# -*- coding: utf-8 -*-
"""Global control parameters for DFTB+
"""
import logging
import dftbplus_step
import seamm
logger = logging.getLogger(__name__)
[docs]
class OptimizationParameters(dftbplus_step.EnergyParameters):
"""The control parameters for optimization using DFTB+"""
parameters = {
"optimization method": {
"default": "Rational Function",
"kind": "enumeration",
"default_units": "",
"enumeration": (
"Rational Function",
"Limited-memory Broyden-Fletcher-Goldfarb-Shanno (LBFGS)",
"Fast inertial relaxation engine (FIRE)",
"Steepest descent",
),
"format_string": "s",
"description": "Method:",
"help_text": ("The optimization method to use."),
},
"MaxForceComponent": {
"default": 1.0e-04,
"kind": "float",
"default_units": "hartree/bohr",
"enumeration": tuple(),
"format_string": ".1f",
"description": "Convergence criterion:",
"help_text": (
"Optimisation is stopped if the force component with the "
"maximal absolute value goes below this threshold."
),
},
"MaxSteps": {
"default": 200,
"kind": "integer",
"default_units": "",
"enumeration": tuple(),
"format_string": "",
"description": "Maximum number of steps:",
"help_text": (
"Maximum number of steps after which the optimisation should "
"stop (unless already stopped by achieving convergence). "
"Setting this value as -1 runs a huge() number of iterations."
),
},
"LatticeOpt": {
"default": "Yes",
"kind": "string",
"default_units": "",
"enumeration": ("Yes", "No"),
"format_string": "",
"description": "Optimize the cell (if periodic):",
"help_text": (
"Allow the lattice vectors to change during optimisation. "
"MovedAtoms can be optionally used with lattice optimisation "
"if the atomic coordinates are to be co-optimised with the "
"lattice."
),
},
"pressure": {
"default": 0.0,
"kind": "float",
"default_units": "",
"enumeration": tuple(),
"format_string": ".1f",
"description": "Pressure:",
"help_text": ("The applied pressure."),
},
"DiagLimit": {
"default": 0.01,
"kind": "float",
"default_units": "",
"enumeration": tuple(),
"format_string": "",
"description": "Limit for diagonal Hessian elements:",
"help_text": (
"The lower limit for the diagonal Hessian elements in the BFGS-like "
"update step in the rational optimizer."
),
},
"Memory": {
"default": 20,
"kind": "integer",
"default_units": "",
"enumeration": tuple(),
"format_string": "",
"description": "Number of steps to remember:",
"help_text": (
"Number of last steps which are saved and used to calculate "
"the next step via the LBFGS algorithm. The literature "
"recommends that Memory should between 3 and 20."
),
},
"StepSize": {
"default": 100.0,
"kind": "float",
"default_units": "a_u_time",
"enumeration": tuple(),
"format_string": "",
"description": "Step size:",
"help_text": (
"Step size (dt) along the forces. The displacement dxi along "
"the ith coordinate is given for each atom as "
"dxi = fi2mdt2, where fi is the appropriate force component "
"and m is the mass of the atom."
),
},
"nMin": {
"default": 5,
"kind": "integer",
"default_units": "",
"enumeration": tuple(),
"format_string": "",
"description": "Steps before increasing stepsize:",
"help_text": (
"The minimum number of steps before the step size is increased."
),
},
"aPar": {
"default": 0.1,
"kind": "float",
"default_units": "",
"enumeration": tuple(),
"format_string": "",
"description": "Velocity update parameter:",
"help_text": "Parameter for the update of the velocities.",
},
"fInc": {
"default": 1.1,
"kind": "float",
"default_units": "",
"enumeration": tuple(),
"format_string": "",
"description": "Stepsize increase factor:",
"help_text": "The factor to increase the step size",
},
"fDec": {
"default": 0.5,
"kind": "float",
"default_units": "",
"enumeration": tuple(),
"format_string": "",
"description": "Stepsize decrease factor:",
"help_text": "The factor to decrease the step size",
},
"fAlpha": {
"default": 0.99,
"kind": "float",
"default_units": "",
"enumeration": tuple(),
"format_string": "",
"description": "Alpha update on reset:",
"help_text": "The factor for the update the alpha parameter on reset.",
},
}
def __init__(self, defaults={}, data=None):
"""Initialize the instance, by default from the default
parameters given in the class"""
super().__init__(
defaults={
**OptimizationParameters.parameters,
**seamm.standard_parameters.structure_handling_parameters,
**defaults,
},
data=data,
)
# Do any local editing of defaults
tmp = self["structure handling"]
tmp.description = "Structure handling:"
tmp = self["system name"]
tmp._data["enumeration"] = (*tmp.enumeration, "optimized with {Hamiltonian}")
tmp.default = "keep current name"
tmp = self["configuration name"]
tmp._data["enumeration"] = ["optimized with {Hamiltonian}", *tmp.enumeration]
tmp.default = "optimized with {Hamiltonian}"
[docs]
def update(self, data):
"""Update values from a dict
This version filters out old, obsolete parameters for compatibility
"""
for key in (
"constrain_cell",
"fix_a",
"fix_b",
"fix_c",
"MaxAtomStep",
"MaxLatticeStep",
"stop_if_scc_fails",
"Alpha",
"Generations",
"LineSearch",
):
if key in data:
del data[key]
for key in ("system name", "configuration name"):
if key in data and data[key] == "optimized with <Hamiltonian>":
data[key] = "optimized with {Hamiltonian}"
super().update(data)
