Jacob B. Socolar

@syclik Thanks for taking a look! > Just curious... how'd you come to this conclusion? And which gradient did you think was "infinite"? (The gradient has two elements.) It'd be...

@andrjohns see also https://discourse.mc-stan.org/t/rstan-segfaults-on-macos-ventura-13-4/31547

This refers to a situation where we care about the posterior probability that a particular observation arises from the latent zero-inflation state. This probability of being in the zero-inflation state...

You're absolutely correct that we're conditioning on the observed data (not necessarily new data; the commoner use case will be the data used in model fitting). I think what you're...

Just a note on naming: I guess `posterior_zi_prob()` is confusing and not well differentiated from `posterior_linpred(..., dpar = "zi", transform = T)` and `posterior_epred(..., dpar = "zi")`. Maybe something like...

Sounds good. Note that in hurdle models the mixture weights are always either 0 or 1 and depend only on data, with no dependency on parameters. Do you think it's...

Edge case: what is the desired return of `pp_mixture` if a zero-inflated or hurdle (or zero-one-inflated) model is included as a mixture component in an explicit mixture model, e.g. ```...

Doesn't `mo(I(x+y))` do this? If not, you could always just define a new column in your input data and populate it with the values `x + y`.

silly me, sorry for the noise. Note that `I` doesn't mean "force equality of coefficients" but rather "isolate the expression and evaluate using standard R syntax and not formula-specific syntax".

Just a note, there's some kind of numerical instability when calculating the zero inflation mixture components, but it's only cropped up for numerically extreme inputs. For example: ```r remotes::install_github("jsocolar/brms", ref...