This is a non-comprehensive log of changes to yt over its many releases.
The CREDITS file contains the
most up-to-date list of everyone who has contributed to the yt source code.
Version 3.3 is the first major release of yt since July 2015. It includes more
than 3000 commits from 41 contributors, including 12 new contributors.
- Raw and processed data from selections, projections, profiles and so forth can
now be saved in a ytdata format and loaded back in by yt. See
Saving Reloadable Data.
- Totally re-worked volume rendering API. The old API is still available for users
who prefer it, however. See 3D Visualization and Volume Rendering.
- Support for unstructured mesh visualization. See
Unstructured Mesh Slices and Unstructured Mesh Rendering.
- Interactive Data Visualization for AMR and unstructured mesh datasets. See
Interactive Data Visualization.
- Several new colormaps, including a new default, ‘arbre’. The other new
colormaps are named ‘octarine’, ‘kelp’, and ‘dusk’. All these new colormaps
were generated using the viscm package and should do a better job of
representing the data for colorblind viewers and when printed out in
grayscale. See Colormaps for more detail.
- New frontends for the ExodusII,
GAMER, and Gizmo data
- The unit system associated with a dataset is now customizable, defaulting to
CGS. See Unit Systems.
- Enhancements and usability improvements for analysis modules, especially the
light_ray modules. See
- Data objects can now be created via an alternative Numpy-like API. See
Slicing Syntax for Selecting Data.
- A line integral convolution plot modification. See
Overplot Line Integral Convolution.
- Many speed optimizations, including to the volume rendering, units, tests,
covering grids, the absorption spectrum and photon simulator analysis modules,
and ghost zone generation.
- Packaging and release-related improvements: better install and setup scripts,
automated PR backporting.
- Readability improvements to the codebase, including linting, removing dead
code, and refactoring much of the Cython.
- Improvements to the CI infrastructure, including more extensible answer tests
and automated testing for Python 3 and Windows.
- Numerous documentation improvements, including formatting tweaks, bugfixes,
and many new cookbook recipes.
- Support for geographic (lat/lon) coordinates.
- Several improvements for SPH codes, including alternative smoothing kernels,
add_smoothed_particle_field function, and particle type-aware octree
construction for Gadget data.
- Roundtrip conversions between Pint and yt units.
- Added halo data containers for gadget_fof frontend.
- Enabled support for spherical datasets in the BoxLib frontend.
- Many new tests have been added.
- Better hashing for Selector objects.
Minor enhancements and bugfixes
- Fixed many bugs related to Python 3 compatibility
- Fixed bugs related to compatibility issues with newer versions of numpy
- Added the ability to export data objects to a Pandas dataframe
- Added support for the fabs ufunc to YTArray
- Fixed two licensings issues
- Fixed a number of bugs related to Windows compatability.
- We now avoid hard-to-decipher tracebacks when loading empty files or
- Fixed a bug related to ART star particle creation time field
- Fixed a bug caused by using the wrong int type for indexing in particle deposit
- Fixed a NameError bug in comparing temperature units with offsets
- Fixed an API bug in YTArray casting during coercion from YTQuantity
- Added loadtxt and savetxt convenience functions for
- Fixed an issue caused by not sort species names with Enzo
- Fixed a units bug for RAMSES when
boxlen > 1.
process_chunk function for non-cartesian geometry.
scale_factor attribute to cosmological simulation datasets
- Fixed a bug where “center” vectors are used instead of “normal” vectors in
- Fixed issues involving invalid FRBs when uses called _setup_plots in their
- Added a
text_args keyword to
- Added a print_stats function for RAMSES
- Fixed a number of bugs in the Photon Simulator
- Added support for particle fields to the [Min,Max]Location derived quantities
- Fixed some units bugs for Gadget cosmology simulations
- Fixed a bug with Gadget/GIZMO StarFormationRate units
- Fixed an issue in TimeSeriesData where all the filenames were getting passed
load on each processor.
- Fixed a units bug in the Tipsy frontend
- Ensured that ARTIOIndex.get_smallest_dx() returns a quantity with units
- Ensured that plots are valid after invalidating the figure
- Fixed a bug regarding code unit labels
- Fixed a bug with reading Tipsy Aux files
- Added an effective redshift field to the Light Ray analysis module for use in
- Fixed a bug with the redshift calculation in LightRay analysis module
- Fixed a bug in the Orion frontend when you had more than 10 on-disk particle
fields in the file
- Detect more types of ART files
- Update derived_field_list in add_volume_weighted_smoothed_field
- Fixed casting issues for 1D and 2D Enzo simulations
- Avoid type indirection when setting up data object entry points
- Fixed issues with SIMPUT files
- Fixed loading athena data in python3 with provided parameters
- Tipsy cosmology unit fixes
- Fixed bad unit labels for compound units
- Making the xlim and ylim of the PhasePlot plot axes controllable
- Adding grid_arrays to grid_container
- An Athena and a GDF bugfix
- A small bugfix and some small enhancements for sunyaev_zeldovich
- Defer to coordinate handlers for width
- Make array_like_field return same units as get_data
- Fixing bug in ray “dts” and “t” fields
- Check against string_types not str
- Closed a loophole that allowed improper LightRay use
- Enabling AbsorptionSpectrum to deposit unresolved spectral lines
- Fixed an ART byte/string/array issue
- Changing AbsorptionSpectrum attribute lambda_bins to be lambda_field for
- No longer require user to save to disk when generating an AbsorptionSpectrum
- ParticlePlot FRBs can now use save_as_dataset and save attributes properly
- Added checks to assure ARTIO creates a metal_density field from existing metal
- Added mask to LightRay to assure output elements have non-zero density (a
problem in some SPH datasets)
- Added a “fields” attribute to datasets
- Updated the TransferFunctionHelper to work with new profiles
- Fixed a bug where the field_units kwarg to load_amr_grids didn’t do anything
- Changed photon_simulator’s output file structure
- Fixed a bug related to setting output_units.
