2.3. ConformerOptimizer

The ConformerOptimizer is not meant to be directly called by the user, but exists as a component of the Generator classes. It is also used under the hood by the UniqueConformers.optimize() method.

This class organizes parallel geometry optimizations for different geometries of a single molecule

class ConformerOptimizer(mol=None, engine_settings=None, nproc=1)

Machine that can optimize the geometry of a set of structures

A simple example of (parallel) use:

>>> from scm.plams import Molecule
>>> from scm.plams import init, finish
>>> from scm.conformers import UniqueConformersCrest, ConformerOptimizer

>>> # Set up the molecular data
>>> mol = Molecule('mol.xyz')
>>> conformers = UniqueConformersCrest()
>>> conformers.prepare_state(mol)

>>> # Set up PLAMS settings
>>> init()

>>> # Create the optimizer and run
>>> optimizer = ConformerOptimizer(mol, nproc=4)
>>> geometries = [mol.as_array()]
>>> geometries, energies = optimizer.optimize_geometries(geometries, level='good')

>>> finish()

The default AMS engine used is the UFF engine. A different engine can be provided upon initiation.

>>> engine_settings = Settings()
>>> engine_settings.DFTB.Model = 'GFN1-xTB'
>>> optimizer = ConformerOptimizer(geometries, engine_settings=engine, nproc=4)
__init__(mol=None, engine_settings=None, nproc=1)

Initiates an instance of the Optimizer class

  • engine_settings – PLAMS Settings object:

    engine_settings = Settings() engine_settings.DFTB.Model = ‘GFN1-xTB’

  • nproc – Number of processors to use in total for the optimizations

property last_timings

Stores the timings of the last optimization call performed

set_atom_types()

In case the GAFF ForceField will be called, make sure atom-typing happens only once

FIXME: This also requires a patch file!

set_cleaning_preference(keep=None)

Sets the preferences for the files saved to disc

  • keep – None or ‘all’. The former will clean PLAMS folders of all files after runtime, the latter will leave all files as they are. See PLAMS for more details.

keep_ams_running(keep=True)

Technical setting for a set of (serial) geometry optimizations

  • keep – Each of the parallel processes will most likely be passed multiple geometries. If keep=True, AMS will be kept running in the background, and geometries will be passed via a pipe. If keep=False, each geometry optimization will start a new AMS process.

optimize_geometries(geometries, level=None, name='go', watch=True)

Optimize a set of geometries

  • geometries – An iterator over molecular coordinates (matching self.mol)

optimize_molecules(molecules, level=None, name='go', watch=True)

Optimize a set of molecules (they have to represent different conformers of the same molecule)

geometry_writer(results, num_jobs, event, t)

Writes the optimized geometries to file during the optimization process

convert_key_to_camelcase(key)

Convert the key to json format

convert_key_to_underscore_case(key)

Revert back to original value, and see if it works

convert_keys_to_camelcase(settings)

Convert all keys to CamelCase (except the last json ones like _type, _default, etc)

convert_keys_to_underscore_case(settings)

Revert all the keys to the Python settings

static create_json_entry(typename, value, unique=True, choices=None, include=None)

Create a single entry

get_json_settings()

Convert the settings object in self to a json style settings object

handle_input(inp, names, handle_nested_objects)

Get the settings from the input, convert to underscore case, and insert them

  • inp - Input object

  • names - The names (self._name) of all the classes in the family of classes of self

  • handle_nested_objects - Boolean specifying if there can be encapsulated objects of the same family,

    for which settings are needed

pass_settings(settings, encaps_settings=False)

Set the settings object into self

read_settings(inp, names)

Get the settings for the conformers object from the input in the required format

  • inp - Input object

  • names - The classnames of all the classes in the family of classes of obj

Note: This is complicated by the fact that a single family of classes may be spread over multiple input blocks.

set_blocknames(blocknames)

Provide the relevant blocknames for the family of classes

  • blocknames – List of strings representing the input blocks relevant for this family of classes

to_json()

Create a json settings object from self.settings