Update NEWS for 1.4.1.

Author: paciorek

UserManual/NimbleUserManual.pdf

@@ -4018,7 +4018,7 @@ sampler_partial_mvn_pp <- nimbleFunction(
 #' 
 #' @section partial_mvn sampler:
 #'
-#' The partial_mvn sampler is designed to sample multivariate normal distributions that are partially observed.  That is, some dimensions of the target node are observed data values, some dimensions are not data. For dimensions that have no data nodes dependent on them, sampling is done from the conditional multivariate normal predictive distribution. Otherwise, sampling is accomplished using either univariate or multivariate random walk Metropolis Hastings of the unobserved dimensions, as determined by the \code{multivariateNodesAsScalars} argument (or using the Barker sampler based on the `code{MCMCuseBarkerAsDefaultMV} nimble option).
+#' The partial_mvn sampler is designed to sample multivariate normal distributions that are partially observed.  That is, some dimensions of the target node are observed data values, some dimensions are not data. For dimensions that have no data nodes dependent on them, sampling is done from the conditional multivariate normal predictive distribution. Otherwise, sampling is accomplished using either univariate or multivariate random walk Metropolis Hastings of the unobserved dimensions, as determined by the \code{multivariateNodesAsScalars} argument (or using the Barker sampler based on the \code{MCMCuseBarkerAsDefaultMV} nimble option).
 #'
 #' The \code{partial_mvn} sampler accepts the following control list elements:
 #' \itemize{

packages/nimble/R/MCMC_samplers.R

@@ -1,3 +1,28 @@
+#              CHANGES IN VERSION 1.4.1 (February 2026)
+
+## USER LEVEL CHANGES
+
+- Add a new sampler to sample portions of a multivariate normal node when some 
+  elements are observed (i.e., data) and some elements are unknown. This 
+  sampler uses the exact conditional normal to sample from the posterior 
+  predictive distribution when the unknown elements have no dependent data nodes
+  (PR #1612).
+  
+- Add ability to change the MCMC thinning interval when continuing an MCMC
+  (PR #1609).
+  
+## BUG FIXES
+
+- Fix a bug in the conjugacy size processing system when dependent nodes 
+  (of the target conjugate node) themselves have multivariate parameters 
+  with a different size from the dependent node itself (PR #1605).
+  
+## DEVELOPER LEVEL CHANGES
+
+- Turn off an old work-around that prevented RStudio from hanging but 
+  which modified the global environment and no longer appears to be 
+  necessary (PR #1606).
+
 #              CHANGES IN VERSION 1.4.0 (December 2025)
 
 ## USER LEVEL CHANGES