- Implemented ptp operation.
- Added effects of transverse doppler redshift to LightRay
- Fixed a casting error for float and int64 multiplication in sdf class
- Added ability to read and write YTArrays to and from groups within HDF5 files
- Made ftype of “on-disk” stream fields “stream”
- Fixed a strings decoding issue in the photon simulator
- Fixed an incorrect docstring in load_uniform_grid
- Made PlotWindow show/hide helpers for axes and colorbar return self
- Made Profile objects store field metadata.
- Ensured GDF unit names are strings
- Tought off_axis_projection about its resolution keyword.
- Reintroduced sanitize_width for polar/cyl coordinates.
- We now fail early when load_uniform_grid is passed data with an incorrect shape
- Replaced progress bar with tqdm
- Fixed redshift scaling of “Overdensity” field in yt-2.x
- Fixed several bugs in the eps_writer
- Fixed bug affecting 2D BoxLib simulations.
- Implemented to_json and from_json for the UnitRegistry object
- Fixed a number of issues with ds.find_field_values_at_point[s]
- Fixed a bug where sunrise_exporter was using wrong imports
- Import HUGE from utilities.physical_ratios
- Fixed bug in ARTIO table look ups
- Adding support for longitude and latitude
- Adding halo data containers for gadget_fof frontend.
- Can now compare YTArrays without copying them
- Fixed several bugs related to active particle datasets
- Angular_momentum_vector now only includes space for particle fields if they
- Image comparison tests now print a meaningful error message if they fail.
- Fixed numpy 1.11 compatibility issues.
- Changed _skip_cache to be True by default.
- Enable support for spherical datasets in the BoxLib frontend.
- Fixed a bug in add_deposited_particle_field.
- Fixed issues with input sanitization in the point data object.
- Fixed a copy/paste error introduced by refactoring WeightedMenParticleField
- Fixed many formatting issues in the docs build
- Now avoid creating particle unions for particle types that have no common
- Patched ParticlePlot to work with filtered particle fields.
- Fixed a couple corner cases in gadget_fof frontend
- We now properly normalise all normal vectors in functions that take a normal
vector (for e.g get_sph_theta)
- Fixed a bug where the transfer function features were not always getting
- Made the Chombo frontend is_valid method smarter.
- Added a get_hash() function to yt/funcs.py which returns a hash for a file
- Added Sievert to the default unit symbol table
- Corrected an issue with periodic “wiggle” in AbsorptionSpectrum instances
ds.field_list sorted by default
- Bug fixes for the Nyx frontend
- Fixed a bug where the index needed to be created before calling derived
- Made latex_repr a property, computed on-demand
- Fixed a bug in off-axis slice deposition
- Fixed a bug with some types of octree block traversal
- Ensured that mpi operations retain ImageArray type instead of downgrading to
YTArray parent class
- Added a call to _setup_plots in the custom colorbar tickmark example
- Fixed two minor bugs in save_annocated
- Added ability to specify that DatasetSeries is not a mixed data type
- Fixed a memory leak in ARTIO
- Fixed copy/paste error in to_frb method.
- Ensured that particle dataset max_level is consistent with the index max_level
- Fixed an issue where fields were getting added multiple times to
- Enhanced annotate_ray and annotate_arrow callbacks
- Added GDF answer tests
- Made the YTFieldTypeNotFound exception more informative
- Added a new function, fake_vr_orientation_test_ds(), for use in testing
- Ensured that instances of subclasses of YTArray have the correct type
- Re-enabled max_level for projections, ProjectionPlot, and OffAxisProjectionPlot
- Fixed a bug in the Orion 2 field definitions
- Fixed a bug caused by matplotlib not being added to install_requires
- Edited PhasePlot class to have an annotate_title method
- Implemented annotate_cell_edges
- Handled KeyboardInterrupt in volume rendering Cython loop
- Made old halo finders now accept ptype
- Updated the latex commands in yt cheatsheet
- Fixed a circular dependency loop bug in abar field definition for FLASH
- Added neutral species aliases as described in YTEP 0003
- Fixed a logging issue: don’t create a StreamHandler unless we will use it
- Correcting how theta and phi are calculated in
- Fixed a bug related to the field dictionary in
- Allowed for the special case of supplying width as a tuple of tuples
- Made yt compile with MSVC on Windows
- Fixed a bug involving mask for dt in octree
- Merged the get_yt.sh and install_script.sh into one
- Added tests for the install script
- Allowed use axis names instead of dimensions for spherical pixelization
- Fixed a bug where close() wasn’t being called in HDF5FileHandler
- Enhanced commandline image upload/delete
- Added get_brewer_cmap to get brewer colormaps without importing palettable at
the top level
- Fixed a bug where a parallel_root_only function was getting called inside
another parallel_root_only function
- Exit the install script early if python can’t import ‘_ssl’ module
- Make PlotWindow’s annotate_clear method invalidate the plot
- Adding int wrapper to avoid deprecation warning from numpy
- Automatically create vector fields for magnetic_field
- Allow users to completely specify the filename of a 1D profile
- Force nose to produce meaningful traceback for cookbook recipes’ tests
- Fixed x-ray display_name and documentation
- Try to guess and load particle file for FLASH dataset
- Sped up top-level yt import
- Set the field type correctly for fields added as particle fields
- Added a position location method for octrees
- Fixed a copy/paste error in uhstack function
- Made trig functions give correct results when supplied data with dimensions of
angle but units that aren’t radian
- Print out some useful diagnostic information if check_for_openmp() fails
- Give user-added derived fields a default field type
- Added support for periodicity in annotate_particles.
