如何将列按功能应用于矩阵列表

Question

How would I go about applying a function by column to a list of matrices? For example I have a list like below.

[[1]]
[[1]][[1]]
    [,1] [,2] [,3]
[1,] "b"  "c"  "d" 
[2,] "y"  "y"  "y" 
[3,] "z"  "z"  "z" 

[[1]][[2]]
    [,1] [,2] [,3] 
[1,] "b"  "b"  "c" 
[2,] "c"  "d"  "d" 
[3,] "y"  "y"  "y" 
[4,] "z"  "z"  "z" 


[[2]]
    [,1] [,2]
[1,] "y"  "z" 

This works fine:

apply(p[[1]][[1]],2,gen.fmla,y="q")

[[1]]
log(q) ~ b + y + z
<environment: 0x920732c>

[[2]]
log(q) ~ c + y + z
<environment: 0x912e66c>

[[3]]
log(q) ~ d + y + z
<environment: 0x85b608c>

But I can't figure out how to apply it to the list. lapply alone doesn't work as it applies the function to the entire matrix. I was trying to use a combo of apply and lapply, but couldn't figure it out.

Answer 1

To get a better answer, you need to supply a reproducible example. For a general answer to your problem, you can use lapply twice. For example:

##Create some data
R> l = list()
R> l[[1]] = matrix(rnorm(10), 2); l[[2]] = matrix(rnorm(10), 2)*10
R> L = list()
R> L[[1]] = l; L[[2]] = l
R> f = function(l) lapply(l, apply, 2, sum) 
R> lapply(L, f)
[[1]]
[[1]][[1]]
[1]  1.1923  0.5275  0.4957  0.6848 -0.2776

[[1]][[2]]
[1] -13.984  15.435 -16.362   8.799   4.186

<snip>

Or using the rapply function:

#Gives the same as above
R> rapply(L, function(i) apply(i, 2, sum), how="replace")

Answer 2

Your problem is not simply addressed by lapply since it is not a simple list. The first element has two lists each of which has as its first element a matrix. The second element is just a matrix. There is an rapply function, which could be used if you provide a sensible test case of list and function.