Runtime Parameters

Runtime Parameters#

Runtime parameters and options in LibRPA are generally managed through the LibrpaOptions object.

When using the API, you can directly assign values to the corresponding attributes to control the runtime behavior of LibRPA.
When using the driver, parameters are read from a file named librpa.in in the working directory, using the syntax key = value. For example:
```
task = rpa
input_dir = ../librpa_dataset
nfreq = 16
cs_threshold = 1e-4
```
If librpa.in is not found, the driver stops with an error. Unlike the API, the driver requires task to be explicitly specified. For all other parameters, default values are used when the corresponding keywords are not provided. Please refer to manual page of driver usage for more information.

Parameter Name	Description	Type	Default Value (Options)
`task`	Task type	string	(required: rpa, g0w0, exx, g0w0_band, exx_band)
`input_dir`	Input directory to find and read the AO dataset	string	`.`
`cs_threshold`	Screening threshold when reading the RI coefficient data	double	1e-6

Parameter Name	Description	Type	Default Value (Options)
`output_dir`	Output directory for results	string	`.`
`parallel_routing`	Parallel scheme of LibRPA	string	auto (auto, atompair, rtau, libri)
`output_level`	Verbosity level	int	2 (0=silent, 1=critical, 2=info, 3=warn, 4=debug)
`vq_threshold`	Real-space Coulomb matrices screening threshold	double	0.0
`use_kpara_scf_eigvec`	Flag for parallel distribution of SCF eigenvectors	bool	`false`
`tfgrids_type`	Type of time-frequency integration grid	string	minimax (gl, gci, gcii, minimax, evenspaced, evenspaced_tf)
`nfreq`	Number of frequency integration grid points	int	6
`tfgrids_freq_min`	Minimum frequency for grid (Hartree)	double	0.0
`tfgrids_freq_interval`	Frequency interval for even-spaced grid (Hartree)	double	0.1
`tfgrids_freq_max`	Maximum frequency for grid (Hartree)	double	10.0
`tfgrids_time_min`	Minimum time for grid (Hartree^-1)	double	0.0
`tfgrids_time_interval`	Time interval for even-spaced grid (Hartree^-1)	double	0.1

Parameter Name	Description	Type	Default Value (Options)
`gf_threshold`	Real-space Green’s function screening threshold for response function	double	1e-4
`use_scalapack_ecrpa`	Flag to use ScaLapack for calculating \(E_\text{c}^{\text{RPA}}\)	bool	`true`

Parameter Name	Description	Type	Default Value (Options)
`n_params_anacon`	Number of parameters for analytic continuation	int	-1 (will use all `nfreq` data)
`use_scalapack_gw_wc`	Flag to use ScaLAPACK for computing Wc from chi0	bool	`true`
`replace_w_head`	Flag to replace head of dielectric matrix by macroscopic dielectric function	bool	`false`
`option_dielect_func`	Option for computing dielectric function on imaginary axis	int	0 (0=direct, 1=model fit, 2=spline)
`sqrt_coulomb_threshold`	Threshold for eigenvalues to perform square root of Coulomb matrices	double	1e-6

Parameter Name	Description	Type
`libri_chi0_threshold_C`	Threshold of LRI triple coefficients for response function	double
`libri_chi0_threshold_G`	Threshold of Green’s function for response function	double
`libri_exx_threshold_C`	Threshold of LRI triple coefficients for exact exchange	double
`libri_exx_threshold_D`	Threshold of density matrices for exact exchange	double
`libri_exx_threshold_V`	Threshold of Coulomb matrices for exact exchange	double
`libri_g0w0_threshold_C`	Threshold of LRI triple coefficients for G0W0 correlation self-energy	double
`libri_g0w0_threshold_G`	Threshold of Green’s function for G0W0 correlation self-energy	double
`libri_g0w0_threshold_Wc`	Threshold of screened Coulomb matrix for G0W0 correlation self-energy	double

Parameter Name	Description	Type	Default Value (Options)
`output_gw_sigc_mat`	Output correlation self-energy matrix (k-space, imaginary frequencies)	bool	false
`output_gw_sigc_mat_rt`	Output correlation self-energy matrix (real-space, imaginary time)	bool	false
`output_gw_sigc_mat_rf`	Output correlation self-energy matrix (real-space, imaginary frequencies)	bool	false

For details on all parameters, you can visit the API documentation of struct LibrpaOptions.