sc::LMP2 Class Reference

Computes the local second order perturbation theory energy. More...

#include <chemistry/qc/lmp2/lmp2.h>

Inheritance diagram for sc::LMP2:

Public Member Functions
	LMP2 (const sc::Ref< sc::KeyVal > &)
	Construct an LMP2 object from KeyVal input. More...
	LMP2 (sc::StateIn &)
void	save_data_state (sc::StateOut &)
	Save the base classes (with save_data_state) and the members in the same order that the StateIn CTOR initializes them. More...
void	compute (void)
	Recompute at least the results that have compute true and are not already computed. More...
int	nelectron (void)
	Returns the number of electrons.
sc::RefSymmSCMatrix	density (void)
	Returns the SO density.
double	magnetic_moment () const
	Computes the S (or J) magnetic moment of the target state(s), in units of . More...
int	value_implemented (void) const
Public Member Functions inherited from sc::LCorr
	LCorr (const sc::Ref< sc::KeyVal > &)
	Construct an LCorr object from KeyVal input. More...
	LCorr (sc::StateIn &)
Public Member Functions inherited from sc::Wavefunction
	Wavefunction (StateIn &)
	Wavefunction (const Ref< KeyVal > &)
	The KeyVal constructor. More...
void	save_data_state (StateOut &)
	Save the base classes (with save_data_state) and the members in the same order that the StateIn CTOR initializes them. More...
double	density (const SCVector3 &)
double	density_gradient (const SCVector3 &, double *)
double	natural_orbital (const SCVector3 &r, int iorb)
double	natural_orbital_density (const SCVector3 &r, int orb, double *orbval=0)
double	orbital (const SCVector3 &r, int iorb, const RefSCMatrix &orbs)
void	orbitals (const SCVector3 &r, const RefSCMatrix &orbs, RefSCVector &values)
double	orbital_density (const SCVector3 &r, int iorb, const RefSCMatrix &orbs, double *orbval=0)
double	total_charge () const
	Returns the total charge of the system.
virtual RefSymmSCMatrix	ao_density ()
	Returns the AO density.
virtual RefSCMatrix	natural_orbitals ()
	Returns the natural orbitals, in SO basis.
virtual RefDiagSCMatrix	natural_density ()
	Returns the natural density (a diagonal matrix).
int	spin_polarized ()
	Return 1 if the magnetic moment != 0.
int	dk () const
	Returns the level the of the Douglas-Kroll approximation.
virtual RefSymmSCMatrix	alpha_density ()
	Return alpha electron densities in the SO basis.
virtual RefSymmSCMatrix	beta_density ()
	Return beta electron densities in the SO basis.
virtual RefSymmSCMatrix	alpha_ao_density ()
	Return alpha electron densities in the AO basis.
virtual RefSymmSCMatrix	beta_ao_density ()
	Return beta electron densities in the AO basis.
virtual RefSCMatrix	nao (double *atom_charges=0)
	returns the ao to nao transformation matrix
virtual RefSymmSCMatrix	overlap ()
	Returns the SO overlap matrix.
virtual RefSymmSCMatrix	core_hamiltonian_for_basis (const Ref< GaussianBasisSet > &bas, const Ref< GaussianBasisSet > &pbas=0)
	Returns the SO core Hamiltonian in the given basis and momentum basis. More...
virtual RefSymmSCMatrix	core_hamiltonian ()
	Returns the SO core Hamiltonian.
RefSymmSCMatrix	core_hamiltonian_nr (const Ref< GaussianBasisSet > &bas)
virtual double	nuclear_repulsion_energy ()
	Returns the nuclear repulsion energy. More...
void	nuclear_repulsion_energy_gradient (double *g)
	Computes the nuclear repulsion gradient. More...
virtual void	nuclear_repulsion_energy_gradient (double **g)
	Computes the nuclear repulsion gradient. More...
RefSCDimension	ao_dimension ()
	Atomic orbital dimension.
RefSCDimension	so_dimension ()
	Symmetry adapted orbital dimension.
RefSCDimension	oso_dimension ()
	Orthogonalized symmetry adapted orbital dimension.
Ref< SCMatrixKit >	basis_matrixkit ()
	Matrix kit for AO, SO, orthogonalized SO, and MO dimensioned matrices.
Ref< Molecule >	molecule () const
	Returns the Molecule.
Ref< GaussianBasisSet >	basis () const
	Returns the basis set.
