Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let $f(\cdot):\mathbb{R}^n\rightarrow \mathbb{R}$ be a convex function. At $x$ denote the subdifferential by $\partial f(x)$ which is compact and closed. Now, define the approximation of $f$ around a point $x_0$ as follows, \begin{equation} \hat{f}(x)=f(x_0)+\sup_{g\in \partial f(x_0)} <x-x_0,g> \end{equation}

For all $x$, $\hat{f}(x)\leq f(x)$. Also, it is known that, for any $v\in\mathbb{R}^n$, the following holds, \begin{equation} \lim_{t\rightarrow 0^+} \frac{f(x_0+tv)-\hat{f}(x_0+tv)}{t}=0 \end{equation} which is why $\hat{f}$ is an approximation. I am wondering whether the above limit is true uniformly over a bounded set of $v$'s. In particular, is the following true?

Claim: Let $S^{n-1}$ be the unit sphere in $\mathbb{R}^n$. Then, given $\epsilon>0$, there exists $t_0>0$ such that, for all $v\in S^{n-1}$ and for all $t\leq t_0$, one has, \begin{equation} \frac{f(x_0+tv)-\hat{f}(x_0+tv)}{t}<\epsilon \end{equation}

share|improve this question
add comment

Your Answer

 
discard

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

Browse other questions tagged or ask your own question.