segmentation

This release provides the following functions:

• pcf: Segments a single sample. Chooses the fastest method (exact, fast, expanding_fast) based on the input size.
• multipcf: Segments multiple samples simultaneously. Chooses the fastest method (exact, fast, expanding_fast) based on the input size.
• exact_pcf: Segments a single sample with no optimisation heuristics.
• exact_multipcf Segments multiple samples simultaneously with no optimisation heuristics.
• fast_pcf: Segments a single sample. Uses a fast prefiltering method to narrow down possible breakpoints (convolution with a high-pass filter).
• fast_multipcf: Segments multiple samples simultaneously. Uses a fast prefiltering method to narrow down possible breakpoints. Uses convolution with a high-pass filter.

Utility functions:
mark_positions: Run the filter convolution step on its own.
restricted_pcf: Run a single sample segmentation that is restricted to an input set of potential breakpoints.

Also provided are implementations of the suggested optimisation from the paper:

An additional way to speed up the computations on long sequences is to initially divide the
sequence into overlapping subsequences, and iteratively find the solution.

• expanding_fast_pcf, expanding_fast_multipcf: Break the input sequence(s) into overlapping subsections, run fast_pcf or fast_multipcf on each, merge the results.

segmentation

Initial release of the package.