NAME

sc::FinDispMolecularHessian - Computes the molecular hessian by finite displacements of gradients.

SYNOPSIS

#include <fdhess.h>

Inherits sc::MolecularHessian.

Public Member Functions

FinDispMolecularHessian (const Ref< MolecularEnergy > &)
FinDispMolecularHessian (const Ref< KeyVal > &)
The FinDispMolecularHessian KeyVal constructor is used to generate a FinDispMolecularHessian object from the input. FinDispMolecularHessian (StateIn &)
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. RefSymmSCMatrix compute_hessian_from_gradients ()
These members are used to compute a cartesian hessian from gradients at finite displacements. int ndisplace () const
int ndisplacements_done () const
RefSCMatrix displacements (int irrep) const
void displace (int disp)
void original_geometry ()
void set_gradient (int disp, const RefSCVector &grad)
void checkpoint_displacements (StateOut &)
void restore_displacements (StateIn &)
RefSymmSCMatrix cartesian_hessian ()
This returns the cartesian hessian. void set_checkpoint (int c)
Set checkpoint option. int checkpoint () const
Return the current value of the checkpoint option. void set_energy (const Ref< MolecularEnergy > &energy)
Some MolecularHessian specializations require a molecular energy object. MolecularEnergy * energy () const
This returns a MolecularEnergy object, if used by this specialization. Ref< SCMatrixKit > matrixkit () const
RefSCDimension d3natom () const

Protected Member Functions

void get_disp (int disp, int &irrep, int &index, double &coef)
void do_hess_for_irrep (int irrep, const RefSymmSCMatrix &dhessian, const RefSymmSCMatrix &xhessian)
void init ()
void restart ()

Protected Attributes

Ref< MolecularEnergy > mole_
Ref< PointGroup > displacement_point_group_
Ref< PointGroup > original_point_group_
RefSCVector original_geometry_
double disp_
double accuracy_
int ndisp_
int nirrep_
int restart_
char * restart_file_
int checkpoint_
char * checkpoint_file_
int only_totally_symmetric_
int eliminate_cubic_terms_
int do_null_displacement_
int debug_
RefSCMatrix symbasis_
RefSCVector * gradients_

Additional Inherited Members

Detailed Description

Computes the molecular hessian by finite displacements of gradients.

This will use the minimum number of displacements, each in the highest possible point group.

Constructor & Destructor Documentation

sc::FinDispMolecularHessian::FinDispMolecularHessian (const Ref< KeyVal > &)

The FinDispMolecularHessian KeyVal constructor is used to generate a FinDispMolecularHessian object from the input. It reads the keywords below.

KeywordTypeDefaultDescription

energyMolecularEnergynoneThis gives an object which will be used to compute the gradients needed to form the hessian. If this is not specified, the object using FinDispMolecularHessian will, in some cases, fill it in appropriately. However, even in these cases, it may be desirable to specify this keyword. For example, this could be used in an optimization to compute frequencies using a lower level of theory.

debugbooleanfalseIf true, print out debugging information.

point_groupPointGroupnoneThe point group to use for generating the displacements.

restartbooleantrueIf true, and a checkpoint file exists, restart from that file.

restart_filestring basename.ckpt.hessThe name of the file where checkpoint information is written to or read from.

checkpointbooleantrueIf true, checkpoint intermediate data.

only_totally_symmetricbooleanfalse If true, only follow totally symmetric displacments. The hessian will not be complete, but it has enough information to use it in a geometry optimization.

eliminate_cubic_termsbooleantrueIf true, then cubic terms will be eliminated. This requires that two displacements are done for each totally symmetric coordinate, rather than one. Setting this to false will reduce the accuracy, but the results will still probably be accurate enough for a geometry optimization.

do_null_displacementbooleantrueRun the calculation at the given geometry as well.

displacementdouble1.0e-2The size of the displacement in Bohr.

gradient_accuracydouble<td>displacement / 1000The accuracy to which the gradients will be computed.

Member Function Documentation

Author

Generated automatically by Doxygen for MPQC from the source code.