Avoid floating point zero when summing scaled set of numbers in R

Question

I am rescaling a set of numbers and want to avoid getting a floating point zero for the sum of the rescaled numbers:

x <- c(-5, 1, 8)
y <- scale(x)
sum(y)
# [1] 1.249001e-16

Is there a way to around this to force the sum to zero? I do not care about precision beyond a three decimal places.

Answer 1

I think you should not just "switch" to integers at some point. The scaling was computed with floating point numbers and is therefore not 100% precise. Forcing some values to 0 would indicate a precision that is not available and should thus be avoided.

If you need to compare floating point values use isTRUE(all.equal(...)) as suggested by R documentation. https://stat.ethz.ch/R-manual/R-devel/library/base/html/all.equal.html