You are on page 1of 6

Section 12.2.

Homework #7 Masaya Sato

1. Prove that similar linear transformations of V (or n × n matrices) have the same charac-
teristic and the same minimal polynomials.
Proof. Let ϕ and ψ be similar transformations on V with representation matrices A and B,
resepectively. Then observe by assumption that there exists some invertible matrix T such
that
A = T −1 BT .
So for an indeterminate x
xI − A = xI − T −1 (B)T ⇒ xI − A = xT −1 T − T −1 (B)T
⇒ xI − A = T −1 (xI − B)T
and thus
det (xI − A) = det (T −1 (xI − B)T ) ⇒ det (xI − A) = det T (det T )−1 det (xI − B)
⇒ det (xI − A) = det (xI − B).
This shows that any similar linear transformations on V have the same characteristic polyno-
mial. Now let mϕ (x) be the minimal polynomial for the transformation ϕ, whose dimension
is m < n. By the division algorithm for polynomials there exist a unique pair of polynomials
q(x) adn r(x) such that
det (xI − B) = q(x)mϕ (x) + r(x),
where 0 ≤ deg r(x) ≤ m − 1. Since det (xI − A) = det (xI − B),
det (xI − A) = q(x)mϕ (x) + r(x).
By minimality of mϕ (x), r(x) is the zero polynomial. Moreover,
0 = det (BI − B) = q(B)mϕ (B)
and therefore mϕ (B) = 0 since F [x] is an integral domain. Hence mϕ (x) is also the minimal
polynomial for the transformation ψ.
3. Prove that two 2 × 2 matrices over F which are not scalar matrices are similar if and only
if they have the same characteristic polynomial.
Proof. (⇒) Suppose that 2 × 2 non-scalar matrices A and B are similar. Then by result
from Exercise 1, A and B share the common characteristic polynomial.
(⇐) Conversely suppose that A and B have the same characteristic polynomial p(x) of
degree 2, i.e. p(x) = x2 + ax + b, where a, b ∈ F . Observe that p(x) is not of the form
p(x) = (x − λ)2 , where λ ∈ F , since A and B are non-scalar matrices. Thus p(x) is also the
minimal polynomial. Therefore both A and B are similar to the companion matrix
 
0 −b
1 −a
and hence A and B are similar.

Abstract Algebra by Dummit and Foote 1


Section 12.2. Homework #7 Masaya Sato

4. Prove that two 3 × 3 matrices are similar if and only if they have the same characteristic
and same minimal polynomials. Give an explicit counterexample to this assertion for 4 × 4
matrices.
Proof. (⇒) Suppose that 3 × 3 matrices A and B are similar. So, similar to Exercise 3, A
and B have the same characteristic and minimal polynomials by result from Exercise 1.
(⇐) Conversely suppose that A and B have the same characteristic and minimal polynomials,
p(x) and m(x). Observe that deg p(x) = 3 and consider the following 3 cases.
Case 1. Suppose that deg m(x) = 1. Then m(x) = x − a and p(x) = (x − a)3 for some a ∈ F .
So invariant factors are ai (x) = x − a for i = 1, 2, 3.
Case 2. Suppose that deg m(x) = 2. Then there exists some linear factor a1 (x) = x − a,
where a ∈ F , such that (x − a)m(x) = p(x). Moreover a2 (x) = m(x).
Case 3. Suppose that deg m(x) = 3. Then a1 (x) = p(x) = m(x).
In any cases discussed above, the characteristic and minimal polynomial uniquely determine
invariant factors. Therefore A and B have the same rational canonical form and hence A
and B are similar.
In terms of a counterexample for 4 × 4 matrices, consider the following distinct lists of
invariant factors

(i) a1 (x) = x, a2 (x) = x, a3 (x) = x2 , and

(ii) a1 (x) = x2 and a2 (x) = x2 .

Observe that both (i) and (ii) have the same characteristic and minimal polynomials x4 and
x2 , respectively. For (i), the matrix is given by
 
0 0 0 0
0 0 0 0
 
0 0 0 0
0 0 1 0

and, and on the other hand (ii) gives the matrix


 
0 0 0 0
1 0 0 0
 .
0 0 0 0
0 0 1 0

However those matrices are not similar.


5. Prove directly from the fact that the collection of all linear transformations of an n
dimensional vector space V over F to itself form a vector space over F of dimension n2 that
the minimal polynomial of a linear transformation T has degree at most n2 .

