The MBPT2_R12 class implements several linear R12 second-order perturbation theory methods. More...
#include <mbptr12.h>
Public Member Functions | |
MBPT2_R12 (StateIn &) | |
MBPT2_R12 (const Ref< KeyVal > &) | |
The KeyVal constructor. | |
void | save_data_state (StateOut &) |
Ref< GaussianBasisSet > | aux_basis () const |
Ref< GaussianBasisSet > | vir_basis () const |
bool | gbc () const |
bool | ebc () const |
LinearR12::ABSMethod | abs_method () const |
LinearR12::StandardApproximation | stdapprox () const |
bool | spinadapted () const |
R12IntEvalInfo::StoreMethod | r12ints_method () const |
const std::string & | r12ints_file () const |
double | corr_energy () |
double | r12_corr_energy () |
RefSymmSCMatrix | density () |
Returns the SO density. | |
void | obsolete () |
Marks all results as being out of date. | |
int | gradient_implemented () const |
int | value_implemented () const |
Information about the availability of values, gradients, and hessians. | |
void | print (std::ostream &o=ExEnv::out0()) const |
Print information about the object. | |
Protected Member Functions | |
void | compute () |
Recompute at least the results that have compute true and are not already computed. |
The MBPT2_R12 class implements several linear R12 second-order perturbation theory methods.
The KeyVal constructor.
gbc
This boolean specifies whether Generalized Brillouin Condition (GBC) is assumed to hold. The default is "true". This keyword is only valid if stdapprox=A'. The effect of setting this keyword to true is rather small -- hence it is not recommended to use this keyword.
ebc
This boolean specifies whether Extended Brillouin Condition (EBC) is assumed to hold. The default is "true". This keyword is only valid if stdapprox=A'. The effect of setting this keyword to true is rather small -- hence it is not recommended to use this keyword.
stdapprox
This gives a string that must take on one of the values below. The default is A'.
A
Use second order Møller-Plesset perturbation theory with linear R12 terms in standard approximation A (MP2-R12/A). Only energies can be computed with the MP2-R12/A method.
A'
Use second order Møller-Plesset perturbation theory with linear R12 terms in standard approximation A' (MP2-R12/A'). This will cause MP2-R12/A energies to be computed also. Only energies can be computed with the MP2-R12/A' method.
B
Use second order Møller-Plesset perturbation theory with linear R12 terms in standard approximation B. This method is not implemented yet.
spinadapted
This boolean specifies whether to compute spin-adapted or spin-orbital pair energies. Default is to compute spin-adapted energies.
aux_basis
This specifies the auxiliary basis to be used for the resolution of the identity. Default is to use the same basis as for the orbital expansion.
vir_basis
This specifies the basis to be used for the virtual orbitals. Default is to use the same basis as for the orbital expansion.
include_mp1
This specifies whether to compute MP1 correction to the MP2 and MP2-R12 energies. This option only has effect if vir_basis is not the same as basis. MP1 correction is a perturbative estimate of the difference between the HF energy computed in vir_basis and basis. Usually, it is a very poor estimate -- therefore this keyword should be avoided by non-experts. Default is false.
abs_method
This string specifies whether the old ABS method, introduced by Klopper and Samson, or the new ABS variant, CABS, introduced by Valeev, should be used. Valid values are "ABS" (Klopper and Samson), "ABS+", "CABS", and "CABS+", where the "+" labels a method where the union of OBS and ABS is used to construct the RI basis. The default is "ABS". The default in 2.3.0 and later will be "CABS+".
lindep_tol
The tolerance used to detect linearly dependent basis functions in the RI basis set. The precise meaning depends on the orthogonalization method. The default value is 1e-8.
r12ints
This specifies how to store transformed MO integrals. Valid values are:
mem-posix
Store integrals in memory for single-pass situations and in a binary file on task 0's node using POSIX I/O for multipass situations. posix
is usually less efficient than mpi
for distributed parallel multipass runs since the I/O is performed by one task only. However, this method is guaranteed to work in all types of environments, hence mem-posix
is the default.
posix
Store integrals in a binary file on task 0's node using POSIX I/O. This method is different from mem-posix
in that it forces the integrals out to disk even if they could be stored in memory. posix
should only be used for benchmarking and testing purposes.
mem-mpi
Store integrals in memory for single-pass situations and in a binary file using MPI-I/O for multipass situations. This method assumes the availability of MPI-I/O. mem-mpi
is the preferred choice in distributed environments which have MPI-I/O available.
mpi
Store integrals in a binary file using MPI-I/O. This method is different from mem-mpi
in that it forces the integrals out to disk even if they could be stored in memory. mpi
should only be used for benchmarking and testing purposes.
mem
Store integrals in memory. Can only be used with single-pass transformations for MP2-R12/A and MP2-R12/A' methods. This method should only be used for testing purposes.
If r12ints
is not specified, then mem-posix
method will be used. If user wishes to use MPI-I/O, pending its availability, for higher parallel efficiency, r12ints
should be explicitly set to mem-mpi
.
r12ints_file
This specifies the prefix for the transformed MO integrals file if r12ints
is set to posix
, mpi
, mem-posix
or mem-mpi
is used. Default is "./<i>inputbasename</i>.r12ints", where inputbasename is the name of the input file without ".in". If MPI-I/O is used then it is user's responsibility to ensure that the file resides on a file system that supports MPI-I/O.
twopdm_grid_aa
This optional keyword specifies a TwoBodyGrid object which to use for coordinates at which to compute alpha-alpha part of 2-PDM.
twopdm_grid_ab
This optional keyword specifies a TwoBodyGrid object which to use for coordinates at which to compute alpha-beta part of 2-PDM.
void sc::MBPT2_R12::compute | ( | ) | [protected, virtual] |
Recompute at least the results that have compute true and are not already computed.
This should only be called by Result's members.
Reimplemented from sc::MBPT2.
void sc::MBPT2_R12::obsolete | ( | ) | [virtual] |
Marks all results as being out of date.
Any subsequent access to results will cause Compute::compute() to be called.
Reimplemented from sc::MBPT2.
int sc::MBPT2_R12::value_implemented | ( | ) | const [virtual] |
Information about the availability of values, gradients, and hessians.
Reimplemented from sc::MBPT2.