postprocess: Post-process of a psd object

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

Post-process of a psd object

Description

This function allows to discard a specified number of samples as burn-in period and apply thinning factor to a psd object.

Usage

postprocess(x, burnin, thin = 1)

Arguments

x	a psd object
burnin	number of initial iterations to be discarded
thin	thinning number (post-processing)

Value

A list with S3 class ‘psd’ containing the following updated components:

psd.median,psd.mean	psd estimates: (pointwise) posterior median and mean
psd.p05,psd.p95	90% pointwise credibility interval
psd.u05,psd.u95	90% uniform credibility interval
fpsd.sample	posterior power spectral density estimates
tau,phi,delta,V	posterior traces of model parameters
ll.trace	trace of log likelihood
DIC	deviance information criterion
count	acceptance probabilities for the weigths

See Also

plot.psd

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);

# Run MCMC (may take some time)
mcmc = gibbs_pspline(data, 5000, 0);
mcmc = postprocess(mcmc, burnin = 500, thin = 10);

require(beyondWhittle)  # For psd_arma() function
freq = 2 * pi / n * (1:(n / 2 + 1) - 1)[-c(1, n / 2 + 1)]  # Remove first and last frequency
psd.true = psd_arma(freq, ar = 0.9, ma = numeric(0), sigma2 = 1)  # True PSD
plot(mcmc)  # Plot log PSD (see documentation of plot.psd)
lines(freq, log(psd.true), col = 2, lty = 3, lwd = 2)  # Overlay true PSD


## End(Not run)

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