1
$\begingroup$

Inspire from Kantorovich Inequality and my previous question. I am looking for a proof of the nice inequality as following:

Let $f(x)$ is a real continuous function that is strictly convex on $[m, M]$, let $m \le x_i \le M$, for $i=1,2,\ldots,n$ then show that:

$$nf\left(\frac{x_1+\cdots+x_n}{n}\right)+n\left(f(M)+f(m)-2f\left(\frac{M+m}{2}\right)\right) \ge f(x_1)+\cdots+f(x_n)$$

Equality holds if only if $m=x_1=x_2=\cdots=x_n=M$

$\endgroup$
1
  • $\begingroup$ Why $n$ twice on the left rather than once in the denominator on the right? $\endgroup$ Aug 21, 2016 at 4:44

1 Answer 1

2
$\begingroup$

This is true. Choose a linear function $\ell(x)$ such that $\ell(m)=f(m),\ell(M)=f(M)$. Denoting $g=f-\ell$ we get $g(m)=g(M)=0$ and have to prove $$ng\left(\frac{x_1+\cdots+x_n}{n}\right)-2ng\left(\frac{M+m}{2}\right) \geqslant g(x_1)+\cdots+g(x_n).$$ But $g$ is convex, thus $g$ is non-positive on $[m,M$]. So, RHS is non-positive, and it suffices to prove that LHS is non-negative. Indeed, if we denote $t=\frac{x_1+\dots+x_n}n$, $s=\frac{m+M}2$, then $g(t)\geqslant g(t)+g(2s-t)\geqslant 2g(s)$.

$\endgroup$
6
  • $\begingroup$ Dear Dr Fedor Petrov, Thank to You very much for your proof. I think the inequality is true with weighted version. And I think it is a generalization of Kantorovic inequality. What do you think with my remarks? Thank to You again. $\endgroup$ Aug 21, 2016 at 14:36
  • $\begingroup$ Let $f(x)$ is a real continuous function that is strictly convex on $[m, M]$, let $m \le x_i \le M$, for $i=1,2,\ldots,n$, $\lambda_i > 0$ and $\sum_1^{n} \lambda_i = 1$ then show that: $$f\left(\lambda_1x_1+\cdots+\lambda_nx_n \right)+\left(f(M)+f(m)-2f\left(\frac{M+m}{2}\right)\right) \ge \lambda_1f(x_1)+\cdots+\lambda_nf(x_n)$$ Equality holds if only if $m=x_1=x_2=\cdots=x_n=M$ $\endgroup$ Aug 22, 2016 at 1:56
  • $\begingroup$ Yes, the proof is the same. $\endgroup$ Aug 22, 2016 at 7:19
  • $\begingroup$ Dear Dr @FedorPetrov , The inequality above is a generalization of the Kantorovich inequality? $\endgroup$ Aug 22, 2016 at 7:24
  • $\begingroup$ I do not think so. It would be if you replace coefficient of $f(M)+f(m)-2f((m+M)/2)$ from $n$ to $n/2$. But such inequality is not true in general. $\endgroup$ Aug 22, 2016 at 13:19

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.