[R] Strange behavior in Lattice when superimposing

[Adam Lyon]

Hi,

I am trying to do a simple superimposing of two density plots and am 
seeing strange behavior (I haven't seen this reported elsewhere). This 
little snipit will make it apparent...

a = rnorm(10)
b = rnorm(100, mean=-2, sd=0.5)
densityplot( ~ a + b)

This is supposed to superimpose the two distributions on the plot, as 
if they were in different "groups". But the resulting density plot is 
goofy -- it looks like it is mixing the two distributions instead of 
plotting each one separately. In fact I can prove that the 
distributions are being mixed...

groups = latticeParseFormula(~ a+b, parent.frame(), multiple=T)$groups

If I understand things correctly, the groups vector has length = the 
combined size of the input vectors (e.g. 110 in this case). The values 
in "groups" are factors indicating which input vector is being pulled 
from. So there should be 10 a's and 100 b's. But in fact I get...

sum(groups=="a")
[1] 55

sum(groups=="b")
[1] 55

So indeed Lattice is for some reason mixing the vectors and that's why 
I'm seeing the incorrect plots.

I'm using Lattice 0.10-14 on R 2.0.1 for MacOSX. Details are below...

  platform = powerpc-apple-darwin6.8
  arch = powerpc
  os = darwin6.8
  system = powerpc, darwin6.8
  status =
  major = 2
  minor = 0.1
  year = 2004
  month = 11
  day = 15
  language = R

I'm not going to have time to try R 2.1.0 for a little while. Is this 
problem fixed in that version? If so, then I'll try it sooner.

Thanks in advance for any assistance!

--- Adam

Adam Lyon
Fermi National Accelerator Laboratory
Computing Division / D0 Experiment