Simple hash table for a conversion R script - r

Question

I would like to use hash table to implement simple coversion script.
An input should be multiplied with a factor using its symbol, i.e. y = x * 1E-12, should be called e.g. y <- X2Y(x,"p") with "p" being the symbol for 1E-12.
library(hash)
sym2num <- function(x) {
h <- hash( c("f"=1E-15,"p"=1E-12,"n"=1E-9,"mu"=1E-6,"m"=1E-3,"c"=1E-2) )
return(h$x)
}
X2Y <- function(X,x) {
xNum <- sym2num(x)
Y <- X * xNum
return(Y)
}
# y = x * 1E-12
y <- X2Y(x,"p")
print(y)
With the above code I get numeric(0) as result. Any idaes where it goes wrong?

Answer 1

There’s no benefit to using the {hash} library here. Indeed, since you rehash your vector before each subsetting, this will be substantially less efficient than a direct lookup.
Even if you only constructed the hash table once instead of repeatedly, it would probably still be faster not to use it: the hash table implementation carries a substantial constant overhead. It’s only faster than direct vector or list subsetting for fairly large tables.
Instead, just do this:
sym2num <- function(x) {
c(f = 1E-15, p = 1E-12, n = 1E-9, mu = 1E-6, m = 1E-3, c = 1E-2)[x]
}
This is idiomatic, efficient R code.
Fundamentally, the mistake in your R code was the subsetting, h$x. This fails because the subset operator $ doesn’t work with variables, it expects an unevaluated name on its right-hand side. The code will thus always look up the literal name x inside h. thc’s answer shows how to avoid the issue.

Answer 2

Your function sym2num always returns the hash of "x", which is NULL.
h$x is a shortcut for h[["x"]], but what you want is h[[x]].
Instead use this:
sym2num <- function(x) {
h <- hash( c("f"=1E-15,"p"=1E-12,"n"=1E-9,"mu"=1E-6,"m"=1E-3,"c"=1E-2) )
return(h[[x]])
}