#include <tensortrain.h>

Inheritance diagram for madness::TensorTrain< T >:

Collaboration diagram for madness::TensorTrain< T >:

Public Member Functions
	TensorTrain ()
	empty constructor More...

	TensorTrain (const long &ndims, const long *dims)
	ctor for a TensorTrain, set up only the dimensions, no data More...

	TensorTrain (const std::vector< long > &dims)
	ctor for a TensorTrain, set up only the dimensions, no data More...

	TensorTrain (const std::vector< Tensor< T > > &c)
	ctor for a TensorTrain, with core tensors explicitly given More...

	TensorTrain (const Tensor< T > &t, double eps)
	ctor for a TensorTrain, with the tolerance eps More...

	TensorTrain (const Tensor< T > &t, double eps, const std::vector< long > dims)
	ctor for a TensorTrain, with the tolerance eps More...

	TensorTrain (const TensorTrain &other)
	copy constructor, shallow More...

void	decompose (const Tensor< T > &t, double eps, const std::vector< long > &dims)
	decompose the input tensor into a TT representation More...

TensorTrain< T > &	emul (const TensorTrain< T > &other)
	compute the Hadamard product of two TensorTrains More...

void	fusedim (const long i)
	merge two dimensions into one More...

TensorTrain< T > &	gaxpy (const std::array< Slice, TENSOR_MAXDIM > &s1, const TensorTrain< T > &rhs, T beta, const std::array< Slice, TENSOR_MAXDIM > &s2)
	Inplace generalized saxpy with slices and without alpha. More...

TensorTrain< T > &	gaxpy (T alpha, const TensorTrain< T > &rhs, T beta)
	Inplace generalized saxpy ... this = thisalpha + otherbeta. More...

Tensor< T > &	get_core (const int i)
	reference to the internal core More...

const Tensor< T > &	get_core (const int i) const
	const reference to the internal core More...

bool	is_operator () const
	check if this is an operator (r,k',k,r) More...

bool	is_tensor () const
	check if this is a tensor (r,k,r) More...

bool	is_zero_rank () const
	if rank is zero More...

TensorTrain< T > &	make_operator ()
	convert this into an operator representation (r,k',k,r) More...

TensorTrain< T > &	make_tensor ()
	convert this into a tensor representation (r,k,r) More...

float_scalar_type	normf () const
	returns the Frobenius norm More...

template<class Q >
	operator TensorTrain< Q > () const
	Type conversion makes a deep copy. More...

TensorTrain< T >	operator* (const T &factor) const
	return this multiplied by a scalar More...

TensorTrain< T > &	operator+= (const TensorTrain< T > &rhs)
	inplace addition of two Tensortrains; will increase ranks of this More...

TensorTrain< T > &	operator-= (const TensorTrain< T > &rhs)
	inplace subtraction of two Tensortrains; will increase ranks of this More...

TensorTrain< T > &	operator= (const T &number)
	assign a number to this tensor More...

TensorTrain &	operator= (const TensorTrain &other)
	assigment operator More...

T *	ptr (const int ivec=0)
	Returns a pointer to the internal data. More...

const T *	ptr (const int ivec=0) const
	Returns a pointer to the internal data. More...

std::vector< long >	ranks () const
	return the TT ranks More...

long	ranks (const int i) const
	return the TT ranks for dimension i (to i+1) More...

long	real_size () const
	return the size of this instance, including static memory for vectors and such More...

Tensor< T >	reconstruct (const bool flat=false) const
	reconstruct this to a full representation More...

void	scale (T fac)
	scale this by a number More...

template<typename Archive >
void	serialize (Archive &ar)
	serialize this More...

void	set_size_and_dim (std::vector< long > dims)

long	size () const
	return the number of coefficients in all core tensors More...

TensorTrain< T >	splitdim (long idim, long k1, long k2, const double eps) const

template<class Q >
std::enable_if<(TensorTypeData< T >::iscomplex or TensorTypeData< Q >::iscomplex), TENSOR_RESULT_TYPE(T, Q)>::type	trace (const TensorTrain< Q > &B) const

template<class Q >
std::enable_if<!(TensorTypeData< T >::iscomplex or TensorTypeData< Q >::iscomplex), TENSOR_RESULT_TYPE(T, Q)>::type	trace (const TensorTrain< Q > &B) const
	Return the trace of two tensors, no complex conjugate involved. More...

template<typename R = T>
std::enable_if<!std::is_arithmetic< R >::value, void >::type	truncate (double eps)
	recompress and truncate this TT representation More...

template<typename R = T>
std::enable_if< std::is_arithmetic< R >::value, void >::type	truncate (double eps)
	recompress and truncate this TT representation More...

void	two_mode_representation (Tensor< T > &U, Tensor< T > &VT, Tensor< typename Tensor< T >::scalar_type > &s) const
	construct a two-mode representation (aka unnormalized SVD) More...

bool	verify () const
	return the dimensions of this tensor More...

void	zero_me ()
	turn this into an empty tensor with all cores properly shaped More...

void	zero_me (const std::vector< long > &dim)
	turn this into an empty tensor with all cores properly shaped More...

