demo/ctap.R

In abarbour/psd: Adaptive, Sine-Multitaper Power Spectral Density and Cross Spectrum Estimation

##
## Show usage of taper constaint procedures
##

# fake taper series
set.seed(1234)
nd <- 2e2
x <- rnorm(nd, sd=10) + c(1:(nd/2),(nd/2 + 1):2)
ntap <- as.tapers(rep(x,3))

layout(matrix(1:4,ncol=2))
op <- par(no.readonly = TRUE)
ylim <- c(0,130)

## Raw tapers
plot(ntap, ylim=ylim, main="Tapers: Raw")

## Constrained tapers:
## minspan
plot(as.tapers(minspan(ntap)), ylim=ylim, main="minspan")

## loess
plot(ntap, ylim=ylim, main="loess") # OK, but a bit too smooth: reduce span
# will complain if if has to assume a sequence -- this is OK though
lines(suppressWarnings(ctap_loess(ntap)), col="greenyellow", lwd=2)
lines(suppressWarnings(ctap_loess(ntap, loess.span=0.1)), col="red", lwd=2)

## first difference
plot(ntap, ylim=ylim, main="simple")
lines(ctap_simple(ntap, maxslope=5), col="greenyellow", lwd=2) #resembles noise: use min slope (1, the default)
lines(ctap_simple(ntap), col="red", lwd=2)

par(op)
layout(matrix(1))