For a random variable $X = (x_1,x_2,\ldots,x_n)^T$, I understand that the entries of the covariance matrix would just be the covariance of $x_i$ and $x_j$, but how do I find the eigenvalues and eigenvectors after that, and how does that turn into the eigendecomposition of the covariance matrix?

  • $\begingroup$ Why isn't the answer accepted? $\endgroup$
    – Ramanujan
    Jul 15 '20 at 13:29

Say the covariance matrix is $C$. The eigenvectors $\{v_1, ... v_n\}$ and eigenvalues $\{\lambda_1, ..., \lambda_n\}$ are the set of vectors/values such that $C v_i = \lambda_i v_i$. For numerical computations, these values can be found using an eigensolver, which is included as a standard function in many software libraries. There are various ways to compute them, which may be more or less efficient in different contexts.

The eigendecomposition is a way of expressing a matrix in terms of its eigenvectors and eigenvalues. Let $V$ be a matrix containing the eigenvectors of $C$ along its columns. Let $\Lambda$ be a matrix containing the corresponding eigenvalues along the diagonal, and zeros elsewhere. The eigendecomposition of $C$ is:

$$C = V \Lambda V^T$$


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.