Input file

入力ファイルcryspy.inの解説.

File format

CrySPY uses the configparser module to read input file, cryspy.in . cryspy.in consists of sections, led by a [section] header and followed by name = value or name : value entries. Section names and values are case sensitive, but names are not. Lines beginning with # or ; are ignored and may be used to provide comments. Accepted bool values are 1, yes, true, and on, which cause this method to return True, and 0, no, false, and off, which cause it to return False. These string values for bool are checked in a case-insensitive manner. Some values are given in a space-separated manner.

情報

See configparser in detail.

メモ

section name: case sensitive
name: case insensitive
value: case sensitive except for bool

[basic] section

Name	Value	Description
`algo`	`RS`, `EA`, `BO`, `LAQA`	Algorithm
`calc_code`	`VASP`, `QE`, `OMX`, `soiap`, `LAMMPS`	Caluculation code for structure optimization
`tot_struc`	int	The total number of structures
`nstage`	int	The number of stages
`njob`	int	The number of jobs running at the same time.
`jobcmd`	str	Command to submit jobs such as qsub and sbatch.
`jobfile`	str	File name of the job file.

[structure] section

2024 May 22, updated

Name	Value	Default	Description
`struc_mode`	`crystal`, `mol`, `mol_bs`	`crystal`	Structure generation mode
`natot`	int		The total number of atoms.
`atype`	atomic symbol [atomic symbol …]		Atom type. e.g. `atype = Na Cl`.
`nat`	int [int …]		The number of atoms in each atom type. e.g. `nat = 8 8`.
`mindist` (`mindist_?`)	float [float …]	`None`	Constraint on minimum interatomic distance [Å].
`mindist_factor`	float	1.0	Scaling factor for `mindist`.
`vol_factor`	float	1.0	Minimum and maximum values of volume factor.
`vol_mu`	float	`None`	Mean of volume if you want specify the volume of cells.
`vol_sigma`	float	`None`	Standard deviation of volume if you want specify the volume of cells.
`symprec`	float	0.01	Precision for symmetry finding.
`spgnum`	`all`, space group number, 0	`all`	Constraint on space group. If `all`, 1–230. If 0, random structure without space group information (no symmetry).
`use_find_wy`	bool	`False`	Structure generation with find_wy.

mindist

Features > Restriction on interatomic distances

if algo is EA-vc (EA-vc is still beta version)

Name	Value	Default	Description
`ll_nat`	int [int …]		Lower limit of `nat`. e.g. `ll_nat = 1 1`.
`ul_nat`	int [int …]		Upper limit of `nat`. e.g. `ul_nat = 8 8`.

if struc_mode is mol or mol_bs

Name	Value	Default	Description
`mol_file`	str [str …]		Path of molecule files or molecule names.
`nmol`	int [int …]		The number of molecules.
`timeout_mol`	float	`None`	Time out for molecular structure generation.
`rot_mol`	`random`, `random_mol`, `random_wyckoff`	`random_wyckoff`	Only used in `mol_bs`. Mode for rotation of molecules.
`nrot`	int	20	Only used in `mol_bs`. Maximum number of trials to rotate molecules.
`mindist_mol_bs` (`mindist_mol_bs_?`)	float [float …]	`None`	Only used in `mol_bs`. Constraint on minimum intermolecular distance [Å].
`mindist_mol_bs_factor`	float	1.0	Only used in `mol_bs`. Scaling factor for `mindist_mol_bs`.

if use_find_wy is True or spgnum = 0

Name	Value	Default	Description
`fwpath`	str	`None`	Only used with find_wy. Path of find_wy. If `None`, `fwpath` is automatically searched in your $PATH.
`minlen`	float		Only used with find_wy or `spgnum = 0`. Minimum length of lattice vector [Å].
`maxlen`	float		Only used with find_wy or `spgnum = 0`. Maximum length of lattice vector [Å].
`dangle`	float		Only used with find_wy or `spgnum = 0`. Delta angle for alpha, beta, and gamma in degree unit.
`maxcnt`	int	50	Only used with find_wy or `spgnum = 0`. Maximum number of trials to determine atom positions.

