This release adds a few features:
-
openPMD-viewer
can now read complex datasets (See #353) -
Avoid errors in LPA diagnostics in the absence of selected particles (See #358)
This release fixes a few miscellaneous bugs:
-
Better 3D reconstruction for
theta=None
(See #344) -
Better support for ADIOS data (See #355)
-
Support for group-based encoding (See #346)
This new release introduces several improvements:
-
The changes introduced in 1.3.0 caused a major slowdown when reading certain types of data. This has been fixed in this new release. (See #340 for more details.)
-
openPMD-viewer
now supportsthetaMode
geometry with data written usingr
as the fastest index (as written by e.g. WarpX) in addition to the previously supported data format which usedz
as the fastest index (as written by e.g. fbpic). (See 337) -
openPMD-viewer
will raise an exception if the user asks for an iteration that is not part of the dataset (instead of printing a message and reverting to the first iteration, which can be confusing) (See 336)
This new release introduces preliminary support for MR datasets (see #332).
This new release introduces several bug-fixes and miscellaneous features:
-
There is a new function
get_energy_spread
that returns the energy spread of the beam. This is partially redundant withget_mean_gamma
, which is kept for backward compatibility. (see #304 and #317) -
The 3D field reconstruction from
ThetaMode
data now has an optionmax_resolution_3d
that limits the resolution of the final 3D array. This is added in order to limit the memory footprint of this array. (see #307) The 3D reconstruction is now also more accurate, thanks to the implementation of linear interpolation. (see #311) -
A bug that affected reading
ThetaMode
data with theopenpmd-api
backend has been fixed. (see #313) -
A bug that affected
get_laser_waist
has been fixed: (see #320)
This new release introduces the option to read openPMD
files with different backends. In addition to the legacy h5py
backend (which can read only HDF5 openPMD file), openPMD-viewer
now has the option to use the openpmd-api
backend (which can read both HDF5 and ADIOS openPMD files). Because the openpmd-api
backend is thus more general, it is selected by default if available (i.e. if installed locally).
The user can override the default choice, by passing the backend
argument when creating an OpenPMDTimeSeries
object, and check which backend has been chosen by inspecting the .backend
attribute of this object.
In addition, several smaller changes were introduced in this PR:
- The method
get_laser_envelope
can now take the argumentlaser_propagation
in order to support lasers that do not propagates along thez
axis. openPMD-viewer
can now properly readgroupBased
openPMD files (i.e. files that contain several iterations) #301.- Users can now pass arrays of ID to the
ParticleTracker
#283
This is a bug-fix release.
-
Unreadable files are now skipped (instead of crashing the whole timeseries) ; see #262.
-
A bug related to units (microns vs meters) was fixed in
get_emittance
andget_current
(see #276) -
The quick-start notebook (
openPMD-visualization
) raised a warning sayingNotebook validation failed
in some cases. This was fixed (see #274) -
When using the option
plot=True
with Python 2, openPMD-viewer crashed. This is now fixed (see #271).
This version introduces major changes and breaks backward compatibility.
Here is a list of the changes:
- The import statement now uses
openpmd_viewer
instead ofopmd_viewer
, e.g.
from openpmd_viewer import OpenPMDTimeSeries
- For consistency,
ts.get_particle
now return particle positions in meters, instead of microns. For instance, in the code below,x
,y
,z
will be in meters
x, y, z = ts.get_particle(['x', 'y', 'z'], iteration=1000)
- In
ts.get_field
, slicing can now be done in several directions, and for 1d, 2d, 3d, and circ geometries. As a consequence, this breaks backward compatibility for 3d field:get_field(field=field, coord=coord, iteration=iteration)
used to return the central slice alongy
while it now returns the full 3d field. In addition, the name of the argument ofget_field
that triggers slicing has been changed fromslicing_dir
toslice_across
(andslicing
has been changed toslice_relative_position
). openPMD-viewer
does not rely on Cython anymore. Instead, it usesnumba
for functions that perform a substantial amount of computation.- A new function (
ts.iterate
) was introduced in order to quickly apply a given function to all iterations of a time series. See the docstring ofts.iterate
for more information. - The function
get_laser_envelope
does not support the argumentindex
anymore (which was effectively used in order to perform slicing). Instead, users should use the argumentslicing_dir
. In addition,get_laser_envelope
now supports the argumentplot_range
. - The function
get_laser_waist
does not support the agumentslicing_dir
anymore.
This release adds two features:
- Improved calculation of the laser envelope, using the Hilbert transform.
- Reconstruction of full 3D field from a quasi-3D dataset, when passing
theta=None
.
This is a bug-fix release. It allows the slider to work properly in JupyterLab,
by using the %matplotlib widget
magic.
This version includes minor improvements to the viewer:
- (Experiemental) support for Windows users
- In the interactive Jupyter GUI, the user can now select the scale of the vertical axis.
- The function
get_emittance
has more options (including calculation of the slice emttance) - The default
openPMD_notebook
now avoids warning messages about matplotlib inline, which used to occur even though%matplotlib notebook
was used.
Many thanks to @MaxThevenet and @AngelFP for their contributions to this release!
This version introduces several improvements to the viewer:
- The ability to read files that contain fields in different geometries (e.g. 3D fields and 2D slices).
- Better support for files that do not contain mesh (or do not contain particles), including support for the openPMD 1.1.0 standard.
- Cloud-In-Cell deposition in histograms.
- Better handling of
%matplotlib notebook
for newer version of jupyter.
This version adds better support, when the local installation of matplotlib has issues:
- The
LpaDiagnostics
can now work without matplotlib if needed. - The
MacOSX
matplotlib backend is now avoided, since there can be issues when using it in the latest version of Jupyter.
This version improves support for ipywidgets
version 7, especially in
the layout of the slider.
In addition, with this version of openPMD-viewer
, matplotlib
is not a
strict requirement anymore. This allows lighter installation for users that
need openPMD-viewer
only as a data reader.
Finally, the calculation of the laser envelope in 2D has been improved
(see PR 170). Note
that the function wstd
(which is not documented in the tutorial, but
which some users might still use) has been renamed to w_std
.
This version improves the layout of the Jupyter GUI and allows the user to select a particular region of the plots through this GUI.
In addition, support for massless particle (e.g. photons) was added. In this case, the momenta are returned in kg.m.s^-1, instead of using the dimensionless momenta.
This is version 0.5.4
of openPMD-viewer.
It adds support for Python 3.4 (which erroneously dropped in the past).
This is version 0.5.3
of openPMD-viewer.
It corrects some of the issues with the size of boxes and widgets in the interactive slider. In addition, the iteration number is now read from the hdf5 metadata, and not the name of the file.
This is version 0.5.2
of openPMD-viewer.
It fixes some of the installation issues associated with Cython.
This is version 0.5.1
of openPMD-viewer.
It corrects a minor bug in the dependencies of the package.
This is version 0.5.0
of openPMD-viewer.
This new version includes the ParticleTracker
object, which allows user to track individual particles across different iterations, provided that their id
is stored in the openPMD file. Also, starting with this version, openPMD-viewer now depends on Cython
.
For more information on how to use the ParticleTracker
, see the tutorial notebook.
This is version 0.4.0
of openPMD-viewer.
This new version includes:
- support for 1D data
- an additional option
use_field_mesh
when plotting the particle. When set toTrue
, this option uses information from the field mesh to choose the parameters of the particle histograms (esp. the bins). This is useful in order to avoid plotting/binning artifacts (aliasing) when the particles are evenly spaced.
In addition, the module openpmd_viewer
now has an attribute __version__
.
This is version 0.3.3
of openPMD-viewer.
This version fixed a bug with the executable openPMD_notebook
. More precisely, the executable was not installed, when using pip
or conda
. In addition, it was failing with Python 3.
This is version 0.3.2
of openPMD-viewer. The following changes were introduced:
- The conda recipe in
conda_recipe/
was simplified and streamlined. - The documentation now explains how to install openPMD-viewer with
conda
, the instructions to release the package was put into a documentRELEASING.md
. - A file
MANIFEST.in
was added, to avoid issues with pip and Python 3.
This is version 0.3.1
of openPMD-viewer. This version introduces minor changes in the way the tests are run in setup.py
. The aim of these changes are to prepare a conda release.
This is version 0.3.0
of openPMD-viewer. This version mainly adapts the interactive GUI so that it can be used with the newer version of ipwidgets
(ipywidgets 5.0
), while still being compatible with previous versions of ipwidgets
. A number of other minor changes have been introduced:
- In the method
get_particle
, the argumentspecies
is now optional in the case where there is only one species. - A number of methods in the LPA addons (
LpaDiagnostics
class) now have an optional argumentplot
, which allows to directly plot the data.
This is version 0.2.0
of openPMD-viewer. A number of minor changes and fixes have been made in order to make the package more general and to prepare it for a PyPI release. Here are the main changes:
- Support for the deprecated widget package
IPython.html
has been dropped. From now on, users need to install the widget packageipywidgets
, for the GUI to work. - The initialization of an
OpenPMDTimeSeries
object can now be made faster by setting the optional argumentcheck_all_files
toFalse
. - The data reader can now support
macroWeighted
quantities. As consequence, output files from PIConGPU can now be correctly read. - The package does not assume anymore that all species contain the same particle quantities. For instance, the package will support a file that contains the positions of ions, and the positions, momenta and weighting of electrons. As part of this, the attribute
OpenPMDTimeSeries.avail_ptcl_quantities
has been replaced by a dictionaryOpenPMDTimeSeries.avail_record_components
. - This release introduces automatic PEP8 verification as part of the automatic tests that are run on Travis CI (see CONTRIBUTING.md).
- The evaluation of the waist and duration of the laser is now based on Gaussian fit of the transverse and longtudinal profile respectively.