The rank of a linear transformation $A: E \longrightarrow F$ is the largest integer such that the pullback of A is not zero

77 Views Asked by At

$\textbf{Notation:}$

$\mathcal{A}_r(F)$ is the space of $r$-forms alternating on $F$.

$A^*$ is the pullback of a linear transformation $A$.

$\textbf{Question:}$ The rank of a linear transformation $A: E \longrightarrow F$ is the largest integer $r$ such that the pullback of $A$, i.e., $A^*: \mathcal{A}_r(F) \longrightarrow \mathcal{A}_r(E)$ is different to zero.

$\textbf{My attempt:}$

Let be $\dim F := n$ and the rank($A$) := $m$. We know that every set of $\{ v_i \in E \}$ with at least $m + 1$ elements is linearly dependent, because rank($A$) $ = m$, then given $\omega \in \mathcal{A}_r(F)$ , $v_1, \cdots, v_{m+1} \in E$ and assuming without loss of generality that $v_2 = k v_1$, we know that

$\omega(v_1, v_2, \cdots,v_{m+1}) = \omega(v_1, kv_1,\cdots,v_{m+1}) = k\omega(v_1, v_1, \cdots, v_n) = k 0 = 0$,

then $r < m+1$ by the definition of $r$ given in the statement of the question. We know too that every set of $\{ v_i \in E \}$ with at most $m$ elements is linearly independent, because rank($A$) $ = m$, then given $\omega \in \mathcal{A}_s(F)$ , $v_1, \cdots, v_{s} \in E$ with $s < m$, $\omega(v_1, v_2, \cdots,v_s) \neq 0$ and follows the definition of $r$ that $r = m. \ \square$

I would like to know if my attempt is correct. Thanks in advance!

1

There are 1 best solutions below

1
On BEST ANSWER

There's a nicely written proof of this question in: Relation between rank of a linear transformation and pullback ($r$-linear forms.)

As for your answer, I think the general idea is correct and I would only like to add:

1 - After proving that $r < m + 1$ you say:

"We know too that every set of $\{v_i \in E \}$ with at most element $m$ elements is linearly independent"

This doesn't always hold (just take a set $\{v_i \in E \}$ such that $v_1 = v_2$). But I do believe that we can choose a convenient set $\{v_i \in E \}$ such that this holds and this is enough for our purposes here.

2 - You also say:

"... then given $\omega \in$ ..."

Again, here it is not necessary that what you say holds for every $\omega$. It is enough to find at least one form $\omega$ who does not vanish to conclude what you want (check the link to see how to find a possible $\omega$ that fulfills the conditions we want).