Pre-determine plot ranges?

Given two vectors of the same length is there a way to determine the axes ranges that would result if the data were plotted without plotting the data?

I am looking to update xlim and ylim manually after the plotted data is changed* but with out relying on the automatic resizing as there are other things plotted that I dont want to be used in determining the plot ranges. Also I would rather not redraw the entire figure with just the modified data then redraw the larger thing.

(plotted data is changed*: I am manipulating the XData and YData parameter of plotted data directly with a handle.)

Answer 1

[Adam Danz]

If vector1 and vector1 are the x and y values to be plotted, the xlim and ylim for a tight axis would be,

xlim([min(vector1), max(vector1)]);
ylim([min(vector2), max(vector2)]);

If you'd like to access those values from the line object handle 'h',

xlim([min(h.XData), max(h.XData)]);
ylim([min(h.YData), max(h.YData)]);

If you'd like to add 5% of the data range on each side so your max and min data points are not on the axis edges,

xlim([min(x),max(x)] + [-1,1]*.05*range(x))
ylim([min(y),max(y)] + [-1,1]*.05*range(y))

[Chris]

Tight axes will look bad for this application,

By default ML does some nice rounding/flooring/ceiling-ing/ to get good tickmarks ranges with some buffer around the edges of the plot. I am hoping to precalculate something like what axis returns below.

>> xx = [-0.01, 1, 2.2]; yy = [0.2 3.01, 2];
>> plot(xx,yy)
>> axis
ans =
         -0.5          2.5            0          3.5 

edit:

I think this might work

>> aa = [min(xx), max(xx)] + [-1,1] .* mean(diff(get(gca, 'xtick')))
aa =
        -0.51          2.7
>> bb = min(diff((get(gca, 'xtick')')))
bb =
          0.5
>> round(aa./bb)*bb
ans =
         -0.5          2.5          

Was hoping there was a direct way. Above looks to work with larger magnitudes of data. thanks for the help.

[Adam Danz]

The 3rd block of code in my answer will add a buffer around your data so the axes are not "tight". If you'd like to take that a step further and make sure the axis limits begin and end at a certain interval (like multiples of 0.5) you easily round those limits.

Here's a demo.

int = 0.5; 
xl = [floor(min(x)/int)*int,ceil(max(x)/int)*int]; 
yl = [floor(min(y)/int)*int,ceil(max(y)/int)*int]; 

xlim(xl)
ylim(yl)

[Adam Danz]

Your proposal requires that an axis already exists. If you want to base "int" from my solution on the axis tick interval,

int = mean(diff(get(gca,'xtick'))); 
xl = [floor(min(x)/int)*int,ceil(max(x)/int)*int]; 

[Chris]

yep good edit. initilization wont be a problem and the changed data points will be of the same magnitude so reusing tick should work.