Save in fig file only the data within the limits of the axes

I am plotting large vectors to a figure. After setting the x and y limits, the display data becomes a fraction of the initial vectors. In order to chose the limits, I need to see the full spectrum of the data. When saving I would like to reduce the size of the output file (.fig) to minimum. Therefore I am asking if figures could be save to .fig, containing only the data within the limits of the figure and not the entire initial ploted vectors (resulting in a smaller file). this is an automatic action so I am referring to matlab code for the saving function.

Answer 1

[dpb]

Think you have to write a function to do it from the chosen final limits...not too difficult to do.

xl=xlim; yl=ylim;  % Retrieve the current plot limits
x=x(iswithin(x,xl(1),xl(2));  % select the values within those limits
y=y(iswithin(y,yl(1),yl(2));
set(hL,'xData',x,'yData',y)   % set plot data to the subset above

In the above I have assumed --

x,y are the data you've plotted; your vector variables go here of course

hL is the handle to the line on the plot -- save it with the *plot* call

iswithin is my utility function--

>> type iswithin

function flg=iswithin(x,lo,hi)
% returns T for values within range of input
% SYNTAX:
%  [log] = iswithin(x,lo,hi)
%      returns T for x between lo and hi values, inclusive

    flg= (x>=lo) & (x<=hi);

ERRATUM

Above doesn't select the correlated points; my bad...

xl=xlim; yl=ylim;  % Retrieve the current plot limits
idx=iswithin(x,xl(1),xl(2)); % x values within logical address vector
x=x(ix); y=y(ix);            % select those that are within on x-direction
idx=iswithin(y,yl(1),yl(2)); % now those inside on the y axes, too...
x=x(ix); y=y(ix);            % now have those that are inside both only

set(hL,'xData',x,'yData',y)   % set plot data to the subset above

NB: idx is an addressing logical array used for each direction in turn.

[Cristian JECU]

Hello, Thank you for your idea. This was helpful. I used it to create this code addapted to my use:

clear all
close all
clc
Fig=figure('units','normalized','outerposition',[0 0 1 1]);
set(gca,'FontSize',18);
V(1,:)=0:5e-5:300;
V(2,:)=     sin(V(1,:)*2*pi*50);
V(3,:)=     sin((V(1,:)+  pi/3)*2*pi*50);
V(4,:)=     sin((V(1,:)+2*pi/3)*2*pi*50);
V(5,:)=4.51*sin((V(1,:)+  pi/6)*2*pi*50);
V(6,:)=4.51*sin((V(1,:)+2*pi/3+pi/6)*2*pi*50);
V(7,:)=4.51*sin((V(1,:)+  pi/6+pi/3)*2*pi*50);
[AX,H1,H2]=plotyy(V(1,:),V(2:4,:),V(1,:),V(5:7,:));
axes(AX(1));xlim([1 1.02]);ylim('auto');grid on;
set(H1,'LineWidth',2)
axes(AX(2));xlim([1 1.02]);ylim('auto');
set(gca,'FontSize',18);set(H2,'LineWidth',2,'LineStyle','--')
% save full file
saveas(Fig,strcat(pwd,'\all.fig'))
%limit the data on first axes
axes(AX(1));
xl=xlim;
Xdata=get(H1,'xData');
Ydata=get(H1,'yData');
for  ii=1:length(Xdata)
  temp=Xdata{ii,1};
  Xdata_lim{ii,1}=temp(find(Xdata{1}==xl(1)):find(Xdata{1}==xl(2)));
  temp2=Ydata{ii,1};
  Ydata_lim{ii,1}=temp2(find(Xdata{1}==xl(1)):find(Xdata{1}==xl(2)));
  clear temp*
  set(H1(ii),'xData',Xdata_lim{ii,1},'yData',Ydata_lim{ii,1})
end
%limit the data on second axes
axes(AX(2));
xl=xlim;
Xdata=get(H2,'xData');
Ydata=get(H2,'yData');
for  ii=1:length(Xdata)
  temp=Xdata{ii,1};
  Xdata_lim{ii,1}=temp(find(Xdata{1}==xl(1)):find(Xdata{1}==xl(2)));
  temp2=Ydata{ii,1};
  Ydata_lim{ii,1}=temp2(find(Xdata{1}==xl(1)):find(Xdata{1}==xl(2)));
  clear temp*
  set(H2(ii),'xData',Xdata_lim{ii,1},'yData',Ydata_lim{ii,1})
end
saveas(Fig,strcat(pwd,'\limited.fig'))

[dpb]

Doh, head slap!!!...forgot to select correlating data...replace the loops and find with

ix=iswithin(x,xl(1),xl(2));  % logical indexing array of those in limits
x=x(ix); y=y(ix);            % select the corresponding x,y arrays

Edited the original answer, too, for the oversight...

[Cristian JECU]

The inconvenience with your code was that you tested each value to check if is within the limits while for my use my X is a time (linear growing function) so there was no need to check more than redefining the xData and yData corresponding to the limits of X. I agree that for an all purpose figure only your solution should work. But given my monotonically growing X that is very large, my proposition is faster.

Thanks again for your help!

[dpb]

No problem...I've not timed it and I did look at the loop too quickly; I thought you were doing a find over the data array initially but it's the handles array...

If the data are monotonic then using

... temp(find(Xdata{1}==xl(1),1):find(Xdata{1}==xl(2),1,'last'));

should help at least a little, particularly if the arrays are sizable.

Also, should be able to take the search for X positions out of the loop; the X data are the same for each line and that search actually could be over V and save retrieving the XData at all.

[AX,H1,H2]=plotyy(V(1,:),V(2:4,:),V(1,:),V(5:7,:));
...
xlim(AX(1),[1 1.02]); ylim(AX(1),'auto');grid on;
...
axes(AX(1));
xl=xlim;
ix=find(V(1,:)==xl(1),1):find(V(1,:)==xl(2),1,'last');
Ydata=get(H1,'yData');
for ii=1:length(H1)
  set(H1(ii),'xData',V(1,ix),'yData',Ydata{ii}(ix})
end

and similarly for H2. I think I got the curlies right for the cell array returned for Y; you don't need any for the xdata as you're reference the raw array directly again. This moves invariant stuff out of the loop as much as possible and eliminates some unneeded temporaries as well.