Public Member Functions inherited from madness::BaseTensor
	BaseTensor ()

virtual	~BaseTensor ()

bool	conforms (const BaseTensor *t) const
	Returns true if this and *t are the same shape and size. More...

long	dim (int i) const
	Returns the size of dimension `i`. More...

const long *	dims () const
	Returns the array of tensor dimensions. More...

long	id () const
	Returns the typeid of the tensor (c.f., `TensorTypeData<T>` ) More...

bool	iscontiguous () const
	Returns true if the tensor refers to contiguous memory locations. More...

long	ndim () const
	Returns the number of dimensions in the tensor. More...

long	size () const
	Returns the number of elements in the tensor. More...

long	stride (int i) const
	Returns the stride associated with dimension `i`. More...

const long *	strides () const
	Returns the array of tensor strides. More...

Private Types
typedef TensorTypeData< T >::float_scalar_type	float_scalar_type
	C++ typename of the floating point type associated with scalar real type. More...

typedef TensorTypeData< T >::scalar_type	scalar_type
	C++ typename of the real type associated with a complex type. More...

typedef T	type
	C++ typename of this tensor. More...

Private Attributes
std::vector< Tensor< T > >	core

bool	zero_rank
	true if rank is zero More...

Friends
template<typename Q >
class	TensorTrain

TensorTrain	copy (const TensorTrain &other)
	deep copy of the whole tensor More...

TensorTrain	copy (const TensorTrain &other, const std::array< Slice, TENSOR_MAXDIM > &s)
	deep copy of a slice of the tensor More...

template<typename R , typename Q >
TensorTrain< TENSOR_RESULT_TYPE(R, Q)>	general_transform (const TensorTrain< R > &t, const Tensor< Q > c[])

template<typename R , typename Q >
TensorTrain< TENSOR_RESULT_TYPE(R, Q)>	outer (const TensorTrain< R > &t1, const TensorTrain< Q > &t2)

template<typename R , typename Q >
TensorTrain< TENSOR_RESULT_TYPE(R, Q)>	transform (const TensorTrain< R > &t, const Tensor< Q > &c)

template<typename R , typename Q >
TensorTrain< TENSOR_RESULT_TYPE(R, Q)>	transform_dir (const TensorTrain< R > &t, const Tensor< Q > &c, const int axis)

Additional Inherited Members
Static Public Member Functions inherited from madness::BaseTensor
static int	get_instance_count ()
	Returns the count of all current instances of tensors & slice tensors of all types. More...

Protected Member Functions inherited from madness::BaseTensor
void	cycledim_inplace (long shift, long start, long end)
	Cyclic shift of dimensions. More...

void	flat_inplace ()
	Reshapes the tensor inplace into 1D. More...

void	fusedim_inplace (long i)
	Fuses dimensions `i` and `i+1`. More...

void	mapdim_inplace (const std::vector< long > &map)
	General permutation of dimensions. More...

void	reshape_inplace (const std::vector< long > &d)
	Reshapes the tensor inplace. More...

void	reshape_inplace (int ndimnew, const long *d)
	Reshapes the tensor inplace. More...

void	set_dims_and_size (long nd, const long d[])

void	splitdim_inplace (long i, long dimi0, long dimi1)
	Splits dimension `i`. More...

void	swapdim_inplace (long i, long j)
	Swaps the dimensions. More...

Protected Attributes inherited from madness::BaseTensor
long	_dim [TENSOR_MAXDIM]
	Size of each dimension. More...

long	_id
	Id from TensorTypeData<T> in type_data.h. More...

long	_ndim
	Number of dimensions (-1=invalid; 0=no supported; >0=tensor) More...

long	_size
	Number of elements in the tensor. More...

long	_stride [TENSOR_MAXDIM]
	Increment between elements in each dimension. More...

