Features

Logging

2023 July 10

CrySPY 1.2.0 adopts logging library of Python. CrySPY logs are output to both the screen and files(log_cryspy and err_cryspy).

log –> screen and log_cryspy
error and warning –> screen and err_cryspy

Here is the example:

[2023-07-10 18:40:54,389][cryspy_init][INFO] 


Start CrySPY 1.2.0


[2023-07-10 18:40:54,389][cryspy_init][INFO] # ---------- Read input file, cryspy.in
[2023-07-10 18:40:54,390][read_input][INFO] Save input data in cryspy.stat
[2023-07-10 18:40:54,391][cryspy_init][INFO] # ---------- Initial structure generation
[2023-07-10 18:40:54,391][cryspy_init][INFO] Number of MPI processes: 1
[2023-07-10 18:40:54,391][gen_init_struc][INFO] # ------ mindist
[2023-07-10 18:40:54,395][struc_util][INFO] Cu - Cu: 1.32
[2023-07-10 18:40:54,395][gen_init_struc][INFO] # ------ generate structures
[2023-07-10 18:40:54,481][gen_pyxtal][INFO] Structure ID      0 was generated. Space group:   1 -->   1 P1
[2023-07-10 18:40:54,493][gen_pyxtal][INFO] Structure ID      1 was generated. Space group:  28 -->  28 Pma2
[2023-07-10 18:40:54,498][gen_pyxtal][INFO] Structure ID      2 was generated. Space group:  29 -->  29 Pca2_1
[2023-07-10 18:40:54,704][gen_pyxtal][INFO] Structure ID      3 was generated. Space group: 137 --> 137 P4_2/nmc
[2023-07-10 18:40:54,725][gen_pyxtal][INFO] Structure ID      4 was generated. Space group: 212 --> 214 I4_132
[2023-07-10 18:40:54,800][cryspy_init][INFO] Elapsed time for structure generation: 0:00:00.408367

If you want to run cryspy as a background job, or if you use the auto script (repeat_cryspy), and do NOT want it to output to the screen, run cryspy with the -n option as follow:

cryspy -n

Backup

CrySPY has a simple backup function. The following files are backed up:

cryspy.in
cryspy.stat
log_cryspy
err_cryspy
calc_in/*
data/*
ext/*

work/* are NOT included.

(v1.1.0 or later) above files are copied to a directory named by date and time in “backup” directory. Previous backups are NOT automatically deleted.
(v1.0.0) only one generation is backed up, and previous backups will be deleted.

Auto backup

The timing of the automatic backup is as follows:

before going to next selection (BO, LAQA) or next generation (EA)
append structures

Manual backup

To manually back up, run cryspy with -b or --backup option as:

cryspy -b

This command only performs backups, unlike the normal execution.

Clean

CrySPY has a simple clean (just move files) function. It is useful when you want to start over from the beginning. The following files are cleaned up:

cryspy.stat
log_cryspy
err_cryspy
lock_cryspy
data/*
work/*
ext/*
tmp_calc_FP/*
tmp_gen_struc/*

To clean up, run cryspy with -c or --clean option as:

$ ls
calc_in  cryspy.in  cryspy.stat  data  err_cryspy  log_cryspy

$ cryspy -c
Are you sure you want to clean the data? 'yes' or 'no' [y/n]: y

$ ls
calc_in  cryspy.in  trash

$ ls trash
20230318_100728

Files other than calc_in/* and cryspy.in are moved to trash and grouped into a directory named by date and time. If you do not need them, you can delete them manually.

Restriction on interatomic distances

2024 April 23, updated

You can restrict the interatomic distance in structure generation. Here is an example of [structure] section in the input file to limit minimum interatomic distance for a A-B binary system.

[structure]
natot = 8
atype = A B
nat = 4 4
mindist_1 = 2.0 1.8
mindist_2 = 1.8 1.5

This means that minimum interatomic distances of A-A, A-B, and B-B are limited to 2.0, 1.8, and 1.5 Å, respectively. Structures with interatomic distances shorter than these values are automatically eliminated.

For ternary systems, you will need mindist_1, mindist_2, and mindist_3. Mindist matrix must be a symmetric matrix.

