Block a user
Simulated annealing
1881feb5e2
Merge pull request 'Fix #532 by making
ShapeND
a non-value class' (!522) from bug/defaultStridesCache into dev
201887187d
Make
ShapeND
a usual non-value class. Implement its equals
and hashCode
methods. Deprecate contentEquals
and contentHashCode
.
6f98c3fbe9
Tests and peak finding examples
2e084edc9b
Implement generic algorithm
47a9bf0e9a
Suboptimal implementation for countExtrema
b8bf56bc42
Extrema padding (experimental)
597c27e615
Improved peak finding algorithm
40ffa8d6fc
Remove unnecessary code from outer sift() function
1c5113da29
Allow tail call optimization
WIP: feature/emd
It works. Without tailrec
it does not produce stack overflow when I run sifting with 5000 iterations per mode
WIP: feature/emd
Thank you kindly for the review. I have submitted corrections for the majority of issues. I'll notify you when I make significant improvements to the algorithm or the code in general.
WIP: feature/emd
Also, is it intended that the spline will ignore double extrema? No, I'm planning on improving this function and making it
public
placed inSeriesExtensions.kt
WIP: feature/emd
I also wondered why there is no .sum()
method. I could implement it for a 1-d series, but doing it for a general buffer seems a bit too much if there is need for arbitrary axis like in NumPy
WIP: feature/emd
Is making a one-line data class declaration above considered idiomatic?
data class SignCounter(val prevSign: Double, val zeroCount: Int)
return fold(SignCounter(sign(get(0)), 0))…
283d629398
Empirical Mode Decomposition for Double series
e0125f0b26
mapWithLabel fix