Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Condition: $h,f\in \mathbb{C}^{N\times1}, \text{where}f =\hat{f} + e \text{ and } e^H e \leq 1,\ \ \ Q=h^Hff^Hh$. The function $ Q$ is convex.

Now I want to get the maximum (not minimum), i.e., the upper bound of $Q$ over the variable $e$. How to get it? Will the Lagrangian function of $Q$ be useful in this case?

The Lagrangian function of $Q$ is $\mathcal{L} = h^H(\hat{f} + e)(\hat{f} + e)^Hh + \lambda (e^H e - 1) $ which is convex, where the $\lambda \geq0$.

Thanks!

share|improve this question
    
What variables are you maximizing over? –  copper.hat Oct 2 '13 at 15:33
    
the variable is $e$ –  begforopt Oct 2 '13 at 15:37
    
Didn't asked this question yesterday?: math.stackexchange.com/questions/511146/… –  user91011 Oct 2 '13 at 15:48
    
yes, I did ask.Because I did not give a clear expression about the question yesterday. so I did not get the answer which I want. –  begforopt Oct 2 '13 at 15:52
    
If the KKT conditions are used, How to understand the $\lambda \leq 0$? –  begforopt Oct 2 '13 at 15:57
add comment

1 Answer 1

up vote 1 down vote accepted

The problem can be written as $\mu = \max_{\|e\| \le 1} | \langle h, \hat{f}+e \rangle |= \max_{\|e\| \le 1} | \alpha+\langle h, e \rangle |$, where $\alpha = \langle h, \hat{f}\rangle$.

It should be clear that $\{\langle h, e \rangle \}_{\|e\| \le 1} = \{z \in \mathbb{C} | |z| \le \|h\| \}$.

The problem reduces to $\max_{|z| \le \|h\|} |\alpha + z|$, and a maximizer is easily seen to be $z = \frac{\alpha}{|\alpha|} \|h\|$, if $\alpha \neq 0$, and $z=\|h\|$, otherwise.

Hence $\mu = |\alpha|+\|h\| = | \langle h, \hat{f}\rangle | + \|h\|$.

share|improve this answer
    
Thanks for your help, copper.hat! –  begforopt Oct 3 '13 at 1:24
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.