tqchem.ttconf.factory

Functions

indices_from_rdkit_conformers(→ list[list[int]])

Calculates indices closest to the rdkit structures

solve_atom_collisions(→ list[int])

If atoms collide, search for a new set of indices without atom collisions

initial_tt_indices(→ list[list[int]])

Return indices to initialize the tensor train in TTOpt

molecularGridObjective(...)

Construct a MolecularGridObjective

ttconf_optimizer(→ tqchem.ttconf.ttconf.TTopt_Optimizer)

Create Optimizer for the parameters provided

Module Contents

tqchem.ttconf.factory.indices_from_rdkit_conformers(molgrid: tqchem.internal_coordinates.MolecularGrid, n_conformers: int, indices: list[list[int]], charge: int = 0) → list[list[int]]

Calculates indices closest to the rdkit structures

tqchem.ttconf.factory.solve_atom_collisions(molgrid: tqchem.internal_coordinates.MolecularGrid, initial_indices: list[int]) → list[int]

If atoms collide, search for a new set of indices without atom collisions

tqchem.ttconf.factory.initial_tt_indices(molecules: list[tqchem.internal_coordinates.MolecularSystem], molgrid: tqchem.internal_coordinates.MolecularGrid, rank: int, solve_collisions: bool, charge: int = 0) → list[list[int]]

Return indices to initialize the tensor train in TTOpt

Creates initial indices from provided molecules, rdkit conformers and random selection.

tqchem.ttconf.factory.molecularGridObjective(molgrid: tqchem.internal_coordinates.MolecularGrid, method: str, filter_type: str, charge: int = 0, solvent: str = None, threads: int = 4, reference_energy: float = None, local_optimization: bool = True, ensemble_optimization: bool = False) → tqchem.ttconf.ttconf.MolecularGridObjective

Construct a MolecularGridObjective

tqchem.ttconf.factory.ttconf_optimizer(molecules: tqchem.internal_coordinates.MolecularSystem | list[tqchem.internal_coordinates.MolecularSystem], method: str = 'gfn2-xtb', n_sweeps: int = 2, rank: int = 2, seed: int = 42, threads: int = 4, charge: int = 0, solvent: str = None, filter_type: str = 'energy', reference_energy: float = None, local_optimization: bool = True, initialize_indices: bool = True, ensemble_optimization: bool = False, solve_collisions: bool = True, **bond_grids) → tqchem.ttconf.ttconf.TTopt_Optimizer

Create Optimizer for the parameters provided

Parameters:

  • molecules ((MolecularSystem | list[MolecularSystem])) – Molecules provided as MolecularSystem

  • method (str, default="gfn2-xtb") – Electronic structure method used Options: “GFN1-xTB”, “GFN2-xTB”, “Sage-FF”, “Amber-FF”, case insensitive

  • n_sweeps (int, default=2) – Number of sweeps through the tensor train

  • rank (int, default=2) – Rank used in the tensor train

  • seed (int, default=42) – Seed for random number selection

  • threads (int, default=4) – Number of threads to use in the ensemble optimization

  • charge (int, default=0) – Molecular charge

  • solvent (str, default=None) – Solvent name Options: “water”, “explicit water” and None (i.e. gas phase), case insensitive

  • filter_type (str, default="energy") – Type of function to filter the energy with Options: “energy”: E_i “energy difference”: E_i - E_0 “boltzmann energy difference”: -exp(-(E_i - E_0)) where: E_0 is the energy of the current molecular system E_i is the energy of the system in iteration i

  • reference_energy (float, default=None) – Reference energy for filter

  • initialize_indices (bool, default=True) – Initialize the initial tensor in TTOpt with the given molecules and rdkit Conformers

  • ensemble_optimization (bool, default=False) – Use ensemble optimization of CREST for gradient optimization

  • solve_collisions (bool, default=True) – Check for atom collisions for the initial indices and resolves them

  • bond_grids –

    Keywords specifying which grids to use when setting dihedrals:

    • include_amide_bonds: bool = False,

      Include amide bonds as rotatable bonds

    • add_rotatable_bonds: dict[tuple[int, int], np.array] = None,

      Adds rotatable bonds to the default ones in terms of a dict relating bonds to an array of allowed dihedral values

    • set_rotatable_bonds: dict[tuple[int, int], np.array] = None,

      Overwrites rotatable bonds with a dict relating bonds to an array of allowed dihedral values

    • amide_grid: np.array, default=[0.0, 180.0]

      Grid of allowed dihedral values for amide bonds

    • amino_grid: np.array, default=[0, 60, 120, 180, 240, 300]

      Grid of allowed dihedral values for amino bods

    • alkane_grid: np.array, default=[0, 60, 120, 180, 240, 300]

      Grid of allowed dihedral values for alkane bonds

    • t_butyl_grid: np.array, default=[0.0, 60.0]

      Grid of allowed dihedral values for t_butyl

    • iso_propyl_grid: np.array, default=[0.0, 60.0, 120.0]

      Grid of allowed dihedral values for iso_propyl

    • chalcogen_grid: np.array, default=[0, 60, 120, 180, 240, 300]

      Grid of allowed dihedral values for chalcogen bonds

    • default_grid: np.array, default=[0, 30, 60, 90, 120, 150, 180, 210, 240, 270, 300, 330]

      Grid of allowed dihedral values for all other bonds

    • overwrite_grids: np.array, default=None

      Overwrite all bond specific grids with this grid

Return type:

TToptOptimizer

Raises:

  • NotImplementedError: – if provided method or filter_type does not match the allowed ones

  • TypeError: – if molecule input does not match the allowed types