- Added a check for whether returned field has units in volume-weighted smoothed
- Casting array indices as ints in colormaps infrastructure
- Fixed a bug where the standard particle fields weren’t getting set up
correctly for the Orion frontends
- Enabled LightRay to accept loaded datasets instead of just filenames
- Allowed for adding or subtracting arrays filled with zeros without checking
- Fixed a bug in selection for semistructured meshes.
- Removed ‘io’ from enzo particle types for active particle datasets
- Added support for FLASH particle datasets.
- Silenced a deprecation warning from IPython
- Eliminated segfaults in KDTree construction
- Fixed add_field handling when passed a tuple
- Ensure field parameters are correct for fields that need ghost zones
- Made it possible to use DerivedField instances to access data
- Added ds.particle_type_counts
- Bug fix and improvement for generating Google Cardboard VR in
- Made DarkMatterARTDataset more robust in its _is_valid
- Added Earth radius to units
- Deposit hydrogen fields to grid in gizmo frontend
- Switch to index values being int64
- ValidateParameter ensures parameter values are used during field detection
- Switched to using cythonize to manage dependencies in the setup script
- ProfilePlot style changes and refactoring
- Cancel terms with identical LaTeX representations in a LaTeX representation of
- Only return early from comparison validation if base values are equal
- Enabled particle fields for clump objects
- Added validation checks for data types in callbacks
- Enabled modification of image axis names in coordinate handlers
- Only add OWLS/EAGLE ion fields if they are present
- Ensured that PlotWindow plots continue to look the same under matplotlib 2.0
- Fixed bug in quiver callbacks for off-axis slice plots
- Only visit octree children if going to next level
- Check that CIC always gets at least two cells
- Fixed compatibility with matplotlib 1.4.3 and earlier
- Fixed two EnzoSimulation bugs
- Moved extraction code from YTSearchCmd to its own utility module
- Changed amr_kdtree functions to be Node class methods
- Sort block indices in order of ascending levels to match order of grid patches
- MKS code unit system fixes
- Disabled bounds checking on pixelize_element_mesh
- Updated light_ray.py for domain width != 1
- Implemented a DOAP file generator
- Fixed bugs for 2D and 1D enzo IO
- Converted mutable Dataset attributes to be properties that return copies
- Allowing LightRay segments to extend further than one box length
- Fixed a divide-by-zero error that occasionally happens in
- Make sure we have an index in subclassed derived quantities
- Added an initial draft of an extensions document
- Made it possible to pass field tuples to command-line plotting
- Ensured the positions of coordinate vector lines are in code units
- Added a minus sign to definition of sz_kinetic field
- Added grid_levels and grid_indices fields to octrees
- Added a morton_index derived field
- Added Exception to AMRKDTree in the case of particle of oct-based data
- Particle-Only Plots - a series of new plotting functions for visualizing
particle data. See here for more information.
- Late-stage beta support for Python 3 - unit tests and answer tests pass for
all the major frontends under python 3.4, and yt should now be mostly if not
fully usable. Because many of the yt developers are still on Python 2 at
this point, this should be considered a “late stage beta” as there may be
remaining issues yet to be identified or worked out.
- Now supporting Gadget Friend-of-Friends/Subfind catalogs - see here to learn
how to load halo catalogs as regular yt datasets.
- Custom colormaps can now be easily defined and added - see here to learn how!
- Now supporting Fargo3D data
- Performance improvements throughout the code base for memory and speed
- Various updates to the following frontends: ART, Athena, Castro, Chombo,
Gadget, GDF, Maestro, Pluto, RAMSES, Rockstar, SDF, Tipsy
- Numerous documentation updates
- Generic hexahedral mesh pixelizer
- Adding annotate_ray() callback for plots
- AbsorptionSpectrum returned to full functionality and now using faster SciPy
- Add a color_field argument to annotate_streamline
- Smoothing lengths auto-calculated for Tipsy Datasets
- Adding SimulationTimeSeries support for Gadget and OWLS.
- Generalizing derived quantity outputs to all be YTArrays or lists of
YTArrays as appropriate
- Star analysis returned to full functionality
- FITS image writing refactor
- Adding gradient fields on the fly
- Adding support for Gadget Nx4 metallicity fields
- Updating value of solar metal mass fraction to be consistent with Cloudy.
- Gadget raw binary snapshot handling & non-cosmological simulation units
- Adding support for LightRay class to work with Gadget+Tipsy
- Add support for subclasses of frontends
- Dependencies updated
- Serialization for projections using minimal representation
- Adding Grid visitors in Cython
- Improved semantics for derived field units
- Add a yaw() method for the PerspectiveCamera + switch back to LHS
- Adding annotate_clear() function to remove previous callbacks from a plot
- Added documentation for hexahedral mesh on website
- Speed up nearest neighbor evaluation
- Add a convenience method to create deposited particle fields
- UI and docs updates for 3D streamlines
- Ensure particle fields are tested in the field unit tests
- Allow a suffix to be specified to save()
- Add profiling using airspeed velocity
- Various plotting enhancements and bugfixes
- Use hglib to update
- Various minor updates to halo_analysis toolkit
- Docker-based tests for install_script.sh
- Adding support for single and non-cosmological datasets to LightRay
- Adding the Pascal unit
- Add weight_field to PPVCube
- FITS reader: allow HDU in auxiliary
- Fixing electromagnetic units
- Specific Angular Momentum [xyz] computed relative to a normal vector
- Adding ability to create union fields from alias fields
- Small fix to allow enzo AP datasets to load in parallel when no APs present
- Use proper cell dimension in gradient function.