Detailed Description

template<typename T>
class madness::TensorTrain< T >

A tensor train is a multi-modal representation of a tensor t

t(i,j,k,l) = \sum G^1_{a1,i,a2} G^2_{a2,j,a3} G^3_{a3,k,a4} G^4_{a4,l,a5}

G

Definition: test_derivative.cc:24

madness::_

static const Slice _(0,-1, 1)

madness::sum

Function< T, NDIM > sum(World &world, const std::vector< Function< T, NDIM > > &f, bool fence=true)

Returns new function — q = sum_i f[i].

Definition: vmra.h:1421

k

static const long k

Definition: rk.cc:44

a2

const double a2

Definition: vnucso.cc:86

a1

const double a1

Definition: vnucso.cc:85

The "core" tensors G are connected via a linear index network, where the first index a1 and the last index a5 are boundary indices and are set to 1.

The tensor train representation is suited for any number of dimensions and in general at least as fast as the 2-way decomposition SVD. If the tensor has full rank it will need about twice the storage space of the full tensor

Member Typedef Documentation

◆ float_scalar_type

template<typename T >

typedef TensorTypeData<T>::float_scalar_type madness::TensorTrain< T >::float_scalar_type

private

C++ typename of the floating point type associated with scalar real type.

◆ scalar_type

template<typename T >

typedef TensorTypeData<T>::scalar_type madness::TensorTrain< T >::scalar_type

private

C++ typename of the real type associated with a complex type.

◆ type

template<typename T >

typedef T madness::TensorTrain< T >::type

private

C++ typename of this tensor.

Constructor & Destructor Documentation

◆ TensorTrain() [1/7]

template<typename T >

madness::TensorTrain< T >::TensorTrain ( )

inline

empty constructor

References madness::TensorTrain< T >::set_size_and_dim().

◆ TensorTrain() [2/7]

template<typename T >

madness::TensorTrain< T >::TensorTrain	(	const Tensor< T > &	t,
		double	eps
	)

inline

ctor for a TensorTrain, with the tolerance eps

The tensor train will represent the input tensor with accuracy || t - this ||_2 < eps

Note that we rely on specific layout of the memory in the tensors, e.g. we pass SliceTensors on to lapack. This will only work if the slices are contiguous.

Parameters

[in]	t	full representation of a tensor
[in]	eps	the accuracy threshold

References d(), madness::TensorTrain< T >::decompose(), madness::BaseTensor::dim(), madness::BaseTensor::dims(), madness::Tensor< T >::flat(), MADNESS_ASSERT, madness::BaseTensor::ndim(), madness::BaseTensor::set_dims_and_size(), and madness::BaseTensor::size().

◆ TensorTrain() [3/7]

template<typename T >

madness::TensorTrain< T >::TensorTrain	(	const Tensor< T > &	t,
		double	eps,
		const std::vector< long >	dims
	)

inline

ctor for a TensorTrain, with the tolerance eps

The tensor train will represent the input tensor with accuracy || t - this ||_2 < eps

Note that we rely on specific layout of the memory in the tensors, e.g. we pass SliceTensors on to lapack. This will only work if the slices are contiguous.

Parameters

[in]	t	full representation of a tensor
[in]	eps	the accuracy threshold
[in]	dims	the tt structure

References madness::TensorTrain< T >::decompose(), madness::BaseTensor::dims(), MADNESS_ASSERT, madness::BaseTensor::ndim(), madness::BaseTensor::set_dims_and_size(), and madness::BaseTensor::size().

◆ TensorTrain() [4/7]

template<typename T >

madness::TensorTrain< T >::TensorTrain	(	const long &	ndims,
		const long *	dims
	)

inline

ctor for a TensorTrain, set up only the dimensions, no data

References madness::BaseTensor::dims(), madness::BaseTensor::set_dims_and_size(), and madness::TensorTrain< T >::zero_rank.

◆ TensorTrain() [5/7]

template<typename T >

madness::TensorTrain< T >::TensorTrain ( const std::vector< long > & dims )

inline

ctor for a TensorTrain, set up only the dimensions, no data

References madness::BaseTensor::dims(), madness::TensorTrain< T >::set_size_and_dim(), and madness::TensorTrain< T >::zero_rank.

◆ TensorTrain() [6/7]

template<typename T >

madness::TensorTrain< T >::TensorTrain ( const std::vector< Tensor< T > > & c )

inline

ctor for a TensorTrain, with core tensors explicitly given

the core tensors must have the shape (k1,r1) (r1,k2,r2) .. (r2,k3)

Parameters

[in] core vector of core tensors, properly shaped

References c, madness::TensorTrain< T >::core, d(), madness::BaseTensor::dim(), madness::TensorTrain< T >::set_size_and_dim(), madness::TensorTrain< T >::zero_me(), and madness::TensorTrain< T >::zero_rank.

◆ TensorTrain() [7/7]

template<typename T >

madness::TensorTrain< T >::TensorTrain ( const TensorTrain< T > & other )

inline

copy constructor, shallow

References madness::BaseTensor::dims(), madness::BaseTensor::ndim(), and madness::BaseTensor::set_dims_and_size().

Member Function Documentation

◆ decompose()

template<typename T >

void madness::TensorTrain< T >::decompose	(	const Tensor< T > &	t,
		double	eps,
		const std::vector< long > &	dims
	)

inline

decompose the input tensor into a TT representation

Parameters

[in]	t	tensor in full rank
[in]	eps	the precision threshold
[in]	dims	the tt structure

References madness::_(), aa, b, c, madness::copy(), madness::TensorTrain< T >::copy, madness::TensorTrain< T >::core, d(), madness::BaseTensor::dims(), k, max, madness::SRConf< T >::max_sigma(), madness::print(), madness::BaseTensor::size(), madness::svd_result(), T(), u(), and madness::TensorTrain< T >::zero_rank.

Referenced by madness::TensorTrain< T >::TensorTrain().

◆ emul()

template<typename T >

TensorTrain<T>& madness::TensorTrain< T >::emul ( const TensorTrain< T > & other )

inline

compute the Hadamard product of two TensorTrains

see Sec. 4.2 of the TT paper

Returns: *this with the ranks of the input tensors multiplied!

References madness::_(), a1, madness::TensorTrain< T >::copy, madness::TensorTrain< T >::core, madness::BaseTensor::dim(), madness::BaseTensor::dims(), madness::TensorTrain< T >::fusedim(), k, MADNESS_ASSERT, MADNESS_EXCEPTION, madness::BaseTensor::ndim(), madness::TensorTrain< T >::outer, madness::TensorTrain< T >::ranks(), madness::Tensor< T >::reshape(), madness::TensorTrain< T >::size(), and madness::TensorTrain< T >::zero_rank.

◆ fusedim()

template<typename T >

void madness::TensorTrain< T >::fusedim ( const long i )

inline

merge two dimensions into one

merge dimension i and i+1 into new dimension i

Parameters

[in] i the first dimension

(r1=0, k1*k2, r2=0)

References madness::TensorTrain< T >::core, d(), madness::BaseTensor::dim(), madness::BaseTensor::fusedim_inplace(), madness::inner(), and madness::TensorTrain< T >::zero_rank.

Referenced by madness::TensorTrain< T >::emul(), and madness::TensorTrain< T >::make_tensor().

◆ gaxpy() [1/2]

template<typename T >

TensorTrain<T>& madness::TensorTrain< T >::gaxpy	(	const std::array< Slice, TENSOR_MAXDIM > &	s1,
		const TensorTrain< T > &	rhs,
		T	beta,
		const std::array< Slice, TENSOR_MAXDIM > &	s2
	)

inline

Inplace generalized saxpy with slices and without alpha.

return this = this(s1) + other(s2) * beta

References madness::_(), beta, madness::TensorTrain< T >::core, d(), madness::BaseTensor::dim(), k, MADNESS_ASSERT, MADNESS_EXCEPTION, madness::BaseTensor::ndim(), madness::TensorTrain< T >::verify(), and madness::TensorTrain< T >::zero_rank.

◆ gaxpy() [2/2]

template<typename T >

TensorTrain<T>& madness::TensorTrain< T >::gaxpy	(	T	alpha,
		const TensorTrain< T > &	rhs,
		T	beta
	)

inline

Inplace generalized saxpy ... this = this*alpha + other*beta.

References madness::_(), alpha, beta, madness::TensorTrain< T >::core, d(), k, MADNESS_ASSERT, MADNESS_EXCEPTION, madness::BaseTensor::ndim(), madness::TensorTrain< T >::scale(), madness::TensorTrain< T >::verify(), and madness::TensorTrain< T >::zero_rank.

Referenced by madness::TensorTrain< T >::operator+=(), and madness::TensorTrain< T >::operator-=().

◆ get_core() [1/2]

template<typename T >

Tensor<T>& madness::TensorTrain< T >::get_core ( const int i )

inline

reference to the internal core

References madness::TensorTrain< T >::core, MADNESS_ASSERT, and madness::BaseTensor::ndim().

Referenced by madness::apply(), and madness::tt_identity().

◆ get_core() [2/2]

template<typename T >

const Tensor<T>& madness::TensorTrain< T >::get_core ( const int i ) const

inline

const reference to the internal core

References madness::TensorTrain< T >::core, MADNESS_ASSERT, and madness::BaseTensor::ndim().

◆ is_operator()

template<typename T >

bool madness::TensorTrain< T >::is_operator ( ) const

inline

check if this is an operator (r,k',k,r)

References madness::TensorTrain< T >::is_tensor().

Referenced by madness::TensorTrain< T >::make_operator().

◆ is_tensor()

template<typename T >

bool madness::TensorTrain< T >::is_tensor ( ) const

inline

check if this is a tensor (r,k,r)

References madness::TensorTrain< T >::core, and madness::BaseTensor::ndim().

Referenced by madness::TensorTrain< T >::is_operator(), and madness::TensorTrain< T >::make_tensor().

◆ is_zero_rank()

template<typename T >

bool madness::TensorTrain< T >::is_zero_rank ( ) const

inline

if rank is zero

References madness::TensorTrain< T >::zero_rank.

◆ make_operator()

template<typename T >

TensorTrain<T>& madness::TensorTrain< T >::make_operator ( )

inline

convert this into an operator representation (r,k',k,r)

References madness::TensorTrain< T >::core, madness::TensorTrain< T >::is_operator(), k0, k2, MADNESS_ASSERT, and madness::BaseTensor::ndim().

Referenced by madness::SeparatedConvolution< Q, NDIM >::make_tt_representation().

◆ make_tensor()

template<typename T >

TensorTrain<T>& madness::TensorTrain< T >::make_tensor ( )

inline

convert this into a tensor representation (r,k,r)

References madness::TensorTrain< T >::core, madness::TensorTrain< T >::fusedim(), madness::TensorTrain< T >::is_tensor(), and madness::BaseTensor::ndim().

Referenced by madness::SeparatedConvolution< Q, NDIM >::make_tt_representation().

◆ normf()

template<typename T >

float_scalar_type madness::TensorTrain< T >::normf ( ) const

inline

returns the Frobenius norm

References std::abs(), and madness::TensorTrain< T >::trace().

◆ operator TensorTrain< Q >()

template<typename T >

template<class Q >

madness::TensorTrain< T >::operator TensorTrain< Q > ( ) const

inline

Type conversion makes a deep copy.

References c, madness::TensorTrain< T >::core, madness::BaseTensor::dims(), and madness::TensorTrain< T >::zero_rank.

◆ operator*()

template<typename T >

TensorTrain<T> madness::TensorTrain< T >::operator* ( const T & factor ) const

inline

return this multiplied by a scalar

Returns: new tensor

References madness::TensorTrain< T >::copy, and madness::TensorTrain< T >::scale().

◆ operator+=()

template<typename T >

TensorTrain<T>& madness::TensorTrain< T >::operator+= ( const TensorTrain< T > & rhs )

inline

inplace addition of two Tensortrains; will increase ranks of this

inefficient if many additions are performed, since it requires many calls of new.

Parameters

[in] rhs a TensorTrain to be added

References madness::TensorTrain< T >::gaxpy().

◆ operator-=()

template<typename T >

TensorTrain<T>& madness::TensorTrain< T >::operator-= ( const TensorTrain< T > & rhs )

inline

inplace subtraction of two Tensortrains; will increase ranks of this

inefficient if many subtractions are performed, since it requires many calls of new.

Parameters

[in] rhs a TensorTrain to be added

References madness::TensorTrain< T >::gaxpy().

◆ operator=() [1/2]

template<typename T >

TensorTrain<T>& madness::TensorTrain< T >::operator= ( const T & number )

inline

assign a number to this tensor

References madness::TensorTrain< T >::core, madness::BaseTensor::dim(), madness::BaseTensor::ndim(), and madness::TensorTrain< T >::zero_rank.

◆ operator=() [2/2]

template<typename T >

TensorTrain& madness::TensorTrain< T >::operator= ( const TensorTrain< T > & other )

inline

assigment operator

References madness::TensorTrain< T >::core, madness::BaseTensor::dims(), madness::BaseTensor::ndim(), madness::BaseTensor::set_dims_and_size(), and madness::TensorTrain< T >::zero_rank.

◆ ptr() [1/2]

template<typename T >

T* madness::TensorTrain< T >::ptr ( const int ivec = 0 )

inline

Returns a pointer to the internal data.

Parameters

[in] ivec index of core vector to which the return values points

References madness::TensorTrain< T >::core.

