Matlab Live Scripts displaying figures in seemingly random order when multiple code blocks with subplots are used

I am working in a Live Script where I am trying to align signals within different data sets and plot the results for each set. This is in a long script with many previous figures. The signal alignment/plotting code is as follows:

% for dataset 1
d1s1_aligned = alignsignals(d1s1, d1s2);

figure
ax1(1) = subplot(2,1,1);
plot(d1s1_aligned)
grid on
title('s1 aligned')
axis tight
ax1(2) = subplot(2,1,2);
plot(d1s2)
grid on
title('s2')
axis tight
linkaxes(ax1, 'xy')

This is for the first data set (as denoted by d1). If I only try to do this for one data set everything works fine, but I want to repeat this code block multiple times for d2, d3, etc. If, for instance, I add this block underneath:

% for dataset 2
d2s1_aligned = alignsignals(d2s1, d2s2);

figure
ax2(1) = subplot(2,1,1);
plot(d2s1_aligned)
grid on
title('d2s1 aligned')
axis tight
ax2(2) = subplot(2,1,2);
plot(d2s2)
grid on
title('d2s2')
axis tight
linkaxes(ax2, 'xy')

(the same code but d1 -> d2, ax1 -> ax2) all figures in the document end up being rendered at the bottom, out of order, and the two code blocks get merged. It seems maybe live scripts fail when there are multiple subplots in a script? Any advice would be much appreciated.

[Emilie Herpain]

For me, what worked is to add the command 'close all' before a new figure that I wanted to generate.

Answer 1

[Walter Roberson]

Because Live Script renders changes in-line, it postpones rendering until one of the following occurs:

• the axes is cleared or deleted (in which case the content immediately prior to deletion is displayed); or
• there is a return to the command line (such as the end of the Live Script -- but also because of debugging.)

Notice that I did not say "or figure() is called". Regular (non-Live) scripts update graphics when drawnow() or pause() or uiwait() or waitfor() or figure() are called, or the command line is returned to, but the same rules are not followed for Live Script. That is because you might have later statements in your code that switch back to the original figure and draw more in the figure.

figure(1)
plot(rand(1,20))
figure(2)
plot(randn(1,20))
cla()         %figure 2 will draw at this point
figure(1)
hold on
plot(rand(1,20).^2);

%figure 1 draws because you reached the end of the script

[Dylan Green]

Thanks for this - as a bit of noob, my interpretation was to add in cla() at the end of each of the code blocks mentioned in the question. This cleared the second axis in each subplot and the other figures still display in the same random order at the bottom. Is there an easy fix to just force each of these to plot inline and not affect earlier plots?

Thanks again!

[Walter Roberson]

Hmmm... I don't know anymore ;-)

In Live Script, the below code renders figure 2 first if and only if the code does not randomly put the text label in figure 2. But at the place it renders figure 2, it cannot know whether or not there is going to be later activity. So we have to conclude that if objects are not cleared or deleted, that the rendering is inserted at the place where there was the last activity for that figure, and that in turn implies that in such cases the figures are not drawn until the end of the script.

The question would then become what we can do to trigger earlier release deliberately, or what we need to do in order to avoid triggering earlier release, and I am no longer sure :(

figure(1)
plot(rand(1,20), 'b')
title('1')
figure(2)
plot(randn(1,20), 'r')
title('2')
hold on   %figure 2

figure(1)
hold on   %figure 1
plot(randn(1,20), 'g');
hold off

figure(2)
if rand() < 0.5
    text(5,.5, 'randomly in 2')
end

[Dylan Green]

What is confusing to me is that I do essentially the same thing all the time. There are many instances where in a code block I use plot(), the figure displays, and then the next code block does the same and everything works. It is not until this particular block which uses subplot() that things fail, and further it is only when I have more than one of them. Given that Live Scripts are Matlab's answer to Jupyter notebooks, it is frustrating that they do not work in a similarly intuitive way.

Thanks for the help, in any case.

[Dylan Green]

I've realized that the line that breaks this is the call to linkaxes(). If this is remove the problem goes away, although that unfortunately makes the plots quite a bit uglier.

Answer 2

[Diaa]

As a workaround, you can use this MATLAB mechanism in your favor by creating a duplicate dummy figure then delete it to show the intended contents at your specified code location

dummyFig = figure;
copyobj(<handle of some axes in the intended parent figure >, dummyFig);
clf(dummyFig)

[Duncan Carlsmith]

This doesn't work for me.