Correlated Splines in BRMS?

Interfaces brms
1

Hi All,

A question: Is it possible to fit a multivariate model in brms with two correlated splines?

For example, rather than fit something like this:

bf(y1 ~ 1 + x1 + (1 |a| year)) +
bf(y2 ~ 1 + x2 + (1 |a| year))

We fit something like this:
bf(y1 ~ 1 + x1 + s(year)) +
bf(y2 ~ 1 + x2 + s(year))

But where the two year splines share information?

2

How should such a correlation be conceptually realized?

3

One option might be to follow the bs = 'fs' basis approach of mgcv and have the two s(year) terms share a single basis and smoothness parameter?