Unofficial archive — problems, solutions & results © IMC, reproduced with permission.

IMC / 1996 / Problems / Day 1, P3

IMC 1996 · Day 1 · P3

linear algebraworth 15 pts

The linear operator AA on the vector space VV is called an involution if A2=EA^2 = E where EE is the identity operator on VV. Let dimV=n<\dim V = n < \infty.

(i) Prove that for every involution AA on VV there exists a basis of VV consisting of eigenvectors of AA.

(ii) Find the maximal number of distinct pairwise commuting involutions on VV.

Solution (official)

(i) Let B=12(A+E)B = \frac{1}{2}(A + E). Then B2=14(A2+2AE+E)=14(2AE+2E)=12(A+E)=B.B^2 = \frac{1}{4} (A^2 + 2AE + E) = \frac{1}{4} (2AE + 2E) = \frac{1}{2} (A + E) = B. Hence BB is a projection. Thus there exists a basis of eigenvectors for BB, and the matrix of BB in this basis is of the form diag(1,,1,0,,0)\operatorname{diag}(1, \dots, 1, 0, \dots, 0).

Since A=2BEA = 2B - E the eigenvalues of AA are ±1\pm 1 only.

(ii) Let {Ai:iI}\{ A_i : i \in I \} be a set of commuting diagonalizable operators on VV, and let A1A_1 be one of these operators. Choose an eigenvalue λ\lambda of A1A_1 and denote Vλ={vV:A1v=λv}V_\lambda = \{ v \in V : A_1 v = \lambda v \}. Then VλV_\lambda is a subspace of VV, and since A1Ai=AiA1A_1 A_i = A_i A_1 for each iIi \in I we obtain that VλV_\lambda is invariant under each AiA_i. If Vλ=VV_\lambda = V then A1A_1 is either EE or E-E, and we can start with another operator AiA_i. If VλVV_\lambda \ne V we proceed by induction on dimV\dim V in order to find a common eigenvector for all AiA_i. Therefore {Ai:iI}\{ A_i : i \in I \} are simultaneously diagonalizable.

If they are involutions then I2n|I| \le 2^n since the diagonal entries may equal 1 or 1-1 only.

Similar problems