wt.filter.shift {wavelets}

Wavelet Transform Filter Phase Shift

Description

Computes phase shifts for wavelet transform coefficients corresponding to a particular filter.

Usage

wt.filter.shift(filter, J, wavelet=TRUE, coe=FALSE, modwt=FALSE)

Arguments

filter

A wt.filter object, a character string indicating which wavelet transform filter to compute, or a numeric vector of wavelet (high pass) filter coefficients (not scaling (low pass) coefficients). If a numeric vector is supplied, the length must be even.

J

A vector of positive integers indicating levels for which to compute phase shifts.

wavelet

A logical flag indicating whether to compute the wavelet (high pass) or scaling (low pass) phase shift(s).

coe

A logical value indicating whether to use the center of energy method (see Percival and Walden 2000, page 118) to compute the phase shift(s).

modwt

A logical value indicating whether to compute MODWT phase shift(s).

Details

For wavelet filters of class 'Least Assymetric' or 'Coiflet', the default method for calculating phase shifts is outlined on pages 112-114 and page 124 of Percival and Walden 2000. For wavelet filters of class 'Best Localized', the default shifts are given on page 119 of Percival and Walden 2000. For the Haar filter, both the level j wavelet and scaling phase shifts are half the length of the level j wavelet and scaling filters and the phase shifts for the D(4) filter are determined by specifying \nu = -1 in equations (114a) and (114b) of Percival and Walden 2000.

For all other filters, the center of energy method is the default method for computing phase shifts (thus rendering the coe argument irrelevant). If coe = TRUE, then the center of energy method is used regardless of filter class.

By default the DWT phase shifts are computed, using the MODWT phase shifts and the methods outlined on pages 115-116 of Percival and Walden 2000. If modwt = TRUE, then only the MODWT phase shifts are computed.

Value

A vector of phase shifts, expressed in absolute value. Each element corresponds to the equivalent element in J.

Author(s)

Eric Aldrich. [e-mail].

References

Percival, D. B. and A. T. Walden (2000) Wavelet Methods for Time Series Analysis, Cambridge University Press.

See Also

wt.filter, align,

Examples

wt.filter.shift("la14", J=1:6)