Griddata gives NaN values

Hello

In my code griddata gives NaN values, I checked if my query points are outside the convex hull of the sample data with the following code :

Xv=topologym_vert(:,1)/D;
Yv=topologym_vert(:,2)/D;
Zv=topologym_vert(:,3)/D;
ch=convhull(Xv,Zv)
plot(Xv(ch),Zv(ch),'r',xv,zv,'g')

and it seems everything is ok there,

So I wonder what else I have to change to get rid of NaN values?

the following is my code:

[xv,zv]=meshgrid(linspace((1/3),12,100),linspace(0.14,2.4,50));
aaa =
   -0.0434   -0.0122    0.0507    0.2726    0.2433    0.2366    0.2452   -0.0344   -0.0225    0.0504    0.2446    0.2414
    0.2426    0.2365   -0.0636   -0.0299    0.1221    0.2021    0.2335    0.2447    0.2399   -0.0342   -0.0002    0.0855
    0.1676    0.2263    0.2327    0.2397    0.0827    0.0961    0.1150    0.1465    0.2199    0.2319    0.2346    0.0958
    0.1031    0.1428    0.1659    0.2065    0.2332    0.2373    0.0896    0.0767    0.1249    0.1763    0.2160    0.2321
    0.2334    0.1084    0.1224    0.1356    0.1621    0.2262    0.2353    0.0656    0.1133    0.1299    0.1539    0.2001
    0.2315    0.2340    0.1426    0.1399    0.1321    0.2093    0.2289    0.1369    0.1675    0.1644    0.2151    0.2308
    0.1828    0.1668    0.1648    0.1755    0.2038    0.2264    0.2319    0.2012    0.1875    0.2016    0.1872    0.2070
    0.2210    0.2334
uvD=griddata(Xv,Zv,aaa,xv,zv);

Xv and Zv are attached.

Thanks.

[John D'Errico]

You do not provide the actual data, so it is impossible to know if the (xv,zv) points truly fall inside the convex hull of the data. All we have is your claim that the points appear to do so, when looking at a plot that we have not been given.

There are TWO possibilities, neither of which can be ruled out.

1. One or more of your data points may have a NaN in it.
2. the numbers you show with 2 significant digits 0.14 and 2.4, for example, are only approximations to the real range of your data.

I see one other inconsistency,

Yv=topologym_vert(:,2)/D;

I note that Yv is never used, yet I wonder if it may have been important.

[mehra]

yes Yv is not necessary here

I attached the actual data for Xv and Zv in my original question

thanks

Answer 1

[darova]

I think you are using griddata in a wrong way

clc,clear

load Xv.mat
load Zv.mat
load data.txt

xx = linspace(min(Xv),max(Xv),20);
zz = linspace(min(Zv),max(Zv),20);
[xv,zv] = meshgrid(xx,zz);          % create 2d mesh
yv = griddata(Xv,Zv,data,xv,zv);    % interpolate data

plot3(Xv,Zv,data,'.r')
surface(xv,zv,yv)
axis vis3d

[mehra]

I cant understand whats wrong with the way I use griddata, I am using it the same as yours but i use the required meshgrid and limits

[xv,zv]=meshgrid(linspace((1/3),12,100),linspace(0.2,2.4,50)); % This is Ok

Xv and Zv are also Ok,

But why the NaN values apear?

[darova]

Don't see NaN