- Minor memory optimization for smoothed particle fields
- Fix thermal_energy for Enzo HydroMethod==6
- Make sure annotate_particles handles unitful widths properly
- Improvements for add_particle_filter and particle_filter
- Specify registry in off_axis_projection’s image finalization
- Apply fix for particle momentum units to the boxlib frontend
- Avoid traceback in “yt version” when python-hglib is not installed
- Expose no_ghost from export_sketchfab down to _extract_isocontours_from_grid
- Fix broken magnetic_unit attribute
- Fixing an off-by-one error in the set x/y lim methods for profile plots
- Providing better error messages to PlotWindow callbacks
- Updating annotate_timestamp to avoid auto-override
- Updating callbacks to consistently define coordinate system
- Fixing species fields for OWLS and tipsy
- Fix extrapolation for vertex-centered data
- Fix periodicity check in FRBs
- Rewrote project_to_plane() in PerspectiveCamera for draw_domain()
- Fix intermittent failure in test_add_deposited_particle_field
- Improve minorticks for a symlog plot with one-sided data
- Fix smoothed covering grid cell computation
- Absorption spectrum generator now 3.0 compliant
- Fix off-by-one-or-more in particle smallest dx
- Fix dimensionality mismatch error in covering grid
- Fix curvature term in cosmology calculator
- Fix geographic axes and pixelization
- Ensure axes aspect ratios respect the user-selected plot aspect ratio
- Avoid clobbering field_map when calling profile.add_fields
- Fixing the arbitrary grid deposit code
- Fix spherical plotting centering
- Make the behavior of to_frb consistent with the docstring
- Ensure projected units are initialized when there are no chunks.
- Removing “field already exists” warnings from the Owls and Gadget frontends
- Various photon simulator bugs
- Fixed use of LaTeX math mode
- Fix upload_image
- Enforce plot width in CSS when displayed in a notebook
- Fix cStringIO.StringIO -> cStringIO in png_writer
- Add some input sanitizing and error checking to covering_grid initializer
- Fix for geographic plotting
- Use the correct filename template for single-file OWLS datasets.
- Fix Enzo IO performance for 32 bit datasets
- Adding a number density field for Enzo MultiSpecies=0 datasets.
- Fix RAMSES block ordering
- Updating ragged array tests for NumPy 1.9.1
- Force returning lists for HDF5FileHandler
This is a scheduled feature release. Below are the itemized, aggregate changes
since version 3.0.
- The RADMC-3D export analysis module has been updated. PR 1358, PR 1332.
- Performance improvements for grid frontends. PR 1350. PR 1382, PR 1322.
- Added a frontend for Dark Matter-only NMSU Art simulations. PR 1258.
- The absorption spectrum generator has been updated. PR 1356.
- The PerspectiveCamera has been updated and a new SphericalCamera has been
added. PR 1346, PR 1299.
- The unit system now supports unit equivalencies and has improved support for MKS units. See Unit Equivalencies. PR 1291, PR 1286.
- Data object selection can now be chained, allowing selecting based on multiple constraints. PR 1264.
- Added the ability to manually override the simulation unit system. PR 1236.
- The documentation has been reorganized and has seen substantial improvements. PR 1383, PR 1373, PR 1364, PR 1351, PR 1345. PR 1333, PR 1342, PR 1338, PR 1330, PR 1326, PR 1323, PR 1315, PR 1305, PR 1289, PR 1276.
Minor or bugfix changes:
- The Ampere unit now accepts SI prefixes. PR 1393.
- The Gadget InternalEnergy and StarFormationRate fields are now read in with the correct units. PR 1392, PR 1379.
- Substantial improvements for the PPVCube analysis module and support for FITS dataset. PR 1390, PR 1367, PR 1347, PR 1326, PR 1280, PR 1336.
- The center of a PlotWindow plot can now be set to the maximum or minimum of any field. PR 1280.
- Fixes for yt testing infrastructure. PR 1388, PR 1348.
- Projections are now performed using an explicit path length field for all
coordinate systems. PR 1307.
- An example notebook for simulations using the OWLS data format has been added
to the documentation. PR 1386.
- Fix for the camera.draw_line function. PR 1380.
- Minor fixes and improvements for yt plots. PR 1376, PR 1374, PR 1288, PR 1290.
- Significant documentation reorganization and improvement. PR 1375, PR 1359.
- Fixed a conflict in the CFITSIO library used by the x-ray analysis module. PR 1365.
- Miscellaneous code cleanup. PR 1371, PR 1361.
- yt now hooks up to the python logging infrastructure in a more standard
fashion, avoiding issues with yt logging showing up with using other
libraries. PR 1355, PR 1362, PR 1360.
- The docstring for the projection data object has been corrected. PR 1366
- A bug in the calculation of the plot bounds for off-axis slice plots has been fixed. PR 1357.
- Improvements for the yt-rockstar interface. PR 1352, PR 1317.
- Fix issues with plot positioning with saving to postscript or encapsulated postscript. PR 1353.
- It is now possible to supply a default value for get_field_parameter. PR 1343.
- A bug in the interpretation of the units of RAMSES simulations has been fixed. PR 1335.
- Plot callbacks are now only executed once before the plot is saved. PR 1328.
- Performance improvements for smoothed covering grid alias fields. PR 1331.
- Improvements and bugfixes for the halo analysis framework. PR 1349, PR 1325.
- Fix issues with the default setting for the
center field parameter. PR 1327.
- Avoid triggering warnings in numpy and matplotlib. PR 1334, PR 1300.
- Updates for the field list reference. PR 1344, PR 1321, PR 1318.
- yt can now be run in parallel on a subset of available processors using an MPI subcommunicator. PR 1340
- Fix for incorrect units when loading an Athena simulation as a time series. PR 1341.
- Improved support for Enzo 3.0 simulations that have not produced any active particles. PR 1329.
- Fix for parsing OWLS outputs with periods in the file path. PR 1320.
- Fix for periodic radius vector calculation. PR 1311.
- Improvements for the Maestro and Castro frontends. PR 1319.
- Clump finding is now supported for more generic types of data. PR 1314
- Fix unit consistency issue when mixing dimensionless unit symbols. PR 1300.
- Improved memory footprint in the photon_simulator. PR 1304.
- Large grids in Athena datasets produced by the join_vtk script can now be optionally split, improving parallel performance. PR 1304.
- Slice plots now accept a
data_source keyword argument. PR 1310.
- Corrected inconsistent octrees in the RAMSES frontend. PR 1302
- Nearest neighbor distance field added. PR 1138.
- Improvements for the ORION2 frontend. PR 1303
- Enzo 3.0 frontend can now read active particle attributes that are arrays of any shape. PR 1248.
- Answer tests added for halo finders. PR 1253
setup_function has been added to the LightRay initializer. PR 1295.
- The SPH code frontends have been reorganized into separate frontend directories. PR 1281.
- Fixes for accessing deposit fields for FLASH data. PR 1294
- Added tests for ORION datasets containing sink and star particles. PR 1252
- Fix for field names in the particle generator. PR 1278.
- Added wrapper functions for numpy array manipulation functions. PR 1287.
- Added support for packed HDF5 Enzo datasets. PR 1282.
This release of yt features an entirely rewritten infrastructure for
data ingestion, indexing, and representation. While past versions of
yt were focused on analysis and visualization of data structured as
regular grids, this release features full support for particle
(discrete point) data such as N-body and SPH data, irregular
hexahedral mesh data, and data organized via octrees. This
infrastructure will be extended in future versions for high-fidelity
representation of unstructured mesh datasets.
Highlighted changes in yt 3.0:
- Units now permeate the code base, enabling self-consistent unit
transformations of all arrays and quantities returned by yt.
- Particle data is now supported using a lightweight octree. SPH
data can be smoothed onto an adaptively-defined mesh using standard
- Support for octree AMR codes
- Preliminary Support for non-Cartesian data, such as cylindrical,
spherical, and geographical
- Revamped analysis framework for halos and halo catalogs, including
direct ingestion and analysis of halo catalogs of several different
- Support for multi-fluid datasets and datasets containing multiple
- Flexible support for dynamically defining new particle types using
filters on existing particle types or by combining different particle
- Vastly improved support for loading generic grid, AMR, hexahedral
mesh, and particle without hand-coding a frontend for a particular
- New frontends for ART, ARTIO, Boxlib, Chombo, FITS, GDF, Subfind,
Rockstar, Pluto, RAMSES, SDF, Gadget, OWLS, PyNE, Tipsy, as well as
rewritten frontends for Enzo, FLASH, Athena, and generic data.
- First release to support installation of yt on Windows
- Extended capabilities for construction of simulated observations,
and new facilities for analyzing and visualizing FITS images and cube
- Many performance improvements
This release is the first of several; while most functionality from
the previous generation of yt has been updated to work with yt 3.0, it
does not yet have feature parity in all respects. While the core of
yt is stable, we suggest the support for analysis modules and volume
rendering be viewed as a late-stage beta, with a series of additional
releases (3.1, 3.2, etc) appearing over the course of the next year to
improve support in these areas.
For a description of how to bring your 2.x scripts up to date to 3.0,
and a summary of common gotchas in this transition, please see
What’s New and Different in yt 3.0?.
This is a scheduled release, bringing to a close the development in the 2.x
series. Below are the itemized, aggregate changes since version 2.5.
- yt is now licensed under the 3-clause BSD license.
- HEALPix has been removed for the time being, as a result of licensing
- The addition of a frontend for the Pluto code
- The addition of an OBJ exporter to enable transparent and multi-surface
exports of surfaces to Blender and Sketchfab
- New absorption spectrum analysis module with documentation
- Adding ability to draw lines with Grey Opacity in volume rendering
- Updated physical constants to reflect 2010 CODATA data
- Dependency updates (including IPython 1.0)
- Better notebook support for yt plots
- Considerably (10x+) faster kD-tree building for volume rendering
- yt can now export to RADMC3D
- Athena frontend now supports Static Mesh Refinement and units (
- Fix long-standing bug for plotting arrays with range of zero
- Adding option to have interpolation based on non-uniform bins in
- Upgrades to most of the dependencies in the install script
- ProjectionPlot now accepts a data_source keyword argument
Minor or bugfix changes:
- Fix for volume rendering on the command line
- map_to_colormap will no longer return out-of-bounds errors
- Fixes for dds in covering grid calculations
- Library searching for build process is now more reliable
- Unit fix for “VorticityGrowthTimescale” field
- Pyflakes stylistic fixes
- Number density added to FLASH
- Many fixes for Athena frontend
- Radius and ParticleRadius now work for reduced-dimensionality datasets
- Source distributions now work again!
