Skip to content

Release 1.2.0

Choose a tag to compare

View all tags
@CozySocksAlways CozySocksAlways released this 04 Mar 13:03
· 9 commits to master since this release
v1.2.0
2c13f60
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

This release of Elephant marks the change to Numpy 2.x, while Numpy 1.x is no longer supported. Along with this change, a number of new features and improvements for existing functionality were added, in part related to the new concept for managing experimental trials using elephant.trials objects.

New Features

  • For the spike times extraction methods in the spike_train_generation module, added multichannel support for neo.AnalogSignal and introduced the always_as_list parameter to ensure spike trains are returned as a list. (#614)
  • Extended statistics.time_histogram to accept both a single spike train and lists of spike trains. (#650)
  • Extended statistics.fano_factor to accept elephant.trials objects. (#645)
  • Added a datasets.load_data function to simplify tutorial and analysis code by allowing dataset retrieval and generation. (#687)

Bug Fixes

  • Fixed numerical stability issues in van_rossum_distance by preventing small negative floating-point before square root evaluation. (#680)
  • Fixed issue in statistics.instantaneous_rate function related to unexpected pooling behavior and kernels that exceed the signal length. (#649, #688)
  • Fixed BinnedSpikeTrain slicing issues caused by changes in SciPy (>=1.15.0) sparse matrix index validation. (#653, #685)
  • Resolved GPU backend detection inconsistencies when CUDA libraries are present but PyCUDA is unavailable. (#666)
  • Fixed GPU kernel launch resource errors in CUDA-accelerated ASSET computations by adapting thread selection dynamically. (#667)

Documentation

  • Improved general documentation quality, including parameter descriptions and tutorial clarity. (#641)
  • Fixed documentation build failures by aligning python version(=3.12) for docs test and ReadTheDocs build CI jobs. (#677)
  • Replaced hidden notebook plotting cells with equivalent Viziphant functions. (#677)
  • Added a new documentation section for example datasets. (#687)
  • Fixed documentation build issues related to sphinx-tabs compatibility. (#681)
  • Fixed ReadTheDocs MPI-related build failures. (#682)

Breaking changes

  • Removed deprecated verbose parameter in the ASSET module in favor of using Python logging module. (#669)

Other changes

  • Added Python 3.13 CI runner to ensure compatibility with the latest Python language features (#654)
  • Updated README with maintainer contact information. (#683)
  • Note: macOS wheels for Elephant are currently distributed as pure Python wheels. The C++ accelerated spade module falls back to a Python implementation on macOS, which may result in reduced performance compared to builds using the C++ backend. Full macOS compiled-acceleration support will be restored in a future patch release.

Selected dependency changes

  • Support for Python 3.13 (#654)
  • NumPy ≥ 2.0 (#656)
  • Support for SciPy 1.17 (#685)
Assets 2
Loading