R help - Oct 2011 - Selecting multiple vectors from a list of lists of matrices

Hi all,

I was unable to find a solution to my problem in the archives, but this might be
due to a lack knowledge on the correct terminology on my part. Please forgive me
if this has been explained before and please forgive me my probably clumsy way
of explaining things.

This is what I want to do:

I have a list made up of 6 lists containing 7  4x4 matrices each.
My goal is to select a large number of rows from all those matrices (thus giving
vectors of length 4), and stack them into a matrix or data frame.
Creating a single such vector is easy using the normal extract functionality,
say I want the 2nd row of the 5th matrix in the 3rd list, I use:
my.list[[3]] [[5]] [2, ]

Problem is, I need to get a matrix or data frame of about 8000 of these rows...

My idea was to create three vectors for each of the three indices, and put those
vector names where the 3, 5 and 2 are in my example. I did something similar
before with a 3-dimensional array, where you can use a matrix with dimensions
[n,3] as an index to create a vector of length n. This, however, does not work
with lists. I created a workaround using the lapply function (shown below) which
works but is incredibly slow.

extract <- function (selector) {
        matrix.list[[selector[1]]][[selector[2]]][selector[3],]
}

output <- function(var1,var2,var3) {
        selector <-  as.data.frame(rbind(var1,var2,var3))
        stack <- do.call('rbind',lapply(selector,FUN=extract))
        stack
}


The object matix.list is the said object consisting of 6 lists containing 7 4x4
matrices each. Var 1 2 and 3 are three vectors of length 8000.

My problem that, although this is working, it is really slow. The output is used
in a monte carlo simulation with many iterations, so the function is called over
and over again. What parts of this code are slowing it down the most, and how
can I speed things up?

Thanks for all the help!

Cheers,
Gimon

  ________________________________
De inhoud van dit bericht is vertrouwelijk en alleen bes...{{dropped:15}}

This is somewhat faster on my machine:

t(sapply(seq(var1), function(i) my.list[[var1[i]]] [[var2[i]]] [var3[i], 
]))

Jean


Graaf, G de wrote on 10/14/2011 09:23:24 AM:
> 
> Hi all,
> 
> I was unable to find a solution to my problem in the archives, but 
> this might be due to a lack knowledge on the correct terminology on 
> my part. Please forgive me if this has been explained before and 
> please forgive me my probably clumsy way of explaining things.
> 
> This is what I want to do:
> 
> I have a list made up of 6 lists containing 7  4x4 matrices each.
> My goal is to select a large number of rows from all those matrices 
> (thus giving vectors of length 4), and stack them into a matrix or data 
frame.
> Creating a single such vector is easy using the normal extract 
> functionality, say I want the 2nd row of the 5th matrix in the 3rd 
> list, I use:
> my.list[[3]] [[5]] [2, ]
> 
> Problem is, I need to get a matrix or data frame of about 8000 of 
> these rows...
> 
> My idea was to create three vectors for each of the three indices, 
> and put those vector names where the 3, 5 and 2 are in my example. I
> did something similar before with a 3-dimensional array, where you 
> can use a matrix with dimensions [n,3] as an index to create a 
> vector of length n. This, however, does not work with lists. I 
> created a workaround using the lapply function (shown below) which 
> works but is incredibly slow.
> 
> extract <- function (selector) {
>         matrix.list[[selector[1]]][[selector[2]]][selector[3],]
> }
> 
> output <- function(var1,var2,var3) {
>         selector <-  as.data.frame(rbind(var1,var2,var3))
>         stack <- do.call('rbind',lapply(selector,FUN=extract))
>         stack
> }
> 
> 
> The object matix.list is the said object consisting of 6 lists 
> containing 7 4x4 matrices each. Var 1 2 and 3 are three vectors of 
> length 8000.
> 
> My problem that, although this is working, it is really slow. The 
> output is used in a monte carlo simulation with many iterations, so 
> the function is called over and over again. What parts of this code 
> are slowing it down the most, and how can I speed things up?
> 
> Thanks for all the help!
> 
> Cheers,
> Gimon
	[[alternative HTML version deleted]]