2
$\begingroup$

Suppose we have a set of vectors $\{e_1, e_2, \ldots, e_m\}$ in a Euclidean space $\mathbb E^n$ such that for all distinct $i$ and $j$, the inner product satisfies the following property: $\langle e_i, e_j\rangle\in[-1; 1]$.

Does there always exist a set of unit vectors $\{e'_1, e'_2, \ldots, e'_m\}$ in $\mathbb E^n$ such that for all distinct $i$ and $j$, the inner product satisfies the following property: $\langle e_i, e_j\rangle = \langle e'_i, e'_j\rangle$?

$\endgroup$

1 Answer 1

1
$\begingroup$

No, not always. Consider in $\Bbb{R}^n$ the three vectors: \begin{align*} e_1 &= (1, 0, 0) \\ e_2 &= (-1, 0, 1) \\ e_3 &= (-1, 0, -1). \end{align*} Then \begin{align*} e_1 \cdot e_2 &= -1 \\ e_1 \cdot e_3 &= -1 \\ e_2 \cdot e_3 &= 0. \end{align*} If there were unit vectors $e'_1, e'_2, e'_3$ with the desired properties, then $$e'_1 \cdot e'_2 = e'_1 \cdot e'_3 = -1,$$ which would imply that $e'_2 = e'_3 = -e'_1$. We would therefore be forced to have $e'_2 \cdot e'_3 = 1 \neq 0$.

$\endgroup$
5
  • $\begingroup$ (Also, I assumed $e_i \cdot e_j = e'_i \cdot e'_j$ is only required for all distinct $i$ and $j$.) $\endgroup$ Commented Aug 30, 2019 at 0:39
  • 1
    $\begingroup$ What are the arguments for your claim that $e'_1 \cdot e'_2 = e'_1 \cdot e'_3 = -1$ implies $e'_2=e'_3=-e'_1$? $\endgroup$ Commented Aug 30, 2019 at 5:53
  • $\begingroup$ @JensSchwaiger For any unit Euclidean vector $u$, there exists a unique unit Euclidean vector $v$ such that $\langle u, v\rangle = -1$. The vector $v$ equals $-u$. $\endgroup$ Commented Aug 30, 2019 at 15:21
  • $\begingroup$ @TheoBendit Thank you, this is a compelling counterexample. I updated the original post to include “distinct”. $\endgroup$ Commented Aug 30, 2019 at 15:21
  • 2
    $\begingroup$ @JensSchwaiger Since the $e'_i$s are all norm $1$, we obtain Cauchy Schwarz equality: $(-e'_1) \cdot e'_2 = \|-e'_1\| \|e'_2\|$. This occurs if and only if $-e'_1$ is a positive multiple of $e'_2$. Since both are norm $1$, this positive multiple must be $1$. The same argument works for $-e'_1$ and $e'_3$. $\endgroup$ Commented Aug 31, 2019 at 9:38

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.