File I/O

This page describes the steps needed to read simulation data from disk and to write computation results back to disk.

Data import

The most common way of importing simulations into OVITO is to load a file from disk using the import_file() function:

from ovito.io import import_file
pipeline = import_file("simulation.dump")

This high-level function works similar to the Load File function from OVITO’s graphical user interface and automatically detects the format of the specified file. See the list of supported file formats. import_file() returns a new Pipeline object, which is connected to a FileSource for loading the specified data file as input to the pipeline.

In case you would like to re-use an existing Pipeline, returned by an earlier call to import_file(), it is possible to subsequently change the input file with a call to the FileSource.load() method:

pipeline.source.load("other_simulation.dump")

The load() method accepts the same parameters as the global import_file() function.

Note

Unlike the Load File function in OVITO’s graphical user interface, the import_file() function does not automatically insert the loaded dataset into the three-dimensional scene. That means the dataset will initially not appear in rendered images or in the interactive viewports. Thus, if you want the imported dataset to be visible, you need to explicitly invoke the method Pipeline.add_to_scene() to place the pipeline’s output into the current visualization scene. More on this topic can be found in the Rendering & visualization section.

Loading simulation trajectories

The FileSource class and the import_file() function support loading a series of simulation snapshots (a trajectory). This happens automatically if the trajectory frames are all stored in a single file. Many simulation codes, however, produce a series of files containing one frame each. To load such a file series, which follows a naming pattern such as frame.0.dump, frame.1000.dump, frame.2000.dump, etc., pass a wildcard pattern to the import_file() function:

pipeline = import_file("/path/frame.*.dump")

OVITO automatically finds all files matching the pattern (must all be in one directory) and loads them as one trajectory. The third option is to specify the list of files explicitly:

file_list = [
    "dir_a/simulation.dump",
    "dir_b/simulation.dump",
    "dir_c/simulation.dump"
]
pipeline = import_file(file_list)

The num_frames property of the pipeline source tells you how many frames are in the loaded simulation trajectory:

print(pipeline.source.num_frames)

Note

To save memory and time, OVITO does not load all frames of a trajectory at once. The call to import_file() lets OVITO quickly scan the directory or the multi-frame file to discover all frames belonging to the trajectory. The actual data of a frame will only be loaded on demand, one at a time, whenever the pipeline is evaluated at a certain animation time, e.g., when jumping to a new frame in the animation or rendering a movie.

Some MD simulation codes store the topology of a molecular system (i.e., the definition of atom types, bonds, etc.) and the atomic trajectories in two separate files. In this case, load the topology file first using import_file(). Then create and apply a LoadTrajectoryModifier, which will load the time-dependent atomic positions from the separate trajectory file:

pipeline = import_file("topology.data")
traj_mod = LoadTrajectoryModifier()
traj_mod.source.load('trajectory.dump')
pipeline.modifiers.append(traj_mod)

Accessing individual frames of a trajectory

Once a simulation trajectory was loaded using import_file(), we can step through the individual frames of the sequence using a for-loop:

for frame in range(pipeline.source.num_frames):
    data = pipeline.compute(frame)
    ...

In the loop, the Pipeline.compute() method is called with the frame number as argument at which the pipeline should be evaluated. As part of this compute() call, the pipeline’s FileSource will fetch the input data of the requested frame from the external simulation file(s). Note that frame numbering starts at 0 in OVITO.

File column to property mapping

When loading a simulation file containing atoms or other types of particles, OVITO needs to map the stored per-particle information to corresponding particle properties within OVITO’s internal data model. Typically, this mapping happens automatically. Certain file formats, however, do not contain sufficient information to perform it automatically. For instance, when loading a legacy XYZ file, which can contain any number of file columns with user-defined meanings, the mapping of these file columns to OVITO’s particle properties needs to be explicitly specified using the columns keyword:

pipeline = import_file("simulation.xyz", columns =
         ["Particle Type", "Position.X", "Position.Y", "Position.Z", "My Property"])

The number of entries in the columns list must match the number of data columns of the XYZ input file. See the documentation of the import_file() function for more information on this.

Data export

Exporting data to an output file is typically done using the global ovito.io.export_file() function. For example, to export the particles and their properties, some of which may have been computed by a Pipeline, one would write:

from ovito.io import export_file

export_file(pipeline, "outputfile.dump", "lammps/dump",
    columns = ["Position.X", "Position.Y", "Position.Z", "My Property"])

The second and third function parameters specify the output filename and the file format. For a list of supported file formats, see the export_file() documentation. Depending on the selected format, additional keyword arguments may need to be specified. For instance, in the example above, the columns parameter specifies the list of particle properties to be exported to the output file.

In addition to particles, export_file() can also export other types of data computed by OVITO. One example are attributes, which are global quantities computed by modifiers in a pipeline. In other words, attributes are global information associated with the dataset as a whole. For example, the ExpressionSelectionModifier outputs the ExpressionSelection.count attribute to report the number of particles that matched the given selection criterion.

We can export the value of this dynamically computed attribute to a text file, typically for all frames of a trajectory as a table. Such a table could then be used to produce a chart of the time evolution of the quantity using an external plotting program. For this purpose the export_file() function supports the txt/attr output format:

pipeline = import_file("simulation*.dump")

modifier = ExpressionSelectionModifier(expression = "PotentialEnergy < -3.9")
pipeline.modifiers.append(modifier)

export_file(pipeline, "potenergy.txt", "txt/attr", multiple_frames = True,
         columns = ["Frame", "ExpressionSelection.count"])

The multiple_frames option tells export_file() to evaluate the pipeline for all frames of the loaded trajectory. Without it, only the first frame (frame 0) would have been exported to the output file. The program above produces a text file containing one line per animation frame:

# "Frame" "ExpressionSelection.count"
0 531
1 540
2 522
3 502
...

The first column contains the animation frame number (starting at 0) and the second column contains the value of the ExpressionSelection.count attribute calculated by the ExpressionSelectionModifier as part of the data pipeline.

Typically, global attributes are dynamically computed by modifiers in the pipeline, but some may also be directly read from the input data file. For example, an attribute named Timestep is automatically generated by OVITO when importing a LAMMPS dump file, reporting the simulation timestep number at the current animation frame. This makes it possible, for example, to replace the animation frame number in the first file column above with the actual timestep of the MD simulation. See DataCollection.attributes for further information on global attributes.