- Athena data now 64 bits everywhere
- Grids displays on plots are now shaded to reflect the level of refinement
- show_colormaps() is a new function for displaying all known colormaps
- PhasePlotter by default now adds a colormap.
- System build fix for POSIX systems
- Fixing domain offsets for halo centers-of-mass
- Removing some Enzo-specific terminology in the Halo Mass Function
- Addition of coordinate vectors on volume render
- Pickling fix for extracted regions
- Addition of some tracer particle annotation functions
- Better error message for “yt” command
- Fix for radial vs poloidal fields
- Piernik 2D data handling fix
- Fixes for FLASH current redshift
- PlotWindows now have a set_font function and a new default font setting
- Colorbars less likely to extend off the edge of a PlotWindow
- Clumps overplotted on PlotWindows are now correctly contoured
- Many fixes to light ray and profiles for integrated cosmological analysis
- Improvements to OpenMP compilation
- Typo in value for km_per_pc (not used elsewhere in the code base) has been
- Enable parallel IPython notebook sessions (
- Change (~1e-6) to particle_density deposition, enabling it to be used by
FLASH and other frontends
- Addition of is_root function for convenience in parallel analysis sessions
- Additions to Orion particle reader
- Fixing TotalMass for case when particles not present
- Fixing the density threshold or HOP and pHOP to match the merger tree
- Reason can now plot with latest plot window
- Issues with VelocityMagnitude and aliases with velo have been corrected in
the FLASH frontend
- Halo radii are calculated correctly for domains that do not start at 0,0,0.
- Halo mass function now works for non-Enzo frontends.
- Bug fixes for directory creation, typos in docstrings
- Speed improvements to ellipsoidal particle detection
- Updates to FLASH fields
- CASTRO frontend bug fixes
- Fisheye camera bug fixes
- Answer testing now includes plot window answer testing
- Athena data serialization
- load_uniform_grid can now decompose dims >= 1024. (#537)
- Axis unit setting works correctly for unit names (#534)
- ThermalEnergy is now calculated correctly for Enzo MHD simulations (#535)
- Radius fields had an asymmetry in periodicity calculation (#531)
- Boolean regions can now be pickled (#517)
Many below-the-surface changes happened in yt 2.5 to improve reliability,
fidelity of the answers, and streamlined user interface. The major change in
this release has been the immense expansion in testing of yt. We now have over
2000 unit tests (run on every commit, thanks to both Kacper Kowalik and Shining
Panda) as well as answer testing for FLASH, Enzo, Chombo and Orion data.
The Stream frontend, which can construct datasets in memory, has been improved
considerably. It’s now easier than ever to load data from disk. If you know
how to get volumetric data into Python, you can use either the
load_uniform_grid function or the
load_amr_grid function to create an
in-memory dataset that yt can analyze.
yt now supports the Athena code.
yt is now focusing on providing first class support for the IPython notebook.
In this release, plots can be displayed inline. The Reason HTML5 GUI will be
merged with the IPython notebook in a future release.
Install Script Changes:
- SciPy can now be installed
- Rockstar can now be installed
- Dependencies can be updated with “yt update –all”
- Cython has been upgraded to 0.17.1
- Python has been upgraded to 2.7.3
- h5py has been upgraded to 2.1.0
- hdf5 has been upgraded to 1.8.9
- matplotlib has been upgraded to 1.2.0
- IPython has been upgraded to 0.13.1
- Forthon has been upgraded to 0.8.10
- nose has been added
- sympy has been added
- python-hglib has been added
We’ve also improved support for installing on OSX, Ubuntu and OpenSUSE.
Most Visible Improvements
- Nearly 200 pull requests and over 1000 changesets have been merged since yt
2.4 was release on August 2nd, 2012.
- numpy is now imported as np, not na. na will continue to work for the
- You can now get a yt cheat sheet <http://yt-project.org/docs/2.5/cheatsheet.pdf>!
- yt can now load simulation data created by Athena.
- The Rockstar halo finder can now be installed by the install script
- SciPy can now be installed by the install script
- Data can now be written out in two ways:
- Sidecar files containing expensive derived fields can be written and
implicitly loaded from.
- GDF files, which are portable yt-specific representations of full
simulations, can be created from any dataset. Work is underway on
a pure C library that can be linked against to load these files into
- The “Stream” frontend, for loading raw data in memory, has been greatly
expanded and now includes initial conditions generation functionality,
particle fields, and simple loading of AMR grids with
- Spherical and Cylindrical fields have been sped up and made to have a
uniform interface. These fields can be the building blocks of more advanced
- Coordinate transformations have been sped up and streamlined. It is now
possible to convert any scalar or vector field to a new cartesian, spherical,
or cylindrical coordinate system with an arbitrary orientation. This makes it
possible to do novel analyses like profiling the toroidal and poloidal
velocity as a function of radius in an inclined disk.
- Many improvements to the EnzoSimulation class, which can now find many
different types of data.
- Image data is now encapsulated in an ImageArray class, which carries with it
provenance information about its trajectory through yt.
- Streamlines now query at every step along the streamline, not just at every
- Surfaces can now be extracted and examined, as well as uploaded to
Sketchfab.com for interactive visualization in a web browser.
- allsky_projection can now accept a datasource, making it easier to cut out
regions to examine.
