Professional Documents
Culture Documents
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.
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
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
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
which is the monic polynomial of degree n. Evaluating the above identity in terms of the
indeterminate x at x = 0 ∈ F ,
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
and the expansion along the 1st column gives the determinant of minors, i.e.
To evaluate det Ã21 , keep expanding along the 1st column and then
det Ã21 = a0 .
Classifying Q(x) as the product of primes factors, Q(x) has the following forms.
(1) x3 + x + 1,
(2) x3 + x2 + 1,
(4) (x − 1)3 .
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,
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)).