/usr/include/trilinos/fei_EqnCommMgr.hpp is in libtrilinos-dev 10.4.0.dfsg-1ubuntu2.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 | #ifndef _fei_EqnCommMgr_hpp_
#define _fei_EqnCommMgr_hpp_
/*--------------------------------------------------------------------*/
/* Copyright 2005 Sandia Corporation. */
/* Under the terms of Contract DE-AC04-94AL85000, there is a */
/* non-exclusive license for use of this work by or on behalf */
/* of the U.S. Government. Export of this program may require */
/* a license from the United States Government. */
/*--------------------------------------------------------------------*/
#include "fei_fwd.hpp"
#include "fei_mpi.h"
#include "snl_fei_PointBlockMap.hpp"
#include <fei_CSRMat.hpp>
#include <fei_CSVec.hpp>
#include <fei_CommUtils.hpp>
#include "fei_ProcEqns.hpp"
#include "fei_EqnBuffer.hpp"
/**
The EqnCommMgr (Equation communication manager) class is responsible
for keeping track of equations that require communications. There
are two types of equations in this class:
1. Local equations which remote processors contribute to (e.g., because
they share some local active nodes).
2. Remote equations that the local processor contributes to (the mirror of
case 1.).
Usage Notes:
1. You can't call exchangeEqns until after exchangeIndices has been called.
2. You can't call addSolnValues until after exchangeIndices has been
called.
In general, usage will proceed like this:
in snl_fei::Structure::initComplete:
addLocalEqn (a bunch of times, probably)
addRemoteIndices (also a bunch of times)
exchangeIndices
getNumLocalEqns
localEqnNumbers
localIndicesPtr
in Filter::sumInElem/sumInElemMatrix/sumInElemRHS
addRemoteEqn and/or addRemoteRHS
in Filter::exchangeRemoteEquations
exchangeEqns
getNumLocalEqns
recvEqnNumbersPtr
localIndicesPtr
localCoefsPtr
localRHSsPtr
in Filter::unpackSolution
getNumLocalEqns
localEqnNumbers
addSolnValues
exchangeSoln
in Filter various getSoln functions
getNumSendEqns
sendEqnNumbersPtr
sendEqnSolnPtr
This class also provides general support functions for exchanging equations
among processors. At the start of an all-to-all exchange, all processors
usually know which equations they need to receive, and which
processors they'll be receiving from, OR they know which equations they need to
send, and which processors they'll be sending to. Usually they don't know both
the sending and receiving information though.
This eqn-processor pairing information (for either the "recv-equations" or the
"send-equations") is held in a ProcEqns object. Thus, an all-to-all exchange
of equation data requires two ProcEqns objects -- one holding the recv info, the
other holding the send info.
An EqnCommMgr function (mirrorProcEqns) is provided which does the
communications necessary to populate the 'send' ProcEqns object given a
populated 'recv' ProcEqns object, or vice-versa. Note that at this point, the
equation-lengths need not be known by either the sending or the receiving
processors.
Once the ProcEqns objects have been correctly mirrored, another EqnCommMgr
function (mirrorProcEqnLengths) is available for mirroring the eqn-length
data from one ProcEqns object to the other.
The next step is: given equations (with all associated data) that we need to
send, along with previously known equation-numbers that we need to receive, an
EqnCommMgr function (exchangeEqnBuffers) is provided to first send the
equation-lengths to the receiving processors, followed by the actual equation
data. At this point the exchange is complete. The equation data is
supplied/returned in EqnBuffer objects.
*/
class EqnCommMgr {
public:
/** Constructor.
@param localProc The MPI rank of 'this' processor.
*/
EqnCommMgr(MPI_Comm comm, bool accumulate = true);
/** copy constructor */
EqnCommMgr(const EqnCommMgr& src);
/** assignment operator */
EqnCommMgr& operator=(const EqnCommMgr& src);
/** Destructor. */
virtual ~EqnCommMgr();
/** Produce a clone of 'this' object, including all of its internal data.*/
EqnCommMgr* deepCopy();
/** return the number of processors that share (contribute to) equations that
are locally owned.
*/
size_t getNumSharingProcs() {return(recvProcEqns_->getNumProcs());};
std::vector<int>& sharingProcsPtr() {return(recvProcEqns_->procsPtr());};
/** return the number of processors that own equations that we share
(equations that we contribute to).
*/
size_t getNumOwnerProcs() {return(sendProcEqns_->getNumProcs());};
std::vector<int>& ownerProcsPtr() {return(sendProcEqns_->procsPtr());};
/** add a local equation to which a remote processor will be contributing. */
void addLocalEqn(int eqnNumber, int srcProc);
void addSolnValues(int* eqnNumbers, double* values, int num);
#ifdef FEI_HAVE_IOSFWD
int exchangeIndices(std::ostream* dbgOut=NULL);
int exchangeEqns(std::ostream* dbgOut=NULL);
#else
int exchangeIndices(ostream* dbgOut=NULL);
int exchangeEqns(ostream* dbgOut=NULL);
#endif
void exchangeSoln();
/** Support function to set up information needed for exchanging equation
info among processors.
