2
$\begingroup$

Consider a strictly increasing convex function $f(x)$ defined on the interval $[0,1)$ such that $f(0)=1$ and $\lim_{x\to 1^{-}}{f(x)}=+\infty$.

My question: Is the function $f(x)$ logarithmically convex (also called super-convex) in the interval $(1-\epsilon,1)$ for $\epsilon$ sufficiently small? In other words, is $\log f(x)$ a convex function in this interval?

Thanks!

$\endgroup$
3
$\begingroup$

Some complicated answers have been posted! Here's a dead simple one. Let $f(x) = 1 + x$ for $x \in [0, \frac12]$. You can always extend it to go to $+\infty$ in a convex way over $[\frac12, 1)$; how you do it doesn't matter. The point is that over $[0,\frac12]$, $\log f(x) = \log(1+x)$ is a translated version of $\log$, and it's clearly not convex.

$\endgroup$
  • $\begingroup$ I'd be happy to hear from whoever downvoted about what is wrong with this answer. $\endgroup$ – Rahul Jan 27 '12 at 23:23
  • $\begingroup$ Undownvoted${}{}$ $\endgroup$ – Norbert May 15 '12 at 16:51
  • $\begingroup$ This function is not logarithmically convex in the interval $[0,1/2]$ but does not answers my question that is more related with the behavior near 1. $\endgroup$ – ght May 22 '12 at 0:21
2
$\begingroup$

A counterexample is $$f(x) = \frac1{1-x} -x^2 =1+ x + \sum_{k=3}^\infty x^k.$$

It is easy to check that $\frac{d^2}{dx^2} \log f(0) = -1$.

(I hope it fulfills all the requirements)

$\endgroup$
  • $\begingroup$ Thanks! My question is more related to what happens when $x$ is close to 1. $\endgroup$ – ght Apr 10 '12 at 13:39

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.