From "Linear algebra done right" chapter 3:

Linear maps can be constructed that take on arbitrary values on a basis. Specifically, given a basis $\{v_1, \dots, v_n\}$ of $V$ and any choice of vectors $w_1, \dots, w_n \in W$, we can construct a linear map $T: V \to W$ such that $T(v_j) = w_j$ for $j = 1, \dots, n$. There is no choice of how to do this — we must define $T$ by $T(a_1v_1 + \dots + a_nv_n) = a_1w_1 + \dots + a_nw_n$, where $a_1, \dots, a_n$ are arbitrary elements of $\mathbb{F}$. Because $\{v_1, \dots, v_n\}$ is a basis of $V$, the equation above does indeed define a function $T$ from $V$ to $W$. You should verify that the function $T$ defined above is linear and that $T(v_j) = w_j$ for $j = 1, \dots, n$.

It seems obvious, and we can prove $\{w_1, \dots, w_j\}$ is a basis for $W$, but how to prove $T(v_j) = w_j$ exactly? Thank you.

  • 1
    $\begingroup$ $v_j=0v_1+0v_2+\cdot+1v_j+0v_{j+1}+\cdots+0v_n$. $\endgroup$ Nov 15 '13 at 5:48
  • $\begingroup$ Got it. Thanks. $\endgroup$
    – iwbabn
    Nov 16 '13 at 14:57
  • $\begingroup$ Good. You can write it up as an answer, help clear up the Unanswered Questions list. $\endgroup$ Nov 17 '13 at 4:25

First of all, you can't prove that $\{w_1, \dots, w_n\}$ is a basis for $W$. This is false as $\{w_1, \dots, w_n\}$ is any choice of $n$ vectors in $W$, they may not even be distinct. For example, the construction outlined above works perfectly well for $w_1 = \dots = w_n = 0$.

Regarding your question, as Gerry Myerson pointed out in his comment, given $T : V \to W$ defined by

$$T(a_1v_1 + \dots + a_nv_n) = a_1w_1 + \dots + a_nw_n$$

we can see that $T(v_j) = w_j$ by noting that

\begin{align*}T(v_j) &= T(0v_1 + \dots + 0v_{j-1} + 1v_j + 0v_{j+1} + \dots + 0v_n)\\ &= 0w_1 + \dots + 0w_{j-1} + 1w_j + 0w_{j+1} + \dots + 0w_n\\ &= w_j.\end{align*}


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.