Page 123 - Matrices theory and applications
P. 123

6. Matrices with Entries in a Principal Ideal Domain; Jordan Reduction
                              106
                              6.3.1 Example: The Companion Matrix of a Polynomial
                              Given a polynomial
                                               P(X)= X + a 1 X
                              there exists a matrix B ∈ M n (k) such that the list of invariant factors of
                              the matrix XI n − B is (1,... , 1,P). We may take the companion matrix
                              associated to P to be      n      n−1  + ··· + a n ,
                                                       0  ···  ···  0  −a n
                                                                          
                                                          .         .   .
                                                          .        .   .  
                                                       1    .       .   .
                                                                          
                                                                          
                                                          .    .    .   .
                                                                           
                                                          .    .   .   .   .
                                                       0    .    .  .   .
                                              B P := 
                                                     .   .    .        .  
                                                    
                                                     . .  . .  . .  0  . .  
                                                                           
                                                       0  ···   0  1   −a 1
                              Naturally, any matrix similar to B P would do as well, because if B =
                              Q −1 B P Q,then XI n −B is similar, hence equivalent, to XI n −B P .In order
                              to show that the invariant factors of B P are the polynomials (1,... , 1,P),
                              we observe that XI n −B P possesses a minor of order n−1 that is invertible,
                              namely, the determinant of the submatrix
                                                                        
                                                   −1   X    0   ···  0
                                                        .    .   .
                                                        .   .    .    . . 
                                                    0    .    .    .   .  
                                                
                                                    .
                                                                        
                                                    .    .   .    .       .
                                                       .    .   .       
                                                   .    .    .    .  0
                                                                        
                                                    .
                                                    .        .    .
                                                            .   .       
                                                   .         .    .  X  
                                                    0   ···  ···  0   −1
                              We thus have D n−1 (XI n − B P ) = 1, so that the invariant factors
                              d 1 ,... ,d n−1 are all equal to 1. Hence d n = D n (XI n − B P )=det(XI n −
                              B P ), the characteristic polynomial of B P ,namely P.
                                In this example P is also the minimal polynomial of B P .Infact,if Q is
                              a polynomial of degree less than or equal to n − 1,
                                                 Q(X)= b 0 X n−1  + ··· + b n−1 ,
                                             1
                              the vector Q(A)e reads
                                                                     1
                                                        n
                                                     b 0 e + ··· + b n−1e .
                              Hence Q(A) = 0 and deg Q ≤ n − 1imply Q = 0. The minimal polynomial
                              is thus of degree at least n. It is thus equal to the characteristic polynomial.
                              6.3.2 First Canonical Form of a Square Matrix
                              Let M ∈ M n (k)beasquare matrix and P 1 ,... ,P n ∈ k[X] its similarity
                              invariants. The sum of their degrees n j (1 ≤ j ≤ n)is n. Let us denote
   118   119   120   121   122   123   124   125   126   127   128