Let $n\in \mathbb{N}$ and $f:\mathbb{R}\rightarrow\mathbb{R}$
$$f(\sum\limits_{k=1}^n \lambda_k x_k) \leq \sum\limits_{k=1}^n \lambda_k f(x_k)$$
where $\sum\limits_{k=1}^n \lambda_k = 1$ and arbitrary $x_k \in \mathbb{R}$ and fixed $\lambda_k \in \mathbb{R} $ with the fixed sum of 1.
Do I understand that correctly, that $\sum\limits_{k=1}^n \lambda_k = 1$ for any $n$ or does it hold only for a specific $n$ and not necessarily $\sum\limits_{k=1}^{n+1} \lambda_k = 1$ for example?
Furthermore, I'd use induction to prove it, the start is trivial so I'll only show the step from n to n+1.
$$f(\sum\limits_{k=1}^{n+1} \lambda_k x_k)=f(\sum\limits_{k=1}^n \lambda_k x_k+\lambda_{n+1}x_{n+1})$$ but here's the problem because in order to use the hypothesis, I'd have to look at $f(\sum\limits_{k=1}^n \lambda_k x_k)$ alone, and I don't know how to "split it". I know I got to use the fact, that the function is convex, so it's increasing, continous and I got the defintion of a convex function. Can somebody give me a hint on what tool to use?