Ref< GaussianBasisSet >	momentum_basis () const
	Returns the basis used for p^2 in the DK correction.
Ref< GaussianBasisSet >	atom_basis () const
	Returns the basis set describing the nuclear charge distributions.
const double *	atom_basis_coef () const
	Returns the coefficients of the nuclear charge distribution basis functions.
Ref< Integral >	integral ()
	Returns the integral evaluator.
void	symmetry_changed ()
	Call this if you have changed the molecular symmetry of the molecule contained by this MolecularEnergy.
RefSCMatrix	so_to_orthog_so ()
	Returns a matrix which does the default transform from SO's to orthogonal SO's. More...
RefSCMatrix	so_to_orthog_so_inverse ()
	Returns the inverse of the transformation returned by so_to_orthog_so.
OverlapOrthog::OrthogMethod	orthog_method () const
	Returns the orthogonalization method.
virtual void	set_orthog_method (const OverlapOrthog::OrthogMethod &)
	(Re)Sets the orthogonalization method and makes this obsolete. More...
double	lindep_tol () const
	Returns the tolerance for linear dependencies.
void	set_lindep_tol (double)
	Re(Sets) the tolerance for linear dependencies.
void	obsolete ()
	Marks all results as being out of date. More...
void	print (std::ostream &=ExEnv::out0()) const
	Print information about the object.
void	writeorbitals ()
	output orbitals to some files to facilitate plotting, with the help of the WriteOrbital class.
Public Member Functions inherited from sc::MolecularEnergy
	MolecularEnergy (const MolecularEnergy &)
	MolecularEnergy (const Ref< KeyVal > &)
	The KeyVal constructor. More...
	MolecularEnergy (StateIn &)
void	set_checkpoint ()
	Set up checkpointing.
void	set_checkpoint_file (const char *)
void	set_checkpoint_freq (int freq)
bool	if_to_checkpoint () const
	Check if need to checkpoint.
const char *	checkpoint_file () const
int	checkpoint_freq () const
MolecularEnergy &	operator= (const MolecularEnergy &)
virtual double	energy ()
	A wrapper around value();.
virtual RefSCDimension	moldim () const
void	guess_hessian (RefSymmSCMatrix &)
	Compute a quick, approximate hessian.
RefSymmSCMatrix	inverse_hessian (RefSymmSCMatrix &)
int	gradient_implemented () const
	Reports whether gradient is implemented either analytically or using MolecularGradient object. More...
int	hessian_implemented () const
	Reports whether hessian is implemented either analytically or using MolecularHessian object. More...
void	set_desired_gradient_accuracy (double acc)
	These functions overload their Function counterparts. More...
void	set_desired_hessian_accuracy (double acc)
void	set_molhess (const Ref< MolecularHessian > &molhess)
	Use this function to provide MolecularHessian object that will be used to compute hessian. More...
const Ref< MolecularHessian > &	molhess () const
RefSymmSCMatrix	hessian ()
	Will throw if hessian_implemented() returns 0.
void	set_molgrad (const Ref< MolecularGradient > &molgrad)
	Use this function to provide MolecularGradient object that will be used to compute gradient. More...
const Ref< MolecularGradient > &	molgrad () const
RefSCVector	gradient ()
	Will throw if gradient_implemented() returns 0.
void	set_x (const RefSCVector &)
	Set and retrieve the coordinate values.
RefSCVector	get_cartesian_x ()
	Return the cartesian coordinates.
RefSCVector	get_cartesian_gradient ()
	Return the cartesian gradient.
RefSymmSCMatrix	get_cartesian_hessian ()
	Return the cartesian hessian.
Ref< MolecularCoor >	molecularcoor ()
Ref< NonlinearTransform >	change_coordinates ()
	An optimizer can call change coordinates periodically to give the function an opportunity to change its coordinate system. More...
virtual void	purge ()
	This function purges any caches of data in MolecularEnergy. More...
const RefSCVector &	electric_field () const
	returns the electric field vector
void	print_natom_3 (const RefSCVector &, const char *t=0, std::ostream &o=ExEnv::out0()) const
