[R] Inconsistency of 1^NA=1 vs. 1.1^NA=NA

[Duncan Murdoch]

On 17/11/2016 2:19 PM, Da Zheng wrote:
> Hello,
>
> I just realized that 1^NA outputs 1 while 1.1^NA outputs NA in R v3.3.1 and
> R v3.2.3.
> I tried other values such as 0^NA and 2^NA, and they all output NA.
> I don't understand this inconsistency here. Shouldn't 1^NA output NA as
> well? Why does R handle it differently? Or is this a bug in these
> particular versions of R?

Our usual interpretation of NA is "an unknown value".  So 1^NA would 
give NA if there were multiple possible values for it depending on what 
number you substitute for NA, and should give the unique answer if 
there's only one possibility (as for example NA | TRUE gives TRUE).

As far as I can see, 1 raised to any power (even infinite ones) should 
give 1, so the answer looks fine to me.  That's not true of any of the 
other bases you mention (just as NA | FALSE gives NA).

Duncan Murdoch