Can someone please check my proof and my definitions.

Let $A \in \mathbb{R}^{n \times m}$ be my matrix.

The left null space of $A$ is written as,

$\mathcal{N}(A^\top) = \{x \in \mathbb{R}^n| A^\top x = 0\}$

The orthogonal complement of the column space $\mathcal{C}(A)$ is written as,

$\mathcal{C}(A)^\perp = \{x \in \mathbb{R}^n | x^\top y = 0, \forall y \in \mathcal{C}(A)\}$

We want to show that $\mathcal{N}(A^\top) = \mathcal{C}(A)^\perp $

First, we show, $\mathcal{N}(A^\top) \subseteq \mathcal{C}(A)^\perp$

Let $x \in \mathcal{N}(A^\top)$, then $A^\top x = 0 \implies x^\top A = 0^\top \implies x^\top Av= 0^\top v, \forall v \in \mathcal{C}(A) \implies x^\top y = 0 , y = Av$, $\implies x \in C(A)^\perp$.

Next, we show, $\mathcal{N}(A^\top) \supseteq \mathcal{C}(A)^\perp$

Let $x \in C(A)^\perp$, then $x^\top y = 0$, forall $y \in C(A)$. But $y = Av, \forall v \in \mathbb{R}^n$. Hence, $x^\top y = x^\top Av = v^\top A^\top x.$ For all $v \neq 0, A^\top x = 0$, hence $x \in \mathcal{N}(A^\top)$.

I'm pretty confident about the first proof. But the second proof is a bit more rough. Can someone please check for me.


$y \in C(A)$ means that there exists (at least one) $v$ of appropriate dimension such that $y = Av$.

So we can say: For $x \in C(A)^{\perp}$, then $x^T y = 0$ for every $y \in C(A)$. For every $y \in C(A)$, we can express $y = Av$ for some (nonzero) $v$. So we can always express $x^T y$ as $x^T Av$. So $$x^T y = x^T (A v) = (x^T A) v = (A^T x)^T v = 0^T v = 0$$ for $v \neq 0$, so we must have $A^T x = 0$, i.e., $x \in N(A^T)$.


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.