	Nicely print n x 3 data that are stored in a vector.
void	print_natom_3 (double **, const char *t=0, std::ostream &o=ExEnv::out0()) const
void	print_natom_3 (double *, const char *t=0, std::ostream &o=ExEnv::out0()) const
Public Member Functions inherited from sc::Function
int	gradient_needed () const
int	do_gradient (int)
virtual double	actual_gradient_accuracy () const
virtual double	desired_gradient_accuracy () const
AccResultRefSCVector &	gradient_result ()
int	hessian_needed () const
int	do_hessian (int)
virtual double	actual_hessian_accuracy () const
virtual double	desired_hessian_accuracy () const
AccResultRefSymmSCMatrix &	hessian_result ()
virtual bool	desired_value_accuracy_set_to_default () const
virtual bool	desired_gradient_accuracy_set_to_default () const
virtual bool	desired_hessian_accuracy_set_to_default () const
RefSCVector	get_x () const
const RefSCVector &	get_x_no_copy () const
void	print_desired_accuracy (std::ostream &=ExEnv::out0()) const
	similar to print(), but only prins desired accuracies
virtual bool	throw_if_tolerance_exceeded () const
	Overridden Compute member.
	Function ()
	Function (StateIn &)
	Function (const Function &)
	Function (const Ref< KeyVal > &, double funcacc=DBL_EPSILON, double gradacc=DBL_EPSILON, double hessacc=DBL_EPSILON)
	The keyval constructor reads the following keywords: More...
virtual	~Function ()
Function &	operator= (const Function &)
Ref< SCMatrixKit >	matrixkit () const
	Return the SCMatrixKit used to construct vectors and matrices.
RefSCDimension	dimension () const
	Return the SCDimension of the problem.
virtual double	value ()
	Return the value of the function.
int	value_needed () const
	Returns nonzero if the current value is not up-to-date.
int	do_value (int)
	If passed a nonzero number, compute the value the next time compute() is called. More...
AccResultdouble &	value_result ()
virtual void	set_desired_value_accuracy (double)
	Set the accuracy to which the value is to be computed.
virtual double	actual_value_accuracy () const
	Return the accuracy with which the value has been computed.
virtual double	desired_value_accuracy () const
	Return the accuracy with which the value is to be computed.
Public Member Functions inherited from sc::SavableState
SavableState &	operator= (const SavableState &)
void	save_state (StateOut &)
	Save the state of the object as specified by the StateOut object. More...
void	save_object_state (StateOut &)
	This can be used for saving state when the exact type of the object is known for both the save and the restore. More...
virtual void	save_vbase_state (StateOut &)
	Save the virtual bases for the object. More...
Public Member Functions inherited from sc::DescribedClass
	DescribedClass (const DescribedClass &)
DescribedClass &	operator= (const DescribedClass &)
ClassDesc *	class_desc () const MPQC__NOEXCEPT
	This returns the unique pointer to the ClassDesc corresponding to the given type_info object. More...
const char *	class_name () const
	Return the name of the object's exact type.
int	class_version () const
	Return the version of the class.
Ref< DescribedClass >	ref ()
	Return this object wrapped up in a Ref smart pointer. More...
Public Member Functions inherited from sc::RefCount
size_t	identifier () const
	Return the unique identifier for this object that can be compared for different objects of different types. More...
int	lock_ptr () const
	Lock this object.
int	unlock_ptr () const
	Unlock this object.
void	use_locks (bool inVal)
	start and stop using locks on this object
refcount_t	nreference () const
	Return the reference count.
refcount_t	reference ()
	Increment the reference count and return the new count.
refcount_t	dereference ()
	Decrement the reference count and return the new count.
int	managed () const
void	unmanage ()
	Turn off the reference counting mechanism for this object. More...

