17

I want to add elements to an empty list on the fly. Each element in the list should be named automatically after a set of variables which value will vary.

However, I cannot seem to find a way to name list elements on the fly without getting errors. Consider the example below:

L <- list()

var1 <- "wood"
var2 <- 1.0
var3 <- "z4"

varname <- paste(var1, as.character(var2), var3, sep="_")

# This works fine:
L$"wood_1_z4" <- c(0,1)
L$"wood_1_z4"
0 1

# This doesn't!!
L$paste(var1, as.character(var2), var3, sep="_") <- c(0,1)
Error in L$paste(var1, as.character(var2), var3, sep = "_") <- c(0, 1) : 
  invalid function in complex assignment

# Ths doesn't either ... 
L$eval(parse(text = "varname")) <- c(0,1)
Error in L$eval(parse(text = "varname")) <- c(0, 1) : 
target of assignment expands to non-language object

Is there a way to do this?

Neodyme
  • 557
  • 1
  • 9
  • 20

3 Answers3

14

You cannot assign to paste() using the <- operator (and I believe this is true for the eval() function as well). Instead, you need to either use the [[ operator or use the names() function. You can do this like so:

L <- list()
var1 <- "wood"
var2 <- 1.0
var3 <- "z4"
varname <- paste(var1, as.character(var2), var3, sep="_")

# Using [[
L[[varname]] <- c(0,1)

> L
$wood_1_z4
[1] 0 1

# Using names()
L[[1]] <- c(0,1)
names(L)[1] <- varname

> L
$wood_1_z4
[1] 0 1

A more effective way to do this might be to use a function that both creates the value and names the list element, or even one that just creates the value - if you then use sapply you can name each list element using arguments from the call to your function and the USE.NAMES options.

In a general sense, R isn't really well-optimized for growing lists over time when the final size and structure of the list aren't well-known. While it can be done, a more "R-ish" way to do it would be to figure out the structure ahead of time.

TARehman
  • 6,659
  • 3
  • 33
  • 60
  • 1
    While it's true that you can't assign to `paste` and `eval`, the reason why it isn't working is because the primitive `$` doesn't evaluate on the right hand side. Per the comments in the C code for `$`: _"We need to be sure to only evaluate the first argument. The second will be a symbol that needs to be matched, not evaluated."_ – Alex A. May 20 '15 at 14:25
5

An alternative would be to use the wrapper setNames() from the stats package, like so:

var1 <- "wood"
var2 <- 1.0
var3 <- "z4"
varname <- paste(var1, as.character(var2), var3, sep="_")

L <- setNames(list(c(0, 1)), varname)

> L
$wood_1_z4
[1] 0 1
Viktor Horváth
  • 139
  • 2
  • 3
1

Use the `names<-` infix operator as a normal function to return a named list. Then use c() to concatenate this with the list L. This one-liner can be used to continue to add elements to L 'on the fly' as requested.

L <- list()
var1 <- "wood"
var2 <- 1.0
var3 <- "z4"
varname <- paste(var1, as.character(var2), var3, sep="_")

L <- c(L, `names<-`(list(c(0, 1)), varname))

L
#> $wood_1_z4
#> [1] 0 1
cg334
  • 11
  • 1
  • 3