[R] Peculiar behaviour subtracting from named numeric vector

[Rich Grenyer]

Hi - I'm just puzzling over something R's been doing when subtracting a 
constant from a named numeric vector. I hope this isn't anything 
embarrassingly simple, but I've tried and failed the usual searches. 
For the last item in my vector, 40-21=19, but 40-40 = 7.105E-15 . If I 
replace the last item of the vector explicitly with 40 (i.e. 
goodtimees[21]<-40), then 40-40=0. I presume that this means the last 
item in the vector is not 40, but 40 + 7.105E-15. This is possible 
(goodtimees is the output of a function in an R addon package - it 
should be at the same precision as the input to that function, but 
there could be a problem with it, perhaps) but either way, why isn't 
the true value showing when the vector is displayed? Is there a more 
deep-rooted problem?

Thanks in advance...

Rich


 > goodtimees
   -1   -2   -3   -4   -5   -6   -7   -8  -28  -48  -93  -96 -152 -163 
-164 -167 -168 -169 -177 -188 -189
98.1 86.8 77.9 76.3 70.6 61.1 53.6 51.5 47.5 43.8 41.7 72.3 41.8 65.3 
49.4 61.1 54.9 43.0 44.9 53.0 40.0

 > str(goodtimees)
  Named num [1:21] 98.1 86.8 77.9 76.3 70.6 ...
  - attr(*, "names")= chr [1:21] "-1" "-2" "-3" "-4" ...

 > goodtimees-21
   -1   -2   -3   -4   -5   -6   -7   -8  -28  -48  -93  -96 -152 -163 
-164 -167 -168 -169 -177 -188 -189
77.1 65.8 56.9 55.3 49.6 40.1 32.6 30.5 26.5 22.8 20.7 51.3 20.8 44.3 
28.4 40.1 33.9 22.0 23.9 32.0 19.0

 > goodtimees-40
           -1           -2           -3           -4           -5        
    -6           -7           -8
5.810000e+01 4.680000e+01 3.790000e+01 3.630000e+01 3.060000e+01 
2.110000e+01 1.360000e+01 1.150000e+01
          -28          -48          -93          -96         -152        
  -163         -164         -167
7.500000e+00 3.800000e+00 1.700000e+00 3.230000e+01 1.800000e+00 
2.530000e+01 9.400000e+00 2.110000e+01
         -168         -169         -177         -188         -189
1.490000e+01 3.000000e+00 4.900000e+00 1.300000e+01 7.105427e-15

 > version
          _
platform powerpc-apple-darwin7.9.0
arch     powerpc
os       darwin7.9.0
system   powerpc, darwin7.9.0
status
major    2
minor    1.1
year     2005
month    06
day      20
language R



--------------------
Rich Grenyer, Ph.D.
Biology Department - University of Virginia
Gilmer Hall
Charlottesville, Virginia
VA 22904
United States of America

tel: (+1) 434 982 5629
fax: (+1) 434 982 5626
http://faculty.virginia.edu/gittleman/rich