Additional Inherited Members
Static Public Member Functions inherited from sc::Wavefunction
static void	orbitals (const Ref< OrbitalSpace > &orbs, const std::vector< SCVector3 > &r, std::vector< double > &values)
Static Public Member Functions inherited from sc::SavableState
static void	save_state (SavableState *s, StateOut &)
static SavableState *	restore_state (StateIn &si)
	Restores objects saved with save_state. More...
static SavableState *	key_restore_state (StateIn &si, const char *keyword)
	Like restore_state, but keyword is used to override values while restoring.
static SavableState *	dir_restore_state (StateIn &si, const char *objectname, const char *keyword=0)
Protected Member Functions inherited from sc::LCorr
int	n_unique_W () const
sma2::Array< 2 > &	unique_W (const domainmapvirbs_t &i)
std::vector< double > &	unique_eigvals (const domainmapvirbs_t &i)
sma2::Array< 2 > &	unique_F_tilde (const domainmapvirbs_t &i)
void	init_virb_to_bfns (const sma2::Range &vir)
	This must be called before compute_W is called.
void	compute_W (domainmapvirbs_t &virset, const sc::Ref< sc::GaussianBasisSet > &basis, sc::RefSCMatrix &F_vir_mat, sc::RefSCMatrix &S_mat, const sma2::Range &vir, int nocc_act, double bound)
void	transform_array (sma2::Array< 2 > &A, sma2::Array< 2 > &B, sma2::Array< 2 > &C, sma2::Array< 2 > &D, const sc::Ref< sc::MessageGrp > &msg)
	Transform A to D using transformation matrices B,C: D = B^T*A*C.
void	clear ()
	Release stored data.
void	print_parameters () const
	Print input parameters.
Protected Member Functions inherited from sc::Wavefunction
double	min_orthog_res ()
double	max_orthog_res ()
void	copy_orthog_info (const Ref< Wavefunction > &)
Protected Member Functions inherited from sc::MolecularEnergy
void	failure (const char *)
virtual void	set_energy (double)
	This is just a wrapper around set_value().
virtual void	set_gradient (RefSCVector &)
	These are passed gradients and hessian in cartesian coordinates. More...
virtual void	set_hessian (RefSymmSCMatrix &)
void	x_to_molecule ()
void	molecule_to_x ()
virtual bool	analytic_gradient_implemented () const
	must overload this in a derived class if analytic gradient can be computed More...
virtual bool	analytic_hessian_implemented () const
	must overload this in a derived class if analytic hessian can be computed More...
Protected Member Functions inherited from sc::Function
virtual void	set_value (double)
virtual void	set_matrixkit (const Ref< SCMatrixKit > &)
	Set the SCMatrixKit that should be used to construct the requisite vectors and matrices.
virtual void	set_dimension (const RefSCDimension &)
virtual void	set_actual_value_accuracy (double)
virtual void	set_actual_gradient_accuracy (double)
virtual void	set_actual_hessian_accuracy (double)
RefSCVector &	get_x_reference ()
	Get read/write access to the coordinates for modification.
void	do_change_coordinates (const Ref< NonlinearTransform > &)
	Change the coordinate system and apply the given transform to intermediates matrices and vectors.
Protected Member Functions inherited from sc::SavableState
	SavableState (const SavableState &)
	SavableState (StateIn &)
	Each derived class StateIn CTOR handles the restore corresponding to calling save_object_state, save_vbase_state, and save_data_state listed above. More...
Protected Member Functions inherited from sc::RefCount
	RefCount (const RefCount &)
RefCount &	operator= (const RefCount &)
Protected Attributes inherited from sc::Wavefunction
ResultRefSCMatrix	natural_orbitals_
ResultRefDiagSCMatrix	natural_density_
Ref< GaussianBasisSet >	gbs_
int	debug_
Protected Attributes inherited from sc::MolecularEnergy
Ref< PointGroup >	initial_pg_
int	print_molecule_when_changed_
Protected Attributes inherited from sc::Function
Ref< SCMatrixKit >	matrixkit_
	Used to construct new matrices.
RefSCVector	x_
	The variables.
RefSCDimension	dim_
	The dimension of x_.
AccResultdouble	value_
	The value of the function at x_.
AccResultRefSCVector	gradient_
	The gradient at x_.
AccResultRefSymmSCMatrix	hessian_
	The hessian at x_.
bool	desired_value_accuracy_set_to_default_
bool	desired_gradient_accuracy_set_to_default_
bool	desired_hessian_accuracy_set_to_default_
bool	throw_if_tolerance_exceeded_