Abstract Algebra by Dummit and Foote 2


Section 12.2. Homework #7 Masaya Sato

Proof. For an arbitrarily taken linear transformation ϕ on V , let A = (aij ) denote the
representation matrix for ϕ, where 1 ≤ i ≤ n and 1 ≤ j ≤ n. Moreover let Eij be an n × n
matrix defined by (
1 at the (i, j) entry
Eij =
0 elsewhere .
Then n
X
A= aij Eij
i,j=1

and thus A is generated by the set {Eij ∈ Mn×n (F )}ni,j=1 . Moreover for cij ∈ F the following
equation
Xn
cij Eij = 0,
i,j=1

where 0 is the zero matrix, has only the trivial solution since
n
X
cij Eij = 0 ⇒ C = 0 [∵ C = (cij )]
i,j=1

⇒ cij = 0

for all i = 1, . . . , n and j = 1, . . . , n. Therefore {Eij ∈ Mn×n (F )}ni,j=1 is linearly indepen-


dent and a basis for a vector space of dimension n2 . This indicates that the characteristic
polynomial is of degree n2 and therefore the degree of its minimal polynomial is at most
n2 .
6. Prove that the constant term in the characteristic polynomial of the n × n matrix A is
(−1)n det A and that the coefficient of xn−1 is the negative of the sum of the diagonal entries
of A (the sum of the diagonal entries of A is called the trace of A). Prove that det A is the
product of the eigenvalues of A and that the trace of A is the sum of the eigenvalues of A.
Proof. For an n × n matrix A its characteristic polynomial is given by det (xI − A). Then

det (xI − A) = xn + an−1 xn−1 + · · · + a1 x + a0 ,

which is the monic polynomial of degree n. Evaluating the above identity in terms of the
indeterminate x at x = 0 ∈ F ,

a0 = det (−A) = (−1)n det A.

On the other hand, the expansion of det (xI − A) gives


X
det (xI − A) = (σ)(xδσ(1)1 − aσ(1)1 ) · · · (xδσ(n)n − aσ(n)n ),
σ∈Sn

Abstract Algebra by Dummit and Foote 3


Section 12.2. Homework #7 Masaya Sato

dn−1
where δij denotes the Kronecker delta. Moreover, applying nth derivative dxn−1
to both sides
of the identity and comparing the constant terms,
−((n − 1)!)a11 − · · · − ((n − 1)!)ann = ((n − 1)!)an−1 ⇒ −((n − 1)!)(Tr(A)) = ((n − 1)!)an−1
⇒ an−1 = −Tr(A).
The argument above implies that for eigenvalues λ1 , . . . , λn of A
det (λI − A) = xn + (λ1 + · · · + λn )xn−1 + · · · + a1 x + (λ1 · · · λn ).
Therefore n
X
Tr(A) = λi
i=1
and n
Y
det A = λi .
i=1

8. Verify that the characteristic polynomial of the companion matrix


 
0 0 0 ··· 0 −a0
1 0 0 · · · 0 −a1 
 
0 1 0 · · · 0 −a2 
 
 .. .. .. .. .. 
. . . . . 
0 0 0 ··· 1 −an−1
is
xn + an−1 xn−1 + · · · a1 x + a0 .
Proof. For an indeterminate x observe first that xI − A is given by
 
x 0 0 ··· 0 a0
−1 x 0 ··· 0 a1 
 
 0 −1 x ··· 0 a2 
 ,
 .. .. .. .. .. 
 . . . . . 
0 0 0 · · · −1 x + an−1
where A denotes the companion matrix. By induction on the size n of the matrix, suppose
that n = 2. Then det (xI − A) = x2 + a1 x + a0 . So now suppose that the det (xI − A) has
the desired polynomial for n. Then for n + 1
 
x 0 0 ··· 0 a0
−1 x 0 · · · 0 a1 
 
 0 −1 x · · · 0 a2 
 
 .. .. .. .. .. 
 . . . . . 
 
0 0 0 ··· x an−1 
0 0 0 · · · −1 x + an

Abstract Algebra by Dummit and Foote 4


Section 12.2. Homework #7 Masaya Sato

and the expansion along the 1st column gives the determinant of minors, i.e.

det (xI − A) = (−1)1+1 x det Ã11 + (−1)2+1 (−1) det Ã21 ,

where Ã11 and Ã21 denote cofactors. Then

det Ã11 = x(xn + an xn−1 + · · · + a1 ) = xn+1 + an xn + · · · + a1 x.

