jb::fftw Namespace Reference

Wrappers for the FFTW3 library. More...

Namespaces
	detail

Classes
class	allocator
	Define an allocator based on fftw_malloc()/fftw_free() More...
class	plan
	Wrap FFTW3 plan objects. More...
class	tde_result
	A time-delay estimator (TDE) is an algorithm to compare two families of timeseries and return the estimated delay of the first family vs. More...
class	tde_result< boost::multi_array< T, 1, A >, value_t >
	Handles TDE result for a multi array type timeseries with dimensionality 1. More...
class	tde_result< boost::multi_array< T, K, A >, value_t >
	Handles TDE result for a multi array type timeseries. More...
class	time_delay_estimator
	A simple time delay estimator based on cross-correlation. More...
class	time_delay_estimator_many
	A time delay estimator based on cross-correlation. More...
struct	traits
	Wrap fftw_* types and operations to treat floating point values generically. More...
struct	traits< double >
	Wrap the FFTW types and functions for double precision float point numbers. More...
struct	traits< float >
	Wrap the FFTW types and functions for single precision float point numbers. More...
struct	traits< long double >
	Wrap the FFTW types and functions for quad precision float point numbers. More...
struct	traits< std::complex< T > >
	Define traits for std::complex<T> in terms of the traits for T. More...

Typedefs
template<typename value_type , std::size_t num_dims>
using	aligned_multi_array = boost::multi_array< value_type, num_dims, jb::fftw::allocator< value_type > >
	Alias boost::multi_array with properly allocated storage for FFTW3. More...
template<typename T >
using	aligned_vector = std::vector< T, jb::fftw::allocator< T > >
	Alias std::vector with properly allocated storage for FFTW3. More...

Functions
template<typename F >
traits< F >::fftw_complex_type *	fftw_cast_array (std::complex< F > *ptr)
template<typename F >
traits< F >::fftw_complex_type const *	fftw_cast_array (std::complex< F > const *ptr)
template<typename F >
F *	fftw_cast_array (F *ptr)
template<typename F >
F const *	fftw_cast_array (F const *ptr)
template<typename vector >
auto	fftw_cast (vector &in) -> decltype(fftw_cast_array(&in[0]))
template<typename vector >
auto	fftw_cast (vector const &in) -> decltype(fftw_cast_array(&in[0]))
template<typename T , std::size_t K, typename A >
auto	fftw_cast (boost::multi_array< T, K, A > const &a) -> decltype(fftw_cast_array(a.data()))
template<typename T , std::size_t K, typename A >
auto	fftw_cast (boost::multi_array< T, K, A > &a) -> decltype(fftw_cast_array(a.data()))
template<typename in_array_type , typename out_array_type >
plan< in_array_type, out_array_type >	create_forward_plan (in_array_type const &in, out_array_type &out, int flags=default_plan_flags)
	Create a plan to compute many DFTs given the input and output arrays. More...
template<typename in_array_type , typename out_array_type >
plan< in_array_type, out_array_type >	create_backward_plan (in_array_type const &in, out_array_type &out, int flags=default_plan_flags)
	Create a plan to compute many inverse DFT given the input and output arrays. More...

Variables
int constexpr	default_plan_flags

Detailed Description

Wrappers for the FFTW3 library.

---

Class Documentation

jb::fftw::traits

struct jb::fftw::traits

template<typename precision_t>
struct jb::fftw::traits< precision_t >

Wrap fftw_* types and operations to treat floating point values generically.

Type traits to handle single-precision (float / fftwf_*), double-precision (double / fftw_*) and quad-precision (long double / fftwl_*) generically in C++ code.

Template Parameters

precision_t

the type of floating point value, must be either float, double or long double.

Definition at line 22 of file traits.hpp.

Typedef Documentation

aligned_multi_array

template<typename value_type , std::size_t num_dims>

using jb::fftw::aligned_multi_array = typedef boost::multi_array<value_type, num_dims, jb::fftw::allocator<value_type> >

Alias boost::multi_array with properly allocated storage for FFTW3.

FFTW3 recommends using storage aligned for SIMD operations. jb::fftw defines an STL allocator to take advantage of the FFTW3 memory allocation functions that guarantee this alignment. This type alias makes it easy to create boost::multi_arrays using the allocator.

Template Parameters

value_type	the type stored by the multi_array
num_dims	the number of dimensions in the multi array

Definition at line 24 of file aligned_multi_array.hpp.

aligned_vector

template<typename T >

using jb::fftw::aligned_vector = typedef std::vector<T, jb::fftw::allocator<T> >

Alias std::vector with properly allocated storage for FFTW3.

FFTW3 recommends using storage aligned for SIMD operations. jb::fftw defines an STL allocator to take advantage of the FFTW3 memory allocation functions that guarantee this alignment. This type alias makes it easy to create std::vectors using the allocator.

