Restarting distributed tree.

2010-11-19 02:24:57 +00:00
parent 8449fe5766
commit d9c2dd657f
3 changed files with 26 additions and 124 deletions
@@ -213,8 +213,33 @@ class DistributedTable: public boost::noncopyable {

    void IndexData(
      const core::metric_kernels::AbstractMetric & metric_in,
+      boost::mpi::communicator &table_outbox_group_comm,
      double sample_probability_in) {

+      // 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;
+      for(int i = 0; i < owned_table_->n_entries(); i++) {
+        if(core::math::Random() <= sample_probability_in) {
+          sampled_indices.push_back(i);
+        }
+      }
+      // 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.
+      int total_num_samples = 0;
+      if(table_outbox_group_comm.rank() == 0) {
+        boost::mpi::reduce(
+          table_outbox_group_comm, sampled_indices.size(),
+          boost::mpi::sum<int>(), 0);
+      }
+      else {
+        boost::mpi::reduce(
+          table_outbox_group_comm, sampled_indices(),
+          boost::mpi::sum<int>(), 0);
+      }
+

    }

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 2.8)

-# the test for distributed_kde.test.cc
+# the test for distributed_kde
 add_executable(distributed_kde
               EXCLUDE_FROM_ALL
               distributed_kde.cc
@@ -38,16 +38,3 @@ target_link_libraries(prefix_scan
                      ${TRILINOS_LIBS}
                      ${Boost_LIBRARIES}
                     )
-
-# the main driver for parallel_kde
-add_executable(parallel_kde
-               EXCLUDE_FROM_ALL
-               parallel_kde.cc
-              )
-
-# link dependencies of parallel_kde
-target_link_libraries(parallel_kde
-                      core
-                      ${TRILINOS_LIBS}
-                      ${Boost_LIBRARIES}
-                     )
@@ -1,110 +0,0 @@
-#include <iostream>
-#include <string>
-#include <armadillo>
-#include "boost/mpi.hpp"
-#include "boost/serialization/string.hpp"
-#include "core/math/math_lib.h"
-#include <vector>
-#include <cstdlib>
-
-class Particle {
-  private:
-    std::vector<int> position_;
-
-    friend class boost::serialization::access;
-
-  public:
-
-    template<class Archive>
-    void serialize(Archive &ar, const unsigned int version) {
-      ar & position_;
-    }
-
-    const std::vector<int> &position() const {
-      return position_;
-    }
-
-    Particle() {
-      position_.resize(3);
-      for(int i = 0; i < position_.size(); i++) {
-        position_[i] = 0;
-      }
-    }
-
-    void AddParticle(const Particle &particle_in) {
-      for(unsigned int i = 0; i < 3; i++) {
-        position_[i] += particle_in.position()[i];
-      }
-    }
-
-    void Randomize() {
-      for(unsigned int i = 0; i < 3; i++) {
-        position_[i] = core::math::RandInt(0, 3);
-      }
-    }
-
-    void Print() const {
-      std::cout << position_[0] << " " << position_[1] <<
-                " " << position_[2] << "\n";
-    }
-};
-
-int main(int argc, char *argv[]) {
-  boost::mpi::environment env(argc, argv);
-  boost::mpi::communicator world;
-
-  if(world.size() <= 1) {
-    std::cout << "Please specify a process number greater than 1.\n";
-    exit(0);
-  }
-
-  Particle generated_particle;
-
-  // Broadcast the skeleton from the master to the slaves.
-  broadcast(world, boost::mpi::skeleton(generated_particle), 0);
-
-  boost::mpi::content c = boost::mpi::get_content(generated_particle);
-
-  // The master process.
-  if(world.rank() == 0) {
-    std::vector<int> num_messages(world.size(), 0);
-    for(int i = 1; i < world.size(); i++) {
-      num_messages[i] = core::math::RandInt(3, 8);
-
-      // Send the number of messages to receive per each processor.
-      world.isend(i, 0, num_messages[i]);
-      printf("Processor 0 is sending processor %d %d message.\n", i,
-             num_messages[i]);
-    }
-    for(int i = 1; i < world.size(); i++) {
-      for(int j = 0; j < num_messages[i]; j++) {
-        generated_particle.Randomize();
-        std::cout << "Master generated and sending to " << i << ": ";
-        generated_particle.Print();
-        world.isend(i, 1, c);
-      }
-    }
-  }
-
-  // The slave process.
-  else {
-    int num_messages_to_receive;
-    boost::mpi::request main_request =
-      world.irecv(0, 0, num_messages_to_receive);
-    main_request.wait();
-
-    Particle received_particle;
-    for(int j = 0; j < num_messages_to_receive; j++) {
-      // Receive from the master.
-      boost::mpi::request receive_request = world.irecv(0, 1, c);
-      receive_request.wait();
-      std::cout << "Process " << world.rank() << " received: ";
-      generated_particle.Print();
-      received_particle.AddParticle(generated_particle);
-    }
-    std::cout << "Process " << world.rank() << " got: ";
-    received_particle.Print();
-  }
-
-  return 0;
-}