Skip to content

SasView 5.0.5
Compare
Choose a tag to compare
@wpotrzebowski wpotrzebowski released this 03 Jun 04:51
· 6 commits to release_5.0.5 since this release
v5.0.5
1a45610
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

Downloads

Windows installer
MacOS installer
Linux & MacOS from source

New in Version 5.0.5

This is a point release which fixes some issues reported in earlier versions
of 5.0.x. A few highlights are:

  • The long standing issue with the Levenberg-Marquardt optimiser not respecting
    parameter bounds has been resolved by the move to a later version of
    the Bumps package.
  • A bug which prevented the radius_effective parameter from being updated
    in $P(Q)*S(Q)$ models when the data were resolution smeared has been fixed.
  • A bug that prevented the formation of composite mixture models with multiplicity
    (for example, models such as core_multi_shell*hardsphere + cylinder or
    core_multi_shell@hardsphere + cylinder) has been fixed.
  • The button to reset the selected Q-limits for fitting now works again!
  • There has been a technical change to the point in the calculation at which the
    volume normalisation of $P(Q)*S(Q)$ models is applied by the move to a later
    version of the Sasmodels package. This change was actually incorporated into
    v5.0.4 but due to an oversight was omitted from the release notes at the time,
    although a note was added to the web version after the release. In most
    instances this change will go un-noticed unless you happen to be plotting the
    individual contributions of $P(Q)$ or $S(Q)$ and comparing them with similar
    calculations performed in versions of SasView before v5.0.4, at which point the
    scaling of the functions will be seen to be different.

There are also some new features in this version. Most notably:

  • The Generic Scattering Calculator Tool has been overhauled and its
    capabilities significantly expanded. In particular, it will now perform
    magnetic/polarised SANS computations. As part of this upgrade, coordinate data
    in some VTK formats are now also supported.
  • The data loaders have also been improved. SasView will now read CanSAS1D XML
    data files with multiple <SASdata> blocks in a single <SASentry>. And, by
    popular demand, 1D data can now be saved in CSV format. SESANS data files
    with the extension .sesans are now also recognised.
  • A Boucher-type interfacial profile function has been added to the spherical_sld model.

This version of SasView is built with Sasmodels 1.0.6 and Bumps 0.9.0.

Pull Request Changes

Documentation Changes

Build System Changes

The full sasview and sasmodels changelogs, respectively, are available at:

New Models

The following models have been added to the Model Marketplace since v5.0.4 was released:

  • Magnetic Whirl
  • Maier-Saupe distribution
  • Cyclic Gaussian distribution
  • Superball Model - Neither Sphere nor Cube
  • OrientedMagneticChains

Known Issues

Update Oct 18. 2022: This version fails to launch on MacOS when no previous
versions of SasView are present on the system. To fix the issue, create an empty file
named custom_config.py in the ~/.sasview/config directory within your user directory.

Update Sep 1, 2022: We have become aware of an issue with the new
Levenberg-Marquardt (LM) optimiser when you are fitting with a GPU
enabled: if the model you are fitting uses C subroutines then nothing happens!
If you are only using the CPU, or if the model you are fitting is a pure Python model,
then the optimiser still works. If you need to use a GPU for your fitting then the
workaround is simply to select a different optimiser (Fitting > Fit Algorithms).

The ‘rpa’ (Random Phase Approximation, for polymer scattering) model has been
temporarily withdrawn. Some gremlins had crept into how the model parameters
interacted with the Fit Page which meant you were not performing the calculation
you thought you were. This is being investigated. Should you need it, a separate
binary blend model (the most performed calculation) is available from the
Model Marketplace .

All the known bugs/feature requests can be found in the issues on github.
Note the sasmodels issues are now separate from the sasview issues (i.e. different repositories)

sasview: https://github.com/SasView/sasview/milestones

sasmodels: https://github.com/SasView/sasmodels/milestones

Contributors

pkienzle, bmaranville, and 13 other contributors
Assets 4
Loading