Pymodalib

This release fixes an exception raised when using constant padding with the wavelet transform.

This release adds the Matlab implementation of the Bayesian inference algorithm, which is now the default.

This release adds the experimental Bayesian inference algorithm.

The wavelet transform now accepts the parameter `wavelet="Morse-a"`.

Wavelet transform

The wavelet transform can now be run in parallel by passing `parallel=True`. This can provide a modest performance improvement of around 25%, along with an increase in memory usage.

Shared memory is used to reduce the memory usage, so parallelization only works on Python 3.8 and higher. (Parallelization is automatically disabled on Python 3.7 and below, even if `parallel==True`.)

Plotting

The performance of `pymodalib.contourf()` has been greatly improved by automatically subsampling the data. The subsampling can be disabled by passing `subsample=False`, and the resolution can be changed using the `subsample_width` parameter.

Additionally, `log=True` can now be passed to `pymodalib.contourf()` to apply a logarithmic scale to the y-axis.

This release greatly improves the reliability of the wavelet transform. **The Python implementation of the wavelet transform is now the default.**