Merge pull request #27 from osorensen/sorts-issue-26

Topological sorts in memory rather than in files

Author: osorensen

R/RcppExports.R

@@ -11,13 +11,11 @@
 #'   must have two columns, the first of which represents the preferred item
 #'   and the second of which represents the disfavored item.
 #' @param n_items An integer specifying the number of items to sort.
-#' @param output_directory A string specifying the directory where the output files will be saved.
 #' @param save_frac Number between 0 and 1 specifying which fraction of sorts to save.
 #'
 #' @details
 #' The function generates all possible topological sorts for the provided preference matrix
-#' and saves approximately `save_frac` of the sorts as binary file in the specified output directory.
-#' The output files are named sequentially as `sort0.bin`, `sort1.bin`, and so on.
+#' and saves approximately `save_frac` of the sorts in a matrix which is returned.
 #'
 #' @return This function returns the number of topological sorts.
 #'
@@ -27,19 +25,33 @@
 #' prefs <- pairwise_preferences[
 #'  pairwise_preferences$user == 1, c("top_item", "bottom_item"), drop = FALSE]
 #'
-#' # Count the number of sorts without saving them.
-#' precompute_topological_sorts(
+#' # Generate all topological sorts, but don't save them:
+#' sorts <- precompute_topological_sorts(
 #'   prefs = as.matrix(prefs),
 #'   n_items = 5,
-#'   output_directory = tempdir(),
 #'   save_frac = 0
 #' )
+#' # Number of sorts
+#' sorts$sort_count
+#' # Empty matrix
+#' sorts$sort_matrix
 #'
-precompute_topological_sorts <- function(prefs, n_items, output_directory, save_frac) {
-    .Call(`_BayesMallowsSMC2_precompute_topological_sorts`, prefs, n_items, output_directory, save_frac)
+#' # Generate all topological sorts and save them:
+#' sorts <- precompute_topological_sorts(
+#'   prefs = as.matrix(prefs),
+#'   n_items = 5,
+#'   save_frac = 1
+#' )
+#' # Number of sorts
+#' sorts$sort_count
+#' # Matrix with all of them
+#' sorts$sort_matrix
+#'
+precompute_topological_sorts <- function(prefs, n_items, save_frac) {
+    .Call(`_BayesMallowsSMC2_precompute_topological_sorts`, prefs, n_items, save_frac)
 }
 
-run_smc <- function(input_timeseries, input_prior, input_options) {
-    .Call(`_BayesMallowsSMC2_run_smc`, input_timeseries, input_prior, input_options)
+run_smc <- function(input_timeseries, input_prior, input_options, input_sort_matrices, input_sort_counts) {
+    .Call(`_BayesMallowsSMC2_run_smc`, input_timeseries, input_prior, input_options, input_sort_matrices, input_sort_counts)
 }
 

R/compute_sequentially.R

@@ -25,16 +25,9 @@
 #'
 #' @param hyperparameters A list returned from [set_hyperparameters()].
 #' @param smc_options A list returned from [set_smc_options()]
-#' @param topological_sorts_directory Path to a directory where precomputed
-#'   topological sorts can be found. Must contain subdirectories with names
-#'   `user1`, `user2`, etc., exactly matching the `user` column of `data`. See
-#'   [precompute_topological_sorts()]. Defaults to `NULL`, but has to be
-#'   provided when data contain pairwise preferences.
-#'
-#' @param num_topological_sorts Integer vector containing the number of
-#'   topological sorts for each user.
-#' @param file_count Integer vector containing the number of files with
-#'   topological sorts for each user.
+#' @param topological_sorts A list returned from
+#'   [precompute_topological_sorts()]. Only used with preference data, and
+#'   defaults to `NULL`.
 #'
 #' @return An object
 #' @export
@@ -43,9 +36,7 @@ compute_sequentially <- function(
     data,
     hyperparameters = set_hyperparameters(),
     smc_options = set_smc_options(),
-    topological_sorts_directory = NULL,
-    num_topological_sorts = NULL,
-    file_count = NULL
+    topological_sorts = NULL
     ){
   rank_columns <- grepl("item[0-9]+", colnames(data))
   preference_columns <- grepl("top\\_item|bottom\\_item", colnames(data))
@@ -60,34 +51,33 @@ compute_sequentially <- function(
     } else {
       attr(input_timeseries, "type") <- "complete rankings"
     }
+    sort_matrices <- sort_counts <- list()
   } else if(sum(preference_columns) == 2) {
-    if(is.null(topological_sorts_directory)) {
-      stop("topological_sorts_directory must be provided with preference data.")
-    }
-    if(is.null(num_topological_sorts)) {
-      stop("num_topological_sorts must be provided with preference data.")
-    }
-    if(is.null(file_count)) {
-      stop("file_count must be provided with preference data.")
+    if(is.null(topological_sorts)) {
+      stop("topological_sorts must be provided with preference data.")
     }
     input_timeseries <- split(data, f = ~ timepoint) |>
       lapply(split, f = ~ user) |>
       lapply(function(x) lapply(x, function(y) as.matrix(y[preference_columns])))
     attr(input_timeseries, "type") <- "pairwise preferences"
-    attr(input_timeseries, "topological_sorts_directory") <- topological_sorts_directory
-    names(num_topological_sorts) <- names(file_count) <- as.character(lapply(input_timeseries, names))
-    attr(input_timeseries, "num_topological_sorts") <- num_topological_sorts
-    attr(input_timeseries, "file_count") <- file_count
+
+    sort_matrices <- lapply(topological_sorts, function(x) {
+      lapply(x, function(y) y$sort_matrix)
+    })
+
+    sort_counts <- lapply(topological_sorts, function(x) {
+      lapply(x, function(y) y$sort_count)
+    })
   } else {
     stop("Something wrong with data")
   }
 
