Sherpa 4.17.0

This release of Sherpa includes various enhancements, documentation updates, bug fixes, and infrastructure changes.

• enhancements:
  • add calc_model and calc_source functions to return an evaluated model/source array
  • added wstat to plot_pvalue for the likelihood ratio test
  • changed XSpec interface to use FunctionUtility C++ API instead of XSFortran API
  • improved support for PHA data starting at channel 0 and handling of STAT_ERR and SYS_ERR PHA columns which are set to 0
  • improved guess for complex models
  • improved filtering handling
  • several updates to enhance plotting capabilities and layout
• documentation changes:
  • added paper citations to front page of Sherpa Read the Docs documentation
  • cleaned up various typos and URL references
  • added examples such as use of set_x/y_label
• infrastructure and testing:
  • improved test coverage
  • many updates to CI
  • drop support for Python 3.9 and numpy <1.24
  • initial/experimental support for Python 3.12
• bug fixes:
  • fixed an issue with plotting 1D data with asymmetric errs after filter
  • include the default identifier in save_all output if it has been changed

Details

#1757 - Allow the wstat statistic to be used in the likelihood-ratio test Allow the Likelihood Ratio Test (as used by plot_value) to be run when using the WStat statistic. Fix #1745.

#1793 - XSPEC move from the xsFortran API to the FunctionUtility API (where possible) Internal change to the XSPEC interface to use the FunctionUtility C++ API rather than the XSFortran API. Fix #1225.

#1921 - Clean up of the IO code Rework the astro data input and output code so that they take advantage of the new data representation in sherpa.astro.io.types

#1949 - Initial support for Python 3.12 Initial support for building with Python 3.12

#1997 - Add support for XSPEC 12.14.0 Allow building with XSPEC 12.14.0 and include the 50 new models in it.

#2017 - IO: separate pack and set commands Replace the logic of the packup argument to the set I/O commands

#2033 - utils: add typing rules to random and parallel Add typing statements to the sherpa.utils.parallel and sherpa.utils.random modules

#2035 - stats: add typing rules Add typing rules to sherpa.stats and minor internal clean up to the module.

#2036 - Update sherpa citations Add Sherpa 4.16.1 to the citation data.

#2045 - Mark code as deprecated in sherpa.utils Rework sherpa.utils and mark some symbols as deprecated - func_counter, is_in, and mysgn - along with suggested replacements.

#2046- Post 4.16.1 update to add DOI info Adds the Sherpa 4.16.1 DOI info to the README.md.

#2048 - CI: update the weekly arch run to add ppc64le and to use a supported python version Support the weekly test builds using aarch64 and ppc46le architectures. This support is considered experimental.

#2052 - XSPEC: do not test bvvnei unless 12.14.0i or later is used Avoid testing the XSbvvnei model if XSPEC <= 12.14.0i is in use, as the model can cause a crash in this case.

#2054 - Allow a file handle or Path object to be sent to the outfile parameter of fit The fit call can now be sent a Path object or a file handle, as well as a string. Fix issue #2063

#2056 - tests: fix fake_pha test failure on crates Address test failures when using crates after merging #1684 and to address a numerical tolerance issue on macOS ARM.

#2057 - Improve the guess routine for PHA data Allow guess to work for composite models, including those with responses, and improve the guess routine used for the norm parameter of XSPEC additive models.

#2058 - Address an error introduced in PR 2025 Fix an error made when addressing NumPy 2.0 changes in issue #2025.

#2059 - Copy over typing annotations to the sherpa.ui and sherpa.astro.ui routines Ensure that annotations are added to the sherpa.ui and sherpa.astro.ui routines. Fix #2053

#2060 - Simple typing support for sherpa.data Add some basic typing rules to the sherpa.data module.

#2061 - Avoid warning messages from AstroPy about unclosed files Initial cleanup of the astro IO code in preparation for future changes. This fixes an issue when using the AstroPy IO backend (pyfits) which could cause Python to report an unclosed file on exit

#2065 - CI: enforce NumPy does not use version 2.0 for the builds Ensure the CI builds use numpy<2

#2068 - CM-887: Updating macos gitlab runner images to version 12 and update the MacOS SDK to version 11.0 This ups the actions macOS Intel version to 12, and updates the macOS SDK to version 11.0, which is consistent with CIAO.

#2069 - Allow the tests to pass with NumPy 2.0 Allow the tests to pass when using NumPy 2.0, as long as pytest-doctestplus is not installed.

#2070 - Migrate to cxc xspec conda test channel Use the cxc xspec test channel for Sherpa test workflows.

#2072 - Use the correct case for the symbol name in the parameter warning message Change how the warning message for deprecated parameter names for models is displayed

#2078 - Include the default identifier in save_all output if it has been changed Record the default identifier in save_all output if the user has changed it. Fix #2077

#2079 - minor cleanup of sherpa.optmethods Internal clean up of the sherpa.optmethods code.

#2081 - minor plot cleanup Clean up some plot code to make better use of modern Python idioms.