To evaluate det Ã21 , keep expanding along the 1st column and then

det Ã21 = a0 .

Therefore det (xI − A) = xn+1 + an xn + · · · + a1 x + a0 as desired.


11. Find all similarity classes of 6 × 6 matrices over C with characteristic polynomial
(x4 − 1)(x2 − 1).
Solution: Observe that the characteristic polynomial (x4 −1)(x2 −1) factors into irreducibles
in C[x] as (x − i)(x + i)(x − 1)2 (x + 1)2 . To find all similarity classes, consider their minimal
polynomials m(x). Since m(x) and (x4 − 1)(x2 − 1) have the same roots, the candidates for
m(x) are (x − i)(x + i)(x − 1)(x + 1), (x − i)(x + i)(x − 1)2 (x + 1), (x − i)(x + i)(x − 1)(x + 1)2 ,
and (x − i)(x + i)(x − 1)2 (x + 1)2 . So consider all 4 cases given above.
Case 1. Suppose that m(x) = (x − i)(x + i)(x − 1)(x + 1). Then the invariant factors are
a1 (x) = (x − 1)(x + 1) and a2 (x) = m(x).
Case 2. Suppose that m(x) = (x − i)(x + i)(x − 1)2 (x + 1). Then the invariant factors are
a1 (x) = x + 1 and a2 (x) = m(x).
Case 3. Suppose that m(x) = (x − i)(x + i)(x − 1)(x + 1)2 . Then the invariant factors are
a1 (x) = x − 1 and a2 (x) = m(x).
Case 4. Suppose that m(x) = (x − i)(x + i)(x − 1)2 (x + 1)2 . Then a1 (x) = m(x) is the only
invariant factor.
Therefore, according the classification discussed above, there are 4 similarity classes.
17. Determine representatives for the conjugacy classes for GL3 (F2 ). [Compare your answer
with Theorem 15 and Proposition 14 of Chapter 6.]
Solution: For F2 [x], since the only nonzero element in F2 is 1,
(i) F2 [x]/(x − 1) ⊕ F2 [x]/((x − 1)2 ),

(ii) F2 [x]/(x − 1) ⊕ F2 [x]/(x − 1) ⊕ F2 [x]/(x − 1), and

(iii) F2 [x]/(Q(x)), where Q is a cubic polynomial.


For (i) and (ii), the corresponding matrices are given by
   
1 0 0 1 0 0
0 1 0 and 0 1 0 , respectively.
0 1 1 0 0 1

Abstract Algebra by Dummit and Foote 5


Section 12.2. Homework #7 Masaya Sato

Then for (iii) let Q(x) = x3 + ax2 + bx + 1. The matrix is given by


 
0 0 1
1 0 −b  .
0 1 −a

Classifying Q(x) as the product of primes factors, Q(x) has the following forms.

(1) x3 + x + 1,

(2) x3 + x2 + 1,

(3) (x − 1)(x2 + x + 1), and

(4) (x − 1)3 .

Therefore from (1) to (4) the matrices are given by


       
0 0 1 0 0 1 1 0 0 1 0 0
1 0 1 , 1 0
  0 , 0 0 1 , and 1 1 0 .
0 1 0 0 1 1 0 1 1 0 1 1

18. Let V be a finite dimensional vector space over Q and suppose T is nonsingular linear
transformation of V such that T −1 = T 2 + T . Prove that the dimension of V is divisible by
3. If the dimension of V is precisely 3 prove that all such transformations T are similar.
Proof. Let V be an n dimensional vector space over Q. Observe that T −1 = T 2 + T implies
T 3 + T 2 − I = 0. So the minimal polynomial m(x) is given by

m(x) = x3 + x2 − 1,

which is irreducible. Therefore

V ∼
= Q[x]/(m(x)) ⊕ · · · ⊕ Q[x]/(m(x)),

where each quotient space is 3 dimensional. Hence the dimension of V is 3k for some k ∈ Z>0
as desired.
Now suppose that V is 3 dimensional. Then

V ∼
= Q[x]/(m(x)).

So all nonsingular linear transformations T satisfying T −1 = T 2 +T have the same companion


matrix because m(x) is also the characteristic polynomial for T . Thus all such T ’s are
similar.

Abstract Algebra by Dummit and Foote 6