Releases: BASTAcode/BASTA
Version 1.4.1
Version 1.4.0
Updated features, added minor features and extension of documentation.
Major updates:
- The installation of BASTA has been made easier, by making it pip installable. Users will still need to clone the repository, but the installation from there is essentially a one-liner. If updating from a previous version, remember to remove the environmnet variables (set in e.g. .bashrc), which are no longer needed.
- The frequency glitch fitting module have been updated to reflect the Glitchpy routines. Helium glitch signatures can now be determined from observed frequencies and fitted separately or alongside frequency ratios with estimated correlations between them.
- A new section describing all input control options in BASTA has been added to the documentation, along with a list of all parameters known by BASTA.
- The routine to read and convert ascii files containing individual frequencies have been updated to automatically assign columns based on the file header and checking of values. The example using individual frequencies have been updated accordingly.
Minor changes:
- The
nottrusted
feature used to exclude individual frequency modes have been renamed toexcludemodes
for clarity. - The Sobol routine have been updated to rely on the implementation by Scipy, and does therefore not rely on the compilation of external Fortran routines. However, due to slight differences in the schemes, it is not backwards compatible with previous versions.
New minor features:
- If no radial order of individual frequencies are provided, they can be automatically estimated from division by the large frequency separation.
- The
onlyradial
input key can now be set to exclude all non-radial frequency modes from the fit. - The
nameinplots
input key places the star identifier on the corner plot and Kiel diagram. - The
dnubias
input can be provided to quadraticly add a bias to the error of the automatically computed large frequency separation, based on the radial individual frequency modes.
Version 1.3.1
Version 1.3.0 .. but now with a working documentation on ReadTheDocs!
Version 1.3.0
Support for Python 3.11 and updated interpolation.
The main new feature is the updated and more efficient interpolation scheme.
It also includes restructured routines for handling frequency I/O and various bugfixes related to the dustmaps.
Updated Verma+22
Updated (and final) version of the branch used for the Verma+22 publication. Features included in release v1.4.0. This release is kept for reference.
Version 1.2.1
Figures in the documentation and important fixes.
New features:
- Adds a prior in
dnufit
based on the value ofdnufrac
to a fit using frequencies unless a prior is already specified on dnu. This reduces computation time but does not affect the results, as models excluded by the prior would anyways be discarded based on the dnufrac-test. It is on by default, but can be disabled like the other frequency-fitting parameters.
Bugfixes:
- Asteroseismic fits without correlations works again. The computation of the inverse covariance matrix are generally restructured.
- Properly activates scaling of dnu to the solar value when fitting frequencies (where dnu is only used for computing matching bins and for 'dnufrac' selection).
- The interpolation can handle the new volume weights.
Minor changes:
- The list of quantiles and other stats parameters have been moved to constants for consistency.
- Plotting routines moved to a common driver function.
- Fix crash in corner plots when the auto binning does not work.
- Fix connecting lines in the echelle diagrams.
- Solar (re)scaling reported for the best-fitting models.
- The interpolation-preview example works again.
Documentation changes:
- Figures in the documentation are changed to png-format to be rendered correctly by all browsers.
- A case for epsilon-differences fitting added.
Version 1.2.0
New fitting mode and major refactoring.
New features:
- Epsilon-differences fitting (alternative surface-independent measure similar to ratios). Everything is derived only from the frequencies. Set
e012
infitparams
(and activate correlations) and it runs. Includes new plots. The method will be described in the documentation in the next release and in more detail in an up-coming paper.
Major changes:
- Refactoring of all frequency-input handling, especially the treatment of ratios. Improves readability, reproducibility, and maintainability.
- Refactoring of code to produce frequency-related plots. Common framework for clarity.
- Ratios are interpolated to the observed frequencies (can be turned off).
Minor changes:
- Updated look of ratios plots (and now in single panel).
- New echelle diagram with no pairing lines made by adding
"echelle"
to the plotting options. To get all of the different echelle diagrams in one go, add"allechelle"
. The examples are updated to use"allechelle"
. - New plot of the correlation matrix. Mainly for surface-independent fitting, where we always derive the full covariance matrix, but also works for frequencies if the input-covariances are provided.
- New options for the frequency fitting (
dnufit_in_ratios
,interp_ratios
).
Verma+22
Version of BASTA used in Verma et al. (2022). This contains the new approach for fitting glitches and ratios together consistently (with a joint covariance matrix).
This version is currently incompatible with the main version of BASTA and therefore released separately. The 'ratios and glitches together'-feature will be included in a main release later.
Version 1.1.6
Minor bugfix/infrastructure release.
Changes:
- Update packages to newest versions.
- Fix bug in json example.
- Fix bugs and add default values in frequency fitting.
- Update plotting strings to match newest version of the Sharma/Stello (asfgrid) corrections.
Version 1.1.5
Will make the installation run out-of-the-box again (mitigates known issue with the version of Black). All requirements bumped.
Additional changes:
- Added Gaia eDR3 extinction values
- The shipped 16 Cyg A grid contains the new weights
- Added an option to use 3-point frequency ratios