Page 323 - DMTH502_LINEAR_ALGEBRA
P. 323

Unit 29: Spectral Theory and Properties of Normal Operators




                                     p = (x – c) (x –  c )                                      Notes
                                      j
          where c is a non-real complex number.
          Now let f  = p/p . Then, since  f , . . ., f  are relatively prime, there  exist polynomials  g   with
                  j     j           1     k                                       j
          coefficients in the scalar field such that

                                     1 =    f g  .                                … (6)
                                             j  j
                                          j
          We briefly indicate how such g  may be constructed. If p  = x – c , then f (c )  0, and for g  we take
                                   j                   j    j     j  j          j
          the scalar polynomial 1/f (c ). When every p  is of this form, the f g  are the familiar Lagrange
                               j  j           j                j  j
          polynomials associated with c , . . ., c , and (6) is clearly valid. Suppose some p  = (x – c)(x –  c )
                                  1     k                                j
          with c a non-real complex number. Then V is a real inner product space, and we take
                                         x c   x c
                                     g =
                                      j
                                          s     s
          where s = (c –  c ) f (c). Then
                         j
                                         (s s  )x  (cs cs )
                                     g =
                                      j
                                               ss
          so that g  is a polynomial with real coefficients. If p has degree n, then
                 j
                                       1 –   f g  j
                                              j
                                           j
          is a polynomial with real coefficients of degree at almost n – 1; moreover, it vanishes at each of
          the n (complex) roots of p, and hence is identically 0.
          Now let   be an arbitrary vector in V. Then by (16)

                                       =    f  ( )g  ( )
                                              T
                                                  T
                                             j   j
                                          j
          and since p (T) f (T) = 0, it follows that f (T) g (T)  is in W  for every j. By Lemma 4, W  is orthogonal
                   j  j                 j   j         j                    j
          to W  whenever i  j. Therefore, V is the orthogonal direct sum of W , . . ., W . If   is any vector in
              j                                                 1     k
          W , then
            j
                                p (T) T = Tp (T)  = 0;
                                 j        j
          thus W  is invariant under T. Let T  be the restriction of T to W . Then p (T ) = 0, so that p  is divisible
                j                    j                    j     j  j         j
          by the minimal polynomial for T . Since p  is irreducible over the scalar field, it follows that p  is
                                    j      j                                        j
          the minimal polynomial for  T .
                                   j
          Next, let e  = f g  and E  = e (T). Then for every vector   in V, E  is in W , and
                  j  j  j   j  j                           j       j
                                       =   E j
                                          j
          Thus   – E  =  E   since W  is orthogonal to W  when j   i, this implies that   – E  is in W . It
                   j      j       j               j                         j      i
                      j i
          now follows from Theorem 4 of unit 24 that E  is the orthogonal projection of V on W .
                                               i                               i
          Definition: We call the subspaces W  (1   j   k) the primary components of V under T.
                                       j
          Corollary: Let T be a normal operator on a finite-dimensional inner product space  V and W ,
                                                                                     1
          …, W  the primary components of V under T. Suppose W is a subspace of V which is invariant
               k
          under T.

                                           LOVELY PROFESSIONAL UNIVERSITY                                   317
   318   319   320   321   322   323   324   325   326   327   328