Updating plot in loop takes increasing amount of time

When I update a plot within a loop, each iteration takes longer after a while. Is there a way to avoid the slowdown? See example below:

f = figure;
a = axes(f);
hold(a,"on")

% note: y is supposed to represent real-time data that I receive with some
% frequency. The full vector Y is not available right before plotting
y = randn(1,100000);
for jj = 1:numel(y)
    tt = tic;
    plot(a,jj,y(jj),'pg');
    drawnow;
    pause(0.02)
    toc(tt)
end

I understand the x and y arguments of the plot are getting larger. Is that the cause? After 6,000 iterations, it already takes three times as long as when the loop it started. Thank you

[Matt J]

I assume there's a reason that you don't simply do this in a single plot command?

f = figure;
a = axes(f);
plot(a,y,'pg')

[dleal]

Hi Matt,

yes, I should have clarified. "y" is supposed to represent real time data that I poll with some frequency. So the full array is not available before plotting

Answer 1

[Walter Roberson]

Use animatedline()

[dleal]

Hi Walter, thanks! this is perfect. After 100,000 iterations, updating the animated line takes about the same for the first few updates as for the last ones. This works just great.

Answer 2

[Matt J]

I understand the x and y arguments of the plot are getting larger. Is that the cause?

The arguments aren't getting larger, but the number of objects in the plot is getting larger and larger. You can reduce the overall time, though, if the graph doesn't need to be redrawn as frequently.

Maybe update in sub-chunks of y:

f = figure;
a = axes(f);
hold(a,"on")
y = randn(1,100000);
for jj = reshape( 1:numel(y),1000,[]) %<--- loop over sub-chunks of length 1000
    tt = tic;
    plot(a,jj,y(jj),'pg');
    drawnow;
    pause(0.02)
    toc(tt)
end

[dleal]

thank you Matt! I did some additional tests and verified that drawnow takes a very very long time