hadron (version 3.1.2)

computeacf: Computes The ACF and Integrated AC Time

Description

Computes the ACF and integrated autocorrelation time of a time series. It also estimates the corresponding standard errors.

Usage

computeacf(tseries, W.max, Lambda = 100)

Arguments

tseries

the time series.

W.max

maximal time lag to be used.

Lambda

cut-off needed to estimate the standard error of the ACF.

Value

It returns a list of class hadronacf with members

lags

time lags of the integrated autocorrelation function

Gamma

normalised autocorrelation function

dGamma

error of normalised autocorrelation function

W.max

max time lag used for the call of acf

W

the cut-off up to which the ACF is integrated for the integrated autocorrelation time

tdata

the original time series

tau

the estimated integrated autocorrelation time

dtau

the estimated error of the integrated autocorrelation time

Details

The standard error of the ACF is computed using equation (E.11) of M. Luescher, hep-lat/0409106. The error of the integrated autocorrelation time using the Madras Sokal formula, see also hep-lat/0409106.

References

'Monte Carlo errors with less errors', Ulli Wolff, http://arxiv.org/abs/hep-lat/0306017hep-lat/0306017

'Schwarz-preconditioned HMC algorithm for two-flavour lattice QCD', Martin Luescher, http://arxiv.org/abs/hep-lat/0409106hep-lat/0409106

N. Madras, A. D. Sokal, J. Stat. Phys. 50 (1988) 109

See Also

uwerr, acf bootstrap.analysis

Examples

# NOT RUN {
data(plaq.sample)
myacf <- computeacf(plaq.sample, 300)
plot(myacf)
summary(myacf)

# }