[VASP] section

VASPを使う場合は(calc_code = VASP)[VASP]セクションが必要になる．

Name	Value	Default	Description
`kppvol`	int [int …]		各ステージにおける逆格子空間でのÅ**(-3)あたりのグリッド密度．
`force_gamma`	bool	`False`	Trueなら常にガンマ点を通るメッシュを使う．

kppvol and force gamma

Input file > Kpoint

[QE] section

[QE] section is required only if you use QE (calc_code = QE)

Name	Value	Default	Description
`kppvol`	int [int …]		Grid density per Å**(-3) of reciprocal cell in each stage
`qe_infile`	str		File name of QE input file.
`qe_outfile`	str		File name of QE output file.
`pv_term`	bool	`False`	If true, read enthalpy instead of total energy.

kppvol

Input file > Kpoint

pv_term

Features > Enthalpy > QE

[OMX] section

[OMX] section is required only if you use OpenMX (calc_code = OMX)

Name	Value	Description
`kppvol`	int [int …]	Grid density per Å**(-3) of reciprocal cell in each stage
`OMX_infile`	str	File name of OpenMX input file.
`OMX_outfile`	str	File name of OpenMX output file.
`ValenceElectrons`	str float float [str float float …]	The number of initial charges for up and down spin states.

kppvol

Input file > Kpoint

ValenceElectrons

e.g. in NaCl: ValenceElectrons = Na 4.5 4.5 Cl 3.5 3.5.

[soaip] section

[soiap] section is required only if you use soiap (calc_code = soiap)

Name	Value	Description
`soiap_infile`	str	File name of soiap input file.
`soiap_outfile`	str	File name of soiap output file.
`soiap_cif`	str	File name of soiap CIF-formatted initial structure.

[LAMMPS] section

[LAMMPS] section is required only if you use LAMMPS (calc_code = LAMMPS)

Name	Value	Default	Description
`lammps_infile`	str		File name of LAMMPS input file.
`lammps_outfile`	str		File name of LAMMPS output file.
`lammps_potential`	str [str …], `None`	`None`	Potential.
`lammps_data`	str		File name of LAMMPS data file.

[ASE] section

[ASE] section is required only if you use ASE (calc_code = ASE)

Name	Value	Default	Description
`ase_python`	str		File name of ASE input file.

[EA] section

2024年5月22日　更新

[EA] section is required only if you use EA (algo = EA)

Name	Value	Default	Description
`n_pop`	int		Population from second generation.
`n_crsov`	int		Number of structures generated by crossover.
`n_perm`	int		Number of structures generated by permutation.
`n_strain`	int		Number of structures generated by strain.
`n_rand`	int		Number of structures generated randomly.
`n_elite`	int		Number of elite structures.
`fit_reverse`	bool	`False`	If `False`, minimal search.
`n_fittest`	int	`None`	Number of structures which can survive.
`slct_func`	`TNM`, `RLT`		Select function.
`t_size`	int	3	Only used with `slct_func = TNM`. Tournament size.
`a_rlt`	float	10.0	Only used with `slct_func = RLT`. Parameter for linear scaling.
`b_rlt`	float	1.0	Only used with `slct_func = RLT`. Parameter for linear scaling.
`crs_lat`	`equal`, `random`	`random`	How to mix lattice vectors.
`nat_diff_tole`	int	4	Tolerance for difference in the number of atoms in crossover.
`ntimes`	int	1	Number of times in permutation.
`sigma_st`	float	0.5	Standard deviation for strain.
`maxcnt_ea`	int	50	Maximum number of trials in EA.
`maxgen_ea`	int	0	Maximum generation.
`emax_ea`	float	`None`	Upper limit of energy in selecting parents.
`emin_ea`	float	`None`	Lower limit of energy in selecting parents.

if algo is EA-vc

Name	Value	Default	Description
`n_add`	int		Number of structures generated by addition.
`n_elim`	int		Number of structures generated by elimination.
`n_subs`	int		Number of structures generated by substitution.
`target`	str	`random`	Target. only `random` for now.
`end_point`	float, float		Energy of end points for convex hull.

[BO] section

2024 May 27th, updated

[BO] section is required only if you use BO (algo = BO)

Name	Value	Default	Description
`nselect_bo`	int		The number of structures to be selected at once.
`score`	`TS`, `EI`, `PI`		Acquisition function.
`num_rand_basis`	int	0	If `0`, Gaussian process. The number of basis function.
`cdev`	float	0.001	Cutoff of deviation for standardization.
`dscrpt`	`FP`		Structure descriptor.
`max_select_bo`	int	0	Maximum number of selection.
`manual_select_bo`	int [int …]	`None`	Structure IDs to be selected manually.
`emax_bo`	float	`None`	Upper limit of energy in BO.
`emin_bo`	float	`None`	Lower limit of energy in BO.

if decrpt is FP

CrySPY 1.3.0 or later

fppath and fp_rmin are obsolete.

Name	Value	Default	Description
`fp_rmax`	float	8.0	Only used with `dscrpt = FP`. Maximum cutoff of r in fingerprint.
`fp_npoints`	int	20	Only used with `dscrpt = FP`. Number of discretized points for each pair in fingerprint.
`fp_sigma`	float	0.7	Only used with `dscrpt = FP`. Sigma parameter [Å] in Gaussian smearing function.

