Given symmetric positive definite matrix $A \in \mathbb{R}^{n\times n}$ and symmetric matrix $B\in \mathbb{R}^{n\times n}$, how to prove that $X = A+B+\frac{1}{2}BA^{-1}B$ is positive definite?
This can be easily seen when they are both diagonal: the diagonal entries of $X$ is
$$x_{ii} = a_{ii}+b_{ii}+\frac{b_{ii}^2}{2a_{ii}} > 0$$ for all $b_{ii} \in \mathbb{R}$ and $a_{ii} > 0$. But I am struggling a bit proving the general case.
The proof follows similar pattern that the diagonal case. For example how did you prove it for the diagonal case? One way of doing it is rewriting $x_{ii}$ as:
\begin{align} x_{ii} = \frac12 a_{ii}^{\frac12}\left(1+\left(1+\frac{b_{ii}}{a_{ii}}\right)^2\right)a_{ii}^{\frac12} \end{align}
In the general case you can rewrite $X$ as:
$$X = \frac12 A^{\frac12} \left(I + \left(I + A^{-\frac12}BA^{-\frac12}\right)^2\right)A^{\frac12}$$
Can you complete the proof from here?