This is machine translation

Translated by Microsoft
Mouseover text to see original. Click the button below to return to the English verison of the page.

Note: This page has been translated by MathWorks. Please click here
To view all translated materals including this page, select Japan from the country navigator on the bottom of this page.


Scattered translates collocation matrix


colmat = stcol(centers,x,type)
colmat = stcol(...,'tr')


colmat = stcol(centers,x,type) is the matrix whose (i,j)th entry is

ψj(x(:,i)),    i=1:size(x,2),j=1:n

with the bivariate functions ψj and the number n depending on the centers and the character vector type, as detailed in the description of stmak.

centers and x must be matrices with the same number of rows.

The default for type is the character vector 'tp', and for this default, n equals size(centers,2), and the functions ψj are given by

ψj(x)=ψ(xcenters(:,j)),   j=1:n

with ψ the thin-plate spline basis function


and with |x| denoting the Euclidean norm of the vector x.


See stmak for a description of other possible values for type.

The matrix colmat is the coefficient matrix in the linear system

jajψj(x(:,i))=yi,   i=1:size(x,2)

that the coefficients aj of the function f = Σjajψj must satisfy in order that f interpolate the value yi at the site x(:,i), all i.

colmat = stcol(...,'tr') returns the transpose of the matrix returned by stcol(...).


Example 1. The following evaluates and plots the function


on a regular mesh, with ψ the above thin-plate basis function, and with c1, c2, c3 three points on the unit circle; see the figure below.

a = [0,2/3*pi,4/3*pi]; centers = [cos(a), 0; sin(a), 0];
[xx,yy] = ndgrid(linspace(-2,2,45)); 
xy = [xx(:) yy(:)].';
coefs = [1 1 1 -3.5];
zz = reshape( coefs*stcol(centers,xy,'tr') , size(xx));
surf(xx,yy,zz), view([240,15]), axis off

Example 2. The following also evaluates, on the same mesh, and plots the length of the gradient of the function in Example 1.

zz = reshape( sqrt(...
        ([coefs,0]*stcol(centers,xy,'tp10','tr')).^2 + ...
figure, surf(xx,yy,zz), view([220,-15]), axis off

See Also


Was this topic helpful?