Page 118 - Advanced Linear Algebra
P. 118

102    Advanced Linear Algebra



            Definition Let  4   be a nonempty subset of a vector space  . The annihilator
                                                            =
            4     4  of    is

                                           i
                               4 ~ ¸   = “  ²4³ ~ ¸ ¹¹                    …
            The term annihilator is quite descriptive,  since  4     consists of all linear
                                  (
                                          )
            functionals that annihilate   send to   every vector in  4  . It is not hard to see

            that  4      is a subspace of  =  i  , even when  4   is not a subspace of  .
                                                               =
            The basic properties of annihilators are contained in the following theorem.
            Theorem 3.14
            1 )(Order-reversing )  If 4  5 and   are nonempty subsets of  , then
                                                             =

                                     4‹ 5 ¬    5 ‹ 4
             )
            2   If dim²= ³  B , then for any nonempty subset 4  of   the natural map
                                                          =
                                       ¢ span²4³ š 4
               is an isomorphism from  span²4³ onto   4    . In particular, if    is  a
                                                                       :
               subspace of  , then  :      š  . :
                         =
            3   If   and   are subspaces of  , then
             )
                      ;
                                      =
                 :


                           ²: q;³ ~ : b ;  and  ²: b ;³ ~ : q;



            Proof. We leave proof of part 1  for the reader. For part 2 , since
                                                          )
                                     )
                                    4      ~  ²  ² span  4  ³  ³

            it is sufficient to  prove  that   ¢: š :  is an isomorphism, where   : is a
            subspace of  . Now, we know that   is a monomorphism, so it remains to prove
                      =

            that         . If   : , then   ~   has the property that for all   :   ,
                                     : ~ :
                                       ² ³ ~    ~
            and so   ~ :     , which implies that  :  ‹ :    . Moreover, if # :    , then
            for all   :    we have
                                      ²#³ ~ #² ³ ~
                                                 :
            and so every linear functional that annihilates   also annihilates  . But if  ¤  #  #  , :
            then  there is a linear functional    =  i  for which   ²:³ ~ ¸ ¹  and  ²#³ £   .
            (We leave proof of this as an exercise. ) Hence, #:  and so #~      #  :   and
            so :      ‹   : .
                    )


            For part 3 , it is clear that   annihilates  b  :  ;   if and only if   annihilates both





            :     ; and  . Hence,  ²: b;³ ~ : q ; .  Also, if     ~   b   : b;  where


              :  and    ; , then   Á   ²: q ;³  and so    ²: q ;³ . Thus,
   113   114   115   116   117   118   119   120   121   122   123