GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-08-12 03:39:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

code simplification

Browse Source
This commit is contained in:
Gael Guennebaud 2015-06-09 15:35:34 +02:00
parent cacbc5679d
commit 64753af3b7
1 changed files with 19 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
unsupported/Eigen/src/IterativeSolvers/GMRES.h
View File

@ -73,7 +73,6 @@ bool gmres(const MatrixType & mat, const Rhs & rhs, Dest & x, const Precondition
// residual and preconditioned residual // residual and preconditioned residual
VectorType p0 = rhs - mat*x; VectorType p0 = rhs - mat*x;
VectorType r0 = precond.solve(p0); VectorType r0 = precond.solve(p0);
VectorType t(m), v(m), workspace(m);
const RealScalar r0Norm = r0.norm(); const RealScalar r0Norm = r0.norm();
@ -92,6 +91,9 @@ bool gmres(const MatrixType & mat, const Rhs & rhs, Dest & x, const Precondition
// storage for Jacobi rotations // storage for Jacobi rotations
std::vector < JacobiRotation < Scalar > > G(restart); std::vector < JacobiRotation < Scalar > > G(restart);
// storage for temporaries
VectorType t(m), v(m), workspace(m), x_new(m);
// generate first Householder vector // generate first Householder vector
Ref<VectorType> H0_tail = H.col(0).tail(m - 1); Ref<VectorType> H0_tail = H.col(0).tail(m - 1);
RealScalar beta; RealScalar beta;
@ -105,15 +107,17 @@ bool gmres(const MatrixType & mat, const Rhs & rhs, Dest & x, const Precondition
v = VectorType::Unit(m, k - 1); v = VectorType::Unit(m, k - 1);
// apply Householder reflections H_{1} ... H_{k-1} to v // apply Householder reflections H_{1} ... H_{k-1} to v
// TODO: use a HouseholderSequence
for (Index i = k - 1; i >= 0; --i) { for (Index i = k - 1; i >= 0; --i) {
v.tail(m - i).applyHouseholderOnTheLeft(H.col(i).tail(m - i - 1), tau.coeffRef(i), workspace.data()); v.tail(m - i).applyHouseholderOnTheLeft(H.col(i).tail(m - i - 1), tau.coeffRef(i), workspace.data());
} }
// apply matrix M to v: v = mat * v; // apply matrix M to v: v = mat * v;
t=mat*v; t.noalias() = mat * v;
v = precond.solve(t); v = precond.solve(t);
// apply Householder reflections H_{k-1} ... H_{1} to v // apply Householder reflections H_{k-1} ... H_{1} to v
// TODO: use a HouseholderSequence
for (Index i = 0; i < k; ++i) { for (Index i = 0; i < k; ++i) {
v.tail(m - i).applyHouseholderOnTheLeft(H.col(i).tail(m - i - 1), tau.coeffRef(i), workspace.data()); v.tail(m - i).applyHouseholderOnTheLeft(H.col(i).tail(m - i - 1), tau.coeffRef(i), workspace.data());
} }
@ -158,18 +162,14 @@ bool gmres(const MatrixType & mat, const Rhs & rhs, Dest & x, const Precondition
if (stop || k == restart) if (stop || k == restart)
{ {
// solve upper triangular system // solve upper triangular system
VectorType y = w.head(k); Ref<VectorType> y = w.head(k);
H.topLeftCorner(k, k).template triangularView <Upper>().solveInPlace(y); H.topLeftCorner(k, k).template triangularView <Upper>().solveInPlace(y);
// use Horner-like scheme to calculate solution vector // use Horner-like scheme to calculate solution vector
VectorType x_new = y(k - 1) * VectorType::Unit(m, k - 1); x_new.setZero();
for (Index i = k - 1; i >= 0; --i)
// apply Householder reflection H_{k} to x_new
x_new.tail(m - k + 1).applyHouseholderOnTheLeft(H.col(k - 1).tail(m - k), tau.coeffRef(k - 1), workspace.data());
for (Index i = k - 2; i >= 0; --i)
{ {
x_new += y(i) * VectorType::Unit(m, i); x_new(i) += y(i);
// apply Householder reflection H_{i} to x_new // apply Householder reflection H_{i} to x_new
x_new.tail(m - i).applyHouseholderOnTheLeft(H.col(i).tail(m - i - 1), tau.coeffRef(i), workspace.data()); x_new.tail(m - i).applyHouseholderOnTheLeft(H.col(i).tail(m - i - 1), tau.coeffRef(i), workspace.data());
} }
@ -185,17 +185,17 @@ bool gmres(const MatrixType & mat, const Rhs & rhs, Dest & x, const Precondition
k=0; k=0;
// reset data for restart // reset data for restart
p0 = rhs - mat*x; p0.noalias() = rhs - mat*x;
r0 = precond.solve(p0); r0 = precond.solve(p0);
// clear Hessenberg matrix and Householder data // clear Hessenberg matrix and Householder data
H = FMatrixType::Zero(m, restart + 1); H.setZero();
w = VectorType::Zero(restart + 1); w.setZero();
tau = VectorType::Zero(restart + 1); tau.setZero();
// generate first Householder vector // generate first Householder vector
r0.makeHouseholder(H0_tail, tau.coeffRef(0), beta); r0.makeHouseholder(H0_tail, tau.coeffRef(0), beta);
w(0)=(Scalar) beta; w(0) = Scalar(beta);
} }
} }
} }