I see a bug in pcolor

C = eye(10);
pcolor(C)
colormap summer

This example, which comes straight out of the documentation, produces a 9 x 9 colored block matrix, not a 10 x 10 matrix. This is a bug.

Answer 1

[Image Analyst]

It's not a bug - it's a feature. That's why I never use pcolor. I use imshow instead, which shows every pixel.

m = magic(3) % Create a 3-by-3 matrix

m = 3×3

8 1 6 3 5 7 4 9 2

subplot(2, 1, 1);
pcolor(m)
title('pcolor');
subplot(2, 1, 2)
imshow(m, []);
title('imshow');

Note that pcolor has the vertices (where the black lines are) be the value of the pixel, not the value of the tiles.

I suggest that you use imshow or image or imagesc, with the colormap option if you want a colormap.

[Steven Lord]

@Image Analyst is correct. This is not a bug. pcolor lets you specify the colors at the vertices, image and related functions let you specify the colors of the faces. See the Algorithm section on the pcolor documentation page.

One analogy I've seen is that games like checkers and chess are played with the pieces on the faces of the board, while go and othello are played on the vertices. Neither is wrong, they're just different.

[Mark Mendlovitz]

I'm still not clear how the entries of a 10 x 10 matrix would correspond to a 9 x 9 plot of vertices. Maybe you mean edges? Still, I can think of beter ways to plot a graph. Regardless, I'll use imshow instead.

[Steven Lord]

As another analogy, this time in 1 dimension: it's the fence post versus fence rail problem. If you want to create a fence with two segments you need three posts, with the first rail segment being between posts 1 and 2 and the second between posts 2 and 3. The image-related functions let you say what color the rails are. The pcolor function lets you specify the color of the posts and colors the rails with the color of the post where they start. The color of the last post (or the last row/column in the pcolor case) doesn't affect the color of any rail.