Page 176 - Linear Algebra Done Right
P. 176

Generalized Eigenvectors
                                           T(z 1 ,z 2 ,z 3 ) = (z 2 , 0,z 3 ),
                             2
                      then T (z 1 ,z 2 , 0) = 0 for all z 1 ,z 2 ∈ C. Hence every element of C
                      whose last coordinate equals 0 is a generalized eigenvector of T.As 3                165
                      you should verify,
                                 3
                                C ={(z 1 ,z 2 , 0) : z 1 ,z 2 ∈ C}⊕{(0, 0,z 3 ) : z 3 ∈ C},
                      where the first subspace on the right equals the set of generalized eigen-
                      vectors for this operator corresponding to the eigenvalue 0 and the sec-
                      ond subspace on the right equals the set of generalized eigenvectors
                      corresponding to the eigenvalue 1. Later in this chapter we will prove
                      that a decomposition using generalized eigenvectors exists for every
                      operator on a complex vector space (see 8.23).
                         Though j is allowed to be an arbitrary integer in the definition of a  Note that we do not
                      generalized eigenvector, we will soon see that every generalized eigen-  define the concept of a
                      vector satisfies an equation of the form 8.3 with j equal to the dimen-  generalized eigenvalue
                      sion of V. To prove this, we now turn to a study of null spaces of  because this would not
                      powers of an operator.                                              lead to anything new.
                                                                                                          j
                                                                            k
                         Suppose T ∈L(V) and k is a nonnegative integer. If T v = 0, then  Reason: if (T − λI) is
                                   k
                                                            k
                      T  k+1 v = T(T v) = T(0) = 0. Thus null T ⊂ null T  k+1 . In other words,  not injective for some
                      we have                                                             positive integer j, then
                                                                                          T − λI is not injective,
                                         0
                                                  1
                                                                k
                      8.4     {0}= null T ⊂ null T ⊂· · ·8 null T ⊂ null T k+1  ⊂· · · .  and hence λ is an
                                                                                          eigenvalue of T.
                      The next proposition says that once two consecutive terms in this se-
                      quence of subspaces are equal, then all later terms in the sequence are
                      equal.
                      8.5    Proposition: If T ∈L(V) and m is a nonnegative integer such
                      that null T  m  = null T m+1 , then
                              0
                                       1
                         null T ⊂ null T ⊂ ··· ⊂ null T m  = null T m+1  = null T m+2  =· · · .
                         Proof:   Suppose T ∈L(V) and m is a nonnegative integer such
                      that null T m  = null T m+1 . Let k be a positive integer. We want to prove
                      that
                                           null T m+k  = null T m+k+1 .

                      We already know that null T m+k  ⊂ null T m+k+1 . To prove the inclusion
                      in the other direction, suppose that v ∈ null T  m+k+1 . Then
   171   172   173   174   175   176   177   178   179   180   181