1

Consider an inner product space $V$ and vectors $x_0, y_0\in V$. Let $E := \{z\in V : z\perp x_0\implies z\perp y_0\}$ be dense in $V$. Is it true that $y_0$ must be a scalar multiple of $x_0$?

If $x_0 = 0$, then $E = \{y_0\}^\perp$, whose denseness implies that $y_0 = 0$ (I can produce a proof if you want). Thus, w.l.o.g., assume that $\|y_0\| = 1$.


I am able to prove that the answer is affirmative if $E = V$:

Take $z := y_0 - \langle y_0, x_0\rangle x_0$. Then $z\perp x_0$ and thus $z\perp y_0$ so that $\|y_0\|^2 - |\langle y_0, x_0\rangle|^2 = 0$ and thus $$ \|y_0\| = |\langle y_0, x_0\rangle|\le\|y_0\|\|x_0\| = \|x_0\|. $$ Hence, the inequality becomes an equality, and the claim follows.


However, what about the case when we are only given the denseness of $E$?

Atom
  • 3,905

1 Answers1

1

Let $C$ be the orthogonal complement of $x_0$. Note that every $z\in V\setminus C$ satisfies $z\bot x_0\implies z\bot y_0$, because the antecedent $z\bot x_0$ is false. So your set $\{z: z\bot x_0\implies z\bot y_0\}$ includes $V\setminus C$ and is therefore dense, provided $V$ has dimension at least $2$.

Andreas Blass
  • 71,833
  • Big +1! How did I not see that! Just one question though: Though it seems plausible that $V\setminus C$ is dense, how do I prove it? And if I may ask another question: In general, is the complement of a proper subspace always dense? – Atom Feb 03 '24 at 20:32
  • Btw, I don't think the remark about dim being at least 2 is necessary. – Atom Feb 03 '24 at 20:36