Daniel Turek

@ethan-alt Great question, and this is one aspect of NIMBLE that I actually have some knowledge of. As you noted, writing and using a custom-writtten prior distribution for the covariance...

Doesn't seem ideal that NIMBLE will "find" things in the R global environment. Would be better to enforce they're properly passed in via `constants`. If setting the enclosing environment to...

@Schmidtpk Thanks for the suggestion. We will look into this.

@brendanf This is an exciting development. Thanks for filing the issue and contacting us. Would you mind taking just a moment to elaborate further on the benefits and implications of...

@paciorek is nimble's `any_nan` function the one you're meaning to use?

Additional historical perspective: this current system for handling the LHS-transformations dates **way back**, to the "new model def" system, c 2014. While clear and functional, all our other systems for...

Example of name clash: ``` library(nimble) code

Sure, sounds good. A unique identifier postfix "_UID_##" would be additionally safe.

No rush whatsoever. I think it was only that the legacy HMM-MCMC code no longer works, since it uses custom distributions that have 3D+ parameters. Some folks were trying to...

So, just to check the implications here, if there isn't a work-around for this. Is it that we can't have user-defined distributions, with parameters of dimension > 2 ? On...