reshape data to fit into surf/contour

nodalPositions is a 2*100 double matrix storing the nodal positions of a PDE on a 2D domain with 100 nodes. x=nodalPositions(1,:) and y=nodalPositions(2,:) are the nodal x and y coordinates. The solution of the PDE, u, is a 1*100 double array. One can plot(x,y,'.') to plot the nodes in a plane or plot3(x,y,u,'.') to plot u in 3-space. How to reshape x, y, u so that one can surf(x,y,u) or contour(x,y,u)?

Answer 1

[Walter Roberson]

You cannot do that.

Consider using https://www.mathworks.com/matlabcentral/fileexchange/38858-contour-plot-for-scattered-data

Or,

N = 100;
[minx, maxx] = bounds(x);
[miny, maxy] = bounds(y);
[Xg, Yg] = ndgrid( linspace(minx, maxx, N), linspace(miny, maxy, N));
F = scatteredInterpolant(x(:), y(:), u(:));
Ug = F(Xg, Yg);

surf(Xg, Yg, Ug, 'edgecolor', 'none');

[feynman feynman]

many thanks for your code.

Error using surf
Data dimensions must agree.

Error in untitled2 (line 16)
surf(Xg, Yg, Ug, 'edgecolor', 'none');

[Walter Roberson]

Seems to work fine for me.

nodalPositions = rand(2, 100);
x=nodalPositions(1,:);
y=nodalPositions(2,:);
u = rand(1,100);

N = 100;
[minx, maxx] = bounds(x);
[miny, maxy] = bounds(y);
[Xg, Yg] = ndgrid( linspace(minx, maxx, N), linspace(miny, maxy, N));
F = scatteredInterpolant(x(:), y(:), u(:));
Ug = F(Xg, Yg);
surf(Xg, Yg, Ug, 'edgecolor', 'none');

[feynman feynman]

many thanks. I used to test with y=x rather than y=nodalPositions(2,:), resulting in Error using surf (Data dimensions must agree). May I ask why?

[Stephen23]

"May I ask why?"

You defined colinear data: what do you expect the interpolation to return? Take a look at the size of Ug.

[Walter Roberson]

@Stephen23 is right. When y = x then all of the data falls along a single line. Triangulation of a single line fails.