Page 219 - DMTH502_LINEAR_ALGEBRA
P. 219

Unit 20: Cyclic Decomposition and the Rational Form




          Step 1: There exist non-zero vectors  , ...,   in V such that                         Notes
                                        1    r
          (a)  V = W  + Z ( ; T) + ... + Z( ; T);
                    0     1          r
          (b)  if 1  k  r and
                                     W = W  + Z ( ; T) + ... + Z( ; T)
                                       k    0     1          k
          then the conductor p  = s( ; W  ) has maximum degree among all T-conductors into the subspace
                          k    k  k – 1
          W , i.e., for every k
            k–1
                                                  s
                                  deg p =  max deg ( ;W k  1 )
                                      k     in V
          This step depends only upon the fact that W  is an invariant subspace. If W is a proper T-invariant
                                             0
          subspace, then
                                     0  max deg ( ; W  )  dim V
                                               s
          and we can choose a vector   so that deg s ( ; W) attains that maximum. The subspace W + Z
          ( ; T) is then T-invariant and has dimension larger than dim W. Apply this process to W = W  to
                                                                                    0
          obtain  . If W  = W  + Z( ; T) is still proper, then apply the process to W  to obtain  . Continue
                 1   1    0    1                                    1         2
          is that manner. Since dim W  > dim W  , we must reach W = V is not more than dim V steps.
                                 k       k – 1            r
          Step 2: Let  , ....   be non-zero vectors which satisfy conditions (a) and (b) of Step 1. Fix k, 1  k
                    1    r
            r. Let   be any vector in V and let f = s ( ; W ). If
                                               k–1
                                   f   0      g  i i  ,  i  in W i
                                          1  i  k
          then f divides each polynomial g  and   = f , where  is in W .
                                    i     0   0       0     0
          If k = 1, this is just the statement that W  is T-admissible. In order to prove the assertion for k > 1,
                                         0
          apply the division algorithm:
                  g i  fh i  i , r  r i  0  or  deg r i  deg .                     ...(2)
                                                         f
          We wish to show that r = 0 for each i, Let
                             i
                       k  1
                         h i i .                                                   ...(3)
                        1
          Since   –   is in W  ,
                         k – 1
                  s ( ; W k  1 ) s ( ; W k  1 )  . f
          Furthermore

                         k  1
                  f   0    r i  i .                                                ...(4)
                          1
          Suppose that some r  is different from 0. We shall deduce a contradiction. Let  j be the largest
                           i
          index i for which r  0. Then
                         i
                          j
                  f        r  ,     r  0     and     deg r  deg .
                                                               f
                      0     i  i     j                   j                         ...(5)
                          1
          Let p = s ( ; W  ). Since W   contains W  , the conductor f = s ( ; W  ) must divide p:
                     j – 1     k   1       j   1                 k  1
                                              p = fg.



                                           LOVELY PROFESSIONAL UNIVERSITY                                   213
   214   215   216   217   218   219   220   221   222   223   224