- Many, many improvements to PlotWindow. If you’re still using
PlotCollection, check out
- PlotWindow can now accept a timeseries instead of a dataset.
- Many fixes for 1D and 2D data, especially in FLASH datasets.
- Vast improvements to the particle file handling for FLASH datasets.
- Particles can now be created ex nihilo with CICSample_3.
- Rockstar halo finding is now a targeted goal. Support for using Rockstar
has improved dramatically.
- Increased support for tracking halos across time using the FOF halo finder.
- The command
yt notebook has been added to spawn an IPython notebook
server, and the
yt.imods module can replace
yt.mods in the IPython
Notebook to enable better integration.
- Metallicity-dependent X-ray fields have now been added.
- Grid lines can now be added to volume renderings.
- Volume rendering backend has been updated to use an alpha channel, fixing
parallel opaque volume renderings. This also enables easier blending of
multiple images and annotations to the rendering. Users are encouraged
to look at the capabilities of the
ImageArray for writing out renders,
as updated in the cookbook examples. Volume renders can now be saved with
an arbitrary background color.
- Periodicity, or alternately non-periodicity, is now a part of radius
- The AMRKDTree has been rewritten. This allows parallelism with other than
power-of-2 MPI processes, arbitrary sets of grids, and splitting of
- Fixed Resolution Buffers and volume rendering images now utilize a new
ImageArray class that stores information such as data source, field names,
and other information in a .info dictionary. See the
docstrings for more information on how they can be used to save to a bitmap
or hdf5 file.
The 2.4 release was particularly large, encompassing nearly a thousand
changesets and a number of new features.
To help you get up to speed, we’ve made an IPython notebook file demonstrating
a few of the changes to the scripting API. You can
download it here.
Most Visible Improvements
Threaded volume renderer, completely refactored from the ground up for
speed and parallelism.
The Plot Window (see Slices & Projections) is now fully functional! No
more PlotCollections, and full, easy access to Matplotlib axes objects.
- Many improvements to Time Series analysis:
- EnzoSimulation now integrates with TimeSeries analysis!
- Auto-parallelization of analysis and parallel iteration
- Memory usage when iterating over datasets reduced substantially
- Many improvements to Reason, the yt GUI
- Addition of “yt reason” as a startup command
- Keyboard shortcuts in projection & slice mode: z, Z, x, X for zooms,
hjkl, HJKL for motion
- Drag to move in projection & slice mode
- Contours and vector fields in projection & slice mode
- Color map selection in projection & slice mode
- 3D Scene
Integration with the all new yt Hub ( http://hub.yt-project.org/ ): upload
variable resolution projections, slices, project information, vertices and
plot collections right from the yt command line!
SlicePlot supplant the functionality
- Camera path creation from keyframes and splines
- Ellipsoidal data containers and ellipsoidal parameter calculation for halos
- PyX and ZeroMQ now available in the install script
- Consolidation of unit handling
- HDF5 updated to 1.8.7, Mercurial updated to 2.2, IPython updated to 0.12
- Preview of integration with Rockstar halo finder
- Improvements to merger tree speed and memory usage
- Sunrise exporter now compatible with Sunrise 4.0
- Particle trajectory calculator now available!
- Speed and parallel scalability improvements in projections, profiles and HOP
- New Vorticity-related fields
- Vast improvements to the ART frontend
- Many improvements to the FLASH frontend, including full parameter reads,
speedups, and support for more corner cases of FLASH 2, 2.5 and 3 data.
- Integration of the Grid Data Format frontend, and a converter for Athena
data to this format.
- Improvements to command line parsing
- Parallel import improvements on parallel filesystems
from yt.pmods import *)
- proj_style keyword for projections, for Maximum Intensity Projections
proj_style = "mip")
- Fisheye rendering for planetarium rendering
- Profiles now provide *_std fields for standard deviation of values
- Generalized Orientation class, providing 6DOF motion control
- parallel_objects iteration now more robust, provides optional barrier.
(Also now being used as underlying iteration mechanism in many internal
- Dynamic load balancing in parallel_objects iteration.
- Parallel-aware objects can now be pickled.
- Many new colormaps included
- Numerous improvements to the PyX-based eps_writer module
- FixedResolutionBuffer to FITS export.
- Generic image to FITS export.
- Multi-level parallelism for extremely large cameras in volume rendering
- Light cone and light ray updates to fit with current best practices for
- (yt 2.3 docs)
- Multi-level parallelism
- Real, extensive answer tests
- Boolean data regions (see Combining Objects: Boolean Data Objects)
- Isocontours / flux calculations (see 3D Surfaces and Sketchfab)
- Field reorganization
- PHOP memory improvements
- Bug fixes for tests
- Parallel data loading for RAMSES, along with other speedups and improvements
- WebGL interface for isocontours and a pannable map widget added to Reason
- Performance improvements for volume rendering
- Adaptive HEALPix support
- Column density calculations (see radial-column-density)
- Massive speedup for 1D profiles
- Lots more, bug fixes etc.
- Substantial improvements to the documentation, including
Using the Manual Plotting Interface and a revamped orientation.
- (yt 2.2 docs)
- Command-line submission to the yt Hub (http://hub.yt-project.org/)
- Initial release of the web-based GUI Reason, designed for efficient remote
usage over SSH tunnels
- Absorption line spectrum generator for cosmological simulations (see
Creating Absorption Spectra)
- Interoperability with ParaView for volume rendering, slicing, and so forth
- Support for the Nyx code
- An order of magnitude speed improvement in the RAMSES support
- Quad-tree projections, speeding up the process of projecting by up to an
order of magnitude and providing better load balancing
- “mapserver” for in-browser, Google Maps-style slice and projection
visualization (see Mapserver - A Google-Maps-like Interface to your Data)
- Many bug fixes and performance improvements
- Halo loader
- (yt 2.1 docs)
- HEALPix-based volume rendering for 4pi, allsky volume rendering
- libconfig is now included
- SQLite3 and Forthon now included by default in the install script
- Development guide has been lengthened substantially and a development
bootstrap script is now included.
