spec_pspline: Spectrum estimate via p-splines

In pmat747/psplinePsd: Spectral density estimation using P-spline priors

Spectrum estimate via p-splines

Description

This function allows to estimate the spectrum

Usage

spec_pspline(
  data,
  l,
  p,
  eq = TRUE,
  Ntotal1,
  burnin1,
  thin1,
  Ntotal,
  burnin,
  thin,
  tau.alpha = 0.001,
  tau.beta = 0.001,
  phi.alpha = 1,
  phi.beta = 1,
  delta.alpha = 1e-04,
  delta.beta = 1e-04,
  k = NULL,
  eqSpacedKnots = FALSE,
  degree = 3,
  diffMatrixOrder = 1,
  printIter = 1000,
  recycl = FALSE,
  likePlot = FALSE
)

Arguments

data	numeric vector
l	length of the data subset to be analysed
p	overlapping percentage of the data subsets
eq	if is TRUE, last data subset has length l, even though the percentage is not p
Ntotal1	total number of iterations to run the pilot Markov chain
burnin1	number of initial iterations to be discarded in the pilot mcmc run
thin1	thinning number in the pilot mcmc run (post-processing)
Ntotal	total number of iterations to run the Markov chain
burnin	number of initial iterations to be discarded
thin	thinning number (post-processing)
tau.alpha, tau.beta	prior parameters for tau (Inverse-Gamma)
phi.alpha, phi.beta	prior parameters for phi (Gamma)
delta.alpha, delta.beta	prior parameters for delta (Gamma)
k	number of B-spline densities in the mixture
eqSpacedKnots	logical value indicating whether the knots are equally spaced or defined according to the periodogram
degree	positive integer specifying the degree of the B-spline densities (default is 3)
diffMatrixOrder	positive integer specifying the order of the difference penalty matrix in the P-splines (default is 2)
printIter	positive integer specifying the periodicity of the iteration number to be printed on screen (default 100)
recycl	if TRUE the mcmc analysis for a data subset is used to calibrate the proposal for the succesor data subset. It works for large p values.
likePlot	if TRUE, a likelihood traceplot is generated for each data subset

Value

A list with S3 class ‘psds’ containing the the power spectral estimates for the data subsets and and a list with relevant information about the analysis

See Also

gibbs_pspline

Examples

## Not run: 

set.seed(1)

# Generate AR(1) data with rho = 0.9
n = 128
data = arima.sim(n, model = list(ar = 0.9));
data = data - mean(data);

# Spectrum estimate via p-splines (may take some time)
spec = spec_pspline(data, l=50, p=90, Ntotal1=2000, burnin1=500, thin1=5,
                    Ntotal=1000, burnin=500, thin=5, k=30, recycl = TRUE);

image(spec) # Plot log PSD (see documentation of image.plot.psd)


## End(Not run)

pmat747/psplinePsd documentation built on July 7, 2023, 9:06 p.m.