You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As we work to add multivariate functionality to State Space the representation of structural component variables become non-scalar.
In the univariate case, however, we have certain variables that are scalar like auto_regressive_sigma which will have a shape of () and a Dimension of None.
In contrast, its multivariate counterpart will have shape (2,) and a Dimension of (auto_regressive_endog,).
Having a concrete value for the shape and a dimension could be intuitive for the end user and switching between univariate and multivariate models may be more parsimonious.
Would it make sense for the univariate case to follow suit such that the auto_regressive_sigma would have shape (1,) and a Dimension of (auto_regressive_endog,)?