DistParMetisDistribution.hpp

/*
 * DistParMetisDistribution.hpp
 *
 *  Created on: Nov 19, 2015
 *      Author: Antonio Leo
 */

#ifndef SRC_DECOMPOSITION_DISTPARMETISDISTRIBUTION_HPP_
#define SRC_DECOMPOSITION_DISTPARMETISDISTRIBUTION_HPP_

#include "SubdomainGraphNodes.hpp"
#include "parmetis_dist_util.hpp"
#include "Graph/dist_map_graph.hpp"
#include "Graph/DistGraphFactory.hpp"

template<unsigned int dim, typename T>
class DistParMetisDistribution
{
	//! Vcluster
	Vcluster<> & v_cl;

	//! Structure that store the cartesian grid information
	grid_sm<dim, void> gr;

	//! rectangular domain to decompose
	Box<dim, T> domain;

	//! Processor sub-sub-domain graph
	DistGraph_CSR<nm_v<dim>, nm_e> g;

	//! Convert the graph to parmetis format
	DistParmetis<DistGraph_CSR<nm_v<dim>, nm_e>> parmetis_graph;

	//! Init vtxdist needed for Parmetis
	openfpm::vector<idx_t> vtxdist;

	//! partitions
	openfpm::vector<openfpm::vector<idx_t>> partitions;

	//! Init data structure to keep trace of new vertices distribution in processors (needed to update main graph)
	openfpm::vector<openfpm::vector<size_t>> v_per_proc;

	//! Number of moved vertices in all iterations
	size_t g_moved = 0;

	//! Max number of moved vertices in all iterations
	size_t m_moved = 0;

	//! Flag to check if weights are used on vertices
	bool verticesGotWeights = false;

	/*! \brief Callback of the sendrecv to set the size of the array received
	 *
	 * \param msg_i Index of the message
	 * \param total_msg Total numeber of messages
	 * \param total_p Total number of processors to comunicate with
	 * \param i Processor id
	 * \param ri Request id
	 * \param ptr Void pointer parameter for additional data to pass to the call-back
	 */
	static void * message_receive(size_t msg_i, size_t total_msg, size_t total_p, size_t i, size_t ri, void * ptr)
	{
		openfpm::vector < openfpm::vector < idx_t >> *v = static_cast<openfpm::vector<openfpm::vector<idx_t>> *>(ptr);

		v->get(i).resize(msg_i / sizeof(idx_t));

		return &(v->get(i).get(0));
	}

public:

	/*! Constructor for the ParMetis class
	 *
	 * @param v_cl Vcluster to use as communication object in this class
	 */
	DistParMetisDistribution(Vcluster<> & v_cl) :
			v_cl(v_cl), parmetis_graph(v_cl, v_cl.getProcessingUnits()), vtxdist(v_cl.getProcessingUnits() + 1), partitions(v_cl.getProcessingUnits()), v_per_proc(v_cl.getProcessingUnits())

	{
	}

	/*! \brief Initialize the distribution graph
	 *
	 * /param grid Grid
	 * /param dom Domain
	 */
	void createCartGraph(grid_sm<dim, void> & grid, Box<dim, T> dom)
	{
		//! Set grid and domain
		gr = grid;
		domain = dom;

		//! Create sub graph
		DistGraphFactory<dim, DistGraph_CSR<nm_v<dim>, nm_e>> dist_g_factory;
		g = dist_g_factory.template construct<NO_EDGE, T, dim - 1, 0>(gr.getSize(), domain);
		g.getDecompositionVector(vtxdist);

		if (dim == 2)
			for (size_t i = 0; i < g.getNVertex(); i++)
				g.vertex(i).template get<nm_v_x>()[2] = 0;

	}

	/*! \brief Get the current graph (main)
	 *
	 */
	DistGraph_CSR<nm_v<dim>, nm_e> & getGraph()
	{
		return g;
	}

