One useful prior family that is missing from the current available options is a truncated normal (or mixture of truncated normal and point-mass at zero). The "generalized binary" prior is similar but rather specialized.
Anyhow, perhaps we could start with a basic implementation, building off of the generalized binary prior, and call it ebnm_point_tnormal?
One useful prior family that is missing from the current available options is a truncated normal (or mixture of truncated normal and point-mass at zero). The "generalized binary" prior is similar but rather specialized.
Anyhow, perhaps we could start with a basic implementation, building off of the generalized binary prior, and call it
ebnm_point_tnormal
?