R devel - Feb 2009 - invalid comparison in numeric sequence (PR#13551)

Full_Name: alex
Version: 2.8.1
OS: Ubuntu / MacOSX
Submission from: (NULL) (162.38.183.51)

> 0.6==0.6
[1] TRUE
> seq(0,1,0.1)==0.4
 [1] FALSE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE FALSE FALSE
FALSE
> seq(0,1,0.1)==0.6
 [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
FALSE
> seq(0,1,0.1)==0.8
 [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE  TRUE FALSE FALSE


What is wrong with 0.6 ??? (TRUE is missing)
I tried 3 differents computers (2 Ubuntu with R 2.8.1, and one Mac with R 2.8).

alexandre.courtiol at gmail.com wrote:
>
>> 0.6==0.6
> [1] TRUE
>> seq(0,1,0.1)==0.4
>  [1] FALSE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE FALSE FALSE FALSE
>> seq(0,1,0.1)==0.6
>  [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
>> seq(0,1,0.1)==0.8
>  [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE  TRUE FALSE FALSE
> 
> 
> What is wrong with 0.6 ??? (TRUE is missing)
> I tried 3 differents computers (2 Ubuntu with R 2.8.1, and one Mac with R
2.8).
FAQ 7.31, not a bug. Expect about 10 people to tell you so.

-- 
   O__  ---- Peter Dalgaard             ?ster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark      Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)              FAX: (+45) 35327907

This is one of the joys of floating point arithmetic.  Look at the 7th
element of

  seq(0,1,.1) - 0.6

Also, check the all.equal(x,y,tolerance=epsilon ) function.

John


-----r-devel-bounces@r-project.org wrote: -----


To: r-devel@stat.math.ethz.ch
From: alexandre.courtiol@gmail.com
Sent by: r-devel-bounces@r-project.org
Date: 02/24/2009 08:55AM
cc: R-bugs@r-project.org
Subject: [Rd] invalid comparison in numeric sequence (PR#13551)

Full_Name: alex
Version: 2.8.1
OS: Ubuntu / MacOSX
Submission from: (NULL) (162.38.183.51)

> 0.6==0.6
[1] TRUE
> seq(0,1,0.1)==0.4
 [1] FALSE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE FALSE FALSE
FALSE
> seq(0,1,0.1)==0.6
 [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
FALSE
> seq(0,1,0.1)==0.8
 [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE  TRUE FALSE FALSE


What is wrong with 0.6 ??? (TRUE is missing)
I tried 3 differents computers (2 Ubuntu with R 2.8.1, and one Mac with R
2.8).

______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
	[[alternative HTML version deleted]]

> > seq(0,1,0.1)==0.4
>  [1] FALSE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE FALSE FALSE FALSE
> > seq(0,1,0.1)==0.6
>  [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
> > seq(0,1,0.1)==0.8
>  [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE  TRUE FALSE FALSE
> 
> What is wrong with 0.6 ??? (TRUE is missing)
> I tried 3 differents computers (2 Ubuntu with R 2.8.1, and one Mac with R
2.8).
If you know that all the numbers in a sequence should have a given decimal
precision, then you obtain a better result using round(,digits=...).
For example,

  x <- round(seq(0,1,0.1), digits=1)
  rbind(x == 0.4, x == 0.6, x ==0.8)

produces

        [,1]  [,2]  [,3]  [,4]  [,5]  [,6]  [,7]  [,8]  [,9] [,10] [,11]
  [1,] FALSE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE FALSE FALSE FALSE
  [2,] FALSE FALSE FALSE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE FALSE
  [3,] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE  TRUE FALSE FALSE

This does not mean, that x[7] is now equal to 0.6, however both x[7] and 0.6
are represented by the same 53-bit floating point number

  formatC(x[7], digits=20) # "0.5999999999999999778"
  formatC(0.6, digits=20)  # "0.5999999999999999778"

Petr.