Full Changelog

3.0 (unreleased)

New Features

  • Make X3D output compatible with X_ITE. This includes cleanup up of a few pieces that X3DOM is not strict about. [#253]

  • Add a new class marxs.optics.CircularBaffle that implements a baffle with a circular hole. [#258]

  • Add a scale_factor variable to the X3D visualization module to allow easy changing of the length unit conversion between MARXS and X3D. [#253]

API Changes

  • Remove the mayavi plotting backend. Mayavi is a very heavy dependency that is hard to build and really an analysis system, while marxs just used it as a plotting backend. [#261]

  • marxs.version has been removed. Use marxs.__version__ instead which is default throughout the Python ecosystem. [#260]

Bug fixes

  • Fix test failures with numpy 2.0 [#254]

Other

  • Change included .par file for marx >= 6.0 [#261]

Other changes and additions

2.0 (02-May-2024)

New Features

  • marxs.simulator.Parallel and marxs.simulator.Sequence now have a first_of_class_top_level method to help find certain subelements. [#222]

  • Added marxs.design.tolerancing.run_tolerances_for_energies2 is similar to marxs.design.tolerancing.run_tolerances_for_energies, but with a simplified interface. [#222]

  • Output columns from optical elements can now be None (they will be ignored) or specify the datatype and other Column meta information. [#226]

  • To avoid probabilities <0 or >1, values assigned to this column are now tested and a ValueError may be raised. [#230]

  • Added a new class CaptureResAeff_CCDgap that selects photons differently for calculation of the effective area and resolving power to account for chip gaps. [#235]

  • Add direct X3D visualization backend to avoid the need for Mayavi. [#238]

  • Add several missions under development that were previously developed in separate repros. As it turn out, keeping the code separate lead to significant parallel work that can be simplified by keeping all of them in marxs itself. However, those mission are less fleshed out then marxs itself, less documented and less tested. [#237]

  • Add CircularMeshGrid as a new way to place gratings on the Rowlandtorus. Adapted from previous Lynx code. [#237]

  • Add various helper functions that previously lived in separate repros. [#237]

  • Add methods to fit the LSF to generate a Chandra-style LSFPARM file. [#237]

  • Include code for several missions (Athena, Arcus, Lynx, AXIS) into marxs. This code had been previously developed separately, since none of it is really ready for public consumption (e.g. the Lynx concept has been rejected, Arcus is under continuous redefinition). However, given how few people except me currently use MARXS, “practicality beats purity” and I’m now including this into marxs. Note that the mission code generally is not documented in the docs and not tested as well. It’s simply either not interesting enough for general users (but needed to replicate some of my SPIE proceedings so Moritz Guenther wants it publicly available) or evolving too fast to write good docs and tests (e.g. the Arcus effective area changes every time the coating or layout changes, tests that check Aeff are thus more checks on the CALDB than on the code). [#240]

  • A number of updates that were developed for those individual missions and are now included into marxs proper, e.g.: A class to make certain plots that display the performance loss for misalignments and utilities to run tolerancing etc. [#240]

  • Add new functions to marxs.design.rowland to construct Rowland tori in a double tilted Rowland torus design. [#245]

  • Add more options to PerfectLens to account for non-perfect reflectivity and the option to have the optical axis offset from the geometrical center of the element. [#248]

  • Add method to create archive files of X3D files. [#251]

API Changes

  • For all classes that place elements on the Rowland torus, the API has been standardized. That means changes to almost all of them. In particular:

    • rowland.RectangularGrid replaces RowlandTorusArray and LinearCCDArray. Inputs are specified the y and z ranges now instead of Rowland circle \($\phi\) coordinates that are hard to visualize since it’s measure from the center of the Rowland circle and the origin of \($\phi=0\) depends on the rotation of the torus. rowland.RectangularGrid can be used for 1D or 2D arrangements and thus d_element must always be two numbers for dimension in both directions.

    • All classes now use a minimizer instead of a brackets root finder. So, instead of a bracketing interval, they now need a guess_distance where the minimization starts.

  • load_table2d has been replaced by marxs.utils.tablerows_to_2d, which performs the same reformatting of a table, but does not read it in. [#237]

  • marxs.missions.mitsnl.InterpolateEfficiencyTable now takes a table, instead of a filename as parameter. [#237]

  • effectivearea_from_photonlist now requires a unit for the input of the effective area. [#237]

  • CaptureResAeff_CCDgap and CaptureResAeff are moved to marxs.analysis.gratings. [#237]

  • plot_wiggle and load_and_plot are now summarized in the class to make it easier to derive from them. [#237]

  • Rename files that had a class of the same name, because this confuses the Sphinx autosummary extension. The following files have been renamed: marxs.missions.chandra.chandra to marxs.missions.chandra.chandra_global, marxs.source.source to marxs.source.basesources, and marxs.optics.baffle to marxs.optics.baffles. Users that follow the recommended practice of importing from the higher level such as from marxs.optics import Baffle should not notice any difference. [#424]

  • The unit of photon energy is now set to keV. [#252]

  • CCDRedistNormal now inherits from an optical element and takes the sigma for redistribution as a keyword argument. [#252]

