! ! Parallel Sparse BLAS version 3.5 ! (C) Copyright 2006, 2010, 2015, 2017 ! Salvatore Filippone ! Alfredo Buttari CNRS-IRIT, Toulouse ! ! Redistribution and use in source and binary forms, with or without ! modification, are permitted provided that the following conditions ! are met: ! 1. Redistributions of source code must retain the above copyright ! notice, this list of conditions and the following disclaimer. ! 2. Redistributions in binary form must reproduce the above copyright ! notice, this list of conditions, and the following disclaimer in the ! documentation and/or other materials provided with the distribution. ! 3. The name of the PSBLAS group or the names of its contributors may ! not be used to endorse or promote products derived from this ! software without specific written permission. ! ! THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ! ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED ! TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR ! PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE PSBLAS GROUP OR ITS CONTRIBUTORS ! BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR ! CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF ! SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS ! INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN ! CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ! ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE ! POSSIBILITY OF SUCH DAMAGE. ! ! !********************************************************************** ! * ! The communication step among processors at each * ! matrix-vector product is a variable all-to-all * ! collective communication that we reimplement * ! in terms of point-to-point communications. * ! The data in input is a list of dependencies: * ! for each node a list of all the nodes it has to * ! communicate with. The lists are guaranteed to be * ! symmetric, i.e. for each pair (I,J) there is a * ! pair (J,I). The idea is to organize the ordering * ! so that at each communication step as many * ! processors as possible are communicating at the * ! same time, i.e. a step is defined by the fact * ! that all edges (I,J) in it have no common node. * ! * ! Formulation of the problem is: * ! Given an undirected graph (forest): * ! Find the shortest series of steps to cancel all * ! graph edges, where at each step all edges belonging * ! to a matching in the graph are canceled. * ! * ! An obvious lower bound to the optimum number of steps * ! is the largest degree of any node in the graph. * ! * ! The algorithm proceeds as follows: * ! 1. Build a list of all edges, e.g. copy the * ! dependencies lists keeping only (I,J) with I