Example: Na8Cl8

Without mindist

cryspy.in

[basic]
algo = RS
calc_code = VASP
tot_struc = 5
nstage = 2
njob = 2
jobcmd = qsub
jobfile = job_cryspy

[structure]
natot = 16
atype = Na Cl
nat = 8 8

[VASP]
kppvol = 40 80

[option]

log_cryspy

[2024-04-23 13:46:28,598][cryspy_init][INFO] 


Start CrySPY 1.2.3


[2024-04-23 13:46:28,598][cryspy_init][INFO] # ---------- Read input file, cryspy.in
[2024-04-23 13:46:28,598][read_input][INFO] Save input data in cryspy.stat
[2024-04-23 13:46:28,599][gen_init_struc][INFO] # ------ mindist
[2024-04-23 13:46:28,601][struc_util][INFO] Na - Na: 1.66
[2024-04-23 13:46:28,602][struc_util][INFO] Na - Cl: 1.3399999999999999
[2024-04-23 13:46:28,602][struc_util][INFO] Cl - Cl: 1.02
...

In the default settings of PyXtal, atoms can sometimes be too close to each other, as shown in the figure above, so it is recommended to set the mindist parameter. That would help simplify DFT calculations.

With mindist

cryspy.in

[basic]
algo = RS
calc_code = VASP
tot_struc = 5
nstage = 2
njob = 2
jobcmd = qsub
jobfile = job_cryspy

[structure]
natot = 16
atype = Na Cl
nat = 8 8
mindist_1 = 2.5 1.5
mindist_2 = 1.5 2.5

[VASP]
kppvol = 40 80

[option]

log_cryspy

[2024-04-23 14:06:21,955][cryspy_init][INFO] 


Start CrySPY 1.2.3


[2024-04-23 14:06:21,955][cryspy_init][INFO] # ---------- Read input file, cryspy.in
[2024-04-23 14:06:21,956][read_input][INFO] Save input data in cryspy.stat
[2024-04-23 14:06:21,956][gen_init_struc][INFO] # ------ mindist
[2024-04-23 14:06:21,956][struc_util][INFO] Na - Na: 2.5
[2024-04-23 14:06:21,956][struc_util][INFO] Na - Cl: 1.5
[2024-04-23 14:06:21,956][struc_util][INFO] Cl - Cl: 2.5

In cases like ionic crystals, it is advisable to set up the configuration in such a way that cations and anions are kept apart from each other.

CrySPY_ID in job files

In the job file of CrySPY, the string “CrySPY_ID” is automatically replaced with the structure ID. When you use a job scheduler such as PBS and SLURM, it is useful to set the structure ID to the job name. For example, in the PBS system, #PBS -N Si_CrySPY_ID in ID 10 is replaced with #PBS -N Si_10. Note that starting with a number will result in an error. You should add a prefix like Si_.

#!/bin/sh
#$ -cwd
#$ -V -S /bin/bash
####$ -V -S /bin/zsh
#$ -N Si8_CrySPY_ID
#$ -pe smp 8
####$ -q ibis1.q
####$ -q ibis2.q

mpirun -np $NSLOTS pw.x -nk 4 -nb 2 < pwscf.in > pwscf.out


if [ -e "CRASH" ]; then
    exit 1
fi

sed -i -e '3 s/^.*$/done/' stat_job

Structure generation with MPI parallelization

Oct. 21 2023, update

Random structure generation using MPI has been available since version 1.1.0 ( using CrySPY >= 1.2.3 is better). You need to install mpi4py in your Python environment for MPI parallelization. Of course, an MPI library such as Open MPI, Intel MPI, and MPICH is required for your workstation.

Info

Requirements:

CrySPY ~~1.1.0~~ 1.2.3 or later
mpi4py
MPI library (Open MPI, Intel MPI, MPICH, etc.)

Warning

1.1.0 <= CrySPY <=1.2.2 has a bug. When you use bash (zsh) to run a job with MPI (e.g., jobcmd = zsh, jobfile = job_cryspy), the MPI job does not run. There is no problem when you use a job scheduler (qsub, sbatch). It has already fixed in version 1.2.3.

The figure below shows the relationship between elapsed time and the number of processes for 1000 structures of Si8 with the following setting:

[basic]
algo = RS
calc_code = soiap
tot_struc = 1000
nstage = 1
njob = 2
jobcmd = zsh
jobfile = job_cryspy

[structure]
natot = 8
atype = Si
nat = 8
mindist_1 = 2.2

The structure generation is taking a long time because of a slightly stricter setting like mindset_1 = 2.2. The structure generation was performed 10 times for each number of processes.

Run

mpiexec -n 4 cryspy -p

Info

Enthalpy

2023/10/18

Info

Requirements:

CrySPY 1.2.2 or later
VASP or QE

When performing CSP at high pressure, enthalpy results can be collected instead of total energy. Not yet compatible with softwares other than VASP and QE.

