Single value in colors vector causes error in scatter plot

Hi,

I have the following code that works perfectly, the NaN values in the colors vector are skipped as desired:

>> colors = [NaN, 121, NaN, 150]

colors =

   NaN   121   NaN   150

>> scatter([1,2,3,4], [1,2,3,4], 50, colors, 'filled');

which produces the following graph:

However, when i replace a value in the colors array with NaN it all goes wrong:

>> colors = [NaN, 121, NaN, NaN]

colors =

   NaN   121   NaN   NaN

>> scatter([1,2,3,4], [1,2,3,4], 50, colors, 'filled');
Warning: Error updating Scatter.

 The logical indices contain a true value outside of the
 array bounds.

Which stops the scatter graph from being plotted, it seems to happen when the vector is all NaN but one value.

Why does simply changing one of the values to NaN cause this error?

It doesn't matter which value is changed, but as soon as there is a single value surrounded by NaN's it will break.

Thanks in advance for any suggestions

Answer 1

[DGM]

The code you give works fine in the newest version that I have (R2019b), but not in R2021b. I'm not exactly sure what changed, so I can't do much to troubleshoot that, but I don't know that there's a need.

If you want to mask off points, it's probably better to simply omit the X and Y data directly. The input for the color property is subject to a bunch of conditional operations and scaling, so it kind of makes sense that it might explode if it only has one valid numeric value in the vector.

x = 1:4;
y = 1:4;
colors = [NaN, 121, NaN, NaN];

mk = ~isnan(colors);
scatter(x(mk), y(mk), 50, colors(mk), 'filled');

colorbar
caxis([1 256])

[Fin Cottle]

Great solution thanks very much!