[R] flaw in CRAN package "wavelets": Daubechies "d8" not recognized by function wt.filter

[David Winsemius]

You can look at the code yourself. The d8 filter appears to have a  
deletion mutation making it "different" than the other d-series and  
the fix looks feasible for testing:

d6.filter <- function(mod = F) {
             class <- "Daubechies"
             name <- "d6"
             L <- as.integer(6)
             g <- c(0.332670552950083, 0.806891509311093,  
0.459877502118491,
                 -0.135011020010255, -0.0854412738820267,  
0.0352262918857096)
             if (modwt == TRUE) {
                 g <- g/sqrt(2)
                 transform <- "modwt"
             }
             else transform <- "dwt"
             h <- wt.filter.qmf(g, inverse = TRUE)
             wt.filter <- new("wt.filter", L = L, h = h, g = g,
                 wt.class = class, wt.name = name, transform =  
transform)
             return(wt.filter)
         }
         d8.filter <- function(mod = F) {
             class <- "Daubechies"
             name <- "d8"
             L <- as.integer(8)
             g <- c(0.230377813307443, 0.714846570548406,  
0.630880767935879,
                 -0.0279837694166834, -0.187034811717913,  
0.0308413818353661,
                 0.0328830116666778, -0.0105974017850021)

< missing code>

             h <- wt.filter.qmf(g, inverse = TRUE)
             wt.filter <- new("wt.filter", L = L, h = h, g = g,
                 wt.class = class, wt.name = name, transform =  
transform)
             return(wt.filter)
         }
         d10.filter <- function(mod = F) {
             class <- "Daubechies"
             name <- "d10"
             L <- as.integer(10)
             g <- c(0.160102397974193, 0.60382926979719,  
0.724308528437773,
                 0.138428145901320, -0.242294887066382,  
-0.0322448695846381,
                 0.0775714938400459, -0.0062414902127983,  
-0.012580751999082,
                 0.0033357252854738)
             if (modwt == TRUE) {
                 g <- g/sqrt(2)
                 transform <- "modwt"
             }
             else transform <- "dwt"
             h <- wt.filter.qmf(g, inverse = TRUE)
             wt.filter <- new("wt.filter", L = L, h = h, g = g,
                 wt.class = class, wt.name = name, transform =  
transform)
             return(wt.filter)
         }

Make your own wt.filter with the missing code inserted and see if  
happiness prevails.When I apply the fix, I get:

 > wt.filter("d8")
An object of class “wt.filter”
Slot "L":
[1] 8

Slot "level":
[1] 1

Slot "h":
[1] -0.01059740 -0.03288301  0.03084138  0.18703481 -0.02798377  
-0.63088077  0.71484657 -0.23037781

Slot "g":
[1]  0.23037781  0.71484657  0.63088077 -0.02798377 -0.18703481   
0.03084138  0.03288301 -0.01059740

Slot "wt.class":
[1] "Daubechies"

Slot "wt.name":
[1] "d8"

Slot "transform":
[1] "dwt"


By the way, there also appears to be no d2 and calling wt.filter with  
"d2" also provokes an error. The proper manner for getting this  
addressed is to draw it to the attention of the package maintainer  
rather than posting a "to whom it may concern" message to the list.

-- 
David Winsemius

On Mar 4, 2009, at 12:36 AM, <mauede at alice.it> <mauede at alice.it> wrote:

>> wt.filter("d8")                                #### HOW COME ????
> Error in validObject(.Object) :
>  invalid class "wt.filter" object: invalid object for slot  
> "transform" in class "wt.filter": got class "function", should be or  
> extend class "character"
>
>> wt.filter("d10")                             # OK
> An object of class “wt.filter”
> Slot "L":
> [1] 10
>
>> wt.filter("d6")                               #OK
> An object of class “wt.filter”
> Slot "L":
> [1] 6
>
>
> Thank you in advance to anyone who takes care of reported  bugs in  
> CRAN packages.
> Maura
>
>
> tutti i telefonini TIM!
>
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.