script_make_script.rst

wfns_make_script.py

The script :py:mod:`wfns_make_script.py` runs a calculation using a set of arguments and optional arguments:

wfns_make_script.py [-h] --nelec NELEC --nspin NSPIN
                    --one_int_file ONE_INT_FILE --two_int_file TWO_INT_FILE
                    --wfn_type WFN_TYPE
                    [--nuc_repulsion NUC_NUC] [--optimize_orbs]
                    [--pspace PSPACE_EXC [PSPACE_EXC ...]]
                    [--objective OBJECTIVE] [--solver SOLVER]
                    [--solver_kwargs SOLVER_KWARGS]
                    [--wfn_kwargs WFN_KWARGS] [--load_orbs LOAD_ORBS]
                    [--load_ham LOAD_HAM] [--load_wfn LOAD_WFN]
                    [--load_chk LOAD_CHK] [--save_orbs SAVE_ORBS]
                    [--save_ham SAVE_HAM] [--save_wfn SAVE_WFN]
                    [--save_chk SAVE_CHK] [--memory MEMORY]
                    [--filename FILENAME]

where the arguments inside square brackets are optional.

Mandatory Keywords

--nelec NELEC

Number of electrons in the wavefunction. NELEC is an integer.

--nspin NSPIN

Number of spin orbitals used in the wavefunction and Hamiltonian. NSPIN is an integer.

--one_int_file ONE_INT_FILE

One-electron integrals used to construct the chemical Hamiltonian. The integrals are expressed with respect to spatial orbitals. ONE_INT_FILE is a numpy file that contains a two-dimensional square array of floats. The dimension of each axis is the number of spatial orbitals.

--two_int_file TWO_INT_FILE

Two-electron integrals used to construct the chemical Hamiltonian. The integrals are expressed with respect to spatial orbitals. TWO_INT_FILE is a numpy file that contains a four-dimensional array of floats with equal dimension in all four axis. The dimension of each axis is the number of spatial orbitals.

--wfn_type WFN_TYPE

Type of the wavefunction. WFNTYPE must be one of fci, doci, mps, determinant-ratio, ap1rog, apr2g, apig, apsetg, or apg.

Optional Keywords

--nuc_repulsion NUC_NUC

Nuclear-nuclear repulsion energy (in Hartree). If not provided, then it is set to 0.0.

--optimize_orbs

Flag for optimizing the orbitals. Currently, orbitals can be optimized only when the solver does not use the gradient (i.e. cma). If not provided, then orbitals are not optimized.

--pspace PSPACE_EXC [PSPACE_EXC ...]

Orders of excitations that wil be used to construct the projection space. For the objectives system and least_squares, the projection space corresponds to the equations that will be created. For the objective variational, the projection space correspond to the projection operators that are inserted to the left and right of the Hamiltonian. See :ref:`Objective <objective>` for more details.

PSPACE_EXC is a sequence of integers (orders of excitations) separated by space. Zeroth order excitation (i.e. HF ground state) is always included. If not provided, then first and second order excitations (and HF ground state) are used as the projection space (equivalent to --pspace 1 2).

--objective OBJECTIVE

Objective function that will be used to optimize the parameters. OBJECTIVE must be one of system, least_squares, or variational. If not provided, then least_squares is selected.

See :ref:`Objective <objective>` for more details.

--solver SOLVER

Solver used to optimize the objective. SOLVER must be one of cma, diag, minimize, least_squares, or root. If not provided, then cma is selected.

Solvers cma and minimize can only be used with the objectives least_squares and variational. Solvers least_squares and root can only be used with objective system. Solver diag can only be used with CI wavefunctions (wavefunction types of fci and doci).

See XXX for more details.

--solver_kwargs SOLVER_KWARGS

Keyword arguments that will be passed to the solver method. SOLVER_KWARGS must be provided as a string that will be passed into the solver method, overwriting all keyword arguments. For example, --solver_kwargs 'sigma0=0.001' will change the sigma value in the cma solver.

If not provided, then no keyword argument will be provided to the solver and its default setting will be used. See XXX for more details.

