pyxtal.optimize.base module

A base class for global optimization including:

• WFS: Width First Sampling

• DFS: Depth First Sampling

• QRS: Quasi Random Sampling

class pyxtal.optimize.base.GlobalOptimize(smiles: str, workdir: str, sg: int | list[int], tag: str, info: dict[any, any] | None = None, ff_opt: bool = False, ff_style: str = 'openff', ff_parameters: str = 'parameters.xml', reference_file: str = 'references.xml', ref_criteria: dict[any, any] | None = None, N_cpu: int = 1, cif: str | None = None, block: list[any] | None = None, num_block: list[any] | None = None, composition: list[any] | None = None, lattice: Lattice | None = None, torsions: list[any] | None = None, molecules: list[pyxtal_molecule] | None = None, sites: list[any] | None = None, use_hall: bool = False, skip_ani: bool = True, factor: float = 1.1, eng_cutoff: float = 5.0, E_max: float = 10000000000.0, random_state=None, max_time: float | None = None, matcher: StructureMatcher | None = None, early_quit: bool = True, check_stable: bool = False, use_mpi: bool = False, pre_opt: bool = False)

Bases: object

Base-class for all global optimization methods

Parameters:

• smiles (str) – smiles string

• workdir (str) – path of working directory

• sg (int or list) – space group number or list of spg numbers

• tag (string) – job prefix

• ff_opt (bool) – activate on the fly FF mode

• ff_style (str) – automated force style (gaff or openff)

• ff_parameters (str or list) – ff parameter xml file or list

• reference_file (str) – path of reference xml data for FF training

• N_cpu (int) – number of cpus for parallel calculation (default: 1)

• cif (str) – cif file name to store all structure information

• block – block mode

• num_block – list of blocks

• compositions – list of composition, (default is [1]*Num_mol)

• lattice (bool) – whether or not supply the lattice

• torsions – list of torsion angle

• molecules (list) – list of pyxtal_molecule objects

• sites (list) – list of wp sites, e.g., [[‘4a’]]

• use_hall (bool) – whether or not use hall number (default: False)

• skip_ani (bool) – whether or not use ani or not (default: True)

• eng_cutoff (float) – the cutoff energy for FF training

• E_max (float) – maximum energy defined as an invalid structure

• matcher – structurematcher from pymatgen

• early_quit – whether quit the program early when the target is found

• pre_opt – whether pre_optimize the structure or not

check_ref(reps=None, reference=None, filename='pyxtal.cif')

Check if ground state structure is found.

Parameters:

• reps – list of representations

• refernce – [pmg, eng]

• filename – filename

count_pxrd_match(xtals, matches)

Wrap up the matched PXRD results

Parameters:

• xtals – list of (xtal, tag) tuples

• matches (list) – list of XRD matches

early_termination(success_rate)

Check if the calculation can be terminated early.

export_references(xtals, engs, N_min=50, dE=2.5, FMSE=2.5)

Add trainning data for FF optimization

Parameters:

• xtals – a list of pyxtals

• engs – a list of energies

• N_min (int) – minimum number of configs to add

• dE (float) – the cutoff energy value

• FMSE (float) – the cutoff Force MSE value

gen_summary(t0, gen_results, xtals)

Write the generic summary for each generation.

Parameters:

• t0 (float) – time stamp

• gen_results – list of results (id, xtal, match)

• xtals – list of (xtal, tag) tuples

get_label(i, label='cpu')

local_optimization(xtals, qrs=False, pool=None)

Perform MPI optimization for each structure in each generation.

Parameters:

• xtals – list of (xtal, tag) tuples

• qrs (bool) – Force mutation or not (related to QRS)

local_optimization_mpi(xtals, qrs, pool)

Perform MPI optimization for each structure in each generation.

Parameters:

• xtals – list of (xtal, tag) tuples

• qrs (bool) – Force mutation or not (related to QRS)

local_optimization_mproc(xtals, ncpu, ids=None, qrs=False, pool=None)

Perform optimization for each structure in multiprocess mode.

Parameters:

• xtals – list of (xtal, tag) tuples

• ncpu (int) – number of parallel python processes

• ids (list) – list of ids of the associated xtals

• qrs (bool) – Force mutation or not (related to QRS)

local_optimization_serial(xtals, qrs=False)

Perform optimization for each structure in each generation.

Parameters:

• xtals – list of (xtal, tag) tuples

• qrs (bool) – Force mutation or not (related to QRS)

new_struc(xtal, xtals)

plot_results(save=True, figsize=(8.0, 5.0), figname=None, ylim=None)

Plot the results

Parameters:

• save (bool) – whether or not save the data

• figsize – e.g. (8.5, 5.0)

• figname (str)

• ylim – e.g. (0, 1.0)

print(*args, **kwargs)

Utility method to print only from rank 0.

print_matches(header=None)

Formatted output for the matched structures with xtal rep and eng rank

run(ref_pmg=None, ref_pxrd=None)

The main code to run Sampling

Parameters:

• ref_pmg – reference pmg structure

• ref_pxrd – reference pxrd profile in 2D array

Returns:

success_rate or None

save(filename)

Save the base class

select_xtals(ref_xtals, ids, N_max)

Select only unique structures

success_count(xtals, matches)

Wrap up the matched results and count success rate.

Parameters:

• xtals – list of (xtal, tag) tuples

• matches (list) – list of matches [True, False, ..]

Returns:

success_rate

pyxtal.optimize.base.arr_to_text(arr)

Convert a 2D array to a text

pyxtal.optimize.base.load_xml(filename, tag='GO')

Load the base class

pyxtal.optimize.base.process_task(args)

pyxtal.optimize.base.run_optimizer_with_timeout(args, logger)

Run the optimizer with a timeout. This function will be executed by each process.

pyxtal.optimize.base.setup_worker_logger(log_file)

Set up the logger for each worker process.

pyxtal.optimize.base.text_to_2darr(text, dtype=<class 'float'>)

Convert a text to an 2D array

pyxtal.optimize.base.text_to_arr(text, dtype=<class 'float'>)

Convert a text to an array

pyxtal.optimize.base.text_to_bool(text)

Convert text string to boolean value

pyxtal.optimize.base.text_to_float(text)

Convert a text to a float

pyxtal.optimize.base.text_to_list(text, dtype=<class 'float'>)

Convert a text to a list In the form of “[1 2 3]”