一 一种枚举类型的新型使用方式
///D:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.29.30133\include\xiosbase
enum _Iostate {
// constants for stream states
_Statmask = 0x17
};
static constexpr _Iostate goodbit = static_cast<_Iostate>(0x0);
static constexpr _Iostate eofbit = static_cast<_Iostate>(0x1);
static constexpr _Iostate failbit = static_cast<_Iostate>(0x2);
static constexpr _Iostate badbit = static_cast<_Iostate>(0x4);
_NODISCARD bool __CLR_OR_THIS_CALL good() const {
return rdstate() == ios_base::goodbit;
}
_NODISCARD bool __CLR_OR_THIS_CALL eof() const {
return rdstate() & ios_base::eofbit;
}
_NODISCARD bool __CLR_OR_THIS_CALL fail() const {
return rdstate() & (ios_base::badbit | ios_base::failbit);
}
_NODISCARD bool __CLR_OR_THIS_CALL bad() const {
return rdstate() & ios_base::badbit;
}
_NODISCARD iostate __CLR_OR_THIS_CALL exceptions() const {
return _Except;
}
void __CLR_OR_THIS_CALL exceptions(iostate _Newexcept) {
// set exception mask to argument
_Except = _Newexcept & _Statmask;
clear(rdstate());
}
AI 代码解读
二 Eigen库中的LDLT分解
Eigen-3.4.0\Eigen\src\Cholesky\LDLT.h
template<typename _MatrixType,int _UpLo>
template<bool Conjugate, typename RhsType, typename DstType>
void LDLT<_MatrixType,_UpLo>::_solve_impl_transposed(const RhsType &rhs, DstType &dst) const
{
// dst = P b
dst = m_transpositions * rhs;
// dst = L^-1 (P b)
// dst = L^-*T (P b)
matrixL().template conjugateIf<!Conjugate>().solveInPlace(dst);
// dst = D^-* (L^-1 P b)
// dst = D^-1 (L^-*T P b)
// more precisely, use pseudo-inverse of D (see bug 241)
using std::abs;
const typename Diagonal<const MatrixType>::RealReturnType vecD(vectorD());
// In some previous versions, tolerance was set to the max of 1/highest (or rather numeric_limits::min())
// and the maximal diagonal entry * epsilon as motivated by LAPACK's xGELSS:
// RealScalar tolerance = numext::maxi(vecD.array().abs().maxCoeff() * NumTraits<RealScalar>::epsilon(),RealScalar(1) / NumTraits<RealScalar>::highest());
// However, LDLT is not rank revealing, and so adjusting the tolerance wrt to the highest
// diagonal element is not well justified and leads to numerical issues in some cases.
// Moreover, Lapack's xSYTRS routines use 0 for the tolerance.
// Using numeric_limits::min() gives us more robustness to denormals.
RealScalar tolerance = (std::numeric_limits<RealScalar>::min)();
for (Index i = 0; i < vecD.size(); ++i)
{
if(abs(vecD(i)) > tolerance)
dst.row(i) /= vecD(i);
else
dst.row(i).setZero();
}
// dst = L^-* (D^-* L^-1 P b)
// dst = L^-T (D^-1 L^-*T P b)
matrixL().transpose().template conjugateIf<Conjugate>().solveInPlace(dst);
// dst = P^T (L^-* D^-* L^-1 P b) = A^-1 b
// dst = P^-T (L^-T D^-1 L^-*T P b) = A^-1 b
dst = m_transpositions.transpose() * dst;
}
#endif
AI 代码解读
三 Eigen中的访问者模式
Eigen-3.4.0\Eigen\src\Core\Visitor.h
template<typename Derived>
template<typename Visitor>
EIGEN_DEVICE_FUNC
void DenseBase<Derived>::visit(Visitor& visitor) const
{
if(size()==0)
return;
typedef typename internal::visitor_evaluator<Derived> ThisEvaluator;
ThisEvaluator thisEval(derived());
enum {
unroll = SizeAtCompileTime != Dynamic
&& SizeAtCompileTime * int(ThisEvaluator::CoeffReadCost) + (SizeAtCompileTime-1) * int(internal::functor_traits<Visitor>::Cost) <= EIGEN_UNROLLING_LIMIT
};
return internal::visitor_impl<Visitor, ThisEvaluator, unroll ? int(SizeAtCompileTime) : Dynamic>::run(thisEval, visitor);
}
AI 代码解读
