ExtremaFunctions: Find local extrema of a numeric vector

Description

These functions take a vector of y values and identify local extrema.

Usage

find_local_extrema(
  y,
  x = NULL,
  window_width = NULL,
  window_width_n = NULL,
  window_height = NULL,
  window_width_frac = NULL,
  window_width_n_frac = NULL,
  return = "index",
  return_maxima = TRUE,
  return_minima = TRUE,
  return_endpoints = TRUE,
  subset = NULL,
  na.rm = TRUE,
  width_limit = NULL,
  width_limit_n = NULL,
  height_limit = NULL
)
first_maxima(
  y,
  x = NULL,
  window_width = NULL,
  window_width_n = NULL,
  window_height = NULL,
  window_width_frac = NULL,
  window_width_n_frac = 0.2,
  return = "index",
  return_endpoints = TRUE,
  ...
)
first_minima(
  y,
  x = NULL,
  window_width = NULL,
  window_width_n = NULL,
  window_height = NULL,
  window_width_frac = NULL,
  window_width_n_frac = 0.2,
  return = "index",
  return_endpoints = TRUE,
  ...
)

Value

find_local_extrema returns a vector corresponding to all the found local extrema.

first_maxima returns only the first maxima, so is a shortcut for

find_local_extrema(return_maxima = TRUE, return_minima = FALSE)[1]

first_minima returns only the first minima, so is a shortcut for

find_local_extrema(return_maxima = FALSE, return_maxima = FALSE)[1]

If return = "index", the returned value(s) are the indices corresponding to local extrema in the data

If return = "x", the returned value(s) are the x value(s) corresponding to local extrema in the data

If return = "y", the returned value(s) are the y value(s) corresponding to local extrema in the data

Arguments

y

Numeric vector of y values in which to identify local extrema

x

Optional numeric vector of corresponding x values

window_width, window_width_n, window_height, window_width_frac, window_width_n_frac

Arguments that set the width/height of the window used to search for local extrema.

window_width is in units of x.

window_width_n is in units of number of data points.

window_height is the maximum change in y a single extrema-search step is allowed to take.

window_width_n_frac is as a fraction of the total number of data points.

For example, the function will not pass a peak or valley more than window_width_n data points wide, nor a peak/valley taller or deeper than window_height.

A narrower width will be more sensitive to narrow local maxima/minima, while a wider width will be less sensitive to local maxima/minima. A smaller height will be more sensitive to shallow local maxima/minima, while a larger height will be less sensitive to shallow maxima/minima.

return

One of c("index", "x", "y"), determining whether the function will return the index, x value, or y value associated with the identified extremas

return_maxima, return_minima

logical for which classes of local extrema to return

return_endpoints

Should the first and last values in y be included if they are in the returned vector of extrema?

subset

A vector of logical values indicating which x and y values should be included (TRUE) or excluded (FALSE).

If return = "index", index will be for the whole vector and not the subset of the vector

na.rm

logical whether NA's should be removed before analyzing

width_limit

Deprecated, use window_width instead

width_limit_n

Deprecated, use window_width_n instead

height_limit

Deprecated, use window_height instead

...

(for first_maxima and first_minima), other parameters to pass to find_local_extrema

Details

For find_local_extrema, one of window_width, window_width_n, window_height, or window_width_n_frac must be provided.

For first_minima or first_maxima, set window_width_n_frac = NULL to override default width behavior.

If multiple of window_width, window_width_n, window_height, or window_width_n_frac are provided, steps are limited conservatively (a single step must meet all criteria).

In the case of exact ties in y values within a window, only the first local extrema is returned.