[R] lists as matrix cells ?

[Sarah Goslee]

A matrix may only contain one data type. By not specifying when you
created m, it was filled with logical values of NA.

A logical value can't hold a list.

You can see that with
str(m)
which returns:
> str(m)
 logi [1:3, 1:2] NA NA NA NA NA NA
 - attr(*, "dimnames")=List of 2
  ..$ : chr [1:3] "Row_one" "Row_two" "Row_three"
  ..$ : chr [1:2] "Column_A" "Column_B"

If you were to instead specify that the elements of m should be list objects:


m <- matrix(list(), nrow=3, ncol=2,
  dimnames=list(
    c("Row_one", "Row_two", "Row_three"),
    c("Column_A", "Column_B")
  )
)

(and check with str(m) again)

then you can replace those list objects with other list objects
m[[ 1,1 ]] <- list(1.0, 2.0)


Please also note that [ and [[ are not the same thing.

I'm not sure this is the most efficient solution for your problem, but
since I'm not really sure what your problem is I solved the question
asked rather than the underlying problem.

Sarah

On Wed, Nov 21, 2012 at 11:11 AM, Asis Hallab <asis.hallab at gmail.com> wrote:
> Dear R experts,
>
> since more or less half a year I am using R.
>
> In many of my computations I construct huge matrices. Often I do so using
> 'cbind' on named lists:
>
> do.call( 'cbind',
>   list(
>     "Column_A"=list("Row_one"=1.0, "Row_two"=2.0, "Row_three"=3.0),
>     "Column_B"=list("Row_one"=4.0, "Row_two"=5.0, "Row_three"=6.0)
>   )
> )
>
> # Returns:
>
>           Column_A Column_B
> Row_one   1        4
> Row_two   2        5
> Row_three 3        6
>
> In some cases I even construct matrices with lists as cell content:
>
> do.call( 'cbind',
>   list(
>     "Column_A"=list("Row_one"=list(1.0, 2.0), "Row_two"=list(2.0, 3.0),
> "Row_three"=list(3.0, 4.0)),
>     "Column_B"=list("Row_one"=list(4.0, 5.0), "Row_two"=list(5.0, 6.0),
> "Row_three"=list(6.0, 7.0))
>   )
> )
>
> # Returns:
>           Column_A Column_B
> Row_one   List,2   List,2
> Row_two   List,2   List,2
> Row_three List,2   List,2
>
> Interestingly I seem not to be able to initialize a 3*2 matrix and
> subsequently fill its cells with lists in order to produce the latter
> example:
>
> m <- matrix( nrow=3, ncol=2,
>   dimnames=list(
>     c("Row_one", "Row_two", "Row_three"),
>     c("Column_A", "Column_B")
>   )
> )
> # Returns:
>           Column_A Column_B
> Row_one         NA       NA
> Row_two         NA       NA
> Row_three       NA       NA
>
> # The following expressions produce the same error:
> m[ "Row_one", "Column_A" ]   <- list(1.0, 2.0)
> m[[ "Row_one", "Column_A" ]] <- list(1.0, 2.0)
> m[ 1,1 ]   <- list(1.0, 2.0)
> m[[ 1,1 ]] <- list(1.0, 2.0)
>
> # Error returned by each of the above expressions:
> *number of items to replace is not a multiple of replacement length*
>
> *So, now my questions:*
> *1)* What am I doing wrong? How to get the above to work?
>
> *2)* Or am I misusing matrices in R? Is it just by coincidence ( bug ) that
> my 'cbind' or 'rbind' calls can generate matrices with lists as cell
> content, while I seem not to be able to do so by direct assignment (as in
> above example expressions)?
>
> Any ideas, comments and help will be much appreciated!
> Kind regards!
> Josef
>

--
Sarah Goslee
http://www.functionaldiversity.org