• Non ci sono risultati.

Suppose further that n ≥ 4 and that An− Bn is invertible

N/A
N/A
Protected

Academic year: 2021

Condividi "Suppose further that n ≥ 4 and that An− Bn is invertible"

Copied!
1
0
0

Testo completo

(1)

Problem 11891

(American Mathematical Monthly, Vol.123, February 2016) Proposed by C. Chiser (Romania).

LetM be the set of all 3 × 3 matrices with complex entries. Suppose A, B ∈ M with AB = BA and Tr(Ak) = Tr(Bk) for k ∈ {1, 2, 3}. Suppose further that n ≥ 4 and that An− Bn is invertible.

Prove that there exist complex numbersα, β, γ such that

A2n+ B2n+ AnBn+ αAn+ βBn+ γI = 0.

Solution proposed by Moubinool Omarjee, Lyc´ee Henri IV, Paris, France, and Roberto Tauraso, Dipartimento di Matematica, Universit`a di Roma “Tor Vergata”, Italy.

We will show that the statement is true for any positive integer n.

Since A and B are commuting matrices, they are simultaneously triangularizable: there exist a invertible matrix P ∈ M such that the matrices P−1AP and P−1BP are both upper triangular.

Let

(a1, a2, a3) = diag(P−1AP) and (b1, b2, b3) = diag(P−1BP).

be the complex eigenvalues of A and B.

For any matrix M ∈ M, the characteristic polynomial is equal to det(xI − M ) = x3− e1x2+ e2x− e3

with

e1= Tr(M ), e2=1

2(Tr2(M ) − Tr(M2)), e3= 1

6(Tr3(M ) − 3Tr(M )Tr(M2) + 2Tr(M3)).

Thus, by Tr(Ak) = Tr(Bk) for k ∈ {1, 2, 3} , it follows that A and B have the same characteristic polynomial and therefore there is a permutation σ ∈ S3such that bk= aσ(k) for k ∈ {1, 2, 3}.

Now, since the matrix

(P−1AP)n− (P−1BP)n= P−1AnP− P−1BnP = P−1(An− Bn)P is upper triangular and An− Bn is invertible, we have that

Y

k∈{1,2,3}

(ank − anσ(k)) = Y

k∈{1,2,3}

(ank− bnk) = det((P−1AP)n− (P−1BP)n)

= det(P−1(An− Bn)P ) = det(An− Bn) 6= 0,

which imply that σ(k) 6= k for k ∈ {1, 2, 3}. So σ is a cyclic permutation in S3and the eigenvalues a1, a2, a3 are distinct complex numbers. Therefore A and B are simultaneously diagonalizable by an invertible matrix Q ∈ M and the equation to be solved is equivalent to

Q−1(A2n+ B2n+ AnBn+ αAn+ βBn+ γI)Q = 0, which yields the linear system,









a2n1 + a2nσ(1)+ an1anσ(1)+ αan1 + βanσ(1)+ γ = 0, a2n2 + a2nσ(2)+ an2anσ(2)+ αan2 + βanσ(2)+ γ = 0, a2n3 + a2nσ(3)+ an3anσ(3)+ αan3 + βanσ(3)+ γ = 0.

It is straightforward to verify that the above system is solved by

α= β = −(an1+ an2 + an3) = −t1, γ= an1an2+ an2an3 + an3an1 =1

2(t21− t2)

where t1:= Tr(An) = Tr(Bn) and t2:= Tr(A2n) = Tr(B2n). 

Riferimenti