Skip to content

I
INMOST
Commit 3511b71c authored by Kirill Terekhov's avatar Kirill Terekhov
Browse files
Options

Commit for #6 

Added early exit for internal BiCGStab(l). Added test example based on
#6. Resolved #6 for internal solvers.
parent ed771fa6
Hide whitespace changes
Inline Side-by-side
Showing with 305 additions and 104 deletions
.gitignore
View file @ 3511b71c
......@@ -5,4 +5,3 @@ compile64_vs2013/
examples/Data/
examples/Matrices/
examples/Grids/
tests/
CMakeLists.txt
View file @ 3511b71c
......@@ -226,6 +226,8 @@ add_subdirectory(projects)
endif(COMPILE_PROJECTS)
if(COMPILE_TESTS)
enable_testing()
include(CTest)
add_subdirectory(tests)
endif(COMPILE_TESTS)
......@@ -266,9 +268,9 @@ install(FILES
install(TARGETS inmost EXPORT inmost-targets
ARCHIVE DESTINATION lib
LIBRARY DESTINATION lib
RUNTIME DESTINATION bin
PUBLIC_HEADER DESTINATION include)
LIBRARY DESTINATION lib
RUNTIME DESTINATION bin
PUBLIC_HEADER DESTINATION include)
install(EXPORT inmost-targets DESTINATION cmake)
Tests/CMakeLists.txt 0 → 100644
View file @ 3511b71c
add_subdirectory(solver_test000)
\ No newline at end of file
Tests/solver_test000/CMakeLists.txt 0 → 100644
View file @ 3511b71c
project(solver_test000)
set(SOURCE main.cpp)
add_executable(solver_test000 ${SOURCE})
target_link_libraries(solver_test000 inmost)
if(USE_MPI)
message("linking solver_test000 with MPI")
target_link_libraries(solver_test000 ${MPI_LIBRARIES})
if(MPI_LINK_FLAGS)
set_target_properties(solver_test000 PROPERTIES LINK_FLAGS "${MPI_LINK_FLAGS}")
endif()
endif(USE_MPI)
if(USE_SOLVER_ANI)
message("linking solver_test000 with ani3d")
target_link_libraries(solver_test000 ani3d)
endif()
if(USE_SOLVER_PETSC)
message("linking solver_test000 with PETSc")
target_link_libraries(solver_test000 ${PETSC_LIBRARIES})
endif()
if(USE_SOLVER_TRILINOS)
message("linking solver_test000 with Trilinos")
target_link_libraries(solver_test000 ${Trilinos_LIBRARIES} ${Trilinos_TPL_LIBRARIES})
endif()
add_test(NAME solver_test000_serial_inner_ilu2 COMMAND $<TARGET_FILE:solver_test000> 0 0)
add_test(NAME solver_test000_serial_inner_mliluc COMMAND $<TARGET_FILE:solver_test000> 0 1)
if(USE_SOLVER_PETSC)
add_test(NAME solver_test000_serial_petsc COMMAND $<TARGET_FILE:solver_test000> 0 2)
endif()
if(USE_SOLVER_TIRLINOS)
add_test(NAME solver_test000_serial_trilinos_aztec COMMAND $<TARGET_FILE:solver_test000> 0 3)
add_test(NAME solver_test000_serial_trilinos_ifpack COMMAND $<TARGET_FILE:solver_test000> 0 4)
add_test(NAME solver_test000_serial_trilinos_ml COMMAND $<TARGET_FILE:solver_test000> 0 5)
add_test(NAME solver_test000_serial_trilinos_belos COMMAND $<TARGET_FILE:solver_test000> 0 6)
endif()
if(USE_SOLVER_ANI)
add_test(NAME solver_test000_serial_ani COMMAND $<TARGET_FILE:solver_test000> 2 0 7)
endif()
if( USE_MPI )
if( EXISTS ${MPIEXEC} )
add_test(NAME solver_test000_parallel_normal_inner_ilu2 COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 0 0)
add_test(NAME solver_test000_parallel_permute1_inner_ilu2 COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 1 0)
add_test(NAME solver_test000_parallel_permute2_inner_ilu2 COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 2 0)
add_test(NAME solver_test000_parallel_normal_inner_mliluc COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 0 1)