Beginning assumption: we (the local processor) have a populated ProcEqns
object (inProcEqns) which contains information pairing certain equations
with certain remote processors. These can be equations that we will be
receiving in an all-to-all exchange, or equations that we will be sending
in an all-to-all exchange. In either case, the "mirror" of that
information is needed before the exchange can be performed. i.e., if we
know which eqns we'll be sending, then the mirror info concerns the eqns
we'll be recv'ing, and vice-versa if we already know which eqns we'll be
recv'ing.
This function is to obtain that mirror info, and return it in
outProcEqns. Given a populated ProcEqns object, we want to populate the
mirror ProcEqns object. Note that this function IGNORES any equation-
lengths, if they are present. i.e., the eqn-length info in 'inProcEqns'
is not referenced, and on completion the eqn-length info in 'outProcEqns'
is not populated.
@param inProcEqns Contains the input equation-numbers and associated
processors.
@param outProcEqns Output.
*/
int mirrorProcEqns(ProcEqns& inProcEqns, ProcEqns& outProcEqns);
/** Support function to set up information needed for exchanging equation
info among processors. This function plays a similar role to that of the
above 'mirrorProcEqns' function, but exchanges the length information
rather than the equation-number information. THIS FUNCTION ASSUMES that
both inProcEqns and outProcEqns are already populated with eqn-number
information.
@param inProcEqns Contains equation-numbers, and their associated
lengths.
@param outProcEqns Input/Output. On entry, contains the equation-numbers
(but not their lengths.
@return error non-zero if error occurs.
*/
int mirrorProcEqnLengths(ProcEqns& inProcEqns,
ProcEqns& outProcEqns);
static int exchangeEqnBuffers(MPI_Comm comm, ProcEqns* sendProcEqns,
EqnBuffer* sendEqns, ProcEqns* recvProcEqns,
EqnBuffer* recvEqns, bool accumulate);
int getNumLocalEqns() {return(recvEqns_->getNumEqns());};
std::vector<int>& localEqnNumbers() {return(recvEqns_->eqnNumbers());};
std::vector<fei::CSVec*>& localEqns(){return(recvEqns_->eqns());};
std::vector<std::vector<double>*>* localRHSsPtr()
{return(recvEqns_->rhsCoefsPtr());};
int addRemoteEqn(int eqnNumber, int destProc, const double* coefs,
const int* indices, int num);
int addRemoteEqn(int eqnNumber, const double* coefs,
const int* indices, int num);
int addRemoteEqns(fei::CSRMat& mat, bool onlyIndices);
int addRemoteRHS(fei::CSVec& vec, int rhsIndex);
void setNumRHSs(int numRHSs);
int addRemoteRHS(int eqnNumber, int destProc, int rhsIndex, double value);
int addRemoteRHS(int eqnNumber, int rhsIndex, double value);
void addRemoteIndices(int eqnNumber, int destProc, int* indices, int num);
int getNumRemoteEqns() {return(sendEqns_->getNumEqns());};
std::vector<int>& sendEqnNumbersPtr() {return(sendEqns_->eqnNumbers());};
double* sendEqnSolnPtr() {return(sendEqnSoln_.size()>0? &sendEqnSoln_[0] : NULL);};
void resetCoefs();
int gatherSharedBCs(EqnBuffer& bcEqns);
int exchangeRemEssBCs(int* essEqns, int numEssEqns, double* essAlpha,
double* essGamma, MPI_Comm comm,
std::ostream* dbgOut = NULL);
int getNumRemEssBCEqns() {return(essBCEqns_->getNumEqns());};
std::vector<int>& remEssBCEqnNumbersPtr() {return(essBCEqns_->eqnNumbers());};
std::vector<fei::CSVec*>& remEssBCEqns() {return(essBCEqns_->eqns());};
int exchangePtToBlkInfo(snl_fei::PointBlockMap& blkEqnMapper);
bool newCoefData() {if (recvEqns_->newCoefData_>0) return(true);
else return(false);}
bool newRHSData() {if (recvEqns_->newRHSData_>0) return(true);
else return(false);}
bool accumulate_;
EqnBuffer* getRecvEqns() { return( recvEqns_ ); }
EqnBuffer* getSendEqns() { return( sendEqns_ ); }
ProcEqns* getRecvProcEqns() { return( recvProcEqns_ ); }
ProcEqns* getSendProcEqns() { return( sendProcEqns_ ); }
private:
void deleteEssBCs();
int getSendProcNumber(int eqn);
int consistencyCheck(const char* caller,
std::vector<int>& recvProcs,
std::vector<int>& recvProcTotalLengths,
std::vector<int>& sendProcs,
std::vector<int>& sendProcTotalLengths);
int localProc_;
ProcEqns* recvProcEqns_;
bool exchangeIndicesCalled_; //whether or not the exchangeIndices function
//has been called yet.
EqnBuffer* recvEqns_;
std::vector<double> solnValues_; //solution values we'll need to return to the
//processors that contribute to our equations
ProcEqns* sendProcEqns_;
EqnBuffer* sendEqns_;
std::vector<double> sendEqnSoln_;
//the solution values for the send equations. i.e.,
//we'll recv these solution values for the equations
//that we contributed to (sent) for other processors.
EqnBuffer* essBCEqns_;
MPI_Comm comm_;
};
#endif
|