	/*! \brief Create first decomposition, it divides the graph in slices and give each slice to a processor
	 *
	 */
	void decompose()
	{
		//! Init sub graph in parmetis format
		parmetis_graph.initSubGraph(g);

		//! Decompose
		parmetis_graph.template decompose<nm_v_proc_id>(g);

		//! Get result partition for this processors
		idx_t *partition = parmetis_graph.getPartition();

		for (size_t i = 0, j = g.firstId(); i < g.getNVertex() && j <= g.lastId(); i++, j++)
		{
			if ((size_t)partition[i] != v_cl.getProcessUnitID())
				g.q_move(g.nodeById(j), partition[i]);
		}
		g.redistribute();
	}

	/*! \brief Refine current decomposition
	 *
	 * It makes a refinement of the current decomposition using Parmetis function RefineKWay
	 * After that it also does the remapping of the graph
	 *
	 */
	void refine()
	{
		//! Reset parmetis graph and reconstruct it
		parmetis_graph.reset(g);

		//! Refine
		parmetis_graph.template refine<nm_v_proc_id>(g);

		//! Get result partition for this processors
		idx_t *partition = parmetis_graph.getPartition();

		for (size_t i = 0, j = g.firstId(); i < g.getNVertex() && j <= g.lastId(); i++, j++)
		{
			if ((size_t)partition[i] != v_cl.getProcessUnitID())
				g.q_move(g.nodeById(j), partition[i]);
		}
		g.redistribute();
	}

	/*! \brief Compute the unbalance value
	 *
	 * \return the unbalance value
	 */
	float getUnbalance()
	{
		long t_cost = getProcessorLoad();

		long min, max, sum;
		float unbalance;

		min = t_cost;
		max = t_cost;
		sum = t_cost;

		v_cl.min(min);
		v_cl.max(max);
		v_cl.sum(sum);
		v_cl.execute();

		unbalance = ((float) (max - min)) / (float) (sum / v_cl.getProcessingUnits());

		return unbalance * 100;
	}

	/*! \brief function that return the position of the vertex in the space
	 *
	 * \param id vertex id
	 * \param pos vector that will contain x, y, z
	 *
	 */
	void getSubSubDomainPosition(size_t id, T (&pos)[dim])
	{
#ifdef SE_CLASS1
		if (id >= g.getNVertex())
			std::cerr << __FILE__ << ":" << __LINE__ << " Position - Such vertex doesn't exist (id = " << id << ", " << "total size = " << g.getNVertex() << ")\n";
#endif

		pos[0] = g.vertex(id).template get<nm_v_x>()[0];
		pos[1] = g.vertex(id).template get<nm_v_x>()[1];
		if (dim == 3)
			pos[2] = g.vertex(id).template get<nm_v_x>()[2];
	}

	/*! \brief Function that set the weight of the vertex
	 *
	 * \param id vertex id
	 * \param weight to give to the vertex
	 *
	 */
	inline void setComputationCost(size_t id, size_t weight)
	{
		verticesGotWeights = true;
#ifdef SE_CLASS1
		if (id >= g.getNVertex())
			std::cerr << __FILE__ << ":" << __LINE__ << "Weight - Such vertex doesn't exist (id = " << id << ", " << "total size = " << g.getNVertex() << ")\n";
#endif

		// If the vertex is inside this processor update the value
		g.vertex(id).template get<nm_v_computation>() = weight;

	}

	/*! \brief Checks if weights are used on the vertices
	 *
	 * \return true if weights are used in the decomposition
	 */
	bool weightsAreUsed()
	{
		return verticesGotWeights;
	}

	/*! \brief function that get the weight of the vertex
	 *
	 * \param id vertex id
	 *
	 * \return the weight of the vertex
	 *
	 */
	size_t getVertexWeight(size_t id)
	{
#ifdef SE_CLASS1
		if (id >= g.getNVertex())
			std::cerr << __FILE__ << ":" << __LINE__ << "Such vertex doesn't exist (id = " << id << ", " << "total size = " << g.getTotNVertex() << ")\n";
#endif

		return g.vertex(id).template get<nm_v_computation>();
	}

	/*! \brief Compute the processor load counting the total weights of its vertices
	 *
	 * \return the computational load of the processor graph
	 */
	size_t getProcessorLoad()
	{
		size_t load = 0;

		for (size_t i = 0; i < g.getNVertex(); i++)
		{
			load += g.vertex(i).template get<nm_v_computation>();
		}
		return load;
	}

	/*! \brief return number of moved vertices in all iterations so far
	 *
	 * \return number of moved vertices
	 *
	 */
	size_t getTotalMovedV()
	{
		return g_moved;
	}

	/*! \brief return number of moved vertices in all iterations so far
	 *
	 * \return number of moved vertices
	 *
	 */
	size_t getMaxMovedV()
	{
		return m_moved;
	}

	/*! \brief Set migration cost of the vertex id
	 *
	 * \param id of the vertex to update
	 * \param migration cost of the migration
	 */
	void setMigrationCost(size_t id, size_t migration)
	{
#ifdef SE_CLASS1
		if (id >= g.getNVertex())
			std::cerr << __FILE__ << ":" << __LINE__ << "Migration - Such vertex doesn't exist (id = " << id << ", " << "total size = " << g.getNVertex() << ")\n";
#endif

		g.vertex(id).template get<nm_v_migration>() = migration;
	}

	/*! \brief Set communication cost of the edge id
	 *
	 * \param v_id Id of the source vertex of the edge
	 * \param e i child of the vertex
	 * \param communication Communication value
	 */
	void setCommunicationCost(size_t v_id, size_t e, size_t communication)
	{
		g.getChildEdge(v_id, e).template get<nm_e::communication>() = communication;
	}

	/*! \brief Returns total number of sub-sub-domains in the distribution graph
	 *
	 * \return number od sub-sub-domain
	 *
	 */
	size_t getNSubSubDomains()
	{
		return g.getNVertex();
	}

	/*! \brief Returns total number of neighbors of the sub-sub-domain id
	 *
	 * \param id id of the sub-sub-domain
	 *
	 * \return the number of neighborhood sub-sub-domain
	 *
	 */
	size_t getNSubSubDomainNeighbors(size_t id)
	{
		if (id >= g.getNVertex())
			std::cerr << "Neighbors - Such vertex doesn't exist (id = " << id << ", " << "total size = " << g.getNVertex() << ")\n";

		return g.getNChilds(id);
	}

	/*! \brief Print current graph and save it to file
	 *
	 * \param file file
	 *
	 */
	void write(const std::string & file)
	{
		VTKWriter<DistGraph_CSR<nm_v<dim>, nm_e>, DIST_GRAPH> gv2(g);
		gv2.write(std::to_string(file + ".vtk"));
	}

	/*! \brief copy operator
	 *
	 * \param dist object to copy
	 *
	 * \return itself
	 *
	 */
	const DistParMetisDistribution<dim, T> & operator=(const DistParMetisDistribution<dim, T> & dist)
	{
		gr = dist.gr;
		domain = dist.domain;
		g = dist.g;
		vtxdist = dist.vtxdist;
		partitions = dist.partitions;
		v_per_proc = dist.v_per_proc;
		verticesGotWeights = dist.verticesGotWeights;

		return *this;
	}

	/*! \brief copy operator
	 *
	 * \param dist object to copy
	 *
	 * \return itself
	 *
	 */
	const DistParMetisDistribution<dim, T> & operator=(DistParMetisDistribution<dim, T> && dist)
	{
		gr = dist.gr;
		domain = dist.domain;
		g.swap(dist.g);
		vtxdist.swap(dist.vtxdist);
		partitions.swap(dist.partitions);
		v_per_proc.swap(dist.v_per_proc);
		verticesGotWeights = dist.verticesGotWeights;

		return *this;
	}
}
;

#endif /* SRC_DECOMPOSITION_PARMETISDISTRIBUTION_HPP_ */