Peaks.jl

code

julia

Find peaks (local extrema) of signals

Author

Allen Hill

Published

Mar 2018

Precompiling Peaks
  ✓ SIMD
  ✓ Peaks
  2 dependencies successfully precompiled in 5 seconds. 3 already precompiled.

Peaks.jl is a library for finding peaks (i.e. local maxima and minima) and peak characteristics (e.g. prominence, width, etc) in vector (1D) data signals.

Peak finding has been a foundational analytical step in every analysis during my graduate studies. When I started exclusively using Julia in 2018, there was no package or functionality that met my needs, such as:

Finding peaks in non-continuous data (i.e. sections of NaN or missing data)
Finding peaks for windows wider than the two directly adjacent elements
Finding peak prominence and/or width

Peaks.jl is the product of my efforts to implement similar peak finding capabilities as I depended on in MATLAB. Peaks.jl is extensively tested and has features to:

Find peak (maxima or minima) locations, height, prominence, and width
- Filter peaks by peak spacing (window size), height, prominence, and width (including “Full Width Half Maximum (FWHM)”)
Fully supports NaN/missing with optional tolerance using keyword arg strict:
- Conventional handling/propagation of NaN/missing when strict = true (the default)
- Reasonable alternatives when strict = false
An alternative, SIMD-accelerated, fast peak finding method (with a reduced feature set compared to the standard method)