Identify sustained peaks using pracma::findpeaks - r

Question

I am having problems with the syntax of the peakpat option within the findpeaks function within the pramca R package (v. 2.1.1). I am using R 3.4.3 x64 windows.
I would like the function to identify peaks that may have two repeated values, and I believe the option peakpat is how I can do this.
This question has been asked before, however I haven't been able to come across an example of how to implement the option Hans is referring to. This seems very basic and I am also quite a beginner when it comes to coding. In the help file online, it says the following about peakpat:
define a peak as a regular pattern, such as the default pattern ``[+]1,[-]1,''; if a pattern is provided, the parameters nups and ndowns are not taken into account."
I'm having problems interpreting what "[+]1,[-]1" means. Any ideas? I've tried variations of what I think this means, but each attempt results in NULL. Please see my example below, any help/insight is greatly appreciated.
# Example:
install.packages("pracma")
library(pracma)
subset = c(570,584,500,310,261,265,272,313,314,315,330,360,410,410,360,365,368,391,390,414)
# Plots
plot(subset)
lines(subset)
# findpeaks without defining repeated values;
# the result does not identify the peak at subset[13:14] (repeated 'peak' values)
result = findpeaks(subset)
pks1 = data.matrix(result[,1])
locs1 = data.matrix(result[,2])
# findpeaks with my futile attempt at defining peakpat
result = findpeaks(subset, nups=2, ndowns=nups, zero = "0", peakpat="[+]2,[-]2,")
result = findpeaks(subset, nups=1, ndowns=nups, zero = "0", peakpat="[+]1,[-]1,")
result = findpeaks(subset, nups=1, ndowns=nups, zero = "0", peakpat="[+]{,1},[-]{,1}")
result = findpeaks(subset, nups=1, ndowns=nups, zero = "0", peakpat="[+]{1,},[-]{1,}")
result = findpeaks(subset, nups=2, ndowns=nups, zero = "0", peakpat="[2],[2]")
result = findpeaks(subset, nups=2, ndowns=nups, zero = "0", peakpat="[1],[1]")
# all of the above results in NULL
Thank you!

Answer 1

The documentation isn't too helpful in this case, but you can get some clues by inspecting the function body.
Typing the function name into the console lets you inspect its source. Without going into complete detail, this line is helpful:
peakpat <- sprintf("[+]{%d,}[-]{%d,}", nups, ndowns)
This shows us that the default arguments correspond to a peakpat of "[+]{1,}[-]{1,}".
This should also reinforce why if you specify peakpat, you don't need to specify anything for nups and ndowns.
A pattern that does what you're after, for peaks of two repeated values:
result <- findpeaks(subset, peakpat = "[+]{1,}[0]{1,}[-]{1,}")
The commas specify an interval. So if you wanted to limit your search to peaks that have a repeated value of at most length 3:
result <- findpeaks(subset, peakpat = "[+]{1,}[0]{1,2}[-]{1,}")
The function works by turning your data into a string and applying a regular expression, so the usual rules for regex should apply.

Answer 2

This is my final solution for finding the peaks with the traditional peakpat definition and defining sustained peaks (the answer by Callum):
# EXAMPLE: findpeaks
library(pracma)
subset = c(570,584,500,310,261,265,272,313,314,315,330,360,410,410,360,365,368,391,390,414)
# Plots
plot(subset)
lines(subset)
# findpeaks without sustained peaks:
# the result does not identify the peak at subset[13:14] (repeated 'peak' values)
result = findpeaks(subset)
pks1 = result[,1]
locs1 = result[,2]
# findpeaks by defining sustained peaks (2 or more consecutive values):
result = findpeaks(subset, peakpat = "[+]{1,}[0]{1,}[-]{1,}")
pks2 = result[,1]
locs2 = result[,2]