qimpy.dft.geometry.Dynamics

class Dynamics(*, comm, dt, n_steps, thermostat=None, seed=1234, T0=298.0 K, P0=1.0 bar, stress0=None, t_damp_T=50.0 fs, t_damp_P=100.0 fs, drag_wavefunctions=True, save_history=True, report_callback=None, checkpoint_in=(None, ''))

Bases: TreeNode

Molecular dynamics of ions and/or lattice. Whether lattice changes is controlled by lattice.movable.

Parameters:

• comm (Comm)

• dt (float)

• n_steps (int)

• thermostat (Thermostat)

• seed (int)

• T0 (float)

• P0 (UnitOrFloat)

• stress0 (Tensor)

• t_damp_T (float)

• t_damp_P (float)

• drag_wavefunctions (bool)

• save_history (bool)

• report_callback (Callable[[Dynamics, int], None] | None)

• checkpoint_in (CheckpointPath)

__init__(*, comm, dt, n_steps, thermostat=None, seed=1234, T0=298.0 K, P0=1.0 bar, stress0=None, t_damp_T=50.0 fs, t_damp_P=100.0 fs, drag_wavefunctions=True, save_history=True, report_callback=None, checkpoint_in=(None, ''))

Specify molecular dynamics parameters.

Parameters:

• dt (float) – [Input file] Time step.

• n_steps (int) – [Input file] Number of MD steps.

• thermostat (Thermostat | dict | str | None) – [Input file] Thermostat/barostat method. Specify name of thermostat eg. ‘nose-hoover’ if using default options for that thermostat method, and dictionary of parameters if not.

• seed (int) – [Input file] Random seed for initial velocities.

• T0 (Unit | float) – [Input file] Initial temperature / temperature set point.

• P0 (Unit | float) – [Input file] Pressure set point for NPT, if lattice.movable is True. Note that this is overridden by stress0, if that is specified.

• stress0 (ndarray | Tensor | None) – [Input file] Stress set point for N-stress-T, if lattice.movable is True. If specified and lattice.movable, strain tensor will fluctuate during dynamics, instead of only volume in NPT mode. (Set to None and specify P0 instead for NPT mode.)

• t_damp_T (Unit | float) – [Input file] Thermostat damping time.

• t_damp_P (Unit | float) – [Input file] Barostat damping time.

• drag_wavefunctions (bool) – [Input file] Whether to drag atomic components of wavefunctions.

• save_history (bool) – [Input file] Whether to save history along the trajectory. Saved quantities include positions, forces, velocities, temperature, pressure, potential and total stress (if available), and lattice (if movable).

• report_callback (Callable[[Dynamics, int], None] | None) – Optional function to call at each step during report. Use this to perform additional reporting / data collection. The functional will be called as report_callback(dynamics, i_iter).

• comm (Comm)

• checkpoint_in (CheckpointPath)

Return type:

None

Methods

__init__	Specify molecular dynamics parameters.
add_child	Construct child object self.`attr_name` of type cls.
add_child_one_of	Invoke add_child on one of several child options in args.
create_gradient	Create gradient from ionic part, initializing optional parts correctly.
get_KE	Compute kinetic energy from ion velocity.
get_T	Compute temperature from kinetic energy KE.
get_acceleration	Acceleration due to ionic forces.
get_masses	Collect the masses of all ions as an n_ions x 1 tensor.
get_pressure
get_stress	Compute total stress tensor including ion velocity contributions.
report
run
save_checkpoint	Save self and all children in hierarchy to cp_path.
thermal_velocities	Thermal velocity distribution at T, randomized with seed.

Attributes

nDOF	Number of degrees of freedom in the dynamics.
system	System being optimized currently
masses	n_ions x 1 for bcast)
stepper
comm	Communictaor over which forces consistent
dt	Time step
n_steps	Number of MD steps
thermostat	Thermostat/barostat method
seed	Random seed for initial velocities
T0	Initial temperature / temperature set point
stress0	Stress set point (used only if lattice.movable)
isotropic	Whether lattice change is isotropic (NPT, vs.
t_damp_T	Thermostat damping time
t_damp_P	Barostat damping time
B0	Characteristic bulk modulus for Berendsen barostat
langevin_gamma	Damping rate for Langevin thermostat
drag_wavefunctions	Whether to drag atomic components of wavefunctions
P	Current pressure (available if lattice.compute_stress)
T	Current temperature
KE	Current kinetic energy
stress	Current stress including kinetic contributions
history	Utility to save trajectory data
report_callback	Callback from report
i_iter_start	Starting iteration number (when continuing from checkpoint)
child_names	Names of attributes with child objects.
variant_name	Version of children having variants (if any)

create_gradient(ions)

Create gradient from ionic part, initializing optional parts correctly.

Parameters:

ions (Tensor)

Return type:

Gradient

get_KE(velocity)

Compute kinetic energy from ion velocity.

Parameters:

velocity (Tensor)

Return type:

float

get_T(KE)

Compute temperature from kinetic energy KE.

Parameters:

KE (float)

Return type:

float

get_acceleration()

Acceleration due to ionic forces.

Return type:

Gradient

static get_masses(ions)

Collect the masses of all ions as an n_ions x 1 tensor.

Parameters:

ions (Ions)

Return type:

Tensor

get_stress(velocity)

Compute total stress tensor including ion velocity contributions.

Parameters:

velocity (Tensor)

Return type:

Tensor | None

thermal_velocities(T, seed)

Thermal velocity distribution at T, randomized with seed.

Parameters:

• T (float)

• seed (int)

Return type:

Tensor

B0: float

Characteristic bulk modulus for Berendsen barostat

KE: float

Current kinetic energy

P: float | None

Current pressure (available if lattice.compute_stress)

T: float

Current temperature

T0: float

Initial temperature / temperature set point

comm: Comm

Communictaor over which forces consistent

drag_wavefunctions: bool

Whether to drag atomic components of wavefunctions

dt: float

Time step

history: History | None

Utility to save trajectory data

i_iter_start: int

Starting iteration number (when continuing from checkpoint)

isotropic: bool

Whether lattice change is isotropic (NPT, vs. N-stress-T mode)

langevin_gamma: float

Damping rate for Langevin thermostat

masses: Tensor

n_ions x 1 for bcast)

Type:

Mass of each ion in system (Dim

property nDOF: int

Number of degrees of freedom in the dynamics.

n_steps: int

Number of MD steps

report_callback: Callable[[Dynamics, int], None] | None

Callback from report

seed: int

Random seed for initial velocities

stress: Tensor | None

Current stress including kinetic contributions

stress0: Tensor

Stress set point (used only if lattice.movable)

system: System

System being optimized currently

t_damp_P: float

Barostat damping time

t_damp_T: float

Thermostat damping time

thermostat: Thermostat

Thermostat/barostat method