Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Make the DiagonalEK1 work for SecondOrderODEs #302

Closed
nathanaelbosch opened this issue Feb 19, 2024 · 0 comments · Fixed by #305
Closed

Make the DiagonalEK1 work for SecondOrderODEs #302

nathanaelbosch opened this issue Feb 19, 2024 · 0 comments · Fixed by #305

Comments

@nathanaelbosch
Copy link
Owner

In principle this is quite straight-forward, but the fact that SecondOrderODEs by default always stack [du; u] is a bit annoying for us as that results in weird covariance structure. For example, if we have isometric Kronecker covariances of the form $I_d \otimes \Sigma$ in our state-space model, then [du; u] has covariances of the form $\tilde{Sigma} \cdot I_d$, i.e. the order of the Kronecker changed. For BlockDiagonals its a bit weirder: Instead of having a BlockDiagonal matrix, which has dense blocks on its diagonal, we have a BlockMatrix densely filled with Diagonals.

So basically, properly resolving this issue might mean implementing a way to efficiently switch between the two orderings. Or in the extreme, it migh even mean changing the ordering we have so far, which should then make this and related things easier.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant