hoomd.write#

Overview

Burst

Write the last \(N\) stored frames in the GSD format.

DCD

Writes simulation trajectories in the DCD format.

CustomWriter

User-defined writer.

GSD

Write simulation trajectories in the GSD format.

Table

Write delimiter separated values to a stream.

Details

Writers.

Writers write the state of the simulation, logger quantities, or calculated results to output files or streams:

  • GSD and DCD save the simulation trajectory to a file.

  • Combine GSD with a hoomd.logging.Logger to save system properties or per-particle calculated results.

  • Use Table to display the status of the simulation periodically to standard out.

  • Implement custom output formats with CustomWriter.

Writers do not modify the system state.

Tip

OVITO has native support for GSD files, including logged per-particle array quantities and particle shapes.

See also

Tutorial: Introducing HOOMD-blue

Tutorial: Logging

class hoomd.write.Table(trigger, logger, output=stdout, header_sep='.', delimiter=' ', pretty=True, max_precision=10, max_header_len=None)#

Write delimiter separated values to a stream.

Use Table to write scalar and string hoomd.logging.Logger quantities to standard out or to a file.

Warning

When logger quantities include strings with spaces, the default space delimiter will result in files that are not machine readable.

Important

All attributes for this class are static. They cannot be set to new values once created.

Parameters:
  • trigger (hoomd.trigger.trigger_like) – The trigger to determine when to run the Table backend.

  • logger (hoomd.logging.Logger) – The logger to query for output. The ‘scalar’ categories must be set on the logger, and the ‘string’ categories is optional.

  • output (file-like object , optional) – A file-like object to output the data from, defaults to standard out. The object must have write and flush methods and a mode attribute. Examples include sys.stdout, sys.stderr and the return value of open().

  • header_sep (str, optional) – String to use to separate names in the logger’s namespace, defaults to '.'. For example, if logging the total energy of an hoomd.md.pair.LJ pair force object, the default header would be md.pair.LJ.energy (assuming that max_header_len is not set).

  • delimiter (str, optional) – String used to separate elements in the space delimited file, defaults to ' '.

  • pretty (bool, optional) – Flags whether to attempt to make output prettier and easier to read, defaults to True. To make the output easier to read, the output will compromise on numerical precision for improved readability. In many cases, the precision will still be high with pretty set to True.

  • max_precision (int, optional) – If pretty is not set, then this controls the maximum precision to use when outputing numerical values, defaults to 10.

  • max_header_len (int, optional) – If not None (the default), limit the outputted header names to length max_header_len. When not None, names are grabbed from the most specific to the least. For example, if set to 7 the namespace ‘hoomd.md.pair.LJ.energy’ would be set to ‘energy’. Note that at least the most specific part of the namespace will be used regardless of this setting (e.g. if set to 5 in the previous example, ‘energy’ would still be the header).

trigger#

The trigger to determine when to run the Table backend.

Type:

hoomd.trigger.Trigger

logger#

The logger to query for output. The ‘scalar’ categories must be set on the logger, and the ‘string’ categories is optional.

Type:

hoomd.logging.Logger

output#

A file-like object to output the data from. The object must have write and flush methods and a mode attribute.

Type:

file-like object

header_sep#

String to use to separate names in the logger’s namespace.’. For example, if logging the total energy of an hoomd.md.pair.LJ pair force object, the default header would be md.pair.LJ.energy (assuming that max_header_len is not set).

Type:

str

delimiter#

String used to separate elements in the space delimited file.

Type:

str

pretty#

Flags whether to attempt to make output prettier and easier to read. To make the output easier to read, the output will compromise on outputted precision for improved readability. In many cases, though the precision will still be high with pretty set to True.

Type:

bool

max_precision#

If pretty is not set, then this controls the maximum precision to use when outputing numerical values, defaults to 10.

Type:

int, optional

max_header_len#

Limits the outputted header names to length max_header_len when not None. Names are grabbed from the most specific to the least. For example, if set to 7 the namespace ‘hoomd.md.pair.LJ.energy’ would be set to ‘energy’. Note that at least the most specific part of the namespace will be used regardless of this setting (e.g. if set to 5 in the previous example, ‘energy’ would still be the header).

Type:

int

min_column_width#

The minimum allowed column width.

Type:

int

write()#

Write out data to self.output.

Writes a row from given hoomd.logging.Logger object data.

class hoomd.write.Burst(trigger, filename, filter=hoomd._hoomd.ParticleFilter, mode='ab', dynamic=None, logger=None, max_burst_size=-1, write_at_start=False)#

Bases: GSD

Write the last \(N\) stored frames in the GSD format.

When triggered, Burst stores up to the last \(N\) frames in a buffer. Burst writes the contents of the buffer to the file when dump is called. When the the next frame would result in \(N + 1\) frames being stored, the oldest frame is removed and the new frame is added.

Parameters:
  • trigger (hoomd.trigger.trigger_like) – Select the timesteps to store in the buffer.

  • filename (str) – File name to write.

  • filter (hoomd.filter.filter_like) – Select the particles to write. Defaults to hoomd.filter.All.

  • mode (str) – The file open mode. Defaults to 'ab'.

  • dynamic (list[str]) – Field names and/or field categores to save in all frames. Defaults to ['property'].

  • logger (hoomd.logging.Logger) – Provide log quantities to write. Defaults to None.

  • max_burst_frames (int) – The maximum number of frames to store before between writes. -1 represents no limit. Defaults to -1.

  • write_at_start (bool) – When True and the file does not exist or has 0 frames: write one frame with the current state of the system when hoomd.Simulation.run is called. Defaults to False.

Warning

Burst errors when attempting to create a file or writing to one with zero frames, unless write_at_start is True.

Note

When analyzing files created by Burst, generally the first frame is not associated with the call to Burst.dump.

Note

For more tips and qualifications see the hoomd.write.GSD documentation.

filename#

File name to write.

Type:

str

trigger#

Select the timesteps to store in the buffer.

Type:

hoomd.trigger.Trigger

filter#

Select the particles to write.

Type:

hoomd.filter.filter_like

mode#

The file open mode.

Type:

str

dynamic#

Field names and/or field categores to save in all frames.

Type:

list[str]

max_burst_frames#

The maximum number of frames to store before between writes. -1 represents no limit.

Type:

int

write_diameter#

When False, do not write particles/diameter. Set to True to write non-default particle diameters.

Type:

bool

write_at_start#

When True and the file does not exist or has 0 frames: write one frame with the current state of the system when hoomd.Simulation.run is called.

Type:

bool

dump()#

Write all currently stored frames to the file and empties the buffer.

This method alllows for custom writing of frames at user specified conditions.

class hoomd.write.CustomWriter(trigger, action)#

Bases: CustomOperation, Writer

User-defined writer.

Parameters:

CustomWriter is a hoomd.operation.Writer that wraps a user-defined hoomd.custom.Action object so the action can be added to a hoomd.Operations instance for use with hoomd.Simulation objects.

Writers may read the system state and generate output files or print to output streams. Writers should not modify the system state.

class hoomd.write.DCD(trigger, filename, filter=hoomd._hoomd.ParticleFilter, overwrite=False, unwrap_full=False, unwrap_rigid=False, angle_z=False)#

Bases: Writer

Writes simulation trajectories in the DCD format.

Parameters:
  • trigger (hoomd.trigger.Periodic) – Select the timesteps to write.

  • filename (str) – File name to write.

  • filter (hoomd.filter.filter_like) – Select the particles to write. Defaults to hoomd.filter.All.

  • overwrite (bool) – When False, (the default) an existing DCD file will be appended to. When True, an existing DCD file filename will be overwritten.

  • unwrap_full (bool) – When False, (the default) particle coordinates are always written inside the simulation box. When True, particles will be unwrapped into their current box image before writing to the DCD file.

  • unwrap_rigid (bool) – When False, (the default) individual particles are written inside the simulation box which breaks up rigid bodies near box boundaries. When True, particles belonging to the same rigid body will be unwrapped so that the body is continuous. The center of mass of the body remains in the simulation box, but some particles may be written just outside it. unwrap_rigid is ignored when unwrap_full is True.

  • angle_z (bool) – When True, the particle orientation angle is written to the z component (only useful for 2D simulations)

DCD writes the simulation trajectory to the specified file in the DCD file format. DCD stores only particle positions and the box parameters, in length units, and is limited to simulations where the number of particles is fixed.

Examples:

writer = hoomd.write.DCD("trajectory.dcd", hoomd.trigger.Periodic(1000))
dcd = hoomd.write.DCD(filename="data/dump.dcd",
                      trigger=hoomd.trigger.Periodic(100, 10))

Warning

When you use DCD to append to an existing DCD file:

  • The period must be the same or the time data in the file will not be consistent.

  • DCD will not write out data at time steps that already are present in the DCD file.

filename#

File name to write.

Type:

str

trigger#

Select the timesteps to write.

Type:

hoomd.trigger.Periodic

filter#

Select the particles to write.

Type:

hoomd.filter.filter_like

overwrite#

When False, an existing DCD file will be appended to. When True, an existing DCD file filename will be overwritten.

Type:

bool

unwrap_full#

When False, particle coordinates are always written inside the simulation box. When True, particles will be unwrapped into their current box image before writing to the DCD file.

Type:

bool

unwrap_rigid#

When False, individual particles are written inside the simulation box which breaks up rigid bodies near box boundaries. When True, particles belonging to the same rigid body will be unwrapped so that the body is continuous. The center of mass of the body remains in the simulation box, but some particles may be written just outside it. unwrap_rigid is ignored when unwrap_full is True.

