Question

Here is some basic code and its output. I really can’t say anything more than a logical test for a sequence containing 1.2 is giving an inaccurate result. It is working for many other values.

# Incorrect
> seq(0.5, 1.5, by=0.05) == 1.2
 [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
[13] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE

# Correct
> seq(0.5, 1.5, by=0.05) == 1.15
 [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
[13] FALSE  TRUE FALSE FALSE FALSE FALSE FALSE FALSE FALSE

# Correct
> seq(0.5, 1.5, by=0.05) == 1.25
 [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
[13] FALSE FALSE FALSE  TRUE FALSE FALSE FALSE FALSE FALSE

# Correct
> seq(0.5, 1.5, by=0.05) == 1.3
 [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
[13] FALSE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE FALSE

I tried testing all the values using the following, which does not reproduce the bug:

> sapply(seq(0.5, 1.5, by=0.05), function(x){sum(seq(0.5, 1.5, by=0.05) == x)})
 [1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

I’m using R version 2.13.2 (2011-09-30), Platform: x86_64-pc-linux-gnu (64-bit).

Answer 1

You can duplicate what all.equal is doing by writing a comparison function of your own:

is.nearenough=function(x,y,tol=.Machine$double.eps^0.5){
  abs(x-y)<tol
}

then you can do which(is.nearenough(s,1.2)) where s is your sequence. You may need to tweak the tolerance for your application.