Page 371 - DMTH401_REAL ANALYSIS
P. 371

Unit 30: Riemann's and Lebesgue




          for every n   and hence   ¥ n 1  A     +1   . Thus    is a -algebra on  containing all open  Notes
                                                                      d
                                 =
                                    n
          subsets of  . Hence  = ( ).
                    d
                                  d
          Now, it suffices to show that card() = card(). Since there is an open ball of radius 1 centered at
          each point of  , we have card()  card( )  card(). So it suffices to establish that card() 
                      d
                                             
          card(). Since card(L ) = card() and  =  <   , it is enough to show that card( )  card() for
                                                                         
          each   L .
                  
                                                                             d
          Let  be the collection of all open balls in   with rational radius and center in   . Then  is
                                              d
                                      d
          countable, and any open set U    can be written as a countable union of members of . Hence
                        
          card( )  card( ) = card(). Let   L  and suppose we have proved that card( )  card() for
                                                                         
          every  < . If  is a limit ordinal, then   =   <    is a countable union and hence card( ) 
                                                                                  
          card(). If there is  with  + 1 = , then any A    can be written as A =   ¥  A  with A  A
                                                                        =
                                                                       n 1  n    n   
                                                                                    
                                                       
                                                                                  
              
                                                      
            . This gives a one-one map from   into (A     ) . Hence card( )  card((A    ) ) 
              
                                                      
                                                                                  
                                                                    
                                          
                                                                             
                                                 
          card(). This completes the proof.
          Corollary: For any uncountable set Y   , there is A  Y such that A is not a Borel subset of .
                                                                                     d
                                           d
                             d
          Proof: We have card(( )) = card() = card(Y) < card((Y)).
                          d
                                                              d
          Definition: Let ( ) = {A   : m * L,d [A] = 0}. The members of ( ) are called Lebesgue null sets.
                                  d
                                                                                     d
                               d
                                              d
                                                     d
          The -algebra ( ) on   generated by ( )  ( ) is called the Lebesgue -algebra on  ,
                         d
          and members of ( ) are called Lebesgue measurable subsets of  .
                           d
                                                                 d
                                                            d
                  d
                                                                  d
                             d
                                                                         d
                                        d
          card(( )) = card(( )) = card(( )) > card(). Hence, ( )  ( )  ( ).
          Proof: Let K be the middle-third Cantor set. Then, for any subset A   K, we have  m * L,1  [A] 
                                                        d
                                                              d
          m *  [K] = 0. So m *  [A] = 0 also. This shows that (K)  ( )  ( ). And card((K)) = card(())
            L,1        L,d
          since K is an uncountable subset of .
                                                          d
          [Translation invariance] (i) A + x  ( ) for every A  ( ) and x   .
                                          d
                                                                    d
                                                   d
                         d
                                         d
          (ii)  A + x  ( ) for every A  ( ) and x   .
          (iii)  A + x  ( ) for every A  ( ) and x   .
                                                  d
                         d
                                         d
          Proof: First let us mention a general principle that will be used at many places. To establish that
          the members of a certain -algebra  on a set X satisfies a certain property P, it suffices to do
          the following: show that the collection {A  X: A satisfies property P} is a -algebra, and then
          find a suitable collection   (X) generating  and show that every member of  satisfies the
          property P.
                                d
                      d
          Let  = {A   : A + x  ( ) for every x   }. It is easy to check that  is a -algebra containing
                                              d
          all d-boxes. And recall that the collection of all d-boxes generates  ( ). This proves (i). Next,
                                                                   d
          statement (ii) is a consequence of the translation invariance property of the Lebesgue outer
          measure, and (iii) follows from (i) and (ii) by applying the principle mentioned above.
          We will give other characterizations of the Lebesgue measurable sets shortly, and we will also
          show that ( )  ( ).
                            d
                      d
          Self Assessment
          Fill in the blanks:
          1.   ............................................... is developed through approximations of a finite nature (e.g.:
               one tries to approximate the area of a bounded subset of    by the sum of the areas of
                                                                2
               finitely many rectangles).
          2.   While Riemann’s  theory is  restricted to  the  Euclidean  space,  the  ideas  involved  in
               ............................ are applicable to more general spaces, yielding an abstract measure theory.
                                           LOVELY PROFESSIONAL UNIVERSITY                                   365
   366   367   368   369   370   371   372   373   374   375   376