isLogDenDrift: isLogDenDrift
In twDEMC: parallel DEMC

Description Usage Arguments Details Value Author(s) Examples

check whether first quartile all the logDensities is significantly smaller than last quartile

1	isLogDenDrift(logDenT, dInfos, alpha = 0.05, maxDrift = 0.3)

`logDenT`	numeric array (nStep x nResComp): logDensity (highest are best)
`dInfos`	list of lists with entry resCompPos (integer vector) specifying the position of result components for each density
`alpha`	the significance level for a difference
`maxDrift`	difference in LogDensity, below which no drift is signalled

Because of large sample sizes, very small differences may be significantly different. Use argument minDiff to specify below which difference a significant difference is not regarded as drift.

TRUE if any of the logDensities are a significantly greater in the fourth quantile compared to the first quantile of the samples , attribute resTTest: numeric maxtrix (logDenStart, logDenEnd, pTTest x nDInfo ) , attribute logDenComp: numeric vector (nResComp): average logDenT of all components of last quater

Thomas Wutzler

data(twdemcEx1)
logDenT <- calcTemperatedLogDen( twdemcEx1, getCurrentTemp(twdemcEx1))
    #undebug(isLogDenDrift)
isLogDenDrift(logDenT, twdemcEx1$dInfos )