[Scilab-users] 3D interpolation : comments adding

[tim at wescottdesign.com]

The help says that splin2d generates bicubic patches, so presumably it's
not linear interpolation. 

On 2017-03-25 00:24, paul.carrico at free.fr wrote:

> Hi All
> 
> To go further in 2D/3D interpolation as I started in my previous emails, I built the example here after. 
> 
> As suggested, I had a look to 
> - cshep2d but but seems can not be used here (dimensions issue) 
> - splin2d + interp2d 
> 
> In the later case, Am I right to say that splin2d "cross-sections" the surface in order to define the "best spline" passing through the nodes (in the cross section obviously), then the interpolation/calculation is basically using this new 2D function, right ? 
> 
> if so this is not a linear interpolation between 2 nodes (as I expect), isn't it? 
> 
> Paul 
> 
> ############################################ 
> 
> mode(0)
> 
> n=10;
> x = linspace(0,300,(n+1))';    // abscissa
> t = [0 25 100];                         // temperature
> z = 20*rand((n+1),3);            // ordinate
> // Nota : we must have the same number of data for both x and z
> 
> //  plot3d
> clf()
> a=get("current_axes");
> a.x_label; x_label=a.x_label; x_label.text=" X abscissa";
> a.y_label; y_label=a.y_label; y_label.text=" Temperature T";
> a.z_label; z_label=a.z_label; z_label.text=" y ordinate";
> plot3d(x,t,z)
> 
> // Nota: if I do a cross section normal to XoZ plane, I've the basic curve z=f(x,T)) -> seems correct
> 
> //  interpolation
> xp = [22 103 236]'
> tp = [5 56 85]'
> 
> [Xp,Tp] = ndgrid(xp,tp)
> 
> _______________________________________________
> users mailing list
> users at lists.scilab.org
> http://lists.scilab.org/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.scilab.org/pipermail/users/attachments/20170325/d232332b/attachment.htm>