That of which we cannot speak, we must remain silent: absent symbols and the `frab` package

In frab: How to Add Two R Tables

knitr::opts_chunk$set(echo = TRUE)
suppressMessages(library("frab"))

knitr::include_graphics(system.file("help/figures/frab.png", package = "frab"))

I do not really understand what Wittgenstein was trying to say in proposition 7 of his Tractatus, but the interpretation of symbols not present in a frab object is certainly an interesting problem. In this short unstructured and informal document, I consider how one might consider key-value pairs in associative arrays when a particular key is absent. First, the canonical use-case for frab objects:

(a <- frab(c(u= -5,v=2,w=1)))
(b <- frab(c(v=1,x=2)))
a+b

Above we see that a has no "x" term, and b has "x"=2, the sum being "x"=2; we are justified in asserting that, in object b, "x"=0 even though it is not present. However, consider this:

(x <- frab(c(a=4, b=-3, c=1, d=-1, e=9)))
x[x > -2]

Above we extract every element greater than $-2$. We might say "extract every element from x that is known to exceed $-2$". It has not extracted any absent symbols such as f, and as such we cannot associate f with zero (because zero exceeds $-2$, and no f element was extracted). But this is not consistent with the a+b example above, in which absent symbols have the very definite and very known value of zero.

One might wonder what happens if we try to coerce a named logical vector to a frab:

(logvec <- c(x=TRUE,y=FALSE,z=TRUE))
frab(logvec)

[earlier versions of the package allowed the coercion]. However, we may coerce logvec to numeric (observe that as.numeric() loses the names):

(logvec <- +logvec)   # NB as.numeric() loses names
frab(logvec)

Above, we see that logvec["y"] vanishes, being zero. So logical frab objects do not make sense as FALSE entries are discarded.

frab documentation built on Sept. 12, 2024, 7:36 a.m.