Change-of-basis without using orthogonal as new basis

24 Views Asked by At

I have a question in my exam which is:

Let :v = [3,1] and T : R2 -> R2 denote the transformation such that T(x) is the vector in span{v} nearest to x

Let :0<a<pi/2 denote the angle between the positive x-axis and the vector v. Let R: R2->R2 denote the transformation such that R(x) is the vector x rotated by the angle a clockwise

Find the Linear transformation R and T (Dot product and Differentiation is not allowed)

I know that R is just a clockwise matrix, but for T, the answer says it is found by producing an orthogonal vector w=[-1,3] to form a new basis and then produce a change-of-basis matrix P = $$\begin{bmatrix} 3 & -1 \\ 1 & 3 \\ \end{bmatrix} $$ Then since T(x) is in span(v), so the transformation performed on the new basis = B = $$\begin{bmatrix} 1 & 0 \\ 0 & 0 \\ \end{bmatrix} $$

Then by T=PB(P^-1) we can find out the answer A = $$\begin{bmatrix} 0.9 & 0.3 \\ 0.3 & 0.1 \\ \end{bmatrix} $$

But I have 2 questions neither the professor nor TA answer me

  1. What if I choose any other arbitrary vector to form the new basis instead of using the orthogonal one? says w=[0,3]

I did some try before asking, but the answer is wrong. I guess that happened on the Matrix B transformation performed on a new basis, probably because the basis is no longer orthogonal. But isn't the T simply project x on v when the basis is changed? which means no height and any other vectores, so the matrix shouldn't be same????

  1. TA says there is actually another easier method, he just hint us T(x) = [ ? ? ]

I am extremely confused with this so-called easier form, I referred to David Lay, linear algebra Ch5.4, I guess he wanted to say is the transformation form x -> Ax, which can be calculated by putting a basis in the transformation and attaining a new basis vector, then form a new transformation matrix. Is that the so-called easier form? How does it work??

Really sorry for my ignorance and bothering your time