add_test(NAME solver_test000_parallel_permute1_inner_mliluc COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 1 1)
add_test(NAME solver_test000_parallel_permute2_inner_mliluc COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 2 1)
if(USE_SOLVER_PETSC)
add_test(NAME solver_test000_parallel_normal_petsc COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 0 2)
#add_test(NAME solver_test000_parallel_permute1_petsc COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 1 2)
#add_test(NAME solver_test000_parallel_permute2_petsc COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 2 2)
endif()
if(USE_SOLVER_TRILINOS)
add_test(NAME solver_test000_parallel_normal_trilinos_aztec COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 0 3)
add_test(NAME solver_test000_parallel_permute1_trilinos_aztec COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 1 3)
add_test(NAME solver_test000_parallel_permute2_trilinos_aztec COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 2 3)
add_test(NAME solver_test000_parallel_normal_trilinos_ifpack COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 0 4)
add_test(NAME solver_test000_parallel_permute1_trilinos_ifpack COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 1 4)
add_test(NAME solver_test000_parallel_permute2_trilinos_ifpack COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 2 4)
add_test(NAME solver_test000_parallel_normal_trilinos_ml COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 0 5)
add_test(NAME solver_test000_parallel_permute1_trilinos_ml COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 1 5)
add_test(NAME solver_test000_parallel_permute2_trilinos_ml COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 2 5)
#add_test(NAME solver_test000_parallel_normal_trilinos_belos COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 0 6)
#add_test(NAME solver_test000_parallel_permute1_trilinos_belos COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 1 6)
#add_test(NAME solver_test000_parallel_permute2_trilinos_belos COMMAND ${MPIEXEC} -np 4 $<TARGET_FILE:solver_test000> 2 6)
endif()
endif()
endif()
Tests/solver_test000/main.cpp 0 → 100644
View file @ 3511b71c
#include <cstdio>
#include <cmath>
#include "../../inmost.h"
using namespace INMOST;
int main(int argc,char ** argv)
{
int permut = 0;
int solver = 0; // 0 - INNER_ILU2, 1 - INNER_MLILUC, 2 - PETSc
// 3 - Trilinos_Aztec, 4 - Trilinos_Ifpack,
// 5 - Trilinos_ML, 6 - Trilinos_Belos, 7 - ANI
int rank,procs,newrank;
Solver::Initialize(&argc,&argv,"database.txt"); // Initialize the solver and MPI activity
#if defined(USE_MPI)
MPI_Comm_rank(MPI_COMM_WORLD,&rank); // Get the rank of the current process
MPI_Comm_size(MPI_COMM_WORLD,&procs); // Get the total number of processors used
#else
rank = 0;
procs = 1;
#endif
if (argc > 1) permut = atoi(argv[1]);
if (argc > 2) solver = atoi(argv[2]);
if (permut < procs) newrank = (rank + permut) % procs;
else newrank = (permut - rank) % procs;
std::cout << rank << " -> " << newrank << std::endl;
Solver::Type type;
switch(solver)
{
case 0: type = Solver::INNER_ILU2; break;
case 1: type = Solver::INNER_MLILUC; break;
case 2: type = Solver::PETSc; break;
case 3: type = Solver::Trilinos_Aztec; break;