Detailed Description

Computes the local second order perturbation theory energy.

This class implements a massively parallel algorithm for the quantum mechanical computation of energies of molecular systems using local second-order Møller-Plesset (LMP2) perturbation theory. It is built upon the Massively Parallel Quantum Chemistry program (MPQC). Both the storage requirement and the computational time scale linearly with the molecular size. High parallel efficiency of the algorithm has been demonstrated for applications employing up to 100 processors.

The parallel algorithm is designed to be scalable, employing a distributed data scheme for the two-electron integrals, avoiding communication bottlenecks, and distributing tasks in all computationally significant steps. A sparse data representation and a set of generalized contraction routines have been developed to allow efficient massively parallel implementation using distributed sparse multidimensional arrays.

The implementation of this software and its performance are described in Nielsen, I. M. B.; Janssen, C. L.; J. Chem. Theory and Comput.; 2007; pp. 71-79.

Constructor & Destructor Documentation

LMP2()

sc::LMP2::LMP2

(

const sc::Ref< sc::KeyVal > &

)

Construct an LMP2 object from KeyVal input.

This reads the keywords in the table below. The LCorr input has additional keywords that are also read.

Keyword	Type	Default	Description
reference	object	none	Gives an object that specializes OneBodyWavefunction which can be used as a reference LMP2 wavefunction. This keyword is required and has no default.
max_iter	int	100	The maximum number of iterations that will be used to converge the LMP2 amplitudes.
nfzc	int or auto	0	The number of frozen core orbitals. An integer can be specified or auto can be given if the number of frozen core orbitals is to be automatically determined.
distance_threshold	double	15.0	The distance threshold used to create the domains, in bohr.
completeness_threshold	double	0.02	The completeness threshold used to create the domains.

In addition to the above options, are a variety of options that is useful to developers, listed in the following table.

Keyword	Type	Default	Description
S_threshold	double	1.0e-6	The threshold used to determine which blocks of the overlap matrix are stored.
integral_threshold	double	1.0e-8	A threshold used to prune the number of two electron integrals which are computed.
q1_threshold	double	1.0e-8	A threshold used to prune the number of first quarter transformed integrals which are computed.
q2_threshold	double	1.0e-8	A threshold used to prune the number of second quarter transformed integrals which are computed.
q3_threshold	double	1.0e-8	A threshold used to prune the number of third quarter transformed integrals which are computed.
q4_threshold	double	1.0e-8	A threshold used to prune the number of fourth quarter transformed integrals which are computed.
threshold_factor	double	10.0	The desired energy tolerance is divided by this factor to obtain the threshold that determines which blocks of a variety of arrays are stored.
complete_domains	bool	0	If true, all possible blocks are computed. The result in this case should be the same as the canonical MP2 energy.
single_virtual_block	bool	0	All of the projected atomic virtual orbitals are grouped into a single, large block, rather than atom based blocks.
i_ge_j	bool	1	If true, use i >= j in the formation of K_2occ.
m_ge_n	bool	1	If true, use m >= n in the formation of K_2occ.
minimize_q2	bool	1	If this and i_ge_j are true, use the i >= j symmetry when rearranging the second quarter tranformed integrals.
always_use_dist_t	bool	0	If true, distribute the amplitudes, in which case an extra communication step is required each iteration.
extrap_T	object	DIIS	This gives an object derived from SelfConsistentExtrapolation that will be used to extraplote the amplitudes. By default DIIS(6,8,0.0,2,1) will be used.
occ_orbitals	string	pipek-mezey	The method used to provide the occupied orbitals. This can be pipek-mezey or canonical.
vir_orbitals	string	projected_atomic	The method used to provide the virtual orbitals. This can be projected_atomic, old_projected_atomic or canonical. If canonical then a single block is used for the virtual orbitals, rather than atom-based blocking.

Member Function Documentation

compute()

void sc::LMP2::compute ( void  )

virtual

Recompute at least the results that have compute true and are not already computed.

This should only be called by Result's members.

Implements sc::Compute.

magnetic_moment()

double sc::LMP2::magnetic_moment ( ) const

virtual

Computes the S (or J) magnetic moment of the target state(s), in units of .

Can be evaluated from density and overlap, as;

(this->alpha_density() * this-> overlap()).trace() -
(this->beta_density() * this-> overlap()).trace()

but derived Wavefunction may have this value as user input.

Returns

the magnetic moment

Reimplemented from sc::Wavefunction.

save_data_state()

void sc::LMP2::save_data_state ( sc::StateOut &  )

virtual

Save the base classes (with save_data_state) and the members in the same order that the StateIn CTOR initializes them.

This must be implemented by the derived class if the class has data.

Reimplemented from sc::LCorr.

value_implemented()

int sc::LMP2::value_implemented ( void  ) const

virtual

Returns

1 (value implemented) or 0 (value not implemented, default). Must be overridden for the class to be useful.

Reimplemented from sc::Function.

---

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

• src/lib/chemistry/qc/lmp2/lmp2.h