Type:

bool

angle_z#

When True, the particle orientation angle is written to the z component

Type:

bool

class hoomd.write.GSD(trigger, filename, filter=hoomd._hoomd.ParticleFilter, mode='ab', truncate=False, dynamic=None, logger=None)#

Bases: Writer

Write simulation trajectories in the GSD format.

Parameters:
  • trigger (hoomd.trigger.trigger_like) – Select the timesteps to write.

  • filename (str) – File name to write.

  • filter (hoomd.filter.filter_like) – Select the particles to write. Defaults to hoomd.filter.All.

  • mode (str) – The file open mode. Defaults to 'ab'.

  • truncate (bool) – When True, truncate the file and write a new frame 0 each time this operation triggers. Defaults to False.

  • dynamic (list[str]) – Field names and/or field categores to save in all frames. Defaults to ['property'].

  • logger (hoomd.logging.Logger) – Provide log quantities to write. Defaults to None.

GSD writes the simulation trajectory to the specified file in the GSD format. GSD can store all particle, bond, angle, dihedral, improper, pair, and constraint data fields in every frame of the trajectory. GSD can write trajectories where the number of particles, number of particle types, particle types, diameter, mass, charge, or other quantities change over time. GSD can also store arbitrary, scalar, string, and array quantities provided by a hoomd.logging.Logger instance in logger.

Valid file open modes:

mode

description

'wb'

Open the file for writing. Create the file if needed, or overwrite an existing file.

'xb'

Create a GSD file exclusively. Raise an exception when the file exists.

'ab'

Create the file if needed, or append to an existing file.

To reduce file size, GSD does not write properties that are set to defaults. When masses, orientations, angular momenta, etc… are left default for all particles, these fields will not take up any space in the file, except on frame 1+ when the field is also non-default in frame 0. GSD writes all non-default fields to frame 0 in the file.

To further reduce file sizes, GSD allows the user to select which specific fields will be considered for writing to frame 1+ in the dynamic list. When reading a GSD file, the data in frame 0 is read when a quantity is missing in frame i, so any fields not in dynamic are fixed for the entire trajectory.

The dynamic list can contain one or more of the following strings:

  • 'property'

    • 'configuration/box'

    • 'particles/N'

    • 'particles/position'

    • 'particles/orientation'

  • 'momentum'

    • 'particles/velocity'

    • 'particles/angmom'

    • 'particles/image'

  • 'attribute'

    • 'particles/types'

    • 'particles/typeid'

    • 'particles/mass'

    • 'particles/charge'

    • 'particles/diameter'

    • 'particles/body'

    • 'particles/moment_inertia'

  • 'topology'

    • bonds/*

    • angles/*

    • dihedrals/*

    • impropers/*

    • constraints/*

    • pairs/*

When you set a category string ('property', 'momentum', 'attribute'), GSD makes all the category member’s fields dynamic.

Warning

GSD buffers writes in memory. Abnormal exits (e.g. kill, scancel, reaching walltime limits) may cause loss of data. Ensure that your scripts exit cleanly and call flush() as needed to write buffered frames to the file.

See also

See the GSD documentation, GSD HOOMD Schema, and GSD GitHub project for more information on GSD files.

Note

When you use filter to select a subset of the whole system, GSD writes only the selected particles in ascending tag order and does not write out topology.

Tip

All logged data fields must be present in the first frame in the gsd file to provide the default value. To achieve this, set the logger attribute before the operation is triggered to write the first frame in the file.

Some (or all) fields may be omitted on later frames. You can set logger to None or remove specific quantities from the logger, but do not add additional quantities after the first frame.

filename#

File name to write.

Type:

str

trigger#

Select the timesteps to write.

Type:

hoomd.trigger.Trigger

filter#

Select the particles to write.

Type:

hoomd.filter.filter_like

mode#

The file open mode.

Type:

str

truncate#

When True, truncate the file and write a new frame 0 each time this operation triggers.

Type:

bool

dynamic#

Field names and/or field categores to save in all frames.

Type:

list[str]

write_diameter#

When False, do not write particles/diameter. Set to True to write non-default particle diameters.

Type:

bool

maximum_write_buffer_size#

Size (in bytes) to buffer in memory before writing to the file.

Type:

int

flush()#

Flush the write buffer to the file.

Example:

gsd_writer.flush()

Flush all write buffers:

for writer in simulation.operations.writers:
    if hasattr(writer, 'flush'):
        writer.flush()
property logger#

Provide log quantities to write.

May be None.

Type:

hoomd.logging.Logger

static write(state, filename, filter=hoomd._hoomd.ParticleFilter, mode='wb', logger=None)#

Write the given simulation state out to a GSD file.

Parameters:

The valid file modes for write are 'wb' and 'xb'.