case 4: type = Solver::Trilinos_Ifpack; break;
case 5: type = Solver::Trilinos_ML; break;
case 6: type = Solver::Trilinos_Belos; break;
case 7: type = Solver::ANI; break;
}
{
Solver S(type); // Specify the linear solver
Solver::Matrix A; // Declare the matrix of the linear system to be solved
Solver::Vector x,b; // Declare the solution and the right-hand side vectors
INMOST_DATA_ENUM_TYPE mbeg, mend;
mbeg = newrank * 10;
mend = (newrank+1) * 10;
A.SetInterval(mbeg,mend);
x.SetInterval(mbeg,mend);
b.SetInterval(mbeg,mend);
for( INMOST_DATA_ENUM_TYPE i = mbeg; i != mend; i++ )
{
A[i][i] = 10.0;
b[i] = i;
x[i] = 0.0;
}
if (rank==0) std::cout << "next call S.SetMatrix(A);" << std::endl;
S.SetMatrix(A); // Compute the preconditioner for the original matrix
if (rank==0) std::cout << "next call S.Solve(b,x);" << std::endl;
if( !S.Solve(b,x) ) // Solve the linear system with the previously computted preconditioner
{
if( rank == 0 )
std::cout << S.GetReason() << std::endl;
MPI_Abort(MPI_COMM_WORLD,-1);
}
double err = 0;
for( INMOST_DATA_ENUM_TYPE i = mbeg; i != mend; i++ )
{
err += (x[i] - static_cast<double>(i)/10.0);
}
#if defined(USE_MPI)
double tmp;
MPI_Allreduce(&err,&tmp,1,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD);
err = tmp;
#endif
if (rank==0)
{
std::cout << "done, error " << err << "\t\t\t" << std::endl;
}
if( fabs(err) > 1.0e-5 || err!=err) MPI_Abort(MPI_COMM_WORLD,-1);
x.Save("output.rhs");
b.Save("b.rhs");
A.Save("A.mtx");
}
Solver::Finalize(); // Finalize solver and close MPI activity
return 0;
}
\ No newline at end of file
examples/Solver/main.cpp
View file @ 3511b71c
......@@ -151,6 +151,7 @@ int main(int argc,char ** argv)
#if defined(USE_PARTITIONER)
Partitioner::Initialize(&argc,&argv);
#endif
/*
{
Mesh m;
//m.SetFileOption("VERBOSITY","2");
......@@ -161,7 +162,7 @@ int main(int argc,char ** argv)
Partitioner::Finalize();
Mesh::Finalize();
return 0;
*/
if( argc > 1 )
{
......@@ -520,5 +521,6 @@ int main(int argc,char ** argv)
Partitioner::Finalize();
#endif
Solver::Finalize();
Mesh::Finalize();
return 0;
}
inmost_solver.h
View file @ 3511b71c
......@@ -94,9 +94,9 @@ namespace INMOST
/// Restore initial nonparallel state of the Matrix with no overlap.
void RestoreMatrix(Matrix & m);
/// Prepare parallel state of the Vector.
void PrepareVector(Vector & v);
void PrepareVector(Vector & v) const;
/// Restore initial nonparallel state of the Vector.
void RestoreVector(Vector & v); //Restore initial nonparallel state of the vector
void RestoreVector(Vector & v) const;
/// Retrieve the processor number by binary search for the specified global index.
INMOST_DATA_ENUM_TYPE GetProcessor(INMOST_DATA_ENUM_TYPE gind) const; //retrive processor by binary search in global_to_proc
void GetOverlapRegion(INMOST_DATA_ENUM_TYPE proc, INMOST_DATA_ENUM_TYPE & mbeg, INMOST_DATA_ENUM_TYPE & mend) const;
......@@ -113,9 +113,9 @@ namespace INMOST
/// Sum shared values in parallel vector.
void Accumulate(Vector & x); // sum shared values in parallel vector
/// Get the sum of num elements of real array on all processes.
void Integrate(INMOST_DATA_REAL_TYPE * inout, INMOST_DATA_ENUM_TYPE num);
void Integrate(INMOST_DATA_REAL_TYPE * inout, INMOST_DATA_ENUM_TYPE num) const;
/// Get the communicator which the solver is associated with.
INMOST_MPI_Comm GetComm() {return comm;}
INMOST_MPI_Comm GetComm() const {return comm;}
// Access to arrays below allows to organize manual exchange
INMOST_MPI_Request * GetSendRequests() {assert(!send_requests.empty()); return &send_requests[0];}
INMOST_MPI_Request * GetRecvRequests() {assert(!recv_requests.empty()); return &recv_requests[0];}
......@@ -128,6 +128,9 @@ namespace INMOST
//for debug
//~ void BeginSequentialCode() {for(int i = 0; i < rank; i++) MPI_Barrier(comm);}
//~ void EndSequentialCode() {for(int i = rank; i < size; i++) MPI_Barrier(comm);}
/// Get the scalar product of the specified interval of the distributed vector.
INMOST_DATA_REAL_TYPE ScalarProd(Vector const & left, Vector const & right, INMOST_DATA_ENUM_TYPE index_begin, INMOST_DATA_ENUM_TYPE index_end) const;
};
/// Distributed vector class.
......@@ -176,8 +179,7 @@ namespace INMOST
/// Get the communicator which the vector is associated with.
INMOST_MPI_Comm GetCommunicator() const {return comm;}
/// Get the scalar product of the specified interval of the distributed vector.
INMOST_DATA_REAL_TYPE ScalarProd(Vector const & other, INMOST_DATA_ENUM_TYPE index_begin, INMOST_DATA_ENUM_TYPE index_end) const;
/// Save the distributed vector to a single data file using parallel MPI I/O.
......
solver.cpp
View file @ 3511b71c
......@@ -77,14 +77,14 @@ namespace INMOST
}
return size;
}
void Solver::OrderInfo::Integrate(INMOST_DATA_REAL_TYPE * inout, INMOST_DATA_ENUM_TYPE num)
void Solver::OrderInfo::Integrate(INMOST_DATA_REAL_TYPE * inout, INMOST_DATA_ENUM_TYPE num) const
{
#if defined(USE_MPI)
if( GetSize() == 1 ) return;
int ierr = 0;
INMOST_DATA_REAL_TYPE * temp = new INMOST_DATA_REAL_TYPE [num];
memcpy(temp,inout,sizeof(INMOST_DATA_REAL_TYPE)*num);
GUARD_MPI(MPI_Allreduce(temp,inout,num,INMOST_MPI_DATA_REAL_TYPE,MPI_SUM,comm));
delete [] temp;
dynarray<INMOST_DATA_REAL_TYPE,64> temp(num);
memcpy(temp.data(),inout,sizeof(INMOST_DATA_REAL_TYPE)*num);
GUARD_MPI(MPI_Allreduce(temp.data(),inout,num,INMOST_MPI_DATA_REAL_TYPE,MPI_SUM,comm));
#else
(void) inout;
(void) num;
......@@ -147,7 +147,7 @@ namespace INMOST
{
storage_type temp(size*2);
//assemble array that includes rank
for(int k = 0; k < size; k++)
for(int k = 0; k < size; ++k)
{
temp[2*k+0] = global_overlap[2*k];
temp[2*k+1] = k;
......@@ -158,6 +158,8 @@ namespace INMOST
MPI_Group oldg, newg;
MPI_Comm newcomm;
std::vector<int> ranks(size);
for(int k = 0; k < size; ++k)
ranks[k] = temp[2*k+1];
GUARD_MPI(MPI_Comm_group(comm,&oldg));
GUARD_MPI(MPI_Group_incl(oldg,size,&ranks[0],&newg));
GUARD_MPI(MPI_Comm_create(comm,newg,&newcomm));
......@@ -553,14 +555,14 @@ namespace INMOST
have_matrix = false;
}
void Solver::OrderInfo::PrepareVector(Vector & v)
void Solver::OrderInfo::PrepareVector(Vector & v) const
{
if( !have_matrix ) throw PrepareMatrixFirst;
v.SetInterval(local_vector_begin,local_vector_end);
v.isParallel() = true;
}
void Solver::OrderInfo::RestoreVector(Vector & v)
void Solver::OrderInfo::RestoreVector(Vector & v) const
{
assert(have_matrix);
if( v.isParallel() )
......@@ -703,6 +705,7 @@ namespace INMOST
{
//std::cout << __FUNCTION__ << " start" << std::endl;
#if defined(USE_MPI)
if( GetSize() == 1 ) return;
//use MPI_Put/MPI_Get to update vector
assert(x.isParallel()); //the vector was prepared
INMOST_DATA_ENUM_TYPE i, j = 1, k, l = 0;
......@@ -749,6 +752,7 @@ namespace INMOST
{
//std::cout << __FUNCTION__ << " start" << std::endl;
#if defined(USE_MPI)
if( GetSize() == 1 ) return;
//use MPI_Put/MPI_Get to update vector
assert(x.isParallel()); //the vector was prepared
INMOST_DATA_ENUM_TYPE i, j = 1, k, l = 0;
......@@ -794,17 +798,12 @@ namespace INMOST
//std::cout << __FUNCTION__ << " end" << std::endl;
}
INMOST_DATA_REAL_TYPE Solver::Vector::ScalarProd(Vector const & other, INMOST_DATA_ENUM_TYPE index_begin, INMOST_DATA_ENUM_TYPE index_end) const
INMOST_DATA_REAL_TYPE Solver::OrderInfo::ScalarProd(Vector const & left, Vector const & right, INMOST_DATA_ENUM_TYPE index_begin, INMOST_DATA_ENUM_TYPE index_end) const
{
INMOST_DATA_REAL_TYPE ret = 0;
for(INMOST_DATA_ENUM_TYPE i = index_begin; i < index_end; i++)
ret += data[i]*other[i];
#if defined(USE_MPI)
INMOST_DATA_REAL_TYPE temp;
int ierr = 0;
GUARD_MPI(MPI_Allreduce(&ret,&temp,1,INMOST_MPI_DATA_REAL_TYPE,MPI_SUM,comm));
ret = temp;
#endif
for(INMOST_DATA_ENUM_TYPE i = index_begin; i < index_end; ++i)
ret += left[i]*right[i];
Integrate(&ret,1);
return ret;
}
......@@ -1833,7 +1832,8 @@ namespace INMOST
}
else sol->EnumParameter(":fill") = static_cast<INMOST_DATA_ENUM_TYPE>(preconditioner_fill_level);
sol->EnumParameter("levels") = solver_gmres_substeps;
if( sizeof(KSOLVER) == sizeof(BCGSL_solver) )
sol->EnumParameter("levels") = solver_gmres_substeps;
if (!sol->isInitialized())
{
......@@ -2158,9 +2158,9 @@ namespace INMOST
{
List->set("Num Blocks",100);
List->set("Block Size",1);
List->set("Maximum Iterations",maximum_iterations);
List->set("Maximum Iterations",static_cast<int>(maximum_iterations));
List->set("Maximum Restarts",20);
List->set("Convergence Tolerance",relative_tolerance);
List->set("Convergence Tolerance",static_cast<double>(relative_tolerance));
}
//int verbosity = Belos::Warnings + Belos::Errors + Belos::StatusTestDetails + Belos::Debug + Belos::FinalSummary + Belos::TimingDetails;
//List->set("Verbosity",verbosity);
......@@ -2187,8 +2187,8 @@ namespace INMOST
Teuchos::RCP< Belos::SolverManager<double,Epetra_MultiVector,Epetra_Operator> > BelosSolver =
// Teuchos::rcp(new Belos::BlockCgSolMgr<double,Epetra_MultiVector,Epetra_Operator >);
// Teuchos::rcp(new Belos::BlockGmresSolMgr<double,Epetra_MultiVector,Epetra_Operator >);
Teuchos::rcp(new Belos::PseudoBlockGmresSolMgr<double,Epetra_MultiVector,Epetra_Operator >);
Teuchos::rcp(new Belos::BlockGmresSolMgr<double,Epetra_MultiVector,Epetra_Operator >);
// Teuchos::rcp(new Belos::PseudoBlockGmresSolMgr<double,Epetra_MultiVector,Epetra_Operator >);
BelosSolver->setParameters(List);
BelosSolver->setProblem(Problem);
Belos::ReturnType ret = BelosSolver->solve();
......
solver_bcgsl.hpp
View file @ 3511b71c
......@@ -20,7 +20,7 @@ namespace INMOST
INMOST_DATA_ENUM_TYPE iters, maxits, l, last_it;
INMOST_DATA_REAL_TYPE resid;
INMOST_DATA_REAL_TYPE * tau, * sigma, * gamma, *theta1, * theta2, * theta3;
Solver::Vector r0, t, * u, * r;
Solver::Vector r_tilde, x0, t, * u, * r;
Solver::Matrix * Alink;
Method * prec;
std::string reason;
......@@ -66,7 +66,8 @@ namespace INMOST
{
if (isInitialized()) Finalize();
if (prec != NULL && !prec->isInitialized()) prec->Initialize();
info->PrepareVector(r0);
info->PrepareVector(r_tilde);
info->PrepareVector(x0);
info->PrepareVector(t);
tau = new INMOST_DATA_REAL_TYPE[l * 5 + l*l];
sigma = tau + l*l;
......@@ -134,35 +135,51 @@ namespace INMOST
if (!isFinalized()) Finalize();
if (prec != NULL) delete prec;
}
void ApplyOperator(Solver::Vector & Input, Solver::Vector & Output)
{
if (prec != NULL) //right preconditioning here! for left preconditioner have to reverse order
{
prec->Solve(Input, t);
info->Update(t);
Alink->MatVec(1.0,t,0,Output);
info->Update(Output);
}
else
{
Alink->MatVec(1.0,t,0,Output);
info->Update(Output);
}
}
bool Solve(Solver::Vector & RHS, Solver::Vector & SOL)
{
assert(isInitialized());
INMOST_DATA_ENUM_TYPE vbeg,vend, vlocbeg, vlocend;
INMOST_DATA_REAL_TYPE rho0 = 1, rho1, alpha = 0, beta, omega = 1;
INMOST_DATA_REAL_TYPE rho0 = 1, rho1, alpha = 0, beta, omega = 1, eta;
INMOST_DATA_REAL_TYPE resid0, resid, temp[2];
bool is_parallel = info->GetSize() > 1;
iters = 0;
info->PrepareVector(SOL);
info->PrepareVector(RHS);
if( is_parallel ) info->Update(SOL);
if( is_parallel ) info->Update(RHS);
info->Update(SOL);
info->Update(RHS);
if( prec != NULL ) prec->ReplaceSOL(SOL);
if( prec != NULL ) prec->ReplaceRHS(RHS);
info->GetLocalRegion(info->GetRank(),vlocbeg,vlocend);
info->GetVectorRegion(vbeg,vend);
//r[0] = b
std::copy(RHS.Begin(),RHS.End(),r[0].Begin());
{
// r[0] = r[0] - A x
Alink->MatVec(-1,SOL,1,r[0]); //global multiplication, r probably needs an update
if( is_parallel ) info->Update(r[0]); // r is good
}
std::copy(r[0].Begin(),r[0].End(),r0.Begin());
std::fill(u[0].Begin(),u[0].End(),0);
{
resid = 0;
for(INMOST_DATA_ENUM_TYPE k = vlocbeg; k != vlocend; k++) resid += r[0][k]*r[0][k];
if( is_parallel ) info->Integrate(&resid,1);
info->Update(r[0]); // r is good
std::copy(x0.Begin(),x0.End(),SOL.Begin()); //x0 = x
std::fill(SOL.Begin(),SOL.End(),0.0); //x = 0
}
last_resid = resid = resid0 = sqrt(resid);
std::copy(r[0].Begin(),r[0].End(),r_tilde.Begin()); // r_tilde = r[0]
std::fill(u[0].Begin(),u[0].End(),0); // u[0] = 0
resid = info->ScalarProd(r[0],r[0],vlocbeg,vlocend); //resid = dot(r[0],r[0])
for(INMOST_DATA_ENUM_TYPE k = vbeg; k != vend; k++) // r_tilde = r[0] / dot(r[0],r[0])
r_tilde[k] /= resid;
last_resid = resid = resid0 = sqrt(resid); //resid = sqrt(dot(r[0],r[0])
last_it = 0;
#if defined(REPORT_RESIDUAL)
if( info->GetRank() == 0 )
......@@ -174,6 +191,13 @@ namespace INMOST
fflush(stdout);
}
#endif
if( last_resid < atol || last_resid < rtol*resid0 )
{
reason = "initial solution satisfy tolerances";
goto exit;
}
long double tt, ts, tp, ttt;
INMOST_DATA_ENUM_TYPE i = 0;
while( true )
......@@ -184,57 +208,49 @@ namespace INMOST
tt = Timer();
for(INMOST_DATA_ENUM_TYPE j = 0; j < l; j++)
{
rho1 = 0.0;
for(INMOST_DATA_ENUM_TYPE k = vlocbeg; k < vlocend; ++k)
rho1 += r[j][k]*r0[k];
if( is_parallel ) info->Integrate(&rho1,1);
rho1 = info->ScalarProd(r[j],r_tilde,vlocbeg,vlocend); // rho1 = dot(r[j],r_tilde)
if( fabs(rho0) < 1.0e-54 )
{
reason = "denominator(1) is zero";
goto exit;
}
beta = alpha * rho1/rho0;
beta = alpha * (rho1/rho0);
rho0 = rho1;
for(INMOST_DATA_ENUM_TYPE i = 0; i < j+1; i++)
for(INMOST_DATA_ENUM_TYPE k = vbeg; k < vend; ++k)
u[i][k] = r[i][k] - beta*u[i][k];
ApplyOperator(u[j],u[j+1]); // u[j+1] = A*R*u[j]
eta = info->ScalarProd(u[j+1],r_tilde,vlocbeg,vlocend); //eta = dot(u[j+1],r_tilde)
if( fabs(eta) < 1.0e-54 )
{
ttt = Timer();
if (prec != NULL) prec->Solve(u[j], t);
tp += Timer() - ttt;
if( is_parallel ) info->Update(t);
ttt = Timer();
Alink->MatVec(1.0,t,0,u[j+1]);
ts += Timer() - ttt;
if( is_parallel ) info->Update(u[j+1]);
}
{
temp[0] = 0;
for(INMOST_DATA_ENUM_TYPE k = vlocbeg; k < vlocend; ++k)
temp[0] += u[j+1][k]*r0[k];
if( is_parallel ) info->Integrate(temp,1);
if( fabs(temp[0]) < 1.0e-54 )
{
reason = "denominator(2) is zero";
goto exit;
}
alpha = rho0 / temp[0];
reason = "denominator(2) is zero";
goto exit;
}
alpha = rho0 / eta;
for(INMOST_DATA_ENUM_TYPE k = vbeg; k < vend; ++k)
SOL[k] += alpha*u[0][k];
for(INMOST_DATA_ENUM_TYPE i = 0; i < j+1; i++)
for(INMOST_DATA_ENUM_TYPE k = vbeg; k < vend; ++k)
for(INMOST_DATA_ENUM_TYPE k = vbeg; k < vend; ++k) //r[i] = r[i] - alpha * u[i+1]
r[i][k] -= alpha*u[i+1][k];
resid = info->ScalarProd(r[j],r[j],vlocbeg,vlocend); // resid = dot(r[j],r[j])
resid = sqrt(resid); // resid = sqrt(dot(r[j],r[j]))
if( resid < atol || resid < rtol*resid0 )
{
ttt = Timer();
if (prec != NULL) prec->Solve(r[j], t);
tp += Timer() - ttt;
if( is_parallel ) info->Update(t);
ttt = Timer();
Alink->MatVec(1.0,t,0,r[j+1]);
