nimbleNoBounds package

[David Pleydell]

Dear nimble users.

I have put together a small nimble package called nimbleNoBounds. 

The package provides a collection of common continuous univariate probability distributions transformed to the real line so that bounds on the parameter space are removed. Using these transformed distributions can result in more efficient MCMC, particularly when adaptive Metropolis-Hastings is used to sample in the vicinity of bounds on the parameter-space. 

See vignette("nimbleNoBounds") for several examples.

The package can be installed from github via 

remotes::install_git(url="https://github.com/DRJP/nimbleNoBounds.git", subdir="nimbleNoBounds", build_vignettes = TRUE)

Best regards

David

[David Pleydell]

Dear all

The nimbleNoBounds package is now on CRAN.  Here's a link to the package vignette, which demonstrates how the package can help increase the efficiency of adaptive Metropolis Hastings when sampling near the bounds of a target ditribution. 

Cheers

David