I'm having a bit of trouble solving a homework problem. I've given the specific problem below, but general answers are 100% welcome:
Consider the following three bases for a vector space , which is a subspace of $C^\infty(-\infty,\infty):$ $$B_1 = \{ 1, e^x, e^{2x}\}$$ $$B_2 = \{ 1, 1+e^{2x}, e^x -e^{2x}\}$$ $$B_3 = \{ 1+e^x, 1+e^{2x}, 1+e^x+e^{2x}\}$$
Obtain the following transition matrices: $$P_{B_1 \to B_2}, P_{B_2 \to B_1}, P_{B_2 \to B_3}$$
I've read about change of basis and done problems for questions like this in finite vector spaces ($\mathbb{R}^n$ and $M_{nm}$, etc). But the way we're given to solve equations involves utilizing the inverse of the basis, which I'm not certain on calculating. At first observation, I can see that I could pull out $\vec{x} = \{1,e^x,e^{2x}\}$ in order to at least simplify what I'm looking at:
$$B_1 = A_1\vec{x} = \begin{bmatrix} 1&0&0\\0&1&0\\0&0&1 \end{bmatrix} \begin{bmatrix} 1\\e^x\\e^{2x} \end{bmatrix}$$ $$B_2 = A_2\vec{x} = \begin{bmatrix} 1&0&0\\1&0&1\\0&1&-1 \end{bmatrix} \begin{bmatrix} 1\\e^x\\e^{2x} \end{bmatrix}$$ $$B_3 = A_3\vec{x} = \begin{bmatrix} 1&1&0\\1&0&1\\1&1&1 \end{bmatrix} \begin{bmatrix} 1\\e^x\\e^{2x} \end{bmatrix}$$
This allows me to view each basis at least in the sense of something which isn't one dimensional. I know from my text that a transition vector can be expressed as $P_{A \to B} = [B]^{-1}[A]$ (where A and B are bases), as well as:
$$AA^{-1} = I\\ IA = A\\ (AB)^{-1} = B^{-1}A^{-1}\\ [B] = [\vec{u}_1| \vec{u}_2| ...| \vec{u}_n]\\ A\vec{x}=\vec{b}\\ \vec{x}=A^{-1}\vec{b}$$
With the final two holding true if A is invertible. I've expressed all the bases in the form $\vec{b} = A\vec{x}$, so following the above logic (since every A from our bases is invertible), I believe the following should hold true: $$ P_{B_1 \to B_2} = B_2^{-1}B_1\\ ... = (A_2\vec{x})^{-1}(A_1\vec{x})\\ ... = (\vec{x})^{-1}A_2^{-1}A_1\vec{x}\\ \vec{x}P_{B_1 \to B_2} = A_2^{-1}A_1\vec{x}$$
However, I'm having trouble getting rid of or simplifying that $\vec{x}$ around. I'm also not certain if it's even appropriate to denote a non-square $\vec{x}^{-1}$. But I feel like there must be a way to simplify $(\vec{x})^{-1}M\vec{x} \to M$.
Failing that, I also tried:
$$rank(B_{1...3}) = 3\\ \therefore dim(V) = 3\\ \therefore \forall \vec{v} \in V, \vec{v} \in \mathbb{R}^3$$
So by this, I believe we can determine that while $C^\infty$ is an infinite-dimensional vector space, $V$ is actually a 3-dimensional subspace described by our bases.
But I suppose this also leaves me a littlle confused, as it would mean e.g. $1 + e^x$ (from $B_{3,1}$) is actually a third-dimensional vector, which is what lead me to breaking down the bases given based on the variable inputs over addition.
In short:
- Is my approach flawed or incorrect? If so, how?
- I'm a bit short on terminology since I'm learning, is there a name for what I did in that first step?
- Most importantly: How do I generate the transition vector for this subspace?