Prove that $\sum_{i=1}^n \sum_{j=1}^n \dfrac{a_i a_j}{i+j-1} \ge 0$ for any real numbers $a_1,\cdots, a_n$.
One possible approach might be to find the minimum possible value of $\sum_{i=1}^n \sum_{j=1}^n \dfrac{a_i a_j}{i+j-1}$ or at least find a lower bound for it that's at least 0. It would be useful if we could reduce the number of possibilities we'd need to consider for the $a_i$'s (e.g. for linear functions, we can just maximize or minimize by checking boundary points, but here we don't have any boundary points). The expression is a continuous function of $a_1,\cdots, a_n,$ so it attains a maximum and minimum on any compact subset of $\mathbb{R}^n.$ It might be useful to consider the effect of swapping any two $a_i$'s. Maybe the inequality becomes easier to prove if we assume $a_1\leq a_2\leq \cdots \leq a_n$?
Edit: in hindsight, this is equivalent to showing a Hilbert matrix is positive definite. The main reason why I couldn't solve this question initially is because I didn't realize the fact that the given expression can be rewritten as $\sum_{i=1}^n \sum_{j=1}^n a_i a_j\int_0^1 t^{i-1+j-1} dt = \int_0^1 (\sum_{i=1}^n a_i t^{i-1})^2 dt.$ This notion of writing things as sums of squares is actually a very useful trick for proving inequalities and can be used for math contests.