[R] Unspecified [upper] xlim/ylim?

[Marc Schwartz]

On Jun 15, 2010, at 12:11 PM, Ted Harding wrote:

> On 15-Jun-10 16:01:24, William Dunlap wrote:
>>> -----Original Message-----
>>> From: r-help-bounces at r-project.org 
>>> [mailto:r-help-bounces at r-project.org] On Behalf Of 
>>> Ted.Harding at manchester.ac.uk
>>> Sent: Tuesday, June 15, 2010 8:49 AM
>>> To: r-help at stat.math.ethz.ch
>>> Subject: [R] Unspecified [upper] xlim/ylim?
>>> 
>>> Greetings!
>>> I would like to be able to specify a fixed (say) lower limit
>>> for plotting, while leaving the upper limit "floating, when
>>> plotting. The context is that the maximum in the data to be
>>> plotted is unpredictable, being the consequence of a simulation,
>>> whereas I know that it cannot be less than (say) 0; and I want
>>> to fix the lower limit at 0 in any plot, leaving the upper limit
>>> to be assigned by plot() as a result of the computed values.
>>> 
>>> I know I can do this by determining the max() of the data, and
>>> then computing a "Ymax" to put in (say) ylim = c(0,Ymax). However,
>>> for certain reasons, I would prefer not to have to do this.
>>> (And it's just a preference ... ).
>>> 
>>> Whereas one can leave the whole issue of setting both plotting
>>> limits to plot(), by not specifying ylim (or xlim), or one can
>>> explcitily specify both the upper and lower limits by (say)
>>> ylim=c(Ymin,Ymax), there seems to be no way of fixing one and
>>> leaving the other floating so that plot() would do its own thing.
>>> 
>>> ylim requires two numbers to be given. Things like ylim=c(0,)
>>> or ylim=c(0,NA) would generate an error.
>> 
>> Currently ylim=c(yValueAtBottomOfPlot, yValueAtTopOfPlot), not
>> c(yMin,yMax).  E.g., ylim=c(10,0) means to reverse the y axis,
>> with 0 at the top and 10 at the bottom.  Putting an NA into
>> ylim seems attractive but doesn't it run into problems because
>> ylim doesn't mean c(yMin,yMax)?
>> 
>> Bill Dunlap
>> Spotfire, TIBCO Software
>> wdunlap tibco.com  
>> 
>>> 
>>> Am I chasing a phantom? Or is there a way?
>>> Thanks
>>> Ted.
> 
> Sorry, Bill, but you've misunderstood what I mean by "Ymin" and "Ymax".
> As explained, these are notations for values which I either compute,
> or choose, to put in as the arguments in ylim=c(... , ...). YMin and
> Ymax would be such as to ensure that the range of the axis included
> at least the range (Ymin,Ymax) (or, depending on the choice, possibly
> to exclude certain values from the plot).
> 
> So, indeed, ylim does mean c(Ymin,Tmax).
> 
> Ted.


Ted, perhaps I am being dense here (always a possibility), but by default if, for example, 'ylim' is unspecified, plot() essentially uses range(YVals) as the min/max values for the Y axis. Also, by default, with par(yaxs = "r"), the Y axis range is extended by 4% in both directions. Same for the x axis range.

Here is the snippet of relevant code from plot.default() for 'ylim':

  ylim <- if (is.null(ylim)) 
     range(xy$y[is.finite(xy$y)])
  else ylim


Thus, if you want to explicitly specify the low end of the range for the Y axis and have the upper end of the range left to the default methodology, you would indeed use:

  ylim = c(0, max(YVals))

as the argument syntax. The same would apply for the x axis limits.

Is that what you are after?

HTH,

Marc Schwartz