CrySPY 1.2.5 or earlyer

Name	Value	Default	Description
`fppath`	str	`None`	Only used with `dscrpt = FP`. Path of cal_fingerprint. If `None`, `fwpath` is automatically searched in your $PATH.
`fp_rmin`	float	0.5	Only used with `dscrpt = FP`. Minimum cutoff of r in fingerprint.
`fp_rmax`	float	5.0	Only used with `dscrpt = FP`. Maximum cutoff of r in fingerprint.
`fp_npoints`	int	20	Only used with `dscrpt = FP`. Number of discretized points for each pair in fingerprint.
`fp_sigma`	float	1.0	Only used with `dscrpt = FP`. Sigma parameter [Å] in Gaussian smearing function.

[LAQA] section

LAQA (algo = LAQA)を使う場合，[LAQA]セクションが必要．

Name	Value	Default	Description
`nselect_laqa`	int		1度の選択で選ばれる構造の数．
`wf`	float	0.1	力の項の重み．
`ws`	float	10.0	ストレス項の重み．

情報

[option] section

Name	Value	Default	Description
`stop_chkpt`	int	0	CrySPY stops at a specified check point.
`load_struc_flag`	bool	`False`	If True, load initial structures from `./data/pkl_data/init_struc_data.pkl`.
`stop_next_struc`	bool	`False`	If True, CrySPY does not submit jobs for next structures, but jobs for next stage are submitted.
`recalc`	int [int …]	(empty list)	Specify structure IDs if you want to recalculate or continue optimization.
`append_struc_ea`	bool	`False`	If True, append structures by EA.
`energy_step_flag`	bool	`False`	If True, save energy_step_data in `./data/pkl_data/energy_step_data.pkl`.
`struc_step_flag`	bool	`False`	If True, save struc_step_data in `./data/pkl_data/struc_step_data.pkl`.
`force_step_flag`	bool	`False`	If True, save force_step_data in `./data/pkl_data/force_step_data.pkl`.
`stress_step_flag`	bool	`False`	If True, save stress_step_data in `./data/pkl_data/stress_step_data.pkl`.

Kpoint

2024年4月22日

CrySPYではk-pointのセッティングはpymatgenのpymatgen.io.vasp.Kpoints.automatic_density_by_vol関数を使って自動生成している．以下はcryspy.inでnstage = 2の場合の例：

[VASP]
kppvol = 40 120

stage 1: kppvol = 40
stage 2: kppvol = 120

kppvolは逆格子空間でのÅ ${}^{-3} $あたりのグリッド密度．
VASP：hexagonalとface-centeredセルではガンマ点を通るメッシュが使われ，それ以外ではMonkhorst-Packが使われる．
QE and OMX：kメッシュだけ利用，オフセットは設定しない．

kppvolの値はどれくらいが適切か？

下記を参考に．結晶構造の画像はVESTAを使用．

Primitive cell of diamond Si

a = b = c = 3.836 Å

kppvol	k-mesh
0	[1, 1, 1]
20	[4, 4, 4]
40	[6, 6, 6]
60	[7, 7, 7]
80	[7, 7, 7]
100	[8, 8, 8]
120	[9, 9, 9]
140	[9, 9, 9]
160	[9, 9, 9]
180	[10, 10, 10]
200	[10, 10, 10]
400	[13, 13, 13]
600	[15, 15, 15]
800	[17, 17, 17]

Conventional cell of diamond Si

a = b = c = 5.431 Å

kppvol	k-mesh
0	[1, 1, 1]
20	[3, 3, 3]
40	[3, 3, 3]
60	[4, 4, 4]
80	[4, 4 ,4]
100	[5, 5, 5]
120	[5, 5, 5]
140	[6, 6, 6]
160	[6, 6, 6]
180	[6, 6, 6]
200	[6, 6, 6]
400	[8, 8, 8]
600	[9, 9, 9]
800	[10, 10, 10]

Nd2Fe14B

a = b = 8.804 Å
c = 12.205 Å

kppvol	k-mesh
0	[1, 1, 1]
20	[1, 1, 1]
40	[2, 2, 1]
60	[2, 2, 2]
80	[3, 3 ,2]
100	[3, 3, 2]
120	[3, 3, 2]
140	[3, 3, 2]
160	[3, 3, 2]
180	[4, 4, 2]
200	[4, 4, 3]
400	[5, 5, 3]
600	[6, 6, 4]
800	[6, 6, 4]

Input file

目次

入力ファイルのサブセクション

File format

[basic] section

[structure] section

mindist

[VASP] section

kppvol and force gamma

[QE] section

kppvol

pv_term

[OMX] section

kppvol

ValenceElectrons

[soaip] section

[LAMMPS] section

[ASE] section

[EA] section

[BO] section

[LAQA] section

[option] section

Kpoint

kppvolの値はどれくらいが適切か？

Primitive cell of diamond Si

Conventional cell of diamond Si

Nd2Fe14B