Sorry my bad. I switched to CmdStanR 2.30 based on my collaborator’s advice and it worked. Thank you!
Hi! Sorry for bringing this thread back to life, but I had a question about it for @spinkney: I noticed that in the earlier post you have here (#8, the one that was edited twice) you’re using real boundary_cond = delta > 0 ? 0 : 1;
. In the final version you posted (#11) you don’t use this anymore.
Is it indeed the case that one shouldn’t use the first wiener_rng
version, but rather use the second?
Many thanks,
Roey
If irc it’s this one: Alternative to wiener_rng in Stan - #11 by spinkney
Thanks, @spinkney!
Sorry for reviving this topic… but it would be very cool if this function could be integrated into cmdstan :D
I believe @Franzi is going to implement the rng function for the generalized Wiener distribution that has 5 or 7 parameters, which includes the 4 parameter case here.
If this particular implementation is much faster for the 4 parameter case we may implement it too. Just let us know. It’s fairly easy to add rng functions because we don’t need to differentiate them.