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
Sigma is now an atomic vector, covariance matrix, or list of covariance matrices. Would it not be more practical to use an array? List implies that the covariance matrices can have different dimensions - but this seems like something that should not be allowed.
The text was updated successfully, but these errors were encountered:
Indeed, covariance matrices should have the same dimension in Bain. I used list because 1. it is easier to use than arrary with which users should understand which dimension is for the number of matrices. To me it is easier to construct a list of matrices rather than a three dimension array. 2. it is easier to handle as we can use length(), lapply() et al. functions on it.
I'm not sure I agree that it is easier for us to use. Array is a simpler datatype than list. With three-dimensional arrays, for example, you can just use apply(x, 3, ...) across the third dimension, and instead of length you can use dim()[3].
For end users, however, you might be right that it's easier to understand. I'll keep this issue open for a bit, because I had a bug which could be traced back to the different data types sigma can have.
Sigma is now an atomic vector, covariance matrix, or list of covariance matrices. Would it not be more practical to use an array? List implies that the covariance matrices can have different dimensions - but this seems like something that should not be allowed.
The text was updated successfully, but these errors were encountered: