This major release brings important changes from the code perspective (including a major change in the structure of the library), several performance improvements and a new infrastructure for running timing benchmarks, as well as some new features and bug fixes.
- Major change in the structure of poliastro codebase: We separated the high level, units safe functions from the low level, fast ones, with the subsequent improvement in code quality. With this change we effectively communicate where "core" algorithms should go, make easier for future contributors to add numerical functions, and improved the overall quality of the library.
- Upgrade to new SciPy ODE solvers: We wrote our own version of Dormand-Prince 8(5,3) based on the new IVP framework in SciPy 1.0 to take advantage of event detection, dense output and other fancy features. In particular, the :py
~poliastro.twobody.orbit.Orbit.samplemethod now uses dense output when available, therefore removing the need to propagate the orbit repeatedly. - New infrastructure for benchmarks: We started publishing timing benchmarks results using Airspeed Velocity, a Python framework for writing, running, studying and publishing benchmarks. Besides, we bought a dedicated machine to run them with as much precision as we can. Please check them out and consider adding new benchmarks as well!
- Several performance improvements: Now that we are tracking performance, we dedicated some time during this release to fix some performance regressions that appeared in propagation, improving the behavior near parabolic orbits, and accelerating (even more!) the Izzo algorithm for the Lambert problem as well as some poliastro utilities.
- New Continuous Integration infrastructure: We started to use CircleCI for the Linux tests, the coverage measurements and the documentation builds. This service has faster machines and better support for workflows, which significantly reduced the build times and completely removed the timeouts that were affecting us in Travis CI.
- Plotly backends now stable: We fixed some outstanding issues with the 2D Plotly backend so now it's no longer experimental. We also started refactoring some parts of the plotting module and prepared the ground for the new interactive widgets that Plotly 3.0 brings.
- New continuous thrust/low thrust guidance laws: We brought some continuous thrust guidance laws for orbital maneuvers that have analytical solution, such as orbit raising combined with inclination change, eccentricity change and so forth. This is based on the Master Thesis of Juan Luis Cano, "Study of analytical solutions for low-thrust trajectories", which provided complete validation for all of these laws and which can be found on GitHub.
- More natural perturbations: We finished adding the most common orbital perturbations, namely Solar radiation pressure and J3 perturbation. We could not reach agreement with the paper for the latter, so if you are considering using it please read the discussion in the original pull request and consider lending us a hand to validate it properly!
- New dark mode for matplotlib plots: We added a
darkparameter to :py~poliastro.plotting.OrbitPlotterobjects so the background is black. Handy for astronomical purposes!
Besides some installation issues due to the evolution of dependencies, these code bugs were fixed:
- Issue #345: Bodies had incorrect aspect ratio in OrbitPlotter2D
- Issue #369: Orbit objects cannot be unpickled
- Issue #382: Orbit.from_body_ephem returns wrong orbit for the Moon
- Issue #385: Sun Incorrectly plotted in plot_solar_system
- Now the :py
poliastro.twobody.Orbit.samplemethod returns a tuple of (times, positions). - All the propagator methods changed their signature and now accept :py
~poliastro.twobody.Orbitobjects.
This is the complete list of the people that contributed to this release, with a + sign indicating first contribution.
- Juan Luis Cano
- Nikita Astrakhantsev
- Shreyas Bapat
- jmerskine1+
This is a minor release that fixes one single issue:
- Issue #369: Orbit objects cannot be unpickled
Thanks to Joan Fort Alsina for reporting.
This major release received lots of improvements in the 2D plotting code and propagation functions, introduced the new perturbation framework and paved the way for the Python in Astronomy 2018 workshop and the Google Summer of Code 2018 program.
- New experimental 2D Plotly backend: A new :py
~poliastro.plotting.OrbitPlotter2Dclass was introduced that uses Plotly instead of matplotlib for the rendering. There are still some issues that should be resolved when we take advantage of the latest Plotly version, hence the "experimental" nature. - New propagators: A new Keplerian propagator :py
~poliastro.twobody.propagation.mean_motionwas introduced that has better convergence properties than :py~poliastro.twobody.propagation.kepler, so now the user can choose. - New perturbation functions: A new module :py
poliastro.twobody.perturbationswas introduced that contains perturbation accelerations that can be readily used with :py~poliastro.twobody.propagation.cowell. So far we implemented J2 and atmospheric drag effects, and we will add more during the summer. Check out the User Guide for examples! - Support for different propagators in sampling: With the introduction of new propagators and perturbation accelerations, now the user can easily sample over a period of time using any of them. We are eager to see what experiments you come up with!
- Easy plotting of the Solar System: A new function :py
~poliastro.plotting.plot_solar_systemwas added to easily visualize our inner or complete Solar System in 2D plots.
- poliastro participates in Google Summer of Code thanks to OpenAstronomy! More information in the poliastro blog.
- poliastro will be presented at the Python in Astronomy 2018 workshop to be held at Center for Computational Astrophysics at the Flatiron Institute in New York, USA. You can read more details about the event here.
