Job

The Job class is the base class defining a BigDFT calculation.

class mybigdft.job.Job(name='', inputparams=None, posinp=None, run_dir=None, ref_data_dir=None, skip=False, pseudos=False)

Bases: object

This class is meant to define a BigDFT calculation. run() is its main method and it must be used in a context manager to ensure that the calculation is run the desired directory.

You may pass input parameters and/or initial geometry (posinp). Make sure to at least provide initial positions, either via the posinp or the input parameters.

You may give a name for the calculation, used to name the input and output files written on disk (default naming conventions are used if not). You can also specify the directory where to run the calculation with run_dir.

A reference calculation may be given in order to copy its data directory to the present calculation (main use: restart from the wavefunctions of the reference calculation).

Parameters

• inputparams (InputParams or None) – BigDFT input parameters.

• posinp (Posinp or None) – BigDFT initial geometry file.

• name (str) – Prefix of the BigDFT calculation (used to define the input and output file names).

• run_dir (str or None) – Folder where to run the calculation (default to current directory).

• ref_data_dir (str) – Path to the data directory of a reference BigDFT calculation.

• skip (bool) – If True, the calculation will be skipped. (Note: Might not be useful now, since we check for the existence of the logfile before running, which might be the actual check of the skip option of BigDFT.)

• pseudos (bool) – If True, the pseudopotential files stored in $PSEUDODIR will be used to complete the job.

Raises

ValueError – If no initial positions are given in the posinp or the input parameters.

A Job instance can be initialized by using a posinp only:

>>> from mybigdft import Posinp, Atom
>>> pos = Posinp(
...     [Atom('N', [2.9763078243490115e-23, 6.872205952043537e-23,
...                 0.01071619987487793]),
...      Atom('N', [-1.1043449194501671e-23, -4.873421744830746e-23,
...                 1.104273796081543])], "angstroem", "free"
... )
>>> job = Job(posinp=pos, run_dir="tests")

Default values are therefore used for the input parameters:

>>> job.inputparams
{}

Input and output file names are defined from the name passed as argument. Here, no name is passed, so that default names are used:

>>> job.input_name
'input.yaml'
>>> job.posinp_name
'posinp.xyz'
>>> job.logfile_name
'log.yaml'

The directories are defined from the run_dir argument:

>>> import os
>>> os.getcwd() == job.init_dir
True
>>> os.path.basename(job.init_dir) != 'tests'
True
>>> os.path.basename(job.run_dir)
'tests'

There is no logfile associated to the job yet as it was not run:

>>> job.logfile == {}
True

To run the job, do it from a context manager:

>>> with job as j:
...     j.run()
...
/.../tests
Logfile log.yaml already exists!
<BLANKLINE>

A logfile being found, it is read and not computed again:

>>> job.logfile == {}
False

property name

Returns

Base name of the calculation used to set the names of files and directories as well as the commands.

Return type

str

property inputparams

Returns

Input parameters of the calculation.

Return type

InputParams

property posinp

Returns

Initial positions of the calculation.

Return type

Posinp or None

property logfile

Returns

Logfile of the calculation (output of the bigdft or bigdft-tool executable).

Return type

Logfile or None

property ref_data_dir

Returns

Reference directory where some relevant data (such as wavefunctions) is stored.

Return type

str

property pseudos

Returns

if True, the calculation uses the pseudopotential files in $PSEUDODIR (environment variable).

Return type

bool

property skip

Returns

If True, the calculation will be skipped. (Note: Might not be useful now, since we check for the existence of the logfile before running, which might be the actual check of the skip option of BigDFT.)

Return type

bool

property init_dir

Returns

Absolute path to the initial directory of the calculation (can differ from run_dir()).

Return type

str

property run_dir

Returns

Absolute path to the directory where the calculation is run.

Return type

str

property data_dir

Returns

Absolute path to the data directory of the calculation.

Return type

str

property bigdft_tool_cmd

Returns

Base command to run the bigdft-tool executable.

Return type

list

property bigdft_cmd

Returns

Base command to run the bigdft executable.

Return type

list

property input_name

Returns

Name of the input parameters file.

Return type

str

property posinp_name

Returns

Name of the input position file.

Return type

str

property logfile_name

Returns

Name of the logfile.

Return type

str

property is_completed

Returns

True if the job has already run successfully.

Return type

bool

__enter__()

When entering the context manager:

• create the directory where the calculations must be run,

• go to that directory.

__exit__(*args)

When leaving the context manager, go back to the initial directory.

run(nmpi=1, nomp=1, force_run=False, dry_run=False, restart_if_incomplete=False, timeout=None)

Run the BigDFT calculation if it was not already performed. The number of MPI and OpenMP tasks may be specified.

You may force the calculation to run even though it was previously successful (e.g., a logfile already exists) by setting force_run to True.

If dry_run is set to True, then bigdft-tool is run instead of the BigDFT executable.

If restart_if_incomplete is set to True, the previously existing logfile is removed and the calculation restarts.

Parameters

• nmpi (int) – Number of MPI tasks.

• nomp (int) – Number of OpenMP tasks.

• force_run (bool) – If True, the calculation is run even though a logfile already exists.

• dry_run (bool) – If True, the input files are written on disk, but the bigdft-tool command is run instead of the bigdft one.

• restart_if_incomplete (bool) – If True, the job is restarted if the existing logfile is incomplete.

• timeout (float or int or None) – Number of minutes after which the job must be stopped.

write_input_files()

Write the input files on disk (there might be no posinp to write, since the initial positions can be defined in the input parameters).

clean(data_dir=False, logfiles_dir=False)

Delete all input and output files on disk as well as some directories if required.

Parameters

• data_dir (bool) – If True, removes the data directory that might exist.

• logfiles (bool) – If True, removes the logfiles directory that might exist.

Warning

The directories are forced to be removed when the above- mentioned options are set to True: use with caution.