0
$\begingroup$

Let $\gamma_i$ be such that for even $i$ $\gamma_i=1$ and for odd $i$ $\gamma_i$ shall have absolute value $1$ and the product of all of the odd ones is also on the complex unit circle but not 1 or -1.

I ask:

Is there a lower bound for the real part

$$\Re \sum_{i=0}^n \gamma_i x_i \overline{x_{i+1}}$$ such that $x_{n+1}:=x_0$ which is better than the trivial one $-\left\lVert x \right\rVert^2$ ?

$\endgroup$
2
  • $\begingroup$ To clarify: you want a bound that is expressed in terms of the $\gamma_i$? $\endgroup$
    – Yemon Choi
    Commented Aug 2, 2017 at 15:00
  • $\begingroup$ yes, precisely.. $\endgroup$
    – Zinkin
    Commented Aug 2, 2017 at 18:25

1 Answer 1

1
$\begingroup$

Let $\alpha \in \mathbb{C}$ be a point on the unit circle that is close to $1$, but $\alpha \neq 1$. Take the vector $$x = (1,-1,1,-1,\ldots)$$ of length $n$.Then $$-\Vert x \Vert^2 = -n.$$ Set $\gamma_1 = \alpha$ and $\gamma_i = 1$ for all other odd $i$. Then the product of the $\gamma_i$ is $\pm \alpha$ which is not $\pm 1$ and we get $$\sum_{i=1}^n \gamma_i x_i\overline{x_{i+1}} = -\alpha -(n-1).$$ As we chose $\alpha$ to be close to $1$, the real part of this expression should be close to $-n$.

Formalizing this idea properly with $\epsilon \to 0$ you should be able to show that there is no better bound in general.

$\endgroup$
3
  • $\begingroup$ you are not allowed to choose gamma. Gamma is given and satisfies the above properties. Also, I think your proof is rather a "proof by example" which is of course not admissible. What is also funny is that you want to let epsilon go to zero without defining it anywhere in your text... $\endgroup$
    – Zinkin
    Commented Aug 2, 2017 at 12:00
  • 2
    $\begingroup$ My "proof" is no proof at all, it is an example where the bound $-\Vert x \Vert^2$ is tight. So you can't find a better lower bound in general. If you now say that the $\gamma_i$ are fix, you might look for a bound that depends on them. $\endgroup$
    – Dirk
    Commented Aug 2, 2017 at 12:05
  • $\begingroup$ that is the point. the $\gamma_i$ are fixed and also you have another sum, don't you? we have $x_ix_{i+1}$ $\endgroup$
    – Zinkin
    Commented Aug 2, 2017 at 12:06

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .