A few more details and explanations to the bayesian arguments in the docu.

[christophsax]

It would be nice to have a bit more info on these. Including some hint on what can be accomplished by manipulating it. See @param obs_df for an example.

#' @param trans_prior prior matrix for B in the transition equation. Default is
#'   zeros.
#' @param trans_shrink prior tightness on B matrix in trasition equation
#' @param trans_df prior degree of freedom for transition equation
#' @param obs_prior prior matrix for H (loadings) in the observation equation
#'  Default is zeros.
#' @param obs_shrink prior tightness on H (loadings) in the observation equation
#' @param obs_df named vector (see details). prior degree of freedom
#'   for observables. This is useful to give specific series a larger weight,
#'   e.g. 1. (default 0, method `bayesian` only).

[SethOttoQuant]

@christophsax, let me know if this looks OK and we can close this issue.

[christophsax]

I will do another turn and may have some more question. So let's keep it open and assign to me.

[christophsax]

matrix dimensions

It would be good if the object is specified in the beginning. E.g.,:

#' @param trans_prior m x n matrix; m: number of factors, n: number of lags. Prior B matrix in the 
#' transition equation. Default is a matrix with 0. E.g., to use a random walk prior with m factors and 
#' p lags, set `trans_prior = cbind(diag(1,m,m), matrix(0,m,m*(p-1)))`.

I currently have no idea how to construct these matrices.

orthogonal_shocks

What is this good for?

#' @param orthogonal_shocks return a rotation of the model with orthogonal
#'   shocks and factors. This is useful ....

[SethOttoQuant]

OK... see updates on arguments.

[christophsax]

Ok for now