Definitions: 1 Given an n-dimensional vector space X and basis set V = {v1 , ..., vn } for X , the coordinate mapping crdV : X → Rn is defined by, for x ∈ X , crdV (x) = α = (α1 , ...αn ), where α is the unique linear combination of the elements of V s.t. x = ∑ni=1 αi vi . 2 Let X and Y be n- and m-dimensional vector spaces with basis sets U and V. Given T ∈ L(X , Y ), define M T by for i = 1, ...n, the i’th column of M T is crdV (T (vi )). M T is the matrix representation of T given bases U and V. 3 Define the mapping MtxV,U : L(X , Y ) → Fm×n by for T ∈ L(X , Y ), MtxV,U (T ) = M T , the matrix rep. of T given bases U and V. 1 2 FIRST subscript identifies the basis w.r.t. which points in range of T are represented. SECOND subscript identifies the basis set on which T is defined matrix rep of v w.r.t. std basis v = (1.5, 2) matrix rep of v w.r.t. alt basis v = (2.23, 1.12) b 2 e2 = (0, 1) a vector v ∈ V v 1 b e1 = (1, 0) Figure 1. Different representations of a vector = √ 5 1)/ (2, u2 2 0.5 1 1 -1 -0.5 0.5 1 u1 -2 -1 T (u2 ) T (u1 ) 1 2 -1 -0.5 -2 -1 Figure 2. We know nothing about the matrix A, only know what T does to U 1 v1 T (v2 ) = 0.5 -1 -0.5 T (v1 ) = 2 v2 0.5 1 −1 √1 √ , 2 2 -2 √3 , √3 2 2 1 -1 1 2 -1 -0.5 -2 -1 Figure 3. MtxU,V (T ): the transformation of V by T , expressed in terms of U FIRST subscript—i.e., U—identifies basis (i.e., standard one) in terms of which points in the range of T are represented SECOND subscript—i.e., V—identifies the points in the domain of T for which T is initially defined. ;; 1 v1 T (v2 ) = (0, 1) 0.5 -1 T (v1 ) = (3, 0) 2 v2 -0.5 0.5 1 -2 1 -1 1 2 -1 -0.5 -2 -1 Figure 4. MtxV,V (T ): the transformation of V by T , expressed in terms of V FIRST subscript—i.e., V—identifies basis (i.e., basis of e/v’s) in terms of which points in the range of T are represented SECOND subscript—i.e., V—identifies the points in the domain of T for which T is initially defined. u2 T (u2 ) = (2.1213, 0.7071) 1 2 XD 0.5 -1 -0.5 T (u1 ) = (2.1213, −0.7071) 1 0.5 1 u1 -2 -1 1 2 -1 -0.5 -2 -1 Figure 5. MtxV,U (T ): the transformation of U by T , expressed in terms of V FIRST subscript—i.e., V—identifies basis (i.e., basis of e/v’s) in terms of which points in the range of T are represented SECOND subscript—i.e., U—identifies the points in the domain of T for which T is initially defined. Theorem: Let X , Y and Z be finite dimensional vector spaces with bases U,V and W respectively, and suppose S ∈ L(X , Y ) and T ∈ L(Y , Z ). Then MtxW,V (T ) · MtxV,U (S ) = MtxW,U (T ◦ S ) where for x ∈ X , the composition mapping T ◦ S is defined by, for x ∈ X , T ◦ S (x) = T (S (x)). (1) u2 1 v2 = S(u2 ) T (v2 ) = T ◦ S(u2 ) 0.5 -1 -0.5 u1 0.5 -2 1 T (v1 ) = T ◦ S(u1 ) 2 v1 = S(u1 ) 1 -1 1 2 -1 -0.5 -2 -1 Figure 6. The mapping T : what it does to the new basis set V Example #1: S defined by S (ui ) = vi , i.e., i’th basis element → i’th e/vector 1 0 MtxV,U (S ) = . 0 1 i T defined by T (v ) = Avi = λi vi , i.e., takes e/v’s to their images 3 0 MtxW,V (T ) = . 0 1 T ◦ S takes each ui to λi vi . 3 0 MtxW,U (T ◦ S ) = . 0 1 Example #2: 1 X is the space of 2nd order polynomials on R, i.e., x : R → R ∈ X if for all θ ∈ R, x (θ) = α0 + α1 θ + α2 θ2 , for some α0 , α1 , α2 ∈ R. Let U = {u1 , u2 , u3 } = {1, θ, θ2 }. 2 3 Y is the space of 1st order polynomials on R, i.e., y (θ) = α0 + α1 θ. Let V = {v1 , v2 } = {1, θ}. Z is the space of zero’th order polynomials on R, i.e., z (θ) = α0 . Let W = {w1 } = {1}. x0 1 0 MtxW,V (T ) = 0 1 . 4 S ∈ L(X , Y ): S (x ) = = 5 T ∈ L(Y , Z ): T (y ) = y 0 = dx (θ) . dθ 6 T ◦ S ∈ L(X , Z ): (T ◦ S )(x ) = x 00 = MtxW,U (T ) = 0 0 0 MtxV,U (S ) = 0 dx (θ) . dθ d 2 x (θ) . d θ2 2 = MtxW,V (T ) · MtxV,U (S ) 0 . 2