1

Let $\mathscr S$ be a subspace of $\mathbb R^p$, and let G be positive semidefinite such that $\operatorname{span}(G) \subset \mathscr S$. The following statements are equivalent:

  1. for all $\nu \in \mathscr S, \nu \neq 0,$ we have $\nu^T G \nu >0$;

  2. $\operatorname{span}(G) = \mathscr S$.

Proof.

$3 \Rightarrow 2$. If statement 2 is not true then there exists $\nu \in \mathscr S, \nu \neq 0$, such that $\nu^T G \nu=0$. Then $\nu \perp \operatorname{span}(G)$ and statement 3 is not true.

Question: Given $\nu^T G \nu=0$, how do we get $\nu \perp \operatorname{span}(G)$?

I know by definition, if I want to say $\nu$ is orthogonal of a subspace $W$, then for every vector $w \in W$, we should have $\nu^T w=0$. But here I have span, which confuses me.

Jonathen
  • 1,044

1 Answers1

1

I assume that the "span" of $G$ is its column-space.

The key fact here is that because $G$ is positive semidefinite, $v^TGv = 0$ if and only if $Gv = 0$; I present several proofs of this fact on my post here.

So, the fact that $v^TGv = 0$ implies that $Gv = 0$, which is to say that $v$ is an element of the null space of $G$, which (because $G$ is a symmetric matrix) is the orthogonal complement to the column space of $G$.

Ben Grossmann
  • 225,327