Refactoring the distributed tree builder.

2011-01-05 17:01:06 +00:00
parent f24c11e09b
commit 8f4ae972c2
2 changed files with 105 additions and 0 deletions
@@ -2,6 +2,7 @@ cmake_minimum_required(VERSION 2.8)

 set(SOURCES
  ball_bound.h
+  distributed_tree_builder.h
  general_spacetree.h
  gen_kdtree.h
  gen_metric_tree.h
@@ -0,0 +1,104 @@
+/** @file distributed_tree_builder.h
+ *
+ *  The generic template for building a distributed tree.
+ *
+ *  @author Dongryeol Lee (dongryel@cc.gatech.edu)
+ */
+
+#ifndef CORE_TREE_DISTRIBUTED_TREE_BUILDER_H
+#define CORE_TREE_DISTRIBUTED_TREE_BUILDER_H
+
+#include <algorithm>
+#include <boost/mpi.hpp>
+#include "core/parallel/parallel_sample_sort.h"
+
+namespace core {
+namespace tree {
+template<typename DistributedTableType>
+class DistributedTreeBuilder {
+  public:
+    typedef typename DistributedTableType::TableType TableType;
+
+  private:
+    DistributedTableType *distributed_table_;
+
+    double sampling_rate_;
+
+  private:
+
+    void SetupGatherPointers_(
+      TableType &sampled_table, const std::vector<int> &counts,
+      std::vector<double *> *gather_pointers_out) {
+
+      // Have the pointers point to the right position based on the
+      // prefix sum position.
+      gather_pointers_out->resize(counts.size());
+      double *pointer = sampled_table.data().ptr();
+      for(unsigned int i = 0; i < counts.size(); i++) {
+        (*gather_pointers_out)[i] = pointer;
+        pointer += sampled_table.n_attributes() * counts[i];
+      }
+    }
+
+    void BuildSampleTree_(boost::mpi::communicator &world) {
+
+      // Each process generates a random subset of the data points to
+      // send to the master. This is a MPI gather operation.
+      TableType sampled_table;
+      std::vector<int> sampled_indices;
+      SelectSubset_(sampling_rate_, &sampled_indices);
+
+      // Send the number of points chosen in this process to the
+      // master so that the master can allocate the appropriate amount
+      // of space to receive all the points.
+      std::vector<int> counts;
+      int local_sampled_indices_size = static_cast<int>(
+                                         sampled_indices.size());
+      boost::mpi::gather(
+        table_outbox_group_comm, local_sampled_indices_size, counts, 0);
+
+      // The master process allocates the sample table and gathers the
+      // chosen samples from each process.
+      if(world.rank() == 0) {
+        int total_num_samples = std::accumulate(
+                                  counts.begin(), counts.end(), 0);
+        sampled_table.Init(
+          distributed_table_->n_attributes(), total_num_samples);
+        std::vector<double *> gather_pointers;
+        SetupGatherPointers_(sampled_table, counts, gather_pointers);
+
+
+      }
+    }
+
+    void SelectSubset_(
+      std::vector<int> *sampled_indices_out) {
+
+      std::vector<int> indices(owned_table_->n_entries(), 0);
+      for(unsigned int i = 0; i < indices.size(); i++) {
+        indices[i] = i;
+      }
+      int num_elements =
+        std::max(
+          (int) floor(sampling_rate_ * owned_table_->n_entries()), 1);
+      std::random_shuffle(indices.begin(), indices.end());
+
+      for(int i = 0; i < num_elements; i++) {
+        sampled_indices_out->push_back(indices[i]);
+      }
+    }
+
+  public:
+    void Init(
+      DistributedTableType &distributed_table_in, double sampling_rate_in) {
+      distributed_table_ = &distributed_table_in;
+      sampling_rate_ = sampling_rate_in;
+    }
+
+    void BuildTree(boost::mpi::communicator &world) {
+    }
+};
+};
+};
+
+#endif