◆ ptr() [2/2]

template<typename T >

const T* madness::TensorTrain< T >::ptr ( const int ivec = 0 ) const

inline

Returns a pointer to the internal data.

Parameters

[in] ivec index of core vector to which the return values points

References madness::TensorTrain< T >::core.

◆ ranks() [1/2]

template<typename T >

std::vector<long> madness::TensorTrain< T >::ranks ( ) const

inline

return the TT ranks

References madness::TensorTrain< T >::core, madness::BaseTensor::dim(), madness::BaseTensor::ndim(), and madness::TensorTrain< T >::zero_rank.

Referenced by madness::apply(), madness::TensorTrain< T >::emul(), madness::TensorTrain< T >::splitdim(), test_stuff(), and madness::TensorTrain< T >::truncate().

◆ ranks() [2/2]

template<typename T >

long madness::TensorTrain< T >::ranks ( const int i ) const

inline

return the TT ranks for dimension i (to i+1)

References madness::TensorTrain< T >::core, MADNESS_EXCEPTION, madness::BaseTensor::ndim(), madness::print(), and madness::TensorTrain< T >::zero_rank.

◆ real_size()

template<typename T >

long madness::TensorTrain< T >::real_size ( ) const

inline

return the size of this instance, including static memory for vectors and such

References madness::TensorTrain< T >::core, madness::TensorTrain< T >::size(), and T().

◆ reconstruct()

template<typename T >

Tensor<T> madness::TensorTrain< T >::reconstruct ( const bool flat = false ) const

inline

reconstruct this to a full representation

Parameters

[in] flat return this in flat representation

Returns: this in full rank representation

References madness::BaseTensor::_size, madness::TensorTrain< T >::core, madness::BaseTensor::dim(), madness::BaseTensor::dims(), madness::Tensor< T >::fusedim(), madness::inner(), MADNESS_ASSERT, madness::BaseTensor::ndim(), madness::TensorTrain< T >::size(), and madness::TensorTrain< T >::zero_rank.

Referenced by test_stuff().

◆ scale()

template<typename T >

void madness::TensorTrain< T >::scale ( T fac )

inline

scale this by a number

Parameters

[in] fac the factor to multiply

Returns: *this * fac

References madness::TensorTrain< T >::core, and madness::TensorTrain< T >::zero_rank.

Referenced by madness::TensorTrain< T >::gaxpy(), and madness::TensorTrain< T >::operator*().

◆ serialize()

template<typename T >

template<typename Archive >

void madness::TensorTrain< T >::serialize ( Archive & ar )

inline

serialize this

References madness::BaseTensor::_dim, madness::BaseTensor::_id, madness::BaseTensor::_ndim, madness::BaseTensor::_size, madness::BaseTensor::_stride, madness::TensorTrain< T >::core, TENSOR_MAXDIM, madness::archive::wrap(), and madness::TensorTrain< T >::zero_rank.

◆ set_size_and_dim()

template<typename T >

void madness::TensorTrain< T >::set_size_and_dim ( std::vector< long > dims )

inline

References madness::BaseTensor::dims(), and madness::BaseTensor::set_dims_and_size().

Referenced by madness::TensorTrain< T >::TensorTrain().

◆ size()

template<typename T >

long madness::TensorTrain< T >::size ( ) const

inline

return the number of coefficients in all core tensors

References madness::TensorTrain< T >::core, and madness::TensorTrain< T >::zero_rank.

Referenced by madness::TensorTrain< T >::emul(), madness::TensorTrain< T >::real_size(), madness::TensorTrain< T >::reconstruct(), madness::TensorTrain< T >::truncate(), and madness::TensorTrain< T >::verify().

◆ splitdim()

template<typename T >

TensorTrain<T> madness::TensorTrain< T >::splitdim	(	long	idim,
		long	k1,
		long	k2,
		const double	eps
	)		const

inline

Returns new view/tensor splitting dimension i as dimi0*dimi1 to produce conforming d+1 dimension tensor

Parameters

[in]	idim	the dimension to be split
[in]	k1	new first dimension of idim
[in]	k2	new second dimension of idim
[in]	eps	threshold for SVD (choose negative to keep all terms)

Returns: new deep copy of this with split dimensions

References madness::TensorTrain< T >::TensorTrain, madness::_(), madness::TensorTrain< T >::copy, madness::TensorTrain< T >::core, madness::BaseTensor::dim(), madness::BaseTensor::dims(), k1, k2, MADNESS_ASSERT, madness::SRConf< T >::max_sigma(), madness::BaseTensor::ndim(), madness::r2(), madness::TensorTrain< T >::ranks(), madness::BaseTensor::splitdim_inplace(), madness::svd(), and madness::TensorTrain< T >::zero_rank.

