0
$\begingroup$

I need to prove the following statement

Let $S \subset \mathbb{R}^n$ a nonempty convex set and $f: S \to \mathbb{R}$. Then $f$ is convex in $S$ if and only if $f\left( \sum_{i=1}^m\lambda_ix_i \right) \leq \sum_{i=1}^m\lambda_if(x_i)$ for al $m \in \mathbb{N}$, for all $x_1, \dots, x_m \in S$ and for al $\lambda_1, \dots, \lambda_m > 0$ such that $\sum_{i=1}^m\lambda_i=1$

My try: I think I managed to prove the backwads implication. Let $m=2$, and since $\lambda_1 + \lambda_2=1$ then $\lambda_2 = 1-\lambda_1$, so

$$f(\lambda_1x_1+\lambda_2x_2)=f(\lambda_1x_1+(1-\lambda_1)x_2) \leq \lambda_1f(x_1)+(1-\lambda_1)f(x_2)$$ since $f$ is convex in $S$ by assumption.

For the forward implication, I thought that induction might work. The case when $m=1$ is trivial. However, I'm struggling to prove the general case. Any help with that step will be highly appreciate. Thanks in advance!

$\endgroup$
4
$\begingroup$

Hint: Suppose that $f(c_1x_1+...+c_nx_n)\leq \sum c_if(x_i)$, $\sum_ic_i=1$.

Consider $c_1,...,c_{n+1}: \sum_ic_i=1, c_{n+1}\neq 0$. $f(c_1x_1+...+c_{n+1}x_{n+1})=f(c_1x_1+..+c_{n-1}x_{n-1}+(c_n+c_{n+1})({{c_nx_n+c_{n+1}x_{n+1}}\over {c_n+c_{n+1}}}))\leq c_1f(x_1)+..c_{n-1}f(x_{n-1})+(c_n+c_{n+1})f({{c_nx_n+c_{n+1}x_{n+1}}\over {c_n+c_{n+1}}})$.

We have $f({{c_nx_n+c_{n+1}x_{n+1}}\over c_{n+1}})\leq {c_n\over {c_n+c_{n+1}}}f(x_n)+{c_n\over {c_n+c_{n+1}}}f(x_{n+1})$ since ${c_n\over{c_n+c_{n+1}}}+{c_{n+1}\over{c_n+c_{n+1}}}=1$. This implies the result.

$\endgroup$
  • $\begingroup$ How do you use ${c_n\over{c_n+c_{n+1}}}+{c_{n+1}\over{c_n+c_{n+1}}}=1$ to get $f({{c_nx_n+c_{n+1}x_{n+1}}\over c_{n+1}})\leq {c_n\over {c_n+c_{n+1}}}f(x_n)+{c_n\over {c_n+c_{n+1}}}f(x_{n+1})$? $\endgroup$ – user441848 Dec 31 '17 at 21:11

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.