-  attr(input_timeseries, "updated_users") <- FALSE
   if(max(table(data$user)) > 1 &&
      attr(input_timeseries, "type") != "pairwise preferences") {
     stop("Updated users not supported.")
   }
 
-  ret <- run_smc(input_timeseries, hyperparameters, smc_options)
+  ret <- run_smc(input_timeseries, hyperparameters, smc_options,
+                 sort_matrices, sort_counts)
 }
 

man/compute_sequentially.Rd

@@ -12,36 +12,37 @@ Rcpp::Rostream<false>& Rcpp::Rcerr = Rcpp::Rcpp_cerr_get();
 #endif
 
 // precompute_topological_sorts
-long long int precompute_topological_sorts(arma::umat prefs, int n_items, std::string output_directory, double save_frac);
-RcppExport SEXP _BayesMallowsSMC2_precompute_topological_sorts(SEXP prefsSEXP, SEXP n_itemsSEXP, SEXP output_directorySEXP, SEXP save_fracSEXP) {
+Rcpp::List precompute_topological_sorts(arma::umat prefs, int n_items, double save_frac);
+RcppExport SEXP _BayesMallowsSMC2_precompute_topological_sorts(SEXP prefsSEXP, SEXP n_itemsSEXP, SEXP save_fracSEXP) {
 BEGIN_RCPP
     Rcpp::RObject rcpp_result_gen;
     Rcpp::RNGScope rcpp_rngScope_gen;
     Rcpp::traits::input_parameter< arma::umat >::type prefs(prefsSEXP);
     Rcpp::traits::input_parameter< int >::type n_items(n_itemsSEXP);
-    Rcpp::traits::input_parameter< std::string >::type output_directory(output_directorySEXP);
     Rcpp::traits::input_parameter< double >::type save_frac(save_fracSEXP);
-    rcpp_result_gen = Rcpp::wrap(precompute_topological_sorts(prefs, n_items, output_directory, save_frac));
+    rcpp_result_gen = Rcpp::wrap(precompute_topological_sorts(prefs, n_items, save_frac));
     return rcpp_result_gen;
 END_RCPP
 }
 // run_smc
-Rcpp::List run_smc(Rcpp::List input_timeseries, Rcpp::List input_prior, Rcpp::List input_options);
-RcppExport SEXP _BayesMallowsSMC2_run_smc(SEXP input_timeseriesSEXP, SEXP input_priorSEXP, SEXP input_optionsSEXP) {
+Rcpp::List run_smc(Rcpp::List input_timeseries, Rcpp::List input_prior, Rcpp::List input_options, Rcpp::List input_sort_matrices, Rcpp::List input_sort_counts);
+RcppExport SEXP _BayesMallowsSMC2_run_smc(SEXP input_timeseriesSEXP, SEXP input_priorSEXP, SEXP input_optionsSEXP, SEXP input_sort_matricesSEXP, SEXP input_sort_countsSEXP) {
 BEGIN_RCPP
     Rcpp::RObject rcpp_result_gen;
     Rcpp::RNGScope rcpp_rngScope_gen;
     Rcpp::traits::input_parameter< Rcpp::List >::type input_timeseries(input_timeseriesSEXP);
     Rcpp::traits::input_parameter< Rcpp::List >::type input_prior(input_priorSEXP);
     Rcpp::traits::input_parameter< Rcpp::List >::type input_options(input_optionsSEXP);
-    rcpp_result_gen = Rcpp::wrap(run_smc(input_timeseries, input_prior, input_options));
+    Rcpp::traits::input_parameter< Rcpp::List >::type input_sort_matrices(input_sort_matricesSEXP);
+    Rcpp::traits::input_parameter< Rcpp::List >::type input_sort_counts(input_sort_countsSEXP);
+    rcpp_result_gen = Rcpp::wrap(run_smc(input_timeseries, input_prior, input_options, input_sort_matrices, input_sort_counts));
     return rcpp_result_gen;
 END_RCPP
 }
 
 static const R_CallMethodDef CallEntries[] = {
-    {"_BayesMallowsSMC2_precompute_topological_sorts", (DL_FUNC) &_BayesMallowsSMC2_precompute_topological_sorts, 4},
-    {"_BayesMallowsSMC2_run_smc", (DL_FUNC) &_BayesMallowsSMC2_run_smc, 3},
+    {"_BayesMallowsSMC2_precompute_topological_sorts", (DL_FUNC) &_BayesMallowsSMC2_precompute_topological_sorts, 3},
+    {"_BayesMallowsSMC2_run_smc", (DL_FUNC) &_BayesMallowsSMC2_run_smc, 5},
     {NULL, NULL, 0}
 };