--wfn_kwargs WFN_KWARGS

Keyword arguments that will be passed to the wavefunction instance. WFN_KWARGS must be provided as a string that will be passed into the wavefunction instance, overwriting all keyword arguments. For example, --wfn_kwargs 'dimension=10' will change the dimension of the matrices in the MPS wavefunction.

If not provided, then no keyword argument will be provided to the wavefunction and its default setting will be used. See XXX for more details.

--load_orbs LOAD_ORBS

Transformation matrix that will be used to rotate the integrals in the Hamiltonian. This keyword can be used to port over the orbitals from another calculation (for example, from a different wavefunction calculation).

LOAD_ORBS must be provided as a numpy file of a two-dimension array with correct dimensions. The number of rows must correspond with the dimension of an axis in the integrals.

If the keyword --load_ham is also provided, then the integrals (orbitals) are rotated after instantiating the Hamiltonian. However, it is not recommended to use both keywords --load_orbs and --load_ham.

If not provided, then integrals will not be rotated.

--load_ham LOAD_HAM

Parameters of the Hamiltonian that will be used to instantiate the Hamiltonian. This keyword can be used to port over the Hamiltonian parameters from another calculation (for example, from a different wavefunction calculation).

LOAD_HAM must be provided as a numpy file of a one-dimension array with the correct dimension. The number of parameters must correspond with the number of elements in the upper triangular matrix of the anti-Hermitian matrix in the transformation operator.

If the keyword --load_orbs is also provided, then the integrals (orbitals) are rotated after instantiating the Hamiltonian. However, it is not recommended to use both keywords --load_orbs and --load_ham.

If not provided, then default Hamiltonian parameters (zeros) will be used.

--load_wfn LOAD_WFN

Parameters of the wavefunction that will be used to instantiate the wavefunction. This keyword can be used to port over the wavefunction parameters from another calculation (for example, from a different Hamiltonian/system).

LOAD_WFN must be provided as a numpy file of a one-dimension array with the correct dimension. The number of parameters varies depending on the wavfunction type. If not provided, then the default parameters of the wavefunction will be used (almost always HF ground state).

See XXX for more details.

--load_chk LOAD_CHK

Checkpoint in the optimization process. This keyword can be used to restart a calculation. LOAD_CHK must be provided as a numpy file of one-dimension array with the correct dimension. The number of parameters can vary depending on the number of active (not frozen) parameters in the optimization.

See XXX for more details.

--save_orbs SAVE_ORBS

Transformation matrix that was used to rotate the integrals in the Hamiltonian. This keyword can be used to save the orbitals for use in another calculation. (for example, in a different wavefunction calculation).

SAVE_ORBS is the name of the numpy file used to save the transformation matrix. Since transformation matrix is produced from the Hamiltonian parameters, it is not recommended to use both keywords --save_orbs and --save_ham.

If not provided, then the transformation matrix will not be stored.

--save_ham SAVE_HAM

Parameters of the Hamiltonian that was used in Hamiltonian instance. This keyword can be used to save the Hamiltonian parameters for use in another calculation (for example, in a different wavefunction calculation).

SAVE_HAM is the name of the numpy file used to save the Hamiltonian parameters. Since transformation matrix is produced from the Hamiltonian parameters, it is not recommended to use both keywords --save_orbs and --save_ham.

If not provided, then Hamiltonian parameters are not saved.

--save_wfn SAVE_WFN

Parameters of the wavefunction that was used in wavefunction instance. This keyword can be used to save the wavefunction parameters for use in another calculation (for example, in a different Hamiltonian/system calculation).

SAVE_WFN is the name of the numpy file used to save the wavefunction parameters.

If not provided, then wavefunction parameters are not saved.

--save_chk SAVE_CHK

Checkpoint file that saves the values of all active (not frozen) parameters in the optimization process. This keyword can be used to save the progress of the optimization so that it can be restarted should the optimization fails prematurely.

SAVE_CHK is the name of the numpy file used to save the checkpoint.

If not provided, then wavefunction parameters are not saved.

--memory MEMORY

Memory available for the wavefunction. MEMORY must be a string that ends with mb (for MB) or gb (for GB). If not provided, then no restrictions will be put on cache for the overlaps of wavefunction, which may result in memory overflow.

--filename FILENAME

Name of the script that will be produced. FILENAME must be a string. If not provided, then the script is printed out in the STDOUT