Linear Transformations finding matrix in respect to a basis and coordinate change matrix.

606 Views Asked by At

Define $T: Poly_2 \ to\ Poly_2$ by $$T(at^2+bt+c)=3ct^2 +2at-b$$

1) Show that T is a linear transformation and give a matrix A for T with respect to the basis $B=\{t^2,t,1\}$.

2) Give a matrix $Q$ for $T$ with respect to the basis $B_1=\{t^2+1 , t-2, t+3\}$ directly

3) Lastly show that $PAP^{-1} = Q$ where $P$ is the coordinate change matrix.

For Part 1: Since $T(3t^2+2at−b) = T(3t^2)+T(2t)−T(b)$ and $T(r(3t^2+2at−b)= rT(3t^2+2at−b)$ because $rT(3t^2+2at−b) = T(r(3t^2))+T(r(2t))−T(rb)$. it's a linear transformation.But is there more to show? I feel like this is just the definition in a sense.

For part 2 Since it's a linear transformation, the vectors are linearly independent, therefor there exists an a,b,c for which we can get the zero vector. Meaning we have $a(t^2+1)+b(t-2)+c(t+3)=0$....please let me know if this is correct so far, because I'm stuck and would like to know if Im even going the right direction. Not sure where to bring $3t^2+2at-b$ in but I feel as if it should be in the factored form $(t+1) (3 t^2-1)$

1

There are 1 best solutions below

15
On BEST ANSWER

Show that T is a linear transformation

A linear transformation $f: V \to W$ has two properties:

  1. Homogeneity of degree $1$: $f(\alpha x) = \alpha f(x)$
  2. Additivity: $f(x+y) = f(x)+f(y)$

These can be tested at the same time by confirming that $f(\alpha x + y) = \alpha f(x) + f(y)$ for all $x,y \in V$ and $\alpha$ in your base field (probably $\Bbb R$ in your class).

So let's test your linear transformation: $$\begin{align}T\left[\alpha(a_1t^2+b_1t+c_1) + (a_2t^2+b_2t+c_2)\right] &= T\left[(\alpha a_1+a_2)t^2+(\alpha b_1+b_2)t+(\alpha c_1+c_2)\right] \\ &= 3(\alpha c_1+c_2)t^2 + 2(\alpha a_1+a_2)t-(\alpha b_1+b_2) \\ &= \alpha(3 c_1)t^2 + \alpha(2a_1)t -\alpha(b_1) + (3c_2t^2 +2a_2t-b_2) \\ &= \alpha(3 c_1t^2 + 2a_1t -b_1) + (3c_2t^2 +2a_2t-b_2) \\ &= \alpha T(a_1t^2+b_1t+c_1) + T(a_2t^2+b_2t+c_2)\end{align}$$

Thus $T$ is linear.


Give a matrix $A$ for $T$ with respect to the basis $B=\{t^2,t,1\}$.

So first we need to determine the coordinates of the input vector $\mathbf x = at^2 + bt+c$ and the ouput vector $\mathbf y = 3ct^2+2at-b$. Hopefully you can see that $$[\mathbf x]_B = \pmatrix{a \\ b \\ c},\quad [\mathbf y]_B = \pmatrix{3c \\ 2a \\ -b}$$

So we just need the matrix $A$ which takes $\mathbf x$ to $\mathbf y$. Thus we figure out $$\pmatrix{? & ? & ? \\ ? & ? & ? \\ ? & ? & ?}\pmatrix{a \\ b \\ c} = \pmatrix{3c \\ 2a \\ -b}$$

Clearly the matrix that does the job is $$A = \pmatrix{0 & 0 & 3 \\ 2 & 0 & 0 \\ 0 & -1 & 0}$$

If you need I can walk you through it, but hopefully you're far enough in your linear algebra studies that you can figure out how that matrix was obtained.


Give a matrix $Q$ for $T$ with respect to the basis $B_1=\{t^2+1,t−2,t+3\}$ directly.

