-
Notifications
You must be signed in to change notification settings - Fork 12
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
PriorParameters in sampling are not transformed for parameterScaleUniform #491
Comments
@dweindl @yannikschaelte I am not to sure, if the other two |
Same for |
In the documentation it should be stated that for the parameterScaleX type priors, values are on the parameter scale (original), not linear. |
... which is apparently not clearly the case, thanks for pointing out ... @dweindl do you remember? were parameterScale parameters supposed to be on or off scale? |
That on the other hand would mean, that for the default ( |
I think the motivation here was that in terms of distributions, one wants to specify the distribution parameters on the scale where they are applied. E.g. for a normal N(0,1) one does not want 1;10. But both goes I think. |
My understanding was that everything should be on linear scale, so that any change of I am not sure whether there really has been a deliberate decision. Agreed that docs need clarification. |
In the sampling routines for
ParameterScaleUniform
, the lower/upper bounds are not transformed to the parameter scale. (Clipping then might lead to samples mostly on one bound...).The text was updated successfully, but these errors were encountered: