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

Update custom scalar example, based on unstable/Eigen/AdolcForward .

Browse Source
This commit is contained in:
Jitse Niesen 2012-06-16 20:35:59 +01:00
parent 3c9289129b
commit 148587e229
1 changed files with 23 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

45
doc/I00_CustomizingEigen.dox
View File

@ -123,16 +123,16 @@ Eigen::MatrixBase<Eigen::Matrix<std::complex<float>, 10000, 1, 2, 10000, 1>
By default, Eigen currently supports the following scalar types: \c int, \c float, \c double, \c std::complex<float>, \c std::complex<double>, \c long \c double, \c long \c long \c int (64 bits integers), and \c bool. The \c long \c double is especially useful on x86-64 systems or when the SSE2 instruction set is enabled because it enforces the use of x87 registers with extended accuracy. By default, Eigen currently supports the following scalar types: \c int, \c float, \c double, \c std::complex<float>, \c std::complex<double>, \c long \c double, \c long \c long \c int (64 bits integers), and \c bool. The \c long \c double is especially useful on x86-64 systems or when the SSE2 instruction set is enabled because it enforces the use of x87 registers with extended accuracy.
In order to add support for a custom type \c T you need: In order to add support for a custom type \c T you need:
1 - make sure the common operator (+,-,*,/,etc.) are supported by the type \c T -# make sure the common operator (+,-,*,/,etc.) are supported by the type \c T
2 - add a specialization of struct Eigen::NumTraits<T> (see \ref NumTraits) -# add a specialization of struct Eigen::NumTraits<T> (see \ref NumTraits)
3 - define a couple of math functions for your type such as: internal::sqrt, internal::abs, etc... -# define a couple of math functions for your type such as: internal::sqrt, internal::abs, etc...
(see the file Eigen/src/Core/MathFunctions.h) (see the file Eigen/src/Core/MathFunctions.h)
Here is a concrete example adding support for the Adolc's \c adouble type. <a href="https://projects.coin-or.org/ADOL-C">Adolc</a> is an automatic differentiation library. The type \c adouble is basically a real value tracking the values of any number of partial derivatives. Here is a concrete example adding support for the Adolc's \c adouble type. <a href="https://projects.coin-or.org/ADOL-C">Adolc</a> is an automatic differentiation library. The type \c adouble is basically a real value tracking the values of any number of partial derivatives.
\code \code
#ifndef ADLOCSUPPORT_H #ifndef ADOLCSUPPORT_H
#define ADLOCSUPPORT_H #define ADOLCSUPPORT_H
#define ADOLC_TAPELESS #define ADOLC_TAPELESS
#include <adolc/adouble.h> #include <adolc/adouble.h>
@ -149,7 +149,8 @@ template<> struct NumTraits<adtl::adouble>
enum { enum {
IsComplex = 0, IsComplex = 0,
IsInteger = 0, IsInteger = 0,
IsSigned, IsSigned = 1,
RequireInitialization = 1,
ReadCost = 1, ReadCost = 1,
AddCost = 1, AddCost = 1,
MulCost = 1 MulCost = 1
@ -158,26 +159,26 @@ template<> struct NumTraits<adtl::adouble>
} }
// the Adolc's type adouble is defined in the adtl namespace namespace Eigen {
// therefore, the following internal::* functions *must* be defined namespace internal {
// in the same namespace
namespace adtl {
inline const adouble& internal::conj(const adouble& x) { return x; } inline const adtl::adouble& conj(const adtl::adouble& x) { return x; }
inline const adouble& internal::real(const adouble& x) { return x; } inline const adtl::adouble& real(const adtl::adouble& x) { return x; }
inline adouble internal::imag(const adouble&) { return 0.; } inline adtl::adouble imag(const adtl::adouble&) { return 0.; }
inline adouble internal::abs(const adouble& x) { return fabs(x); } inline adtl::adouble abs(const adtl::adouble& x) { return adtl::fabs(x); }
inline adouble internal::abs2(const adouble& x) { return x*x; } inline adtl::adouble abs2(const adtl::adouble& x) { return x*x; }
inline adouble internal::sqrt(const adouble& x) { return sqrt(x); }
inline adouble internal::exp(const adouble& x) { return exp(x); } using adtl::sqrt;
inline adouble internal::log(const adouble& x) { return log(x); } using adtl::exp;
inline adouble internal::sin(const adouble& x) { return sin(x); } using adtl::log;
inline adouble internal::cos(const adouble& x) { return cos(x); } using adtl::sin;
inline adouble internal::pow(const adouble& x, adouble y) { return pow(x, y); } using adtl::cos;
using adtl::pow;
}
} }
#endif // ADLOCSUPPORT_H #endif // ADOLCSUPPORT_H
\endcode \endcode