E_eV_atom in cryspy_rslt and cryspy_rslt_energy_asc turns into enthalpy (eV/atom). Here is the example of CSP results under 40 GPa pressure for Sr4O4. CsCl-type structure (ID 5) is more stable than NaCl-type (ID 6).

   Spg_num Spg_sym  Spg_num_opt Spg_sym_opt  E_eV_atom  Magmom      Opt
5       26  Pmc2_1          221       Pm-3m  -2.276790     NaN     done
6      225   Fm-3m          225       Fm-3m  -2.244800     NaN     done
1      101  P4_2cm          107        I4mm  -2.181115     NaN     done
4      123  P4/mmm          123      P4/mmm  -2.034509     NaN  not_yet
3       20  C222_1           63        Cmcm  -0.686541     NaN     done
2       75      P4           75          P4  -0.008713     NaN  not_yet
9       51    Pmma           47        Pmmm   0.096430     NaN     done
8       65    Cmmm          123      P4/mmm   1.099657     NaN     done
0      187   P-6m2          187       P-6m2   1.292124     NaN     done
7       53    Pmna           53        Pmna   5.153504     NaN  not_yet

VASP

CrySPY reads energy (enthalpy) from a OSZICAR file. This automatically changes to enthalpy when PSTRESS is set in INCAR_x as follows:

PSTRESS = 400

You do not have to do anything in cryspy.in. energy_step_flag is also supported for enthalpy.

Example:

CrySPY Utility > Examples > vasp_Sr4O4_RS_pv_term

QE

Add pv_term = True in the QE section of cryspy.in to use enthalpy:

[QE]
qe_infile = pwscf.in
qe_outfile = pwscf.out
kppvol =  40  80
pv_term = True

Don’t forget to write press in the QE input:

 &cell
    press = 400
 /

Warning

In QE, energy_step_flag is not supported yet for enthalpy.

Example:

CrySPY Utility > Examples > qe_Sr4O4_RS_pv_term

As library

2024 May 31

Info

Requirements:

CrySPY 1.3.0 or later

Cryspy can be used as a library to generate random structures or structures by evolutionary algoritym. The jupyter notebook is available in CrySPY utility > notebook > as_library.

Random structure generation

####
#### when you change set_logger(), you need to restart the kernel
####
from cryspy.util.utility import set_logger    # optional
set_logger()    # optional
#set_logger(noprint=True, logfile='log_cryspy', errfile='err_cryspy')    # write log and err messages to files

from cryspy.RS.gen_struc_RS import gen_pyxtal

nstruc = 10
atype = ('Na', 'Cl')
nat = (4, 4)
mindist = ((2.0, 1.5),
           (1.5, 2.0))
spgnum = 'all'

init_struc_data = gen_pyxtal.gen_struc(
    nstruc=nstruc,
    atype=atype,
    nat=nat,
    mindist=mindist,
    spgnum=spgnum,
)

You can get init_struc_data (dict: {ID: pymatgen Strcture, …})

Structure generation by evolutionary algorithm

Situation: parent A (, parent B) –> child

Prepare two (one) parent structures as pymatgen Structure object.
In this example, just use the results of RS for Cu4Au4 (see, CrySPY utility > notebook > as_library).

import pickle
with open('./Cu4Au4_sample/opt_struc_data.pkl', 'rb') as f:
    opt_struc_data = pickle.load(f)

Crossover

from cryspy.EA.gen_struc_EA import crossover

# you can change parent_A and parent_B
parent_A = opt_struc_data[0]
parent_B = opt_struc_data[1]

atype = ('Cu', 'Au')
nat = (4, 4)
mindist = ((1.5, 1.5),
           (1.5, 1.5))

child = crossover.gen_child(
    atype=atype,
    nat=nat,
    mindist=mindist,
    parent_A=parent_A,
    parent_B=parent_B,
)

# child: pymatgen Structure

Permutation

from cryspy.EA.gen_struc_EA import permutation

# you can change parent_A
parent_A = opt_struc_data[0]

atype = ('Cu', 'Au')
nat = (4, 4)
mindist = ((1.5, 1.5),
           (1.5, 1.5))
ntimes = 1    # number of times to perform permutatio

child = permutation.gen_child(
    atype=atype,
    mindist=mindist,
    parent_A=parent_A,
    ntimes=ntimes,
)

# child: pymatgen Structure

Strain

from cryspy.EA.gen_struc_EA import strain

atype = ('Cu', 'Au')
nat = (4, 4)
mindist = ((1.5, 1.5),
           (1.5, 1.5))
sigma_st = 0.05    # standard deviation of strain

child = strain.gen_child(
    atype=atype,
    mindist=mindist,
    parent_A=parent_A,
    sigma_st=sigma_st,
)

