Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.


Dalia
Posts:
2
Registered:
2/8/13


Matlab : Test Polynomial Interpolation Program
Posted:
Feb 8, 2013 5:30 PM


Hi, I'm Italian, so I apologize for my bad english...i'm writing here because in this period I'm trying to study matlab, but I think I denied...so I have a program and I ask you what tests can I do to verify its operation, the program is this:
% The function y allows you to calculate the value of the % interpolating polynomial, given some starting points in % arbitrary values of the variable % independent. %It accepts INPUT% in three vectors: % x: whose components constitute the abscissas of the %points interpolating; % y whose components are the ordinates of the %interpolating points; % t: tab vector, whose components are the values on which % we want to calculate the polynomial interpolating the %points (x (i), f (i)) % and a scalar: % TOLL: tolerance within which must be the absolute value % of % difference of the abscissa (or ordinate) of the points % of interpolation. % In OUTPUT instead provides the vector y whose %components %are the calculated values of the polynomial % in the components of the vector t tab.
function y= Newton(x, f, t, TOLL) n= length(x); m=length(f); k =length(t); y=1:1:k; if n~= m % length control vectors x and f error('error:interpolation points unspecified'); end A=zeros(n,n); % initialize matrix designed to contain %the divided differences for i=1:1:n A(i,i)=f(i); end for i=2:1:n for j=i1 :1:1 if abs(x(i)x(j))<= TOLL && abs(f(i)f(j))<= TOLL error('coincident points'); elseif abs(x(i)x(j))<= TOLL error('abscissas coincident') end A(i,j)= (A(i,j+1)  A(i1,j))/(x(i)x(j)); % the matrix A at the end of the cycle will be %inferiorly triangular % and is such that for i> j %A(i,j)=f[x_j,x_(j+1),....,x_i] %with f [, ..] is the divided difference end end for h=1:1:k % calculation of the interpolating %polynomial in the components of t p=A(n,1); for i=n1:1:1 p=(t(h)  x(i))*p + A(i,1); end y(h)=p; end end