ts += Timer() - ttt;
if( is_parallel ) info->Update(r[j+1]);
reason = "early exit in bi-cg block";
last_resid = resid;
goto exit;
}
for(INMOST_DATA_ENUM_TYPE k = vbeg; k < vend; ++k)
SOL[k] += alpha*u[0][k];
ApplyOperator(r[j],r[j+1]); // r[j+1] = A*R*r[j]
}
for(INMOST_DATA_ENUM_TYPE j = 1; j < l+1; j++)
......@@ -244,7 +260,7 @@ namespace INMOST
tau[i-1 + (j-1)*l] = 0;
for(INMOST_DATA_ENUM_TYPE k = vlocbeg; k < vlocend; ++k)
tau[i-1 + (j-1)*l] += r[j][k]*r[i][k];
if(is_parallel) info->Integrate(&tau[i-1 + (j-1)*l],1);
info->Integrate(&tau[i-1 + (j-1)*l],1);
tau[i-1 + (j-1)*l] /= sigma[i-1];
for(INMOST_DATA_ENUM_TYPE k = vbeg; k < vend; ++k)
r[j][k] -= tau[i-1 + (j-1)*l]*r[i][k];
......@@ -256,24 +272,24 @@ namespace INMOST
temp[0] += r[j][k]*r[j][k];
temp[1] += r[0][k]*r[j][k];
}
if(is_parallel) info->Integrate(temp,2);
info->Integrate(temp,2);
sigma[j-1] = temp[0]+1.0e-35; //REVIEW
theta2[j-1] = temp[1]/sigma[j-1];
}
omega = theta1[l-1] = theta2[l-1];
for(INMOST_DATA_ENUM_TYPE j = l-1; j > 0; j--)
{
temp[0] = 0;
eta = 0;
for(INMOST_DATA_ENUM_TYPE i = j+1; i < l+1; i++)
temp[0] += tau[j-1 + (i-1)*l] * theta1[i-1];
theta1[j-1] = theta2[j-1] - temp[0];
eta += tau[j-1 + (i-1)*l] * theta1[i-1];
theta1[j-1] = theta2[j-1] - eta;
}
for(INMOST_DATA_ENUM_TYPE j = 1; j < l; j++)
{
temp[0] = 0;
eta = 0;
for(INMOST_DATA_ENUM_TYPE i = j+1; i < l; i++)
temp[0] += tau[j-1 + (i-1)*l] * theta1[i];
theta3[j-1] = theta1[j] + temp[0];
eta += tau[j-1 + (i-1)*l] * theta1[i];
theta3[j-1] = theta1[j] + eta;
}
for(INMOST_DATA_ENUM_TYPE k = vbeg; k < vend; ++k)
{
......@@ -292,10 +308,7 @@ namespace INMOST
}
last_it = i+1;
{
resid = 0;
for(INMOST_DATA_ENUM_TYPE k = vlocbeg; k != vlocend; k++)
resid += r[0][k]*r[0][k];
if( is_parallel ) info->Integrate(&resid,1);
resid = info->ScalarProd(r[0],r[0],vlocbeg,vlocend);
resid = sqrt(resid);
}
tt = Timer() - tt;
......@@ -337,17 +350,14 @@ namespace INMOST
}
i++;
}
exit:
if (prec != NULL)
{
prec->Solve(SOL, r0);
{
Solver::Vector::iterator itSOL = SOL.Begin();
Solver::Vector::iterator itr0 = r0.Begin(), endr0 = r0.End();
while(itr0 != endr0) *itSOL++ = *itr0++;
}
//std::copy(r0.Begin(), r0.End(), SOL.Begin());
prec->Solve(SOL, r_tilde); //undo right preconditioner
std::copy(r_tilde.Begin(), r_tilde.End(), SOL.Begin());
}
exit:
for(INMOST_DATA_ENUM_TYPE k = vlocbeg; k < vlocend; ++k) //undo shift
SOL[k] += x0[k];
//info->RestoreMatrix(A);
info->RestoreVector(SOL);
info->RestoreVector(RHS);
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment