/****************************************************************/
/* Parallel Combinatorial BLAS Library (for Graph Computations) */
/* version 1.6 -------------------------------------------------*/
/* date: 6/15/2017 ---------------------------------------------*/
/* authors: Ariful Azad, Aydin Buluc  --------------------------*/
/****************************************************************/
/*
 Copyright (c) 2010-2017, The Regents of the University of California
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal
 in the Software without restriction, including without limitation the rights
 to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 copies of the Software, and to permit persons to whom the Software is
 furnished to do so, subject to the following conditions:
 
 The above copyright notice and this permission notice shall be included in
 all copies or substantial portions of the Software.
 
 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 THE SOFTWARE.
 */


#ifndef _FULLY_DIST_H
#define _FULLY_DIST_H

#include <iostream>
#include <algorithm>
#include "myenableif.h"

namespace combblas {

template <class IT, class NT, class Enable=void>
class FullyDist
{};     // dummy generic template


/** 
  * The full distribution is actually a two-level distribution that matches the matrix distribution
  * In this scheme, each processor row (except the last) is responsible for t = floor(n/sqrt(p)) elements. 
  * The last processor row gets the remaining (n-floor(n/sqrt(p))*(sqrt(p)-1)) elements
  * Within the processor row, each processor (except the last) is responsible for loc = floor(t/sqrt(p)) elements. 
  * Example: n=103 and p=16
  * All processors P_ij for i=0,1,2 and j=0,1,2 get floor(floor(102/4)/4) = 6 elements
  * All processors P_i3 for i=0,1,2 get 25-6*3 = 7 elements
  * All processors P_3j for j=0,1,2 get (102-25*3)/4 = 6 elements
  * Processor P_33 gets 27-6*3 = 9 elements  
  * Both derived classes, whether sparse or dense, are distributed
  * to processors based on their "length", so that a conversion does not
  * need any communication between sparse and dense formats
 **/
template <class IT, class NT>
class FullyDist<IT, NT, typename combblas::disable_if< combblas::is_boolean<NT>::value, NT >::type >
{
public:
	explicit FullyDist():glen(0)
	{
        SpParHelper::Print("COMBBLAS Warning: It is dangerous to create (vector) objects without specifying the communicator, are you sure you want to create this object in MPI_COMM_WORLD?\n");
		commGrid.reset(new CommGrid(MPI_COMM_WORLD, 0, 0));
	}
	explicit FullyDist(IT globallen): glen(globallen)
	{
        SpParHelper::Print("COMBBLAS Warning: It is dangerous to create (vector) objects without specifying the communicator, are you sure you want to create this object in MPI_COMM_WORLD?\n");
		commGrid.reset(new CommGrid(MPI_COMM_WORLD, 0, 0));
	}
    	/* ABAB: This clashes with FullyDist(IT globallen) signature on MPICH based systems that #define MPI_Comm to be an INT
	FullyDist( MPI_Comm world):glen(0)
    	{
        	commGrid.reset(new CommGrid(world, 0, 0));
    	}*/
	FullyDist( std::shared_ptr<CommGrid> grid):glen(0)
	{
		commGrid = grid;
	}
	FullyDist( std::shared_ptr<CommGrid> grid, IT globallen): glen(globallen)
	{
		commGrid = grid;
	}
	FullyDist<IT,NT> & operator=(const FullyDist<IT,NT> & rhs)
	{
		glen = rhs.glen;
		commGrid = rhs.commGrid;
		return *this;
	}

	IT LengthUntil() const;
	IT RowLenUntil() const;
	IT RowLenUntil(int k) const;
	IT MyLocLength() const;
	IT MyRowLength() const;
	IT TotalLength() const { return glen; }
	int Owner(IT gind, IT & lind) const;
	int OwnerWithinRow(IT n_thisrow, IT ind_withinrow, IT & lind) const;

protected:
	std::shared_ptr<CommGrid> commGrid;
	IT glen;		// global length (actual "length" including zeros)
};


//! Given global index gind,
//! Return the owner processor id, and
//! Assign the local index to lind
template <class IT, class NT>
int FullyDist<IT,NT, typename combblas::disable_if< combblas::is_boolean<NT>::value, NT >::type>
::Owner(IT gind, IT & lind) const
{
	// C++ implicitly upcasts both operands to 64-bit if one is 64-bit and other is 32-bit
	int procrows = commGrid->GetGridRows();
	IT n_perprocrow = glen / procrows;	// length on a typical processor row
	IT n_thisrow;		// length assigned to owner's processor row	
	int own_procrow;	// owner's processor row
	if(n_perprocrow != 0)
	{
		own_procrow = std::min(static_cast<int>(gind / n_perprocrow), procrows-1);	// owner's processor row
	}
	else	// all owned by the last processor row
	{
		own_procrow = procrows -1;
	}

	IT ind_withinrow = gind - (own_procrow * n_perprocrow);
	if(own_procrow == procrows-1)
		n_thisrow = glen - (n_perprocrow*(procrows-1));
	else
		n_thisrow = n_perprocrow;	

	int proccols = commGrid->GetGridCols();
	IT n_perproc = n_thisrow / proccols;	// length on a typical processor

	int own_proccol;
	if(n_perproc != 0)
	{
		own_proccol = std::min(static_cast<int>(ind_withinrow / n_perproc), proccols-1);
	}
	else
	{
		own_proccol = proccols-1;
	}
	lind = ind_withinrow - (own_proccol * n_perproc);

	// GetRank(int rowrank, int colrank) { return rowrank * grcols + colrank;}
	return commGrid->GetRank(own_procrow, own_proccol);
}

/**
 * @param[in] ind_withinrow {index within processor row}
 * @param[in] n_thisrow {length within this row}
 * @param[out] lind {index local to owning processor}
 * Return the owner processor id (within processor row)
 **/
template <class IT, class NT>
int FullyDist<IT,NT,typename combblas::disable_if< combblas::is_boolean<NT>::value, NT >::type >
::OwnerWithinRow(IT n_thisrow, IT ind_withinrow, IT & lind) const
{
	int proccols = commGrid->GetGridCols();
	IT n_perproc = n_thisrow / proccols;	// length on a typical processor
	
	int own_proccol;
	if(n_perproc != 0)
	{
		own_proccol = std::min(static_cast<int>(ind_withinrow / n_perproc), proccols-1);
	}
	else
	{
		own_proccol = proccols-1;
	}
	lind = ind_withinrow - (own_proccol * n_perproc);

	return own_proccol;
}

template <class IT, class NT>
IT FullyDist<IT,NT,typename combblas::disable_if< combblas::is_boolean<NT>::value, NT >::type >
::LengthUntil() const
{
	int procrows = commGrid->GetGridRows();
	int my_procrow = commGrid->GetRankInProcCol();
	IT n_perprocrow = glen / procrows;	// length on a typical processor row
	IT n_thisrow;	// length assigned to this processor row	
	if(my_procrow == procrows-1)
		n_thisrow = glen - (n_perprocrow*(procrows-1));
	else
		n_thisrow = n_perprocrow;	

	int proccols = commGrid->GetGridCols();
	int my_proccol = commGrid->GetRankInProcRow();
	IT n_perproc = n_thisrow / proccols;	// length on a typical processor
	return ((n_perprocrow * my_procrow)+(n_perproc*my_proccol));
}

// Return the length until this processor, within this processor row only
template <class IT, class NT>
IT FullyDist<IT,NT,typename combblas::disable_if< combblas::is_boolean<NT>::value, NT >::type >
::RowLenUntil() const
{
	int procrows = commGrid->GetGridRows();
	int my_procrow = commGrid->GetRankInProcCol();
	IT n_perprocrow = glen / procrows;	// length on a typical processor row
	IT n_thisrow;	// length assigned to this processor row	
	if(my_procrow == procrows-1)
		n_thisrow = glen - (n_perprocrow*(procrows-1));
	else
		n_thisrow = n_perprocrow;	

	int proccols = commGrid->GetGridCols();
	int my_proccol = commGrid->GetRankInProcRow();
	IT n_perproc = n_thisrow / proccols;	// length on a typical processor
	return (n_perproc*my_proccol);
}

// Return the length until the kth processor, within this processor row only
template <class IT, class NT>
IT FullyDist<IT,NT,typename combblas::disable_if< combblas::is_boolean<NT>::value, NT >::type >
::RowLenUntil(int k) const
{
	int procrows = commGrid->GetGridRows();
	int my_procrow = commGrid->GetRankInProcCol();
	IT n_perprocrow = glen / procrows;	// length on a typical processor row
	IT n_thisrow;	// length assigned to this processor row	
	if(my_procrow == procrows-1)
		n_thisrow = glen - (n_perprocrow*(procrows-1));
	else
		n_thisrow = n_perprocrow;	

	int proccols = commGrid->GetGridCols();
	IT n_perproc = n_thisrow / proccols;	// length on a typical processor
	assert(k < proccols);
	return (n_perproc*k);
}

template <class IT, class NT>
IT FullyDist<IT,NT, typename combblas::disable_if< combblas::is_boolean<NT>::value, NT >::type>
::MyLocLength() const
{
	int procrows = commGrid->GetGridRows();
	int my_procrow = commGrid->GetRankInProcCol();
	IT n_perprocrow = glen / procrows;	// length on a typical processor row
	IT n_thisrow;	// length assigned to this processor row	
	if(my_procrow == procrows-1)
		n_thisrow = glen - (n_perprocrow*(procrows-1));
	else
		n_thisrow = n_perprocrow;	

	int proccols = commGrid->GetGridCols();
	int my_proccol = commGrid->GetRankInProcRow();
	IT n_perproc = n_thisrow / proccols;	// length on a typical processor
	if(my_proccol == proccols-1)
		return (n_thisrow - (n_perproc*(proccols-1)));
	else
		return n_perproc;	
}


template <class IT, class NT>
IT FullyDist<IT,NT,typename combblas::disable_if< combblas::is_boolean<NT>::value, NT >::type>
::MyRowLength() const
{
	int procrows = commGrid->GetGridRows();
	int my_procrow = commGrid->GetRankInProcCol();
	IT n_perprocrow = glen / procrows;	// length on a typical processor row
	IT n_thisrow;	// length assigned to this processor row	
	if(my_procrow == procrows-1)
		n_thisrow = glen - (n_perprocrow*(procrows-1));
	else
		n_thisrow = n_perprocrow;	

	return n_thisrow;
}

}

#endif