This part is really tedious. What we need is the coordinates of the $T(t^2+1)$, $T(t-2)$, and $T(t+3)$ wrt to the basis $B_1$. First let's figure out how to represent an arbitrary vector in this basis. That's equivalent to figuring out $[\mathbf x]_{B_1}$. So let's do that.

$$\begin{align}\mathbf x &= at^2+bt+c = a'(t^2+1)+b'(t-2)+c'(t+3) \\ &= a't^2+a'+b't-2b'+c't+3c' \\ &= a't^2 +(b'+c')t+(a'-2b'+3c')\end{align} \\ \implies \begin{cases} a=a' \\ b=b'+c' \\ c=a'-2b'+3c'\end{cases} \\ \implies \pmatrix{a \\ b \\ c} = \pmatrix{1 & 0 & 0 \\ 0 & 1 & 1 \\ 1 & -2 & 3}\pmatrix{a' \\ b' \\ c'} \\ \implies \pmatrix{a' \\ b' \\ c'} = \frac 15\pmatrix{5 & 0 & 0 \\ 1 & 3 & -1 \\ -1 & 2 & 1}\pmatrix{a \\ b \\ c}$$

That last step required finding the inverse matrix of $$\pmatrix{1 & 0 & 0 \\ 0 & 1 & 1 \\ 1 & -2 & 3}$$

Thus we can see that $[\mathbf x]_{B_1} = \frac 15\pmatrix{5a \\ a+3b-c \\ -a + 2b + c}$

Now that we've found the coordinates of an arbitrary vector wrt $B_1$, let's figure out where the basis vectors of $B_1$ get mapped to:

$$\begin{align}T(t^2+1) &= 3t^2+2t \\ &= \frac 15(15)(t^2+1)+\frac 15(3+6-0)(t-2) + \frac 15(-3+4+0)(t+3) \\ &= 3(t^2+1) + \frac 95(t-2) +\frac 15(t+3) \\ T(t-2) &= -6t^2-1 = -6(t^2+1) -(t-2)+(t+3) \\ T(t+3) &=9t^2-1= 9(t^2+1)+2(t-2)-2(t+3)\end{align}$$

Thus $Q$ will just be the matrix that maps $\pmatrix{1 \\ 0 \\ 0}$ to $\pmatrix{3 \\ \frac 95 \\ \frac 15}$, etc.

Then $$Q = \pmatrix{3 & -6 & 9 \\ \frac 95 & -1 & 2 \\ \frac 15 & 1 & -2}$$


Lastly show that $PAP^{−1}=Q$ where $P$ is the coordinate change matrix.

What we want here is a matrix that'll change the coordinates of a vector with respect to the basis $B_1$ to the coordinates of that vector with respect to the basis $B$. Then once we've done that, the reason why $PAP^{-1}=Q$ is that on the LHA we're just converting the coordinates of an input vector from the $B_1$ basis to the $B$ basis, then using the matrix $A$ to perform the transformation $T$, then converting back to the $B_1$ basis. That is equivalent to just finding a single matrix which performs the transformation $T$ with respect to the basis $B_1$ -- i.e. the RHS should just be the matrix $Q$ found in part $2$.

All we need to do is figure out the coordinates of the basis vectors in $B_1$ with respect to $B$. That should be easy enough:

$$\begin{cases} t^2+1 = (1)t^2 + (1)1 \\ t-2 = (1)t+(-2)1 \\ t+3 = (1)t+(3)1\end{cases}$$

Therefore $$[t^2+1]_B = \pmatrix{1 \\ 0 \\ 1},\quad [t-2]_B = \pmatrix{0 \\ 1 \\ -2},\quad [t+3]_B = \pmatrix{0 \\ 1 \\ 3}$$

I posit that if we were to put those vectors into the columns of a matrix, then that would be the change of basis matrix from $B_1$ to $B$. Let's test it. Remember that $[t^2+1]_{B_1} = \pmatrix{1 \\ 0 \\ 0}$, so let's see where that maps to under this matrix:

$$\pmatrix{1 & 0 & 0 \\ 0 & 1 & 1 \\ 1 & -2 & 3} \pmatrix{1 \\ 0 \\ 0} = \pmatrix{1 \\ 0 \\ 1}$$ But that vector is just the coordinates of $t^2+1$ in the basis $B$. You can likewise confirm that this matrix converts the other two basis vectors in $B_1$ to their coordinates in $B$.

Thus $P^{-1}$ (in your question's notation) is $\pmatrix{1 & 0 & 0 \\ 0 & 1 & 1 \\ 1 & -2 & 3}$. Find its inverse to find $P$ and then take the product to see if you get back the matrix we found the hard way in part $2$.