1
$\begingroup$

Let $(f_n)_{n\in\mathbb{N}}$ be convex $C^2$ functions such that $$f_n(x)\xrightarrow[n\to\infty]{}f(x)\in\mathbb{R}\ \forall x\in\mathbb{R}.$$ Do you think it's true that for Lebesgue-almost every $x\in\mathbb{R}$ the sequence $(f_n''(x))_{n\in\mathbb{N}}$ is bounded?

$\endgroup$
1
$\begingroup$

No. For $n = 2^k + m$, $n,k,m$ integers with $0 \le m < 2^k$, let $g_n$ be continuous with $0 \le g_n(x) \le 1$, $g_n(x) = 1$ for $m 2^{-k} \le x \le (m+1) 2^{-k}$, $g_n(x) = 0$ for $x \le (m-1) 2^{-k}$ or $x \ge (m+1) 2^{-k}$. Let $f_n(x) = \int_0^x (x-s) g_n(s)\ ds$.

Thus $f_n'' = g_n$. Now $0 \le f_n(x) \le |x| \int_0^1 g_n(s)\ ds < 3 \cdot 2^{-k} |x| \le 6 |x|/n$, so $f_n \to 0$. But every $x \in [0,1]$ has $g_n(x) = 1$ for infinitely many $n$.

$\endgroup$
0
$\begingroup$

No, I don't think so. Take a sequence $h_n$ of nonnegative continuous functions on $[0,1]$ such that $h_n\to 0$ in $L^1$ norm, but for every $x$ the sequence $\{h_n(x)\}$ is unbounded. (Think of tall and thin trapezoids). Integrate twice to get $f_n$ such that $f_n''=h_n$ and $f_n(0)=f_n'(0)=0$. Since $h_n$ converge in $L^1$, you have uniform convergence of $f_n'$, and a forteriori of $f_n$.

$\endgroup$

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.