[R] Using apply on a three dimensional matrix and passing multiple arguments to user defined function

[Justin Peter]

Dear Liusfo and Jean,

Thank you both for your help and suggestions, which both work.

As Jean mentioned, there is no real speed up using the apply family, which I thought there would be, so I will stick with the for loop for clarity.

Liusfo, the reason I used c(1,2) in the apply function (i.e. mask_data <- apply(data,c(1,2),mask,y=lsmask)) was because this is what you would do if you wanted to sum all the two-dimensional vectors over time (for instance to produce an average of a field over a year).

i.e. to get the sum you would do sum_data <- apply(data,c(1,2),sum)

I thought it would extend to using my mask function.

Anyway, thanks again for both of your help.

Cheers,
Justin
--
Justin Peter
Research Fellow
International Centre for Applied Climate Sciences,
University of Southern Queensland
West St, Toowoomba, QLD, 4350
Australia

Email: justin.peter at usq.edu.au<mailto:justin.peter at usq.edu.au>
Ph: +61 (0) 7 4631 1181
Fax: +61 (0) 7 4631 5581
Mob: +61 (0)474 774 107


-----Original Message-----
From: Luisfo <luisfo89 at yahoo.es<mailto:Luisfo%20%3cluisfo89 at yahoo.es%3e>>
To: "Adams, Jean" <jvadams at usgs.gov<mailto:%22Adams,%20Jean%22%20%3cjvadams at usgs.gov%3e>>, Justin Peter <Justin.Peter at usq.edu.au<mailto:Justin%20Peter%20%3cJustin.Peter at usq.edu.au%3e>>
CC: r-help at r-project.org <r-help at r-project.org<mailto:%22r-help at r-project.org%22%20%3cr-help at r-project.org%3e>>
Subject: Re: [R] Using apply on a three dimensional matrix and passing multiple arguments to user defined function
Date: Wed, 7 Sep 2016 16:05:48 +0200

Hi,

Jean's example with lapply works fine.

However, if you still want to use apply, I think this works.
One observation first. You were passing c(1,2) as second argument to apply, in your code. And that is what makes you have lots of NAs as a result, since your function is being applied twice, by rows and columns (first and second dimensions) respectively.
Use:
    masked_data <- apply(data,3,mask,y=lsmask)
    # but now masked_data has dim(nlon*nlat,ntime), so change it
    dim(masked_data) <- dim(data)

The apply goes over the third dimension (second parameter '3'), so it takes every nlot*nlat matrix as first argument for function mask.
I think it should work.

Regards,

Luisfo Chiroque
PhD Student | PhD Candidate
IMDEA Networks Institute
http://fourier.networks.imdea.org/people/~luis_nunez/<http://fourier.networks.imdea.org/people/%7Eluis_nunez/>


On 09/07/2016 03:17 PM, Adams, Jean wrote:



Justin,

I don't think you can get the apply() function to return an array.  You
could use lapply() instead, and then use simplify2array() to convert the
list of matrices to an array.  Also, in your mask() function you don't need
the which() and you should return the x.  See my example with toy data
below.

# toy data
nlon <- 2
nlat <- 4
ntime <- 3
data <- array(1:(nlon*nlat*ntime), dim=c(nlon, nlat, ntime))
lsmask <- array(sample(0:1, size=nlon*nlat, replace=TRUE), dim=c(nlon,
nlat))

# newly defined function
mask <- function(x, y) {
  x[y==0] <- NA
  x
}

# doit
data2 <- simplify2array(lapply(1:ntime, function(i) mask(data[, , i],
lsmask)))


You may prefer to stick with the for() loop approach (for clarity or
simplicity or ...)  When I ramped up the toy data to much larger
dimensions, the lapply() approach was only slightly faster than the for()
loop approach on my PC.

data3 <- data
data3[ , , i] <- mask(data3[ , , i], lsmask)

Jean




On Tue, Sep 6, 2016 at 11:33 PM, Justin Peter <Justin.Peter at usq.edu.au><mailto:Justin.Peter at usq.edu.au>
wrote:




Dear R-user,

I have a three-dimensional matrix of atmospheric data. The first two
dimensions are spatial (lon and lat) and the third is time, such that

dim(data) <- c(nlon,nlat,ntime)

I wish to apply a land sea mask data which is a matrix of "0" and "1" if
dim(nlon,nlat)

dim(lsmask) <- c(nlon,nlat)

I wish to set all of the elements in the two-dimensional array of
data[,,ntime] for every 1:length(ntime).

I could do this in a loop:

for (i in 1:ntime){
    data[,,i][which(lsmask == 0)] <- NA
}

I would like to do this using apply, but I need to pass two variables to
the function in apply (data and lsmask), where data is a two-dimensional
array.

I tried:

mask <- function(x,y) {x[which(y==0)] <- NA}

masked_data <- apply(data,c(1,2),mask,y=lsmask)

but I get back a vector of dim(nlon,nlat) populated with NA.

Any clues as to what I am missing?

Thanks in advance for you help.

Kind regards,
Justin



--
Justin Peter
Research Fellow
International Centre for Applied Climate Sciences,
University of Southern Queensland
West St, Toowoomba, QLD, 4350
Australia

Email: justin.peter at usq.edu.au<mailto:justin.peter at usq.edu.au><mailto:justin.peter at usq.edu.au><mailto:justin.peter at usq.edu.au>
Ph: +61 (0) 7 4631 1181
Fax: +61 (0) 7 4631 5581
Mob: +61 (0)474 774 107




_____________________________________________________________
This email (including any attached files) is confidential and is for the
intended recipient(s) only. If you received this email by mistake, please,
as a courtesy, tell the sender, then delete this email.

The views and opinions are the originator's and do not necessarily reflect
those of the University of Southern Queensland. Although all reasonable
precautions were taken to ensure that this email contained no viruses at
the time it was sent we accept no liability for any losses arising from its
receipt.

The University of Southern Queensland is a registered provider of
education with the Australian Government.
(CRICOS Institution Code QLD 00244B / NSW 02225M, TEQSA PRV12081 )


        [[alternative HTML version deleted]]

______________________________________________
R-help at r-project.org<mailto:R-help at r-project.org> mailing list -- To UNSUBSCRIBE and more, see
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.





        [[alternative HTML version deleted]]

______________________________________________
R-help at r-project.org<mailto:R-help at r-project.org> mailing list -- To UNSUBSCRIBE and more, see
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.




_____________________________________________________________
This email (including any attached files) is confidential and is for the intended recipient(s) only. If you received this email by mistake, please, as a courtesy, tell the sender, then delete this email.

The views and opinions are the originator's and do not necessarily reflect those of the University of Southern Queensland. Although all reasonable precautions were taken to ensure that this email contained no viruses at the time it was sent we accept no liability for any losses arising from its receipt.

The University of Southern Queensland is a registered provider of education with the Australian Government.
(CRICOS Institution Code QLD 00244B / NSW 02225M, TEQSA PRV12081 )


	[[alternative HTML version deleted]]