v0.12.1

NIMBLE is a system for building and sharing analysis methods for statistical models, especially for hierarchical models and computationally-intensive methods (such as MCMC and SMC).

Version 0.12.1 follows shortly after 0.12.0 and fixes a bug introduced in conjugacy processing in version 0.11.0 that causes incorrect MCMC sampling only in specific cases. The impacted cases have terms of the form "a[i] + x[i] * beta" (or more simply "x[i] * beta" or "a[i] + beta"), with beta subject to conjugate sampling and either (i) 'x' provided via NIMBLE's constants argument and x[1] == 1 or (ii) 'a' provided via NIMBLE's constants argument and a[1] == 0.

v0.12.0

NIMBLE is a system for building and sharing analysis methods for statistical models, especially for hierarchical models and computationally-intensive methods (such as MCMC and SMC).

Version 0.12.0 provides a variety of new functionality, bug fixes, and improved error trapping, including:

• completely revamping WAIC such that (1) by default WAIC is calculated in an online fashion without the need for any particular monitors, (2) either conditional or marginal (integrating over latent variables) WAIC can be calculated and data nodes can be grouped into joint likelihood terms, and (3) there is a new calculateWAIC() function that can compute (conditional) WAIC on a user-provided samples either in an MCMC object or a matrix;
• adding the LKJ distribution, useful for prior distributions for correlation matrices, with default Metropolis-Hastings samplers executing on an unconstrained trasnformed parameter space;
• fixing a bug in MCMC sampling of the dcar_proper distribution that results in incorrect MCMC results when the mean of the dcar_proper distribution is not the same for all elements of the node assigned the distribution;
• fixing the isData() function to return TRUE whenever any elements of a multivariate data node are flagged as data;
• correctly error trapping cases where a Bayesian nonparametric model has a differing number of dependent stochastic nodes (e.g., observations) or dependent deterministic nodes per group of elements clustered jointly, thereby preventing incorrect MCMC sampling in such cases, which were not previously detected; and
• improving the formatting of standard logging messages produced by nimbleModel() and compileNimble().

v0.11.1

NIMBLE is a system for building and sharing analysis methods for statistical models, especially for hierarchical models and computationally-intensive methods (such as MCMC and SMC).

Version 0.11.1 is primarily a bug fix release that fixes a bug that was introduced in Version 0.11.0 (which was released on April 17, 2021) that affected MCMC sampling in MCMCs using the “posterior_predictive_branch” sampler introduced in version 0.11.0. This sampler would be listed by name when the MCMC configuration object is created and would be assigned to any set of multiple nodes that (as a group of nodes) have no data dependencies and are therefore sampled as a group from their predictive distributions.

v0.11.0

NIMBLE is a system for building and sharing analysis methods for statistical models, especially for hierarchical models and computationally-intensive methods (such as MCMC and SMC).

Version 0.11.0 provides a variety of new functionality (posterior_predictive_branch MCMC sampler, getParents() method, improved conjugate sampling efficiency), improved error trapping, and bug fixes, in particular fixing a bug giving incorrect node names and potentially affecting algorithm behavior for models with more than 100,000 elements in a vector node or any dimension of a multi-dimensional node.

v0.10.1

NIMBLE is a system for building and sharing analysis methods for statistical models, especially for hierarchical models and computationally-intensive methods (such as MCMC and SMC).

Version 0.10.1 is primarily a bug fix release. It fixes a bug in retrieving parameter values from distributions that was introduced in version 0.10.0. The bug can cause incorrect behavior of conjugate MCMC samplers under certain model structures (such as particular state-space models), so we strongly encourage users to upgrade to 0.10.1. In addition, version 0.10.1 restricts use of WAIC to the conditional version of WAIC (conditioning on all parameters directly involved in the likelihood). Previous versions of nimble gave incorrect results when not conditioning on all parameters directly involved in the likelihood (i.e., when not monitoring all such parameters). In a future version of nimble we plan to make a number of improvements to WAIC, including allowing use of marginal versions of WAIC, where the WAIC calculation integrates over random effects.

v0.10.0

NIMBLE is a system for building and sharing analysis methods for statistical models, especially for hierarchical models and computationally-intensive methods (such as MCMC and SMC).

Version 0.10.0 greatly extends NIMBLE's CRP-based BNP functionality by allowing multiple observations to be grouped together, improves the efficiency of various model and algorithm building steps, moves the sequential Monte Carlo (SMC; aka particle filtering) algorithms to the new nimbleSMC package, and fixes a bug that produced incorrect results from runCrossValidate. In addition there are a variety of other improvements and bug fixes. Please see the inst/NEWS file.

v0.9.1

NIMBLE is a system for building and sharing analysis methods for statistical models, especially for hierarchical models and computationally-intensive methods (such as MCMC and SMC).

Version 0.9.1 is primarily a bug fix release that fixes use of NIMBLE in R 4.0 on Windows but also provides some minor improvements in functionality.

v0.9.0

NIMBLE is an R package for hierarchical statistical modeling (aka
graphical modeling). It enables writing general models along with
methods such as Markov chain Monte Carlo (MCMC), particle filtering
(aka sequential Monte Carlo), and other general methods.

This release adds a new SMC algorithm, fixes some bugs in existing
SMC algorithms, improves the speed of MCMC configuration, and fixes
a variety of bugs.

v0.7.1

NIMBLE is an R package for hierarchical statistical modeling (aka
graphical modeling). It enables writing general models along with
methods such as Markov chain Monte Carlo (MCMC), particle filtering
(aka sequential Monte Carlo), and other general methods.

This release is primarily a maintenance release focused on bug fixes, including
fixing an error in MCMC sampling for the dCRP Bayesian
nonparametric distribution and fixing a stack overflow issue.

v0.6.13

NIMBLE is an R package for hierarchical statistical modeling (aka
graphical modeling). It enables writing general models along with
methods such as Markov chain Monte Carlo (MCMC), particle filtering
(aka sequential Monte Carlo), and other general methods.

This release includes additional efficiency and functionality for
Bayesian nonparametric mixture modeling
using Dirichlet process mixtures, additional MCMC samplers
and functionality, and several bug fixes.