WaveThresh Package News

WaveThresh 4.6.1 Information

• Probably the most noticeable change is that the nlevels function has been renamed to be called nlevelsWT. Users will have noticed that with previous versions a message alerting the user to a clash with an already existing function, nlevels which is to do with the number of levels in a statistical factor. Our change of name removes that clash, but legacy code might need to be changed.
• Merging of the cthresh package into WaveThresh. This is because half of the code for this package was effectively within WaveThresh anyway (e.g. wst for complex-valued was in cthresh but the inversion code AvBasis was in WaveThresh!). The way that cthresh was originally meant to work was as a layer on top of WaveThresh, but that is not feasible now, at least in a straightforward way. The new version is a simplification in that there is now only one filter.select, wd, wst, wr.wd function in one package!
• Some functions, such as PsiJ printed out a lot of messages that were not useful. This could be incredibly annoying and these have been removed.
• Some C I/O calls have been removed and changed to calls to Rprintf or REprintf. There are also no longer any calls to exit either.
• Several minor bugfixes.

Enjoy. Guy Nason, Oct 31 2012.

WaveThresh 4.5 Information

WaveThresh, version 4.5, is now available directly from CRAN (or install.packages() from within R itself). This is now the preferred method of using WaveThresh.

Note that when you install WaveThresh then you have to type:
library("wavethresh")
to make it work (ie all lower case).

You might be interested in learning WaveThresh in conjunction with the new Springer book Wavelet methods in Statistics with R

If you use WaveThresh, or my book, please do remember to reference it!

Guy Nason
15th March 2010.

Supplementary material for the book

Guy Nason
4th August 2008.

Book Errata

Semandeni should be Semadeni

This occurs at lines -14 and -10 on Page 154 and line -14 on page 250.

Page 4

The forward reference to "Figure 2.26 on page 29" is incorrect. The correct forward reference is "Figure 2.4 on page 30".

Line -9 on page 20.

A (9*9=)81 is missing. The total should be 300.

Line -7 on page 29.

Delete word "defined".

Section title 2.2.3.

Replace "from-finer scale" by "from finer-scale".

Equation (2.33) on page 32

Quantity is f0-f1, ie reversed. See comments below relating to formula (2.91) and (2.42).

Line 6 on page 33.

Delete word "defined"

Line -3 on page 44

"two" should be "three" (i.e. three vanishing moments)

Formula (2.56) on Page 46.

Second line of the formula. Of course, Sinc[0] = 1.

Formula (2.91) and (2.42)

The formula for the wavelet coefficients are different by a factor of -1. This makes no material difference (If \psi(x) is a wavelet then \psi(-x) is one too). Indeed, the explanation prior to (2.91) makes it clear that it is a generalization and hence need not be exactly the same!

Line -13 on page 160

The line of code that calls the as.numeric should be:

ix <- as.numeric(dimnames(tb)[[1]]) + 1

Line 13 on page 188. The call to diff, constructing the dBabyECG data set is wrong. It shoud read

> dBabyECG <- diff(c(BabyECG[2], BabyECG))

hazest <- function(z, delta, nbins=32, levN=1, levD=1,
    filter.number=8){

Line -10 on page 236. The call to hazest needs another argument: delta (the censoring indicator). The function header on this line should read:

hazest <- function(z, delta, nbins=32, levN=1, levD=1,
    filter.number=8){

Line -7 on page 236

The line of code that calls the function subf should be

fsub <- subf(z=z[delta==1], nbins=nbins, filter.number=filter.number,
    lev=levN)

Survival/Hazard Rate Code