◆ trace() [1/2]

template<typename T >

template<class Q >

std::enable_if<(TensorTypeData<T>::iscomplex or TensorTypeData<Q>::iscomplex), TENSOR_RESULT_TYPE(T,Q)>::type madness::TensorTrain< T >::trace ( const TensorTrain< Q > & B ) const

inline

References MADNESS_EXCEPTION.

Referenced by madness::TensorTrain< T >::normf().

◆ trace() [2/2]

template<typename T >

template<class Q >

std::enable_if<!(TensorTypeData<T>::iscomplex or TensorTypeData<Q>::iscomplex), TENSOR_RESULT_TYPE(T,Q)>::type madness::TensorTrain< T >::trace ( const TensorTrain< Q > & B ) const

inline

Return the trace of two tensors, no complex conjugate involved.

Returns: <this | B>

References d(), madness::BaseTensor::dim(), madness::Tensor< T >::fusedim(), madness::inner_result(), MADNESS_ASSERT, MADNESS_EXCEPTION, max, madness::BaseTensor::ndim(), Q(), madness::Tensor< T >::reshape(), T(), madness::TENSOR_RESULT_TYPE(), and madness::Tensor< T >::trace().

◆ truncate() [1/2]

template<typename T >

template<typename R = T>

std::enable_if<!std::is_arithmetic<R>::value, void>::type madness::TensorTrain< T >::truncate ( double eps )

inline

recompress and truncate this TT representation

this in recompressed TT form with optimal rank

Parameters

[in] eps the truncation threshold

References MADNESS_EXCEPTION.

Referenced by madness::SeparatedConvolution< Q, NDIM >::make_tt_representation().

◆ truncate() [2/2]

template<typename T >

template<typename R = T>

std::enable_if<std::is_arithmetic<R>::value, void>::type madness::TensorTrain< T >::truncate ( double eps )

inline

recompress and truncate this TT representation

this in recompressed TT form with optimal rank

Parameters

[in] eps the truncation threshold

References madness::_(), madness::copy(), madness::TensorTrain< T >::core, d(), madness::BaseTensor::dim(), madness::Tensor< T >::flat(), madness::inner(), k, madness::kmax, L, madness::lq_result(), m, MADNESS_EXCEPTION, max, madness::SRConf< T >::max_sigma(), madness::BaseTensor::ndim(), madness::TensorTrain< T >::ranks(), madness::Tensor< T >::reshape(), madness::TensorTrain< T >::size(), madness::svd_result(), TENSOR_MAXDIM, madness::TensorTrain< T >::verify(), madness::TensorTrain< T >::zero_me(), and madness::TensorTrain< T >::zero_rank.

◆ two_mode_representation()

template<typename T >

void madness::TensorTrain< T >::two_mode_representation	(	Tensor< T > &	U,
		Tensor< T > &	VT,
		Tensor< typename Tensor< T >::scalar_type > &	s
	)		const

inline

construct a two-mode representation (aka unnormalized SVD)

Parameters

[out]	U	The left singular vectors, ({i},rank)
[out]	VT	The right singular vectors, (rank,{i})
[out]	s	Vector holding 1's, (rank)

References madness::TensorTrain< T >::core, madness::BaseTensor::dim(), madness::inner(), MADNESS_ASSERT, MADNESS_EXCEPTION, madness::BaseTensor::ndim(), and madness::TensorTrain< T >::zero_rank.

◆ verify()

template<typename T >

bool madness::TensorTrain< T >::verify ( ) const

inline

return the dimensions of this tensor

check that the ranks of all core tensors are consistent

References c, madness::TensorTrain< T >::core, d(), madness::BaseTensor::dim(), madness::BaseTensor::ndim(), and madness::TensorTrain< T >::size().

Referenced by madness::TensorTrain< T >::gaxpy(), and madness::TensorTrain< T >::truncate().

◆ zero_me() [1/2]

template<typename T >

void madness::TensorTrain< T >::zero_me ( )

inline

turn this into an empty tensor with all cores properly shaped

References madness::BaseTensor::dims(), and madness::BaseTensor::ndim().

Referenced by madness::TensorTrain< T >::TensorTrain(), and madness::TensorTrain< T >::truncate().

