Coordinates and Linear Transformations October 22, 2015 p~(1) . In words, T p~(2) maps a polynomial in the variable t to the column vector of its values at t = 1 and t = 2. For example, 1 1 0 3 T (1) = , T (t) = , T ( 2 + 3t t ) = . 1 2 4 Consider the linear transformation P3 (a) Compute T (t2 ), T (t3 ), and T (3 T / R 2 defined by T (~p) = t + 2t2 + 5t3 ). In order to understand T as completely as possible, we will make use of coordinates. [ ]B / R4 the coordinate mapping, Let B = {1, t, t2 , t3 } be the standard basis for P3 and P3 which just maps a polynomial to the column vector of its coefficients. For example, 2 3 2 3 2 3 1 0 2 6 7 607 617 7 = ~e1 , [t]B = 6 7 = ~e2 , [ 2 + 3t t3 ]B = 6 3 7 = 2~e1 + 3~e2 ~e4 . [1]B = 6 405 405 405 1 0 0 (b) What is dim P3 ? (c) Compute [t2 ]B , [t3 ]B , and [3 t + 2t2 + 5t3 ]B . Using the coordinate mapping, we can view T as a linear transformation R4 P3 [ ]B ✏ R T / R2 > S / R2 : . S 4 S should act the same way on a vector of coordinates as T acts on the polynomial that has those B-coordinates, namely S([~p]B ) = T (~p). Since the domain and codomain of S are vector spaces of the form Rn , we can find a matrix A such that S(~x) = A~x for all ~x in R4 . In fact, the columns of A are S(~e1 ), S(~e2 ), S(~e3 ), S(~e4 ). Let’s compute these vectors. Here are the first two: 1 1 S(~e1 ) = S([1]B ) = T (1) = , S(~e2 ) = S([t]B ) = T (t) = . 1 2 (d) Compute S(~e3 ) and S(~e4 ). (e) Write down the matrix A. (f) Use A and your answer in (c) to compute S([3 your answer matches (a). 2 t + 2t2 + 5t3 ]B ). Check to make sure You now know how S acts, so you can compute the kernel of S, which equals Nul A, and the range of S, which equals Col A. (Recall that the kernel of S is the set of all vectors ~x such that S(~x) = ~0 and the range of S is the set of all outputs S(~x) for all ~x.) (g) Compute a basis for Nul A by row reducing A. (h) Compute a basis for Col A. (Hint: use the pivot columns of A.) (i) Compute the dimensions of the kernel of S and the range of S. (j) Is S one-to-one? Is S onto? Now that you have mastered S, you can use the fact that the coordinate mapping is an isomorphism to deduce the same information about T . (k) Compute a basis for the kernel of T using your basis for the kernel of S. Check your vectors are really in the kernel of T by applying T to them. 3 (l) Without doing any work, why is the range of T equal to the range of S? (m) Compute the dimensions of the kernel of T and the range of T . (n) Is T one-to-one? Is T onto? Now that you have deduced a lot of information about T , you should check whether your results make sense by returning to the definition of T at the beginning. (o) The fact that T is onto means that for any real numbers a and b, there is a polynomial p~ in P3 such that p~(1) = a andp~(2) = b. Is this surprising? Compute such a polynomial a by finding a solution to A~x = using row reduction. b (p) Factor the polynomials in your basis for the kernel of T in (k). (Feel free to use www.wolframalpha.com to help you.) Why are these factorizations not surprising given the way we defined T ? 4