This is the complete list of the people that contributed to this release, with a + sign indicating first contribution.
- Juan Luis Cano
- Pablo Galindo+
- Matt Ettus+
- Shreyas Bapat+
- Ritiek Malhotra+
- Nikita Astrakhantsev+
- Issue #294: Default steps 2D plots were too visible
- Now the :py
poliastro.twobody.Orbit.samplemethod returns a tuple of (times, positions). - All the propagator methods changed their signature and now accept :py
~poliastro.twobody.Orbitobjects.
This is a new major release, focused on bringing 3D plotting functions and preparing the material for the Open Source Cubesat Workshop.
- Sampling method for :py
~poliastro.twobody.Orbitobjects that returns an array of positions. This was already done in the plotting functions and will help providing other applications, such as exporting an Orbit to other formats. - 3D plotting functions: finally poliastro features a new high level object, :py
poliastro.plotting.OrbitPlotter3D, that uses Plotly to represent orbit and trajectories in 3D. The venerable notebook about the trajectory of rover Curiosity has been updated accordingly. - Propagation to a certain date: now apart from specifying the total elapsed time for propagation or time of flight, we can directly specify a target date in :py
poliastro.twobody.orbit.Orbit.propagate. - Hyperbolic anomaly conversion: we implemented the conversion of hyperbolic to mean and true anomaly to complement the existing eccentric anomaly functions and improve the handling of hyperbolic orbits in :py
poliastro.twobody.angles.
- poliastro is now an Astropy affiliated package, which gives the project a privileged position in the Python ecosystem. Thank you, Astropy core developers! You can read the evaluation here.
- poliastro will be presented at the first Open Source Cubesat Workshop to be held at the European Space Operations Centre in Darmstadt, Germany. You can read the full program of the event here.
This is the complete list of the people that contributed to this release, with a + sign indicating first contribution.
- Juan Luis Cano
- Antonio Hidalgo
- mattrossman+
- Roshan Jossey+
- Issue #275: Converting from true to mean anomaly fails for hyperbolic orbits
- The
ephemmodule has been removed in favor of theastropy.coordinates.get_body_barycentric_posvelfunction.
This is a new major release, which adds new packages and modules, besides fixing several issues.
- NEOS package: a new package has been added to poliastro, :py
~poliastro.neospackage. It provides several ways of getting NEOs (Near Earth Objects) data from NASA databases, online and offline. - New patched conics module. New module containing a function to compute the radius of the Sphere of Influence (SOI).
- Use Astropy for body ephemerides. Instead of downloading the SPK files ourselves, now we use Astropy builtin capabilities. This also allows the user to select a builtin ephemerides that does not require external downloads. See #131 for details.
- Coordinates and frames modules: new modules containing transformations between ICRS and body-centered frame, and perifocal to body_centered, :py
~poliastro.coordinatesas well as Heliocentric coordinate frame in :py~poliastro.framesbased on Astropy for NEOs. - Pip packaging: troublesome dependencies have been released in wheel format, so poliastro can now be installed using pip from all platforms.
- Legend plotting: now label and epoch are in a figure legend, which ends with the ambiguity of the epochs when having several plots in the same figure.
- Joined Open Astronomy: we are now part of Open Astronomy, a collaboration between open source astronomy and astrophysics projects to share resources, ideas, and to improve code.
- New constants module: poliastro has now a :py
~poliastro.constantsmodule, with GMs and radii of solar system bodies. - Added Jupyter examples: poliastro examples are now available in the documentation as Jupyter notebooks, thanks to nbsphinx.
- New Code of Conduct: poliastro community now has a Code of conduct.
- Documentation update: documentation has been updated with new installation ways, propagation and NEOs examples, "refactored" code and images, improved contribution guidelines and intersphinx extension.
- New success stories: two new success stories have been added to documentation.
- Bodies now have a parent. It is now possible to specify the attractor of a body.
- Relative definition of Bodies. Now it is possible to define Body parameters with respect to another body, and also add any number of properties in a simple way.
Thanks to the generous SOCIS grant from the European Space Agency, Antonio Hidalgo has devoted three months developing poliastro full time and gained write acces to the repository.
This is the complete list of the people that contributed to this release, with a + sign indicating first contribution.
- Juan Luis Cano
- MiguelHB+
- Antonio Hidalgo+
- Zac Miller+
- Fran Navarro+
- Pablo Rodríguez Robles+
- Issue #205: Bug when plotting orbits with different epochs.
- Issue #128: Missing ephemerides if no files on import time.
- Issue #131: Slightly incorrect ephemerides results due to improper time scale.
- Issue #130: Wrong attractor size when plotting different orbits.
- Non-osculating orbits: removed support for non-osculating orbits.
plotting.plot()calls containingosculatingparameter should be replaced.
This major release was focused on refactoring some internal core parts and improving the propagation functionality.
- Support Python 3.6. See #144.
