I am looking for a proof of weighted version of the inequality as following.

Let $I$ be an interval of the real line and let $f$ denote a real-valued, convex function defined on $I$.

If $x_1, . . . , x_n$ and $y_1, . . . , y_n$ are numbers in $I$ such that:

  1. $x_1 \ge x_2 \ge x_3...\ge x_n,$ and $y_1 \ge y_2 \ge y_3...\ge y_n$

  2. $x_1+...+x_i \ge y_1+...+y_i$ and $x_{i+1}+...+x_n \le y_{i+1}+...+y_n$ for $i=1,...,n-1$

  3. Let $\lambda_i >0$ and $\sum_1^{n} \lambda_i =1$

Then show that:

$$\lambda_1f(x_1)+\lambda_2f(x_2)+...+\lambda_nf(x_n)-f(\lambda_1x_1+\lambda_2x_2+...+\lambda_nx_n) \ge \lambda_1f(y_1)+\lambda_2f(y_2)+...+\lambda_nf(y_n)-f(\lambda_1y_1+\lambda_2y_2+...+\lambda_ny_n) $$

The inequality holds with equality if and only if $x_i=y_i$ for all $i \in {1, 2,...,n}$


If $\lambda_1x_1+\lambda_2x_2+....+\lambda_nx_n=\lambda_1y_1+\lambda_2y_2+....+\lambda_ny_n$

Then we have the weighted version of the Karamata inequality:

$ \lambda_1f(x_1)+\lambda_2f(x_2)+...+\lambda_nf(x_n) \ge \lambda_1f(y_1)+\lambda_2f(y_2)+...+\lambda_nf(y_n) $


This is false by some stupid reasons. Say, if $\lambda_1=\lambda_n=0$, then what we have to prove does not depend on $x_1,x_n,y_1,y_n$, but changing these four variables with fixed other variables we may easily satisfy all the conditions.

  • $\begingroup$ Dear Dr @FedorPetrov . Thank to You, but I don't think so, because when $\lambda_1=\lambda_n=0$ then $x_1, x_n$ and $y_1, y_n$ be removed in two sides. $\endgroup$ – Oai Thanh Đào Aug 27 '16 at 10:53
  • $\begingroup$ From both sides of the inequality, but not from your conditions! $\endgroup$ – Fedor Petrov Aug 27 '16 at 11:01
  • $\begingroup$ Dear Dr. @FedorPetrov , yes you are right $\endgroup$ – Oai Thanh Đào Aug 27 '16 at 11:09

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.