- Installation script now installs Python 2.7 and HDF5 1.8.6
- iyt now tab-completes field names
- Halos can now be stored on-disk much more easily between HaloFinding runs.
- Halos found inline in Enzo can be loaded and merger trees calculated
- Support for CASTRO particles has been added
- Chombo support updated and fixed
- New code contributions
- Contour finder has been sped up by a factor of a few
- Constrained two-point functions are now possible, for LOS power spectra
- Time series analysis (Time Series Analysis) now much easier
- Stream Lines now a supported 1D data type
- Stream Lines now able to be calculated and plotted (Streamlines: Tracking the Trajectories of Tracers in your Data)
- In situ Enzo visualization now much faster
- “gui” source directory reorganized and cleaned up
- Cython now a compile-time dependency, reducing the size of source tree
yt-supplemental repository now checked out by default, containing
cookbook, documentation, handy mercurial extensions, and advanced plotting
examples and helper scripts.
- Pasteboards now supported and available
- Parallel yt efficiency improved by removal of barriers and improvement of
- Major reorganization of the codebase for speed, ease of modification, and maintainability
- Re-organization of documentation and addition of Orientation Session
- Support for FLASH code
- Preliminary support for MAESTRO, CASTRO, ART, and RAMSES (contributions welcome!)
- Perspective projection for volume rendering
- Exporting to Sunrise
- Preliminary particle rendering in volume rendering visualization
- Drastically improved parallel volume rendering, via kD-tree decomposition
- Simple merger tree calculation for FOF catalogs
- New and greatly expanded documentation, with a “source” button
- Direct writing of PNGs
- Multi-band image writing
- Parallel halo merger tree (see Halo Merger Tree)
- Parallel structure function generator (see Two Point Functions)
- Image pan and zoom object and display widget.
- Parallel volume rendering (see 3D Visualization and Volume Rendering)
- Multivariate volume rendering, allowing for multiple forms of emission and
absorption, including approximate scattering and Planck emissions. (see
3D Visualization and Volume Rendering)
- Added Camera interface to volume rendering (See 3D Visualization and Volume Rendering)
- Off-axis projection (See 3D Visualization and Volume Rendering)
- Stereo (toe-in) volume rendering (See 3D Visualization and Volume Rendering)
- DualEPS extension for better EPS construction
- yt now uses Distribute instead of SetupTools
iyt initialization for GUI support
- Rewritten, memory conservative and speed-improved contour finding algorithm
- Speed improvements to volume rendering
- Preliminary support for the Tiger code
- Default colormap is now
- Lightweight projection loading with
- Improvements to yt.data_objects.time_series
- Improvements to
Analyzing an Entire Simulation)
- Fixed bug causing double data-read in projections
- Added Cylinder support to ParticleIO
- Fixes for 1- and 2-D Enzo datasets
- Preliminary, largely non-functional Gadget support
- Speed improvements to basic HOP
- Added physical constants module
- Beginning to standardize and enforce docstring requirements, changing to
autosummary-based API documentation.
- Critical fixes to ParticleIO
- Halo mass function fixes for comoving coordinates
- Fixes to halo finding
- Fixes to the installation script
- “yt instinfo” command to report current installation information as well as
auto-update some types of installations
- Optimizations to the volume renderer (2x-26x reported speedups)
Version 1.6 is a point release, primarily notable for the new parallel halo
finder (see halo_finding)
- (New) Parallel HOP ( http://arxiv.org/abs/1001.3411 , halo_finding )
- (Beta) Software ray casting and volume rendering
(see 3D Visualization and Volume Rendering)
- Rewritten, faster and better contouring engine for clump identification
- Spectral Energy Distribution calculation for stellar populations
(see Synthetic Spectrum Generator)
- Optimized data structures such as the index
- Star particle analysis routines
(see Star Particle Analysis)
- Halo mass function routines
- Completely rewritten, massively faster and more memory efficient Particle IO
- Fixes for plots, including normalized phase plots
- Better collective communication in parallel routines
- Consolidation of optimized C routines into
- Many bug fixes and minor optimizations
Version 1.5 features many new improvements, most prominently that of the
addition of parallel computing abilities (see Parallel Computation With yt) and
generalization for multiple AMR data formats, specifically both Enzo and Orion.
- Rewritten documentation
- Fully parallel slices, projections, cutting planes, profiles,
- Parallel HOP
- Friends-of-friends halo finder
- Object storage and serialization
- Major performance improvements to the clump finder (factor of five)
- Generalized domain sizes
- Generalized field info containers
- Dark Matter-only simulations
- 1D and 2D simulations
- Better IO for HDF5 sets
- Support for the Orion AMR code
- Spherical re-gridding
- Halo profiler
- Disk image stacker
- Light cone generator
- Callback interface improved
- Several new callbacks
- New data objects – ortho and non-ortho rays, limited ray-tracing
- Fixed resolution buffers
- Spectral integrator for CLOUDY data
- Substantially better interactive interface
- Performance improvements everywhere
- Command-line interface to many common tasks
- Isolated plot handling, independent of PlotCollections