Page 210 - Matrix Analysis & Applied Linear Algebra
P. 210

4.4 Basis and Dimension                                                            205

                                        If X and Y are subspaces of a vector space V, then the sum of X and
                                    Y was defined in §4.1 to be

                                                      X + Y = {x + y | x ∈X and y ∈Y},
                                    and it was demonstrated in (4.1.1) that X + Y is again a subspace of V. You
                                    were asked in Exercise 4.1.8 to prove that the intersection X∩ Y is also a
                                    subspace of V. We are now in a position to exhibit an important relationship
                                    between dim (X + Y) and dim (X∩ Y) .


                                                          Dimension of a Sum

                                       If X and Y are subspaces of a vector space V, then
                                             dim (X + Y) = dim X + dim Y− dim (X∩ Y) .         (4.4.19)




                                    Proof.  The strategy is to construct a basis for X + Y and count the number
                                    of vectors it contains. Let S = {z 1 , z 2 ,..., z t } be a basis for X∩ Y. Since
                                    S⊆ X and S⊆ Y, there must exist extension vectors {x 1 , x 2 ,..., x m } and
                                    {y 1 , y 2 ,..., y n } such that


                                                  B X = {z 1 ,..., z t , x 1 ,..., x m } = a basis for X
                                    and
                                                  B Y = {z 1 ,..., z t , y 1 ,..., y n } = a basis for Y.
                                    We know from (4.1.2) that B = B X ∪B Y spans X + Y, and we wish show that
                                    B is linearly independent. If
                                                        t        m         n

                                                          α i z i +  β j x j +  γ k y k = 0,      (4.4.20)
                                                       i=1      j=1       k=1
                                    then
                                                                                 
                                                     n             t        m

                                                       γ k y k = −    α i z i +  β j x j    ∈X.
                                                    k=1           i=1      j=1
                                    Since it is also true that     γ k y k ∈Y, we have that     γ k y k ∈X ∩ Y, and
                                                            k                         k
                                    hence there must exist scalars δ i such that

                                           n          t                        n         t

                                              γ k y k =  δ i z i  or, equivalently,  γ k y k −  δ i z i = 0.
                                          k=1        i=1                      k=1       i=1
   205   206   207   208   209   210   211   212   213   214   215