Summary of changes in lssa 0.1.4:
** Code updates to support install in Octave 6.0
Summary of changes in lssa 0.1.3:
** Mark fastlscomplex BIT as a known failure (Bug #53963)
** Update package with maintainer makefile
** Code updates to support install in Octave 4.4
Summary of changes in lssa 0.1.2:
** All functions now have input checks in place to return useful errors as
opposed to division by zero, etc. Documentation has also been improved.
Summary of status of the intial lssa release, 0.1.1:
Current status:
** lscomplex and lsreal both produce accurate results; they can be slow for
very large datasets.
** fastlscomplex is accurate for the first octave of results; there is still an
error I need to pin down in the merging for additional octaves. fastlsreal
is disabled at the moment as I move to an implementation based on the new
fastlscomplex.
** lscorrcoeff works, although I'm still attempting to understand the initial
author's reasoning. Its generated results are relevant to any given data
set, but it does not appear to be normalized to any great extent.
** There are two wavelet functions under development, but they are not included
in this release as they are currently not functional. For all your wavelet
needs, the specific transformation used is available in the lswaveletcoeff
function, and will generate a single cosine/sine magnitude pair (as a
complex number) for a complex-valued series (this function may be joined by
a companion for real-valued series) and can be looped to simulate a full
wavelet transform.
** For all the working functions, tests have been written and formatted to
Octave coding standards. These tests should pass on any given architecture
(there was some question about that previously) and often provide examples
of how the function operates. For a few functions, there are demo scripts.
Package: lssa