2
$\begingroup$

As the titel is suggesting, i want to calculate one covariance entry given some other entries. lets say i have a multivariate gaussian of the following form,

$p(X_1,X_2,X_3) = N(\begin{bmatrix} \mu_1 \\ \mu_2 \\ \mu_3 \end{bmatrix},\begin{bmatrix} \Sigma_{1,1} & \Sigma_{1,2} & \Sigma_{1,3} \\ \Sigma_{2,1} & \Sigma_{2,2} & \Sigma_{2,3} \\ \Sigma_{3,1} & \Sigma_{3,2} & \Sigma_{3,3} \end{bmatrix})$.

how can i calculate $\Sigma_{1,3}$ given $\Sigma_{1,1}, \Sigma_{2,2}, \Sigma_{3,3}, \Sigma_{1,2}, \Sigma_{2,3}$, involving conditional independence of $X_1$ and $X_3$ given $X_2$.

$\endgroup$
2
$\begingroup$

Calculate the conditional covariance matrix of $X_1$ and $X_3$ given $X_2$ by applying the widely known Schur complement formula, as shown in What is the conditional probability of variables in a multivariate gaussian? . Then set the "1,3" element of the conditional covariance equal to 0, and solve for $\Sigma_{1,3}$

The result is $\Sigma_{1,3} = \Sigma_{1,2} \Sigma_{2,2}^{-1} \Sigma_{3,2}$

$\endgroup$
  • $\begingroup$ i had something similar in mind, but cant figure out to set the element equal to zero. thank you very much! $\endgroup$ – Looper Jul 24 '17 at 16:50

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.