynomial of degree < 3 with
both 2 and 3 as roots. So we just need to show that (T − 2I)(T − 3I) 6= 0, and this
 5

we can do by direct computation.

     
0 0 0 0 −1 0 0 0 0 0 0 0
0 1 0 1 0 0 0 1 0 0 0 1
(T − 2I)(T − 3I) = 
0 0 1
· = 
0  0 0 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0 0 0 0
which is not 0. Thus the minimal polynomial of T is indeed
mT (x) = (x − 2)(x − 3)2 .