Template Parameters

T	the type stored by the vector

Definition at line 22 of file aligned_vector.hpp.

Function Documentation

create_backward_plan()

template<typename in_array_type , typename out_array_type >

plan<in_array_type, out_array_type> jb::fftw::create_backward_plan	(	in_array_type const &	in,
		out_array_type &	out,
		int	flags = default_plan_flags
	)

Create a plan to compute many inverse DFT given the input and output arrays.

Parameters

in	the input timeseries
out	the output timeseries
flags	the FFTW flags for the plan

Exceptions

std::invalid_argument

if the input and output vectors are not compatible

Returns

a jb::fftw::plan<> of the right type.

Template Parameters

in_array_type	the type of the input timeseries
out_array_type	the type of the output timeseries

Definition at line 261 of file plan.hpp.

References jb::fftw::detail::check_plan_inputs(), jb::fftw::plan< in_timeseries_type, out_timeseries_type >::create_backward_impl(), jb::fftw::plan< in_timeseries_type, out_timeseries_type >::create_backward_many_impl(), jb::detail::element_count(), fftw_cast(), and jb::detail::nsamples().

Referenced by jb::fftw::plan< timeseries_type, frequency_timeseries_type >::create_backward_impl(), and jb::fftw::plan< timeseries_type, frequency_timeseries_type >::execute().

create_forward_plan()

template<typename in_array_type , typename out_array_type >

plan<in_array_type, out_array_type> jb::fftw::create_forward_plan	(	in_array_type const &	in,
		out_array_type &	out,
		int	flags = default_plan_flags
	)

Create a plan to compute many DFTs given the input and output arrays.

Parameters

in	the input array of timeseries
out	the output array of timeseries
flags	the FFTW flags for the plan

Exceptions

std::invalid_argument

if the input and output vectors are not compatible

Returns

a jb::fftw::plan<> of the right type.

Template Parameters

in_array_type	the type of the input array of timeseries
out_array_type	the type of the output array of timeseries

Definition at line 231 of file plan.hpp.

References jb::fftw::detail::check_plan_inputs(), jb::fftw::plan< in_timeseries_type, out_timeseries_type >::create_forward_impl(), jb::fftw::plan< in_timeseries_type, out_timeseries_type >::create_forward_many_impl(), jb::detail::element_count(), fftw_cast(), and jb::detail::nsamples().

Referenced by jb::fftw::plan< timeseries_type, frequency_timeseries_type >::create_forward_impl(), and jb::fftw::plan< timeseries_type, frequency_timeseries_type >::execute().

fftw_cast() [1/4]

template<typename vector >

auto jb::fftw::fftw_cast

(

vector &

in

)

-> decltype(fftw_cast_array(&in[0]))

Definition at line 33 of file cast.hpp.

References fftw_cast_array().

Referenced by create_backward_plan(), create_forward_plan(), and jb::fftw::plan< timeseries_type, frequency_timeseries_type >::execute().

fftw_cast() [2/4]

template<typename vector >

auto jb::fftw::fftw_cast

(

vector const &

in

)

-> decltype(fftw_cast_array(&in[0]))

Definition at line 38 of file cast.hpp.

References fftw_cast_array().

fftw_cast() [3/4]

template<typename T , std::size_t K, typename A >

auto jb::fftw::fftw_cast

(

boost::multi_array< T, K, A > const &

a

)

-> decltype(fftw_cast_array(a.data()))

Definition at line 43 of file cast.hpp.

References fftw_cast_array().

fftw_cast() [4/4]

template<typename T , std::size_t K, typename A >

auto jb::fftw::fftw_cast

(

boost::multi_array< T, K, A > &

a

)

-> decltype(fftw_cast_array(a.data()))

Definition at line 49 of file cast.hpp.

References fftw_cast_array().

fftw_cast_array() [1/4]

template<typename F >

traits<F>::fftw_complex_type* jb::fftw::fftw_cast_array

(

std::complex< F > *

ptr

)

Definition at line 12 of file cast.hpp.

Referenced by fftw_cast().

fftw_cast_array() [2/4]

template<typename F >

traits<F>::fftw_complex_type const* jb::fftw::fftw_cast_array

(

std::complex< F > const *

ptr

)

Definition at line 18 of file cast.hpp.

fftw_cast_array() [3/4]

template<typename F >

F* jb::fftw::fftw_cast_array

(

F *

ptr

)

Definition at line 23 of file cast.hpp.

fftw_cast_array() [4/4]

template<typename F >

F const* jb::fftw::fftw_cast_array

(

F const *

ptr

)

Definition at line 28 of file cast.hpp.

Variable Documentation

default_plan_flags

int constexpr jb::fftw::default_plan_flags

Initial value:

=
    (FFTW_ESTIMATE | FFTW_PRESERVE_INPUT | FFTW_UNALIGNED)

Definition at line 13 of file plan.hpp.