pcme.tests {pcme}R Documentation

Solve maximum entropy completion problems on subsets

Description

Given a periodic autocovariance function, declare subsets of coefficients to be missing and maximise the entropy. Mainly for testing and examples.

Usage

pcme.test1(r, ...)
pcme.testcombn(r, w = 1, ...)

Arguments

r the periodic autocovariance function, a numeric matrix.
w size of gaps subsets, a numeric vector, see Details.
... parameters to be passed to the entropy maximisation function.

Details

pcme.test1 solves a sequence of maximum entropy (ME) problems with one gap. The gap is set to each of the non-zero lags, in turn.

pcme.testcombn solves a sequence of maximum entropy problems with number of gaps given by w. All subsets of w elements are considered. If w is a vector, then this is done for each element of w. Subsets of all lengths may be requested by setting w="all".

Value

For pcme.test1, a matrix whose (s,l)th element is the entropy of the ME solution when the gap is set to (s,l).
For pcme.testcombn, a list of the entropies. Each entropy has attibutes gaps and gapslist giving the gaps as a matrix and list, respectively.

Note

We used these functions as a basis for bulk testing.

Setting w="all" is convenient but there may be many, many subsets to go through...

Author(s)

Georgi Boshnakov

References

Boshnakov, Georgi and Lambert-Lacroix, Sophie (2009?) Maximum entropy for periodically correlated processes from nonconsecutive autocovariance coefficients. J. Time Series Anal. (to appear)

See Also

pcme, pcme-package

Examples

data(pcme.paperex)
resex1   <- pcme.test1(pcme.paperex$rex1)
resexAll <- pcme.testcombn(pcme.paperex$rex1,"all")

data(pcme.someex)
pcme.testcombn(pcme.someex$r1,1)
pcme.testcombn(pcme.someex$r1,1:2)
resr1 <- pcme.testcombn(pcme.someex$r1,"all")

[Package pcme version 0.51 Index]