- Introduced ``Orbit`` objects to replace
Stateones. The latter has been simplified, reducing some functionality, now their API has been moved to the former. See the User Guide and the examples for updated explanations. See #135. - Allow propagation functions to receive a callback. This paves the way for better plotting and storage of results. See #140.
This is a new major release, focused on expanding the initial orbit determination capabilities and solving some infrastructure challenges.
- Izzo's algorithm for the Lambert problem: Thanks to this algorithm multirevolution solutions are also returned. The old algorithm is kept on a separate module.
- Documentation on Read the Docs: You can now browse previous releases of the package and easily switch between released and development versions.
- Mailing list: poliastro now has a mailing list hosted on groups.io. Come and join!
- Clarified scope: poliastro will now be focused on interplanetary applications, leaving other features to the new python-astrodynamics project.
- Issue #110: Bug when plotting State with non canonical units
- Drop Legacy Python: poliastro 0.5.x and later will support only Python 3.x. We recommend our potential users to create dedicated virtual environments using conda or virtualenv or to contact the developers to fund Python 2 support.
- Change ``lambert`` function API: The functions for solving Lambert's problem are now _generators, even in the single revolution case. Check out the User Guide for specific examples.
- Creation of orbits from classical elements: poliastro has reverted the switch to the semilatus rectum \(p\) instead of the semimajor axis \(a\) made in 0.4.0, so \(a\) must be used again. This change is definitive.
Fixed packaging problems.
This is a new major release, focused on improving stability and code quality. New angle conversion and modified equinoctial elements functions were added and an important backwards incompatible change was introduced related to classical orbital elements.
- Angle conversion functions: Finally brought back from poliastro 0.1, new functions were added to convert between true \(\nu\), eccentric \(E\) and mean \(M\) anomaly, see #45.
- Equinoctial elements: Now it's possible to convert between classical and equinoctial elements, as well as from/to position and velocity vectors, see #61.
- Numerical propagation: A new propagator using SciPy Dormand & Prince 8(5,3) integrator was added, see #64.
- MIT license: The project has been relicensed to a more popular license. poliastro remains commercial-friendly through a permissive, OSI-approved license.
- Python 3.5 and NumPy 1.10 compatibility. poliastro retains compatibility with legacy Python (Python 2) and NumPy 1.9. Next version will be Python 3 only.
- Issue #62: Conversion between coe and rv is not transitive
- Issue #69: Incorrect plotting of certain closed orbits
- Creation of orbits from classical elements: poliastro has switched to the semilatus rectum \(p\) instead of the semimajor axis \(a\) to define
Stateobjects, and the function has been renamed to :py~poliastro.twobody.State.from_classical. Please update your programs accordingly. - Removed specific angular momentum \(h\) property to avoid a name clash with the fourth modified equinoctial element, use
norm(ss.h_vec)instead.
This is a new minor release, with some bug fixes backported from the main development branch.
- Fixed installation problem in Python 2.
- Issue #49: Fix velocity units in
ephem. - Issue #50: Fixed
ZeroDivisionErrorwhen propagating with time zero.
This is a new major release, focused on switching to a pure Python codebase. Lambert problem solving and ephemerides computation came back, and a couple of bugs were fixed.
- Pure Python codebase: Forget about Fortran linking problems and nightmares on Windows, because now poliastro is a pure Python package. A new dependency, numba, was introduced to accelerate the algorithms, but poliastro will use it only if it is installed.
- Lambert problem solving: New module :py
~poliastro.iodto determine an orbit given two position vectors and the time of flight. - PR #42: Planetary ephemerides computation: New module :py
~poliastro.ephemwith functions to deal with SPK files and compute position and velocity vectors of the planets. - PR #38: New method :py
~poliastro.twobody.State.parabolicto create parabolic orbits. - New conda package: visit poliastro binstar channel!
- New organization and logo.
- Issue #19: Fixed plotting region for parabolic orbits.
- Issue #37: Fixed creation of parabolic orbits.
This is a bugfix release, no new features were introduced since 0.2.0.
- Fixed #35 (failing tests with recent astropy versions), thanks to Sam Dupree for the bug report.
- Updated for recent Sphinx versions.
- Totally refactored code to provide a more pythonic API (see PR #14 and wiki for further information) heavily inspired by Plyades by Helge Eichhorn.
- Mandatory use of physical units through
astropy.units. - Object-oriented approach: :py
~poliastro.twobody.Stateand :py~poliastro.maneuver.Maneuverclasses. - Vector quantities: results not only have magnitude now, but also direction (see for example maneuvers).
- Mandatory use of physical units through
- Easy plotting of orbits in two dimensions using matplotlib.
- Module
examplewith sample data to start testing the library.
These features were removed temporarily not to block the release and will see the light again in poliastro 0.3:
- Conversion between anomalies.
- Ephemerides calculations, will look into Skyfield and the JPL ephemerides prepared by Brandon Rhodes (see issue #4).
- Lambert problem solving.
- Perturbation analysis.