Skip to content

matrix() can't handle NaN (PR#193)

3 messages · Brian Ripley, Peter Dalgaard, Bill Simpson

Original
Brian Ripley
# 
The field is regarded as a character field unless  na.strings is
extended to include NaN.

I am not in favour of treating NaN as a number, which is linguistic nonsense!
(What does NaN stand for?)

First, in IEEE arithmetic, NaN is a class of numbers as I understand it
(and internally in R NA is one of those NaNs).  So just which one do
you mean?  And what are you going to achieve by having essentially a
second NA, except to blur the real purposes of NaNs?

Secondly, on non-IEEE implementations of R, I believe NaN = NA: in any
case we should not lightly be making IEEE features an essential part of the
language. (Or with the gradual demise of Vaxen is IEEE arithmetic now
universal: I still have a Sun running with a special-purpose non-IEEE
floating-point unit.)
Peter Dalgaard
# 
Bill Simpson <wsimpson@gcal.ac.uk> writes:
And actually the logic is a bit warped:
[1] NA
[1] NA
[1] NaN
[1] TRUE
[1] FALSE
[1] TRUE
We do allow them as constants in the language itself, so I suppose we
could do it in read.table as well - this really means inside
as.double, I suppose. The prototype does this:

Splus> is.nan(as.double("NaN"))
[1] T
Nope:
[1] 0+1i

or, equivalently,
[1] 0+1i
Bill Simpson
#
On Tue, 11 May 1999, Prof Brian Ripley wrote:

            

      
      
      
    

        
That's why I said that NaN was a "number" (with quotes). It is a number in
the sense that it is the output of an arithmetic operation.

            

      
      
      
    

        
Yes.

            

      
      
      
    

        
Oh.

            

      
      
      
    

        
Well to me NA means missing value, and it is abuse of NA to use it to
represent things like 0/0.

The purpose of having NaN to me:
I am allowed to get an NaN as a result of some operation in R. Let's say I
then save that in a file, then I read the file in again using read.table.
In the process of saving and reading it in I lose the NaN. I don't like
that. It doesn't seem consistent. If the NaN is to be changed, I would
like to do the changing explicitly myself, so I know what happened to it.

I would also include Inf and -Inf here. These are allowed outputs from R
operations, so why can't they be read in by read.table? So I am proposing
that allowable IEEE entities Inf, -Inf, NaN be read in by read.table as
themselves (not as character strings).

            

      
      
      
    

        
OK I see this point. That's what I meant when I said maybe it would be
opening a can of worms to try to use NaN and Inf consistently throughout
R.

Bill

PS: I remember reading on this list about complex numbers in R. That
brings in a conflict with IEEE (doesn't it?), because in IEEE,
sqrt(negative number)->NaN.
But that is not a NaN if you allow complex numbers (I would think).

            

      
      
      
    

        
Warning: NaNs produced in function "sqrt"
[1] NaN

Is there a need for an "I" symbol in R?

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._