0
$\begingroup$

Let $f_{i}\colon X\rightarrow \mathbb{R}$ be convex, Lipshitz and positive (X - reflexive Banach space) for $i=1,2$. One may easily prove that $f_{i}$ is weak lower semicontinuous. How to prove the following statement:

$lim\, inf\, f_1(x_{n})[f_{2}(x_{n})-f_{2}(x)]\geq0$, where $\{x_{n}\}$ converges weakly to some $x$?

We can write $lim\, inf\, f_1(x_{n})[f_{2}(x_{n})-f_{2}(x)]\geq\lim\, inf\, f_1(x_{n})\cdot lim\, inf\, [f_{2}(x_{n})-f_{2}(x)]$ (it would solve the problem)

but

$(f_{2}(x_{n})-f_{2}(x))$ has to be positive and I don't know that. Maybe convexity has something to do with the proof?

$\endgroup$

1 Answer 1

0
$\begingroup$

EDITED:

If the sequence $x_n$ converges weakly, by the Uniform Boundedness Principle it must be bounded. Since $f_1$ is Lipschitz, that implies $f_1(x_n)$ is bounded, let's say by $B$. Since $\liminf_n f_2(x_n) \ge f_2(x)$, for any $\epsilon > 0$ there is $N$ such that $f_2(x_n) \ge f_2(x) - \epsilon/B$ for all $n \ge N$, and then $f_1(x_n) (f_2(x_n) - f_2(x)) \ge -\epsilon$ for such $n$. Thus $\liminf_n f_1(x_n)(f_2(x_n) - f_2(x)) \ge 0$.

$\endgroup$
1
  • $\begingroup$ Perfect and no convexity used. Thank you so much! $\endgroup$
    – zorro47
    Mar 22, 2017 at 19:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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