LINEAR ALGEBRA (Math 211)
Sheet #7
LINEAR TRANSFORMATIONS

A linear transformation is a map $T: V\to W$ from the vector space V into the vector space W satisfying two conditions.
\fbox{$T(\mbox{\bf x}+ \mbox{\bf y}) = T(\mbox{\bf x}) + T(\mbox{\bf y})$} , and

\fbox{$T({\lambda}\mbox{\bf x})={\lambda}T(\mbox{\bf x})$} for any $\mbox{\bf x}$, $\mbox{\bf y}$ from V and any ${\lambda}$ from ${\mathbb R}$.


Examples. V=W=Pn is the space of polynomials of degree $\leq n$.
(1) $T={d\over dx}$.                 (2)Sifting operator:     $T:f(t)\mapsto f(t+1)$.




Matrix of a linear transformation T. Let $\mbox{\bf e}_1$, $\mbox{\bf e}_2$, ..., $\mbox{\bf e}_n$ be a basis for V and $\mbox{\bf f}_1$, $\mbox{\bf f}_2$, ..., $\mbox{\bf f}_m$ be a basis for W. If

\begin{displaymath}T(\mbox{\bf e}_i)= a_{1,i}\mbox{\bf f}_1 + a_{2,i}\mbox{\bf f...
...t(\begin{array}{c}a_{1,i}\\ \vdots\\ a_{m,i}\end{array}\right)
\end{displaymath}

then the matrix AT= $\left(
\begin{array}{ccc}
a_{1,1}&\dots&a_{1,n}\\
\vdots&\vdots&\vdots\\
a_{m,1}&\dots&a_{m,n}\end{array}\right)$ is called the matrix of the linear transformation T with respect to the chosen bases $\mbox{\bf e}_1$, ..., $\mbox{\bf e}_n$ for V and $\mbox{\bf f}_1$, ..., $\mbox{\bf f}_m$ for W.

Theorem. For a vector $\mbox{\bf x}= x_1\mbox{\bf e}_1 + x_2\mbox{\bf e}_2+\dots+ x_n\mbox{\bf e}_n$ from V the coordinates of the vector $\mbox{\bf y}=T(\mbox{\bf x})$ with respect to the basis $\mbox{\bf f}_1$, ..., $\mbox{\bf f}_m$ can be written as

\begin{displaymath}\left(\begin{array}{c}y_1\\ \vdots\\ y_m\end{array}\right) = ...
...t
\left(\begin{array}{c}x_1\\ \vdots\\ x_n\end{array}\right)\ .\end{displaymath}




Composition of linear transformations. Let $T: V\to W$ and $L: W\to U$ be linear transformations. Their composition $L\circ T$ is a linear transformation from V to U such that $(L\circ T)(\mbox{\bf x})=L\left(T(\mbox{\bf x})\right)$. Then \fbox{$A_{L\circ T} = A_L\cdot A_T$} , where

AL is the matrix of L with respect to the bases $\mbox{\bf e}_1$,..., $\mbox{\bf e}_n$ for V and $\mbox{\bf f}_1$, ..., $\mbox{\bf f}_m$ for W;
AT is the matrix of L with respect to the bases $\mbox{\bf f}_1$,..., $\mbox{\bf f}_m$ for W and $\mbox{\bf g}_1$, ..., $\mbox{\bf g}_k$ for U;
$A_{L\circ T}$ is the matrix of $L\circ T$ with respect to the bases $\mbox{\bf e}_1$,..., $\mbox{\bf e}_n$ for V and $\mbox{\bf g}_1$, ..., $\mbox{\bf g}_k$ for U.





Transition matrix P from an old basis ${\cal B}$ to a new one ${\cal B}'$.

$\textstyle \parbox{8cm}{For vectors:\quad
$\left(\mbox{\bf e}'_1\ \dots\ \mbox{\bf e}'_n\right)=
\left(\mbox{\bf e}_1\ \dots\ \mbox{\bf e}_n\right)\cdot P$}$                  $\textstyle \parbox{8cm}{For coordinates:\quad{
$\left(\begin{array}{c}x_1\\ \vd...
...}=
$P\cdot${
$\left(\begin{array}{c}x'_1\\
\vdots\\ x'_n\end{array}\right)$}}$

Theorem. Let AT be the matrix of the linear transformation $T: V\to W$ with respect to the bases $\mbox{\bf e}_1$,..., $\mbox{\bf e}_n$ for V and $\mbox{\bf f}_1$, ..., $\mbox{\bf f}_m$ for W. Then

$A'_T = P^{-1}_W\cdot A_T\cdot P_V$
is the matrix of T with respect to the new bases $\mbox{\bf e}'_1$,..., $\mbox{\bf e}'_n$ for V and $\mbox{\bf f}'_1$, ..., $\mbox{\bf f}'_m$ for W, where
PV is the transition matrix from the basis $\mbox{\bf e}_1$,..., $\mbox{\bf e}_n$ to the basis $\mbox{\bf e}'_1$,..., $\mbox{\bf e}'_n$ and
PW is the transition matrix from the basis $\mbox{\bf f}_1$,..., $\mbox{\bf f}_m$ to the basis $\mbox{\bf f}'_1$,..., $\mbox{\bf f}'_m$.