Situation: parent group, fitness –> children

Data set

Prepare structure and fitness (energy) data as dict. The key is structure ID. In this example, just use the results of RS for Cu4Au4 (see, CrySPY utility > notebook > as_library)..

e.g.
struc_data = {0: (pymatgen Structure), 1: (pymatgen Structure), …}
fitness = {0: 0.019632287242441926, 1: -0.005437509701440302, …}

import pickle
with open('./Cu4Au4_sample/opt_struc_data.pkl', 'rb') as f:
    opt_struc_data = pickle.load(f)
with open('./Cu4Au4_sample/rslt_data.pkl', 'rb') as f:
    rslt_data = pickle.load(f)

struc_data = opt_struc_data    # dict
fitness = rslt_data['E_eV_atom'].to_dict()    # you may include None or np.nan for values

Survival of the fittest

from cryspy.EA.survival import survival_fittest
from cryspy.EA.gen_struc_EA.select_parents import SelectParents
from cryspy.EA.gen_struc_EA import crossover, permutation, strain

n_fittest = 5    # number of survivors

ranking, _, _ = survival_fittest(
    fitness=fitness,
    struc_data=struc_data,
    elite_struc=None,
    elite_fitness=None,
    n_fittest=n_fittest,
    fit_reverse=False,
    emax_ea=None,
    emin_ea=None,
)

# ranking <-- e.g. [2, 1, 0, 7, 9] without structure duplicaiton

Select parents class

sp = SelectParents(ranking)    # after set_xxx, we can use sp.get_parents(n_parent)
sp.set_tournament(t_size=2)

Crossover

atype = ('Cu', 'Au')
nat = (4, 4)
mindist = ((1.5, 1.5),
           (1.5, 1.5))
n_crsov = 5    # number of structures to be generated by crossover
#id_start = len(init_struc_data)  # next Structure ID
id_start = 10

co_children, co_parents, co_operation = crossover.gen_crossover(
    atype=atype,
    nat=nat,
    mindist=mindist,
    struc_data=struc_data,
    sp=sp,
    n_crsov=n_crsov,
    id_start=id_start,
)

# co_children <-- dict {ID: pymatgen Structure, ID: pymatgen Structure, ...}
# co_parents  <-- e.g. {10: (2, 7), 11: (2, 1), 12: (2, 1), 13: (0, 2), 14: (2, 1)}
# co_operation <-- e.g. {10: 'crossover', 11: 'crossover', ...}

Permutation

n_perm = 5    # number of structures to be generated by permutation
#id_start = len(init_struc_data) + n_crsov   # next Structure ID
id_start = 15
ntimes = 1    # number of times to perform permutation

pm_children, pm_parents, pm_operation = permutation.gen_permutation(
    atype=atype,
    mindist=mindist,
    struc_data=struc_data,
    sp=sp,
    n_perm=n_perm,
    id_start=id_start,
    ntimes=ntimes,
)

# pm_children <-- dict {ID: pymatgen Structure, ID: pymatgen Structure, ...}
# pm_parents  <-- e.g. {15: (2,), 16: (1,), 17: (2,), 18: (1,), 19: (1,)}
# pm_operation <-- e.g. {15: 'permutaion', 16: 'permutation', ...}

Strain

n_strain = 5    # number of structures to be generated by strain
#id_start = len(init_struc_data) + n_crsov + n_perm   # next Structure ID
id_start = 20
sigma_st = 0.05    # standard deviation of strain

st_children, st_parents, st_operation = strain.gen_strain(
    atype=atype,
    mindist=mindist,
    struc_data=struc_data,
    sp=sp,
    n_strain=n_strain,
    id_start=id_start,
    sigma_st=sigma_st,
)

# st_children <-- dict {ID: pymatgen Structure, ID: pymatgen Structure, ...}
# st_parents  <-- e.g. {20: (1,), 21: (2,), 22: (0,), 23: (2,), 24: (2,)}
# st_operation <-- e.g. {20: 'strain', 21: 'strain', ...}

Features

Table of contents

Subsections of Features

Logging

Backup

Auto backup

Manual backup

Clean

Restriction on interatomic distances

Example: Na8Cl8

Without mindist

With mindist

CrySPY_ID in job files

Structure generation with MPI parallelization

Run

Enthalpy

VASP

QE

As library

Random structure generation

Structure generation by evolutionary algorithm

Situation: parent A (, parent B) –> child

Crossover

Permutation

Strain

Situation: parent group, fitness –> children

Data set

Survival of the fittest

Select parents class

Crossover

Permutation

Strain