Perform SSA Decomposition

Description

Performs the SSA decomposition.

Usage

1
2
3
4
5
6
## S3 method for class 'ssa'
decompose(x, neig = NULL, ..., force.continue = FALSE)
## S3 method for class 'toeplitz.ssa'
decompose(x, neig = NULL, ..., force.continue = FALSE)
## S3 method for class 'cssa'
decompose(x, neig = NULL, ..., force.continue = FALSE)

Arguments

x

SSA object holding the decomposition.

neig

number of desired eigentriples or 'NULL' for default value (minimum from 50 and trajectory space dimension).

...

additional arguments passed to SVD routines.

force.continue

logical, if TRUE then continuation of the decomposition is explicitly requested

Details

This is the main function which does the decomposition of the SSA trajectory matrix. Depending on the SVD method selected in the ssa different SVD implementations are called. This might be the ordinary full SVD routines or fast methods which exploit the Hankel / Toeplitz / Hankel with Hankel blocks matrix structure and allow the calculation of first few eigentriples.

Some SVD methods support continuation of the decomposition: if the 'ssa' object already holds some decomposition and more eigentriples are requested, then the decomposition continues using the current values as a starting point reducing the computation time dramatically.

Value

The SSA object.

Note

Usually there is no need to call this function directly. Call to ssa does the decomposition in the end. Other functions do the decomposition when necessary.

See Also

Rssa for an overview of the package, as well as, svd, ssa.

Examples

1
2
3
4
5
6
# Decompose 'co2' series with default parameters and decomposition turned off.
s <- ssa(co2, force.decompose = FALSE, svd.method = "nutrlan")
# Perform the decomposition
decompose(s, neig = 50)
# Continue the decomposition
decompose(s, neig = 100)

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.