#2086 - data: allow axis labels to be changed Allow the independent and dependent axis labels to be changed. Addresses #2083

#2089 - minor cleanup of sherpa.optmethods Internal change to handling of the optional output file option when calling fit. There is no functional change.

#2091 - Asymmetric filter plot Allow 1D data with asymmetric errors to be plotted after a filter. Fixes #2090.

#2092 - Add a name field to the IO backends Add a name field to the I/O backends. This is mainly useful for testing and debugging.

#2094 - Improve messages about invalid file types (PHA) Tweak the warning messages when files can not be read to better indicate the type of file being loaded (for PHA related files).

#2102 - Change the URL for XPA Change the URL used for XPA. Fix #2100.

#2103 - Ensure crates file reads are read-only Ensure that we read in table data as read-only when using the crates I/O backend.

#2105 - docs: typo fixes and remove some references. Fix a typo in a docstring and change how two references are included in the text.

#2107 - XSPEC: add show_xsabund command Add the show_xsabund command to the sherpa.astro.ui module and get_xsabundances to sherpa.astro.xspec.

#2108 - Asymmetric settings Allow the statistic and optimizer to be changed when re-sampling data with asymmetric errors. Fixes #2093.

#2109 - Add a set_xsabundances call for XSPEC users Add a set_xsabundances call to make it easier to change one or more elemental abundances for the XSPEC models and ensure that both the get and set calls are available from the sherpa.astro.ui module.

#2111 - IO: ensure NUMELT/GRP are written out as integers for RMF header Ensure the NUMELT and NUMGRP keywords for RMF files are written out as integers.

#2112 - Minor typing cleanup Clean up some of the recently-added typing rules.

#2115 - doc: add examples of using set_xlabel/ylabel Add documentation showing the new set_xlabel/ylabel functionality from #2086

#2118 - Improve support for PHA data starting at channel 0 Improve the handling of PHA files that start at channel 0. PHA data that are read in from such a file no longer renumber the CHANNEL column to start at 0 rather than 1, which will mean that notice and ignore filter ranges that are given in channel units will need to be reduced by 1.

#2120 - Improve handling of STAT/SYS_ERR PHA columns Improve handling of STAT_ERR and SYS_ERR PHA columns which are set to 0.

#2121 - UI: add calc_model/source commands Add calc_model and calc_source routines to return the model expression evaluated per bin.

#2123 - Data cleanups Internal clean up of some of the data code.

#2125 - Allow per plot kwargs for plot_xxx_components calls Allow per-plot keyword arguments for plot_model_components and plot_source_components.

#2126 - Allow plot and contour to change the plot layout and have per-plot settings. Expand the plot and contour commands to allow per-plot options to be given, matching the plot_model_components/plot_source_components commands.

#2129 - Fix plot overplot Improve the behaviour of plot() and contour() when called with overplot=True or overcontour=True respectively. Fix #2128.

#2130 - data: fixup numpy import Fix up several typing annotations from a recent refactoring.

#2133 - UI: add format and delim arguments to get_filter. Allow the get_filter call to override the format and delim arguments. Fix #1277

#2135 - Tests: tweak tolerances for macOS ARM #2135 Updated macOS ARM test tolerances

#2137 - Fix doctests in sherpa.plot.__init__ Changed documentation to match backends as objects instead of modules

#2138 - UI: add documentation for table to show_xsabund The show_xsabund command now includes a description of the abundance table.

#2139 - Update helpdesk URL Changed helpdesk link - fixes issue #1931

#2140 - Updated docs for asymmetric errors Expanded documentation for load_ascii_with_errors to clarify assumptions for elo and ehi settings.

#2141 - Update LaTeX labels and docs for bokeh Correctly render LaTeX labels in bokeh

#2144 - Fix xspec missing doc Add get_xsabund_doc to the list of functions in the documentation. There is no functional change.

#2147 - Improve corner-case handling of filtering and empty data handling A clean-up pass for the data classes, in particular DataPHA, to normalize the filtering and handling of an empty object

#2151 - updating Sherpa citations Added a citation to the most recent sherpa paper

#2152 - DATA: clean up DataPHA.get_x Internal clean up of the DataPHA.get_x method.

#2153 - updated the front page of Sherpa RTD with references Added references to three sherpa papers in the sherpa docs

#2154 - Update README Added additional text regarding acknowledging/citing sherpa

#2155 - Drop support for Python 3.9 and add 3.12 Drop support for Python 3.9 and numpy < 1.24 and note Python 3.12 experimental support.

#2157 - Improve test coverage of ignore_bad Add more corner-case of quality channel handling for PHA data. There is no functional change here.

#2160 - Fixup test failures from #2118 Fix up a test case when writing out the range of a FITS column.

#2161 - UI: note when a 1D filter command has failed Provide more information for the UI user if a filter command has failed.

#2162 - TEST: ignore fork deprecation warnings Ignore the deprecation warnings about the use of the fork method for multiprocessing.

#2177 - Read the Docs build failures with newer sphinx versions and sphinx_rtd_theme deprecation warning Specify sphinx < 8 for the read the doc environment and remove the call to get_html_theme_path no longer needed.