fixing #421

ocbe-uio · Sep 5, 2024 · 770033c · 770033c
1 parent 771fa30
commit 770033c
Show file tree

Hide file tree

Showing 4 changed files with 12 additions and 5 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: BayesMallows
 Type: Package
 Title: Bayesian Preference Learning with the Mallows Rank Model
-Version: 2.2.2
+Version: 2.2.2.9000
 Authors@R: c(person("Oystein", "Sorensen",
                     email = "oystein.sorensen.1985@gmail.com",
                     role = c("aut", "cre"),

diff --git a/NEWS.md b/NEWS.md
@@ -1,3 +1,7 @@
+# BayesMallows (development versions)
+
+* A bug in the pseudolikelihood proposal for latent rankings has been corrected.
+
 # BayesMallows 2.2.2
 
 * An error in compute_mallows_loglik when the number of clusters is more than

diff --git a/src/rank_proposal.cpp b/src/rank_proposal.cpp
@@ -224,9 +224,11 @@ std::pair<arma::vec, double> PartialPseudoProposal::propose_pseudo(
     vec sample_probs = normalise(exp(log_numerator), 1);
 
     if(forward) {
-      ivec ans(sample_probs.size());
-      R::rmultinom(1, sample_probs.begin(), sample_probs.size(), ans.begin());
-      proposal(span(item_to_rank)) = available_rankings(find(ans == 1));
+      Rcpp::IntegerVector rank_index = Rcpp::sample(
+        sample_probs.size(), 1, true,
+        Rcpp::as<Rcpp::NumericVector>(Rcpp::wrap(sample_probs)),
+        false);
+      proposal(span(item_to_rank)) = available_rankings(rank_index(0));
     }
 
     int ranking_chosen = as_scalar(find(proposal(item_to_rank) == available_rankings));

diff --git a/tests/testthat/test-smc_update_correctness.R b/tests/testthat/test-smc_update_correctness.R
@@ -137,7 +137,8 @@ test_that("update_mallows is correct for new partial rankings", {
 
     expect_equal(
       mean(mod_smc_next$alpha$value),
-      ifelse(aug == "uniform", 10.9348917143992, 10.9279783083453)
+      ifelse(aug == "uniform", 10.9348917143992, 10.910),
+      tolerance = .01
     )
 
     expect_equal(