hpmc.update
Overview
Apply box updates to sample isobaric and related ensembles. |
|
Apply geometric cluster algorithm (GCA) moves. |
|
Quickly compress a hard particle system to a target box. |
Details
HPMC updaters.
- class hoomd.hpmc.update.BoxMC(betaP, trigger=1)
Apply box updates to sample isobaric and related ensembles.
- Parameters
betaP (
float
orhoomd.variant.Variant
) – \(\frac{p}{k_{\mathrm{B}}T}\) \([\mathrm{length}^{-2}]\) in 2D or \([\mathrm{length}^{-3}]\) in 3D.trigger (hoomd.trigger.Trigger) – Select the timesteps to perform box trial moves.
Use
BoxMC
in conjunction with an HPMC integrator to allow the simulation box to undergo random fluctuations at constant pressure.BoxMC
supports both isotropic (all box sides changed equally) and anisotropic volume change moves as well as shearing of the simulation box. Multiple types of box moves can be applied simultaneously during a simulation. For this purpose, each type of box move has an associated weight that determines the relative frequency of a box move happening relative to the others. By default, no moves are applied (weight values for all move types default to 0). After a box trial move is proposed, all the particle positions are scaled into the new box. Trial moves are then accepted, if they do not produce an overlap, according to standard Metropolis criterion and rejected otherwise.Mixed precision
BoxMC
uses reduced precision floating point arithmetic when checking for particle overlaps in the local particle reference frame.- volume
Parameters for isobaric volume moves that scale the box lengths uniformly. The dictionary has the following keys:
weight
(float) - Relative weight of volume box moves.mode
(str) -standard
proposes changes to the box volume andln
proposes changes to the logarithm of the volume. Initially starts off in ‘standard’ mode.delta
(float) - Maximum change in V or ln(V) where V is box area (2D) or volume (3D).
- Type
- aspect
Parameters for isovolume aspect ratio moves. The dictionary has the following keys:
weight
(float) - Relative weight of aspect box moves.delta
(float) - Maximum relative change of box aspect ratio.
- Type
- length
Parameters for isobaric box length moves that change box lengths independently. The dictionary has the following keys:
weight
(float) - Maximum change of HOOMD-blue box parameters Lx, Ly, and Lz.delta
(tuple[float, float, float]) - Maximum change of the box lengths(Lx, Ly, Lz)
\([\mathrm{length}]\).
- Type
- shear
Parameters for isovolume box shear moves. The dictionary has the following keys:
weight
(float) - Relative weight of shear box moves.delta
(tuple[float, float, float]) - maximum change of the box tilt factor(xy, xz, yz)
.reduce
(float) - Maximum number of lattice vectors of shear to allow before applying lattice reduction. Values less than 0.5 disable shear reduction.
- Type
- instance
When using multiple
BoxMC
updaters in a single simulation, give each a unique value forinstance
so they generate different streams of random numbers.- Type
- property aspect_moves
The accepted and rejected aspect moves.
(0, 0) before the first call to
Simulation.run
.(
Loggable
: category=”sequence”)
- property counter
Trial move counters.
The counter object has the following attributes:
volume
:tuple
[int
,int
] - Number of accepted and rejected volume and length moves.shear
:tuple
[int
,int
] - Number of accepted and rejected shear moves.aspect
:tuple
[int
,int
] - Number of accepted and rejected aspect moves.
Note
The counts are reset to 0 at the start of each call to
hoomd.Simulation.run
. Before the first call toSimulation.run
,counter
isNone
.
- property shear_moves
The accepted and rejected shear moves.
(0, 0) before the first call to
Simulation.run
.(
Loggable
: category=”sequence”)
- class hoomd.hpmc.update.Clusters(pivot_move_probability=0.5, flip_probability=0.5, trigger=1)
Apply geometric cluster algorithm (GCA) moves.
- Parameters
The GCA as described in Liu and Lujten (2004), http://doi.org/10.1103/PhysRevLett.92.035504 is used for hard shape, patch interactions and depletants.
Implicit depletants are supported and simulated on-the-fly, as if they were present in the actual system.
Supported moves include pivot moves (point reflection) and line reflections (pi rotation around an axis). With anisotropic particles, the pivot move cannot be used because it would create a chiral mirror image of the particle, and only line reflections are employed. In general, line reflections are not rejection free because of periodic boundary conditions, as discussed in Sinkovits et al. (2012), http://doi.org/10.1063/1.3694271 . However, we restrict the line reflections to axes parallel to the box axis, which makes those moves rejection-free for anisotropic particles, but the algorithm is then no longer ergodic for those and needs to be combined with local moves.
Mixed precision
Clusters
uses reduced precision floating point arithmetic when checking for particle overlaps in the local particle reference frame.
- class hoomd.hpmc.update.QuickCompress(trigger, target_box, max_overlaps_per_particle=0.25, min_scale=0.99)
Quickly compress a hard particle system to a target box.
- Parameters
trigger (Trigger) – Update the box dimensions on triggered time steps.
target_box (Box) – Dimensions of the target box.
max_overlaps_per_particle (float) – The maximum number of overlaps to allow per particle (may be less than 1 - e.g. up to 250 overlaps would be allowed when in a system of 1000 particles when max_overlaps_per_particle=0.25).
min_scale (float) – The minimum scale factor to apply to box dimensions.
Use
QuickCompress
in conjunction with an HPMC integrator to scale the system to a target box size.QuickCompress
can typically compress dilute systems to near random close packing densities in tens of thousands of time steps.It operates by making small changes toward the
target_box
:L_new = scale * L_current
for each box parameter and then scaling the particle positions into the new box. If there are more thanmax_overlaps_per_particle * N_particles
hard particle overlaps in the system, the box move is rejected. Otherwise, the small number of overlaps remain.QuickCompress
then waits until local MC trial moves provided by the HPMC integrator remove all overlaps before it makes another box change.Note
The target box size may be larger or smaller than the current system box, and also may have different tilt factors. When the target box parameter is larger than the current, it scales by
L_new = 1/scale * L_current
QuickCompress
adjusts the value ofscale
based on the particle and translational trial move sizes to ensure that the trial moves will be able to remove the overlaps. It chooses a value ofscale
randomly betweenmax(min_scale, 1.0 - min_move_size / max_diameter)
and 1.0 wheremin_move_size
is the smallest MC translational move size adjusted by the acceptance ratio andmax_diameter
is the circumsphere diameter of the largest particle type.Tip
Use the
hoomd.hpmc.tune.MoveSize
in conjunction withQuickCompress
to adjust the move sizes to maintain a constant acceptance ratio as the density of the system increases.Warning
When the smallest MC translational move size is 0,
QuickCompress
will scale the box by 1.0 and not progress toward the target box.Mixed precision
QuickCompress
uses reduced precision floating point arithmetic when checking for particle overlaps in the local particle reference frame.- max_overlaps_per_particle
The maximum number of overlaps to allow per particle (may be less than 1 - e.g. up to 250 overlaps would be allowed when in a system of 1000 particles when max_overlaps_per_particle=0.25).
- Type
- instance
When using multiple
QuickCompress
updaters in a single simulation, give each a unique value forinstance
so that they generate different streams of random numbers.- Type
- property complete
True when the box has achieved the target.