Source code for gpaw.lcaotddft

import os

import numpy as np
from typing import Optional
from gpaw.typing import Any, Vector

from ase.units import Bohr, Hartree

from gpaw.calculator import GPAW
from gpaw.external import ExternalPotential, ConstantElectricField
from gpaw.lcaotddft.hamiltonian import TimeDependentHamiltonian
from gpaw.lcaotddft.logger import TDDFTLogger
from gpaw.lcaotddft.propagators import create_propagator
from gpaw.tddft.units import attosec_to_autime

[docs]def LCAOTDDFT(filename: str, **kwargs) -> Any: if os.environ.get('GPAW_NEW'): from import RTTDDFT assert kwargs.get('propagator', None) in [None, 'ecn'], \ 'Not implemented yet' assert kwargs.get('rremisison', None) in [None], 'Not implemented yet' assert kwargs.get('fxc', None) in [None], 'Not implemented yet' assert kwargs.get('scale', None) in [None], 'Not implemented yet' assert kwargs.get('parallel', None) in [None], 'Not implemented yet' assert kwargs.get('communicator', None) in [None], \ 'Not implemented yet' new_tddft = RTTDDFT.from_dft_file(filename) return new_tddft return OldLCAOTDDFT(filename, **kwargs)
class OldLCAOTDDFT(GPAW): """Real-time time-propagation TDDFT calculator with LCAO basis. Parameters ---------- filename File containing ground state or time-dependent state to propagate propagator Time propagator for the Kohn-Sham wavefunctions td_potential External time-dependent potential rremission Radiation-reaction potential for Self-consistent Light-Matter coupling fxc Exchange-correlation functional used for the dynamic part of Hamiltonian scale Experimental option (use carefully). Scaling factor for the dynamic part of Hamiltonian parallel Parallelization options communicator MPI communicator txt Text output """ def __init__(self, filename: str, *, propagator: dict = None, td_potential: dict = None, rremission: object = None, fxc: str = None, scale: float = None, parallel: dict = None, communicator: object = None, txt: str = '-'): """""" assert filename is not None self.time = 0.0 self.niter = 0 # TODO: deprecate kick keywords (and store them as td_potential) self.kick_strength = np.zeros(3) self.kick_ext: Optional[ExternalPotential] = None self.tddft_initialized = False self.action = '' tdh = TimeDependentHamiltonian(fxc=fxc, td_potential=td_potential, scale=scale, rremission=rremission) self.td_hamiltonian = tdh self.propagator_set = propagator is not None self.propagator = create_propagator(propagator) self.default_parameters = GPAW.default_parameters.copy() self.default_parameters['symmetry'] = {'point_group': False} GPAW.__init__(self, filename, parallel=parallel, communicator=communicator, txt=txt) self.set_positions() def write(self, filename, mode=''): # This function is included here in order to generate # documentation for LCAOTDDFT.write() with autoclass in sphinx GPAW.write(self, filename, mode=mode) def _write(self, writer, mode): GPAW._write(self, writer, mode) if self.tddft_initialized: w = writer.child('tddft') w.write(time=self.time, niter=self.niter, kick_strength=self.kick_strength, propagator=self.propagator.todict()) self.td_hamiltonian.write(w.child('td_hamiltonian')) def read(self, filename): reader =, filename) if 'tddft' in reader: r = reader.tddft self.time = r.time self.niter = r.niter self.kick_strength = r.kick_strength if not self.propagator_set: self.propagator = create_propagator(r.propagator) else: self.log('Note! Propagator possibly changed!') self.td_hamiltonian.wfs = self.wfs def tddft_init(self): if self.tddft_initialized: return self.log('-----------------------------------') self.log('Initializing time-propagation TDDFT') self.log('-----------------------------------') self.log() assert self.wfs.dtype == complex self.timer.start('Initialize TDDFT') # Initialize Hamiltonian self.td_hamiltonian.initialize(self) # Initialize propagator self.propagator.initialize(self) self.log('Propagator:') self.log(self.propagator.get_description()) self.log() # Add logger TDDFTLogger(self) # Call observers before propagation self.action = 'init' self.call_observers(self.niter) self.tddft_initialized = True self.timer.stop('Initialize TDDFT') def absorption_kick(self, kick_strength: Vector): """Kick with a weak electric field. Parameters ---------- kick_strength Strength of the kick in atomic units """ self.tddft_init() self.timer.start('Kick') self.kick_strength = np.array(kick_strength, dtype=float) magnitude = np.sqrt(np.sum(self.kick_strength**2)) direction = self.kick_strength / magnitude self.log('---- Applying absorption kick') self.log('---- Magnitude: %.8f Hartree/Bohr' % magnitude) self.log('---- Direction: %.4f %.4f %.4f' % tuple(direction)) # Create hamiltonian object for absorption kick cef = ConstantElectricField(magnitude * Hartree / Bohr, direction) # Propagate kick self.propagator.kick(cef, self.time) # Call observers after kick self.action = 'kick' self.call_observers(self.niter) self.niter += 1 self.timer.stop('Kick') def kick(self, ext): """Kick with any external potential. Parameters ---------- ext External potential """ self.tddft_init() self.timer.start('Kick') self.log('---- Applying kick') self.log('---- %s' % ext) self.kick_ext = ext # Propagate kick self.propagator.kick(ext, self.time) # Call observers after kick self.action = 'kick' self.call_observers(self.niter) self.niter += 1 self.timer.stop('Kick') def propagate(self, time_step: float = 10.0, iterations: int = 2000): """Propagate the electronic system. Parameters ---------- time_step Time step in attoseconds iterations Number of propagation steps """ self.tddft_init() time_step *= attosec_to_autime self.maxiter = self.niter + iterations self.log('---- About to do %d propagation steps' % iterations) self.timer.start('Propagate') while self.niter < self.maxiter: # Propagate one step self.time = self.propagator.propagate(self.time, time_step) # Call registered callback functions self.action = 'propagate' self.call_observers(self.niter) self.niter += 1 self.timer.stop('Propagate') def replay(self, **kwargs): # TODO: Consider deprecating this function? self.propagator = create_propagator(**kwargs) self.tddft_init() self.propagator.control_paw(self)