◆ zero_me() [2/2]

template<typename T >

void madness::TensorTrain< T >::zero_me ( const std::vector< long > & dim )

inline

turn this into an empty tensor with all cores properly shaped

References madness::BaseTensor::dim().

Friends And Related Function Documentation

◆ TensorTrain

template<typename T >

template<typename Q >

friend class TensorTrain

friend

Referenced by madness::TensorTrain< T >::splitdim().

◆ copy [1/2]

template<typename T >

TensorTrain copy ( const TensorTrain< T > & other )

friend

deep copy of the whole tensor

if argument has zero rank return a zero-rank tensor of the same dimensions

Referenced by madness::TensorTrain< T >::decompose(), madness::TensorTrain< T >::emul(), madness::TensorTrain< T >::operator*(), and madness::TensorTrain< T >::splitdim().

◆ copy [2/2]

template<typename T >

TensorTrain copy	(	const TensorTrain< T > &	other,
		const std::array< Slice, TENSOR_MAXDIM > &	s
	)

friend

deep copy of a slice of the tensor

this operation does not change the ranks, i.e. the resulting tensor is most likely not in an optimal compression state

Parameters

[in]	other	tensor to be sliced
[in]	s	vector of slices

◆ general_transform

template<typename T >

template<typename R , typename Q >

TensorTrain<TENSOR_RESULT_TYPE(R,Q)> general_transform	(	const TensorTrain< R > &	t,
		const Tensor< Q >	c[]
	)

friend

◆ outer

template<typename T >

template<typename R , typename Q >

TensorTrain<TENSOR_RESULT_TYPE(R,Q)> outer	(	const TensorTrain< R > &	t1,
		const TensorTrain< Q > &	t2
	)

friend

Referenced by madness::TensorTrain< T >::emul().

◆ transform

template<typename T >

template<typename R , typename Q >

TensorTrain<TENSOR_RESULT_TYPE(R,Q)> transform	(	const TensorTrain< R > &	t,
		const Tensor< Q > &	c
	)

friend

◆ transform_dir

template<typename T >

template<typename R , typename Q >

TensorTrain<TENSOR_RESULT_TYPE(R,Q)> transform_dir	(	const TensorTrain< R > &	t,
		const Tensor< Q > &	c,
		const int	axis
	)

friend

Member Data Documentation

◆ core

template<typename T >

std::vector<Tensor<T> > madness::TensorTrain< T >::core

private

holding the core tensors of a tensor train the tensors have the shape (k,r0) (r0,k,r1) (r1,k,r2) .. (rn-1,k)

◆ zero_rank

template<typename T >

bool madness::TensorTrain< T >::zero_rank

private

The documentation for this class was generated from the following file:

tensortrain.h

Public Member Functions

Private Types

Private Attributes

Friends

Additional Inherited Members

Detailed Description

template<typename T> class madness::TensorTrain< T >

Member Typedef Documentation

◆ float_scalar_type

◆ scalar_type

◆ type

Constructor & Destructor Documentation

◆ TensorTrain() [1/7]

◆ TensorTrain() [2/7]

◆ TensorTrain() [3/7]

◆ TensorTrain() [4/7]

◆ TensorTrain() [5/7]

◆ TensorTrain() [6/7]

◆ TensorTrain() [7/7]

Member Function Documentation

◆ decompose()

◆ emul()

◆ fusedim()

◆ gaxpy() [1/2]

◆ gaxpy() [2/2]

◆ get_core() [1/2]

◆ get_core() [2/2]

◆ is_operator()

◆ is_tensor()

◆ is_zero_rank()

◆ make_operator()

◆ make_tensor()

◆ normf()

◆ operator TensorTrain< Q >()

◆ operator*()

◆ operator+=()

◆ operator-=()

◆ operator=() [1/2]

◆ operator=() [2/2]

◆ ptr() [1/2]

◆ ptr() [2/2]

◆ ranks() [1/2]

◆ ranks() [2/2]

◆ real_size()

◆ reconstruct()

◆ scale()

◆ serialize()

◆ set_size_and_dim()

◆ size()

◆ splitdim()

◆ trace() [1/2]

◆ trace() [2/2]

◆ truncate() [1/2]

◆ truncate() [2/2]

◆ two_mode_representation()

◆ verify()

◆ zero_me() [1/2]

◆ zero_me() [2/2]

Friends And Related Function Documentation

◆ TensorTrain

◆ copy [1/2]

◆ copy [2/2]

◆ general_transform

◆ outer

◆ transform

◆ transform_dir

Member Data Documentation

◆ core

◆ zero_rank

template<typename T>
class madness::TensorTrain< T >