Bug fixes

  • marxs.missions.mitsnl.catsupportbars no longer throws an exception if no photon hit a facet before. [#221]

  • To avoid probabilities <0 or >1, the default interpolation for MIT grating efficiency tables is now linear. [#229]

  • Fig bug in design_tilted_torus. The equations implemented previously implicitly assumed 2 * alpha = beta. [#250]

Other changes and additions

  • Previously, MARXS only contained modules for currently operating missions. However, in practice, it’s mostly used to simulate proposed missions. Thus, we changed the policy to include simulation code for proposed missions here. That code may not be of general use, but it simplifies the workflow for developers considerably and serve example on how to build your own instrument from the building blocks provided by the main marxs modules. [#237]

  • Updates of CI and simplify _astropy_init and conftest for new Python and astropy versions. [#236]

1.2 (06-Nov-2020)

New Features

API Changes

  • Sources now require input for flux, energy, and polarization as astropy quantities. Any table-like input must not be an astropy QTable and flux and polarization values are densities not per-bin. [#218]

  • The geometry of a class is now defined as a separate object from the new marxs.math.geometry module. Before, the geometry was backed into the optical components itself which led to significant code duplication. The new scheme allows making e.g. cylindrical and flat gratings with the same gratings objects. However, not all geometries work with all objects and there is currently no way to check automatically - the user has to use caution or check the code. [#182]

  • Geometry objects provide item access like this detector.geometry['center'] instead of a function interface with (...). [#182]

  • Remove utils.MergeIdentical merge strategy since it is no longer used after #189. [#191]

  • Quantity is used in more places, and thus the base classes need to accept quantities as output from “specific_process_photons”. This change requires some care and thus is done gradually only where needed at this point. [#202]

  • RandomGaussianScatter can be initialized without setting scatter, provided that its already defined a class level variable (e.g. for subclasses). [#202]

  • RandomGaussianScatter and Scatter now expect angular quantities instead of plain floats. [#216]

Bug fixes

  • marxs.analysis.gratings.resolvingpower_per_order has been updated to ignore photons with probability 0. [#162]

  • An index mix-up in marxs.simulator.ParallelCalculated.calculate_elempos introduced unintended zoom and shear in the elements. [#164]

  • [#159] left behind an undefined filterfunc. This is fixed and a regression test added. [#165]

  • marxs.analysis.analysis.find_best_detector_position will now change the detector position always along an axis perpendicular to the detector plane. [#171]

  • marxs.optics.scatter.RadialMirrorScatter now works with inplanescatter=0 which is useful for parameters studies. [#174]

  • Plot only “half-box” for elements such as mirrors where the optical interaction occurs on a surface. [#178]

  • With marxs.optics.MultiAperture photons would always be sorted by aperture number. To fix this, apertures now behave more like other optical elements and use process_photons. [#189]

  • marxs.optics.FlatStack now inherits from marxs.simulator.BaseContainer. [#196]

  • Fix bug in distributions generated by marxs.source.SphericalDiskSource. [#202]

  • Objects now fail to generate when zoom is set to 0. [#217]

Other changes and additions

  • Examples are now written for numpy >= 1.14 (which changed some printing formats). [#182]

1.1 (15-Jul-2017)

New Features

  • Add ability to plot cylinders in Mayavi (for support struts) [#147]

  • Add shape "None" (as a string) to avoid plotting an object [#157]

API Changes

  • Remove marxs.source.LabPointSource, which was just a special case of LabPointSourceCone. Instead, set the default values of the later so that it reproduces the behaviour of the former. [#144]

  • MultiLayerEfficiency and MultiLayerMirror now expect all parameters as keyword arguments for consistency with the other elements in MARXS. [#144]

  • marxs.visualization.utils.format_saved_positions is now a method of KeepCol with the new name format_positions() and the atol keyword can be switched off. Additionally, KeepCol now has a __array__ method. This makes the function useful for columns that are not positions, but e.g. polarization vectors. On the other hand, the plot_rays functions do not accept KeepCol objects directly as input any longer. [#149, #152]

  • According to the docs, a pointing could be initialized with either a SkyCoord or a tuple that would initialize the SkyCoord. The later option was broken and has been removed entirely. [#151]

  • Remove parameter filterfunc from resolvingpower_from_photonlist and detected_fraction. Instead, the photon list can be filtered before calling these functions just as easily. [#159]

Bug fixes

  • Added missing keywords in display dict for some objects and fixed exception when plotting things that are not objects. Discovered and fixed as part of [#147].

  • Polarization after reflection from a mirror used to just parallel transport the vector and calculate the probability of the photon based on s and p polarization. This needs to be applied to the outgoing polarization vector, too. [#148]

  • Plotting of Rowland Torus failed in Mayavi due to typo. [#154]

Other changes and additions

  • Improve Documentation [#146]

  • Docs: Add polarization example [#153]

  • Docs: Add example to calculate flux from normalized spectrum [#160]

  • Add experimental data for comparison to the polarization example [#158]

1.0 (14-Apr-2017)

This is the first release intended to use. The Change log will begin starting with this release.

0.1 (experimental release)

This release was not intended to be used, but the ver sioning scheme in the development branch required a tagged commit.