Let $A \in \mathbb{R}^{n \times n}$ be a symmetric, positive definite matrix with $n \geq 2$. Does the following inequality then hold?
$$
\sum_{i=1}^n \frac{1}{\lambda_i} > \frac{1}{\lambda_\min},
$$
where the $\lambda_i$'s are the eigenvalues of $A$ and $\lambda_\min := \min_{i \in \{1, \cdots, n\}} \left( \lambda_i \right)$.
Thoughts so far:
A simpler inequality that clearly holds is
$$
\sum_{i=1}^n \frac{1}{\lambda_i} \geq \frac{n}{\lambda_\max},
$$
but that doesn't seem useful in this situation.
Also, I realize that the left hand size is $\text{trace}(A^{-1})$, not sure what to do with this.
Background:
An inequality in some lecture notes that I'm reading seems to implicitly hinge on this claim. Specifically, the following is stated as a portion of a longer discourse:
$$
\sigma^2 \sum_{i = 1}^p \frac{1}{\lambda_i} > \frac{\sigma^2}{\lambda_\min},
$$
where $(X^T X)$ is a positive definite matrix, $X \in \mathbb{R}^{n \times p}$.
Best Answer
Yes.
Since the eigenvalues are positive, all the terms in the sum are positive. Therefore, the sum is larger than any of its terms.