tomarparambyComp {mixAR}

Translations of my old MixAR Mathematica functions

Description

Translations of some of my MixAR Mathematica functions. Not sure if these are still used.

Usage

tomarparambyComp(params)
tomarparambyType(params)
permuteArpar(params)

Arguments

params

the parameters of the MixAR model, a list, see Details.

Details

tomarparambyComp is for completeness, my Mathematica programs do not have this currently.

The arrangement of the parameters of MixAR models in package "MixAR" is “by type”: slot prob contains the mixture probabilities (weights), shift contains intercepts, and so on.

An alternative representation is “by component”: a list whose k-th elements contains all parameters associated with the k-th mixture component. The functions described here use the following order for the parameter of the k-th component: prob_k, shift_k, arcoeff_k, sigma2_k.

tomarparambyType takes an argument, params, arranged “by component” and converts it to “by type”. tomarparambyComp does the inverse operation, from “by type” to “by component”.

permuteArpar creates all permutaions of the components of a MixAR model. It takes a “by component” argument. The autoregressive orders are not permuted, in that if the input model has AR orders c(2, 1, 3), all permuted models are also c(2, 1, 3). The AR coefficients of shorter or longer components are padded with zeroes or truncated, respectively, see the unexported adjustLengths().

Value

For tomarparambyComp, a list containing the parameters of the model arranged “by component”, see Details.

For tomarparambyType, a list containing the parameters of the model arranged “by type”. It contains the following elements.

prob

mixture probabilities, a numeric vector,

shift

shifts, a numeric vector,

arcoef

autoregressive coefficients,

s2

noise variances, a numeric vector.

For permuteArpar, a list with one element (arranged “by type”) for each possible permutation of the AR parameters.

Author(s)

Georgi N. Boshnakov

See Also

randomArCoefficients

Examples

bycomp <- list(list(0.1, 10,  0.11,                1),
               list(0.2,  20, c(0.11, 0.22),       2),
               list(0.3,  30, c(0.11, 0.22, 0.33), 3) )
bytype <- tomarparambyType(bycomp)
identical(bycomp, tomarparambyComp(bytype)) # TRUE
               
permuteArpar(bycomp)