Detailed Description

template<typename Symmetry, typename Scalar, typename MpoScalar = double>
class UmpsCompressor< Symmetry, Scalar, MpoScalar >

Compressor for uMPS. Needed to obtain various operations containing uMPSs and MPOs with a variational approach.

Template Parameters

Symmetry	: A class providing all relevant functions and infos that are determined by the Symmetry. Click here for more information.
Scalar	: `double` or `complex<double>`
MpoScalar	: `double` or `complex<double>`

Note: Until now, only support for compressing a state.

Warning: This is class is nearly untested and probably just works for Lcell=1 because the parallel algorithm implemented for Lcell>1 does not work well.

Longterm Todo:: Implemented a sequential algorithm and test the comporession more seriously.

Definition at line 22 of file UmpsCompressor.h.

#include <UmpsCompressor.h>

Public Member Functions
	UmpsCompressor (DMRG::VERBOSITY::OPTION VERBOSITY=DMRG::VERBOSITY::SILENT)


string	info () const

double	memory (MEMUNIT memunit=GB) const


void	stateCompress (const Umps< Symmetry, Scalar > &Vin, Umps< Symmetry, Scalar > &Vout, size_t Dinit_input, size_t Qinit_input, double tol_input, size_t max_iterations=100ul, size_t min_iterations=10ul)

Private Member Functions
void	optimize_parallel (const Umps< Symmetry, Scalar > &Vin, Umps< Symmetry, Scalar > &Vout)

void	build_cellEnv (const Umps< Symmetry, Scalar > &Vbra, const Umps< Symmetry, Scalar > &Vket)

void	build_LR (const Umps< Symmetry, Scalar > &Vbra, const Umps< Symmetry, Scalar > &Vket)

void	calc_error (const Umps< Symmetry, Scalar > &Vout)

size_t	minus1modL (size_t l) const

Private Attributes
DMRG::VERBOSITY::OPTION	CHOSEN_VERBOSITY

vector< PivotOverlap1< Symmetry, Scalar > >	Env

size_t	N_sites

size_t	N_iterations

size_t	Dinit

size_t	Qinit

double	err_var

double	tol

complex< double >	lambdaL

double	t_fixedL

double	t_fixedR

Constructor & Destructor Documentation

◆ UmpsCompressor()

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

UmpsCompressor< Symmetry, Scalar, MpoScalar >::UmpsCompressor ( DMRG::VERBOSITY::OPTION VERBOSITY = DMRG::VERBOSITY::SILENT )

inline

Definition at line 26 of file UmpsCompressor.h.

Member Function Documentation

◆ build_cellEnv()

template<typename Symmetry , typename Scalar , typename MpoScalar >

void UmpsCompressor< Symmetry, Scalar, MpoScalar >::build_cellEnv	(	const Umps< Symmetry, Scalar > &	Vbra,
		const Umps< Symmetry, Scalar > &	Vket
	)

private

Definition at line 162 of file UmpsCompressor.h.

◆ build_LR()

template<typename Symmetry , typename Scalar , typename MpoScalar >

void UmpsCompressor< Symmetry, Scalar, MpoScalar >::build_LR	(	const Umps< Symmetry, Scalar > &	Vbra,
		const Umps< Symmetry, Scalar > &	Vket
	)

private

Definition at line 186 of file UmpsCompressor.h.

◆ calc_error()

template<typename Symmetry , typename Scalar , typename MpoScalar >

void UmpsCompressor< Symmetry, Scalar, MpoScalar >::calc_error ( const Umps< Symmetry, Scalar > & Vout )

private

Definition at line 260 of file UmpsCompressor.h.

◆ info()

template<typename Symmetry , typename Scalar , typename MpoScalar >

string UmpsCompressor< Symmetry, Scalar, MpoScalar >::info

Prints some info about this class.

Definition at line 84 of file UmpsCompressor.h.

◆ memory()

template<typename Symmetry , typename Scalar , typename MpoScalar >

double UmpsCompressor< Symmetry, Scalar, MpoScalar >::memory ( MEMUNIT memunit = GB ) const

Calculates the (theoretically) allocated memory (note: by default in GB).

Definition at line 110 of file UmpsCompressor.h.

◆ minus1modL()

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

size_t UmpsCompressor< Symmetry, Scalar, MpoScalar >::minus1modL ( size_t l ) const

inlineprivate

Safely calculates $l-1 mod L$ without overflow for size_t.

Definition at line 72 of file UmpsCompressor.h.

◆ optimize_parallel()

template<typename Symmetry , typename Scalar , typename MpoScalar >

void UmpsCompressor< Symmetry, Scalar, MpoScalar >::optimize_parallel	(	const Umps< Symmetry, Scalar > &	Vin,
		Umps< Symmetry, Scalar > &	Vout
	)

private

Definition at line 218 of file UmpsCompressor.h.

◆ stateCompress()

template<typename Symmetry , typename Scalar , typename MpoScalar >

void UmpsCompressor< Symmetry, Scalar, MpoScalar >::stateCompress	(	const Umps< Symmetry, Scalar > &	Vin,
		Umps< Symmetry, Scalar > &	Vout,
		size_t	Dinit_input,
		size_t	Qinit_input,
		double	tol_input,
		size_t	max_iterations = `100ul`,
		size_t	min_iterations = `10ul`
	)

Compresses a given uMps $V_{out} \approx V_{in}$ .

Parameters

[in]	Vin	: input state to be compressed
[out]	Vout	: compressed output state
[in]	Dinit_input	: matrix size cutoff per site and subspace for `Vout`
[in]	Qinit_input	: symmetry block cutoff per site for `Vout`
[in]	tol_input	: tolerance for the variational error
[in]	max_iterations	: maximal amount of iterations; break if exceeded
[in]	min_iterations	: minimal amount of iterations

Definition at line 128 of file UmpsCompressor.h.

Member Data Documentation

◆ CHOSEN_VERBOSITY

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

DMRG::VERBOSITY::OPTION UmpsCompressor< Symmetry, Scalar, MpoScalar >::CHOSEN_VERBOSITY

private

Definition at line 59 of file UmpsCompressor.h.

◆ Dinit

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

size_t UmpsCompressor< Symmetry, Scalar, MpoScalar >::Dinit

private

Definition at line 76 of file UmpsCompressor.h.

◆ Env

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

vector<PivotOverlap1<Symmetry,Scalar> > UmpsCompressor< Symmetry, Scalar, MpoScalar >::Env

private

Definition at line 62 of file UmpsCompressor.h.

◆ err_var

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

double UmpsCompressor< Symmetry, Scalar, MpoScalar >::err_var

private

Definition at line 77 of file UmpsCompressor.h.

◆ lambdaL

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

complex<double> UmpsCompressor< Symmetry, Scalar, MpoScalar >::lambdaL

private

Definition at line 78 of file UmpsCompressor.h.

◆ N_iterations

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

size_t UmpsCompressor< Symmetry, Scalar, MpoScalar >::N_iterations

private

Definition at line 75 of file UmpsCompressor.h.

◆ N_sites

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

size_t UmpsCompressor< Symmetry, Scalar, MpoScalar >::N_sites

private

Definition at line 74 of file UmpsCompressor.h.

◆ Qinit

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

size_t UmpsCompressor< Symmetry, Scalar, MpoScalar >::Qinit

private

Definition at line 76 of file UmpsCompressor.h.

◆ t_fixedL

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

double UmpsCompressor< Symmetry, Scalar, MpoScalar >::t_fixedL

private

Definition at line 79 of file UmpsCompressor.h.

◆ t_fixedR

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

double UmpsCompressor< Symmetry, Scalar, MpoScalar >::t_fixedR

private

Definition at line 80 of file UmpsCompressor.h.

◆ tol

template<typename Symmetry , typename Scalar , typename MpoScalar = double>

double UmpsCompressor< Symmetry, Scalar, MpoScalar >::tol

private

Definition at line 77 of file UmpsCompressor.h.

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

UmpsCompressor.h

Detailed Description

Public Member Functions

Private Member Functions

Private Attributes

Constructor & Destructor Documentation

◆ UmpsCompressor()

Member Function Documentation

◆ build_cellEnv()

◆ build_LR()

◆ calc_error()

◆ info()

◆ memory()

◆ minus1modL()

◆ optimize_parallel()

◆ stateCompress()

Member Data Documentation

◆ CHOSEN_VERBOSITY

◆ Dinit

◆ Env

◆ err_var

◆ lambdaL

◆ N_iterations

◆ N_sites

◆ Qinit

◆ t_fixedL

◆ t_fixedR

◆ tol