1
$\begingroup$

The entire problem statement is,

Suppose that $V$ and $W$ are finite dimensional and that $U$ is a subspace of $V$. Prove that there exists $T\in\mathfrak{L}(V,W)$ such that $\text{null}(T)=U$ if and only if $\text{dim}(U)\geq\text{dim}(V)-\text{dim}(W)$.

My attempt at the proof is:

Suppose first there exists $T\in\mathfrak{T}(V,W)$ such that $U=\text{null}(T)$. We know that $$\text{dim}(V)=\text{dim null}(T)+\text{dim}(\text{range}(T))$$ and so $$\text{dim}(U)=\text{dim}(V)-\text{dim}(\text{range}(T))\geq\text{dim}(V)-\text{dim}(W).$$ For the other direction, let $(u_1,\dots,u_m)$ be a basis for $U$, which can then be extended to a basis in $V$, namely, $(u_1,\dots,u_m,v_1\dots,v_n)$. Also, let $(w_1,\dots,w_p)$ be a basis for $W$. So then by the hypothesis, $m\geq(m+n)-p$ which means $n\leq p$.

That's as far as I got for the other direction. I'm not sure how else to continue, but I do know I have to figure out a linear map such that $U=\text{null}(T)$. Is there an obvious construction?

Thanks for help or hints!

$\endgroup$

1 Answer 1

3
$\begingroup$

You have basically done it : You know that $n\leq p$, so just define $T: V\to W$ by $$ T(u_i) = 0, 1\leq i\leq m, \text{ and } T(v_j) = w_j, 1\leq j\leq n $$ Then $T$ extends to a linear map $T \in \mathcal{L}(V,W)$. By this construction $$ U \subset \text{null}(T) $$ Furthermore, if $T(v) = 0$, then write $$ v = \sum_{i=1}^m \alpha_i u_i + \sum_{j=1}^n \beta_j v_j $$ Then $$ T(v) = \sum_{j=1}^n \beta_j w_j = 0 $$ By linear independence of $\{w_j\}_{j=1}^n$, it follows that $\beta_j = 0$ for all $j$, and hence $$ v = \sum_{i=1}^m \alpha_i u_i \in U $$ Hence, $U = \text{null}(T)$.

$\endgroup$
3
  • $\begingroup$ Is it because the basis in $W$ is larger that we can define $T(v_j)=w_j$? Can $j$ go up to $p$ or is that not possible? Also, how did you know to construct $v$ the way you did in the second part after "Furthermore,..."? $\endgroup$
    – S.D.
    Aug 25, 2013 at 5:39
  • $\begingroup$ Firstly, $j$ goes up to $n$. Hence, $j$ can go up to $p$ iff $n=p$. Secondly, any $v\in V$ is a linear combination of basis vectors. This is all I have written in that line, except that I have broken the basis up into vectors in $U$ and vectors outside $U$ (as you have chosen the basis vectors that way) $\endgroup$ Aug 25, 2013 at 5:47
  • $\begingroup$ I see. Thanks for the help. $\endgroup$
    – S.D.
    Aug 25, 2013 at 5:54

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .