fix merge conflict in .travis.yml

.travis.yml

@@ -2,7 +2,7 @@ language: r
 r:
     - 4.0
 
-# dist: bionic  ## experimented with removing in Aug 2020 due to travis errors, see: https://travis-ci.community/t/r-install-broken-travis-ci-rstudio-org-returns-403/9640; seems fine as of Sep 2020
+dist: bionic  ## experimented with removing in Aug 2020 due to travis errors, see: https://travis-ci.community/t/r-install-broken-travis-ci-rstudio-org-returns-403/9640; seems fine as of Sep 2020
 sudo: false
 
 cache:

UserManual/src/chapter_BNP.Rmd

@@ -111,13 +111,11 @@ One of the advantages of BNP mixture models is that the number of clusters is tr
 
 ### Extensions {#sec:extensionscrp}
 
-The BNP functionality has been recently extended to more general models. These extensions enable users, for instance, to use a DP or DPM prior for the distribution of the random effects in a generalized linear mixed effects model with mutiple measurements over time or multiple trials per participant. 
-
-#### Example {#sec:exextensionscrp}
+The BNP functionality in NIMBLE was extended in version 0.10.0 to more general models. These extensions enable users, for instance, to use a DP or DPM prior for the distribution of the random effects in a generalized linear mixed effects model with mutiple measurements over time or multiple trials per participant. 
 
 The following example illustrates how to use NIMBLE in a random effects model with repeated measurements per subject using a DP prior for the distribution of the subject's random effects. The model is given by
 
-$$ y_{i,j} \mid \tilde{\boldsymbol{\theta}}, z_i, \sigma^2 \ {\sim}\ N(\tilde{\theta}_{z_i}, {\sigma}^2,)\hspace{0.5cm} i = 1, \ldots, N,\hspace{0.5cm} j = 1, \ldots, J, $$
+$$ y_{i,j} \mid \tilde{\boldsymbol{\theta}}, z_i, \sigma^2 \ {\sim}\ N(\tilde{\boldsymbol{\theta}}_{z_i}, {\sigma}^2,)\hspace{0.5cm} i = 1, \ldots, N,\hspace{0.5cm} j = 1, \ldots, J, $$
 
 $$ \boldsymbol{z} \sim \mbox{CRP}(\alpha), \hspace{0.5cm} \alpha \sim \mbox{Gamma}(1, 1), $$
 
@@ -150,6 +148,8 @@ inits <- list(thetatilde = rnorm(constants$M, 0, 10),
 modelRandEff <- nimbleModel(code, constants, data, inits)
 ```
 
+Alternatively, each subject could have a vector of parameters being clustered. For example in the model above one could instead specify a vector of means, such as `thetaTilde[z[i], j]`, instead of a single mean. 
+
 As before, the model can be fitted through  MCMC sampling. NIMBLE will assign a specialized sampler to update `z` and `alpha`. See Chapter  \@ref(cha:mcmc) for information about NIMBLE's MCMC engine, and Section \@ref(sec:mcmcdcrp) for details on  MCMC sampling of the CRP.
 
 ## Stick-breaking model {#sec:sb}
@@ -238,7 +238,7 @@ NIMBLE's MCMC engine provides specialized samplers for the `dCRP` distribution,
   1. A non-conjugate sampler in the case of the baseline distribution  not being conjugate for the mixture kernel.
 
 
-Note that both samplers are specialized versions that operate on a vector having a CRP distribution. Details of these assignments are strictly internal to the CRP samplers. The current release of NIMBLE supports conjugate sampling for the `dCRP` distribution for the relationships listed in Table \@ref(tab:BNPconjugacy). Additional relationships are provided in Table \@ref(tab:BNPconjugacy2) for normal mixture kernels when both,  mean and  variance, are unknown.
+Note that both samplers are specialized versions that operate on a vector having a CRP distribution. Details of these assignments are strictly internal to the CRP samplers. The current release of NIMBLE supports conjugate sampling for the `dCRP` distribution for the relationships listed in Table \@ref(tab:BNPconjugacy). Additional relationships are provided in Table \@ref(tab:BNPconjugacy2) for normal mixture kernels when both mean and  variance are unknown.
 
 
 ```{r, child = 'tables/BNPconjugacyTable.md'}

packages/nimble/inst/tests/test_utils.R

@@ -208,8 +208,12 @@ test_coreRfeature_batch_internal <- function(input_batch, verbose = nimbleOption
             if(is.null(checkEqual)) checkEqual <- FALSE
             if(is.null(input[['return']])) { ## use default 'out' object
                 if(!checkEqual) {
+                    expect_identical(class(out), class(out_nfC), info = paste('iden tmp test of class', class(out), class(out_nfC)))
+                    expect_identical(dim(out), dim(out_nfC), info = 'iden test of dim')
+                    expect_identical(round(out, 10), round(out_nfC, 10), info='iden test of round')
                     expect_identical(out, out_nfR, info = "Identical test of coreRfeature (direct R vs. R nimbleFunction)")
-                    expect_identical(out, out_nfC, info = "Identical test of coreRfeature (direct R vs. C++ nimbleFunction)")
+                    wh <- which.max(abs(out - out_nfC))
+                    expect_identical(out, out_nfC, info = paste("Identical test of coreRfeature (direct R vs. C++ nimbleFunction)", system('gcc --version', intern = T)[1], ' ', sprintf("%0.20f", out[wh]), " ", sprintf("%0.20f", out_nfC[wh])))
                 } else {
                     expect_equal(out, out_nfR, info = "Equal test of coreRfeature (direct R vs. R nimbleFunction)")
                     expect_equal(out, out_nfC, info = "Equal test of coreRfeature (direct R vs. C++ nimbleFunction)")
@@ -316,8 +320,8 @@ test_coreRfeature_internal <- function(input, verbose = nimbleOptions('verbose')
   if(is.null(checkEqual)) checkEqual <- FALSE
   if(is.null(input[['return']])) { ## use default 'out' object
       if(!checkEqual) {
-          expect_identical(out, out_nfR, info = paste0("Identical test of coreRfeature (direct R vs. R nimbleFunction): ", input$name))
-          expect_identical(out, out_nfC, info = paste0("Identical test of coreRfeature (direct R vs. C++ nimbleFunction): ", input$name))
+          expect_identical(out, out_nfR, info = paste0("FOO Identical test of coreRfeature (direct R vs. R nimbleFunction): ", input$name))
+          expect_identical(out, out_nfC, info = paste0("FOO Identical test of coreRfeature (direct R vs. C++ nimbleFunction): ", input$name))
       } else {
           expect_equal(out, out_nfR, info = paste0("Equal test of coreRfeature (direct R vs. R nimbleFunction): ", input$name) )
           expect_equal(out, out_nfC, info = paste0("Equal test of coreRfeature (direct R vs. C++ nimbleFunction): ", input$name))