I cannot understand the reason why the approximation obtained through finite difference does not converge to the exact solution of the following problem
clear;clc;f=@(x)2*exp(x).*(2*sin(pi*x)-2*pi*cos(pi*x)+pi^2*sin(pi*x));sol=@(x)2*exp(x).*sin(pi*x);%exact solution
a=0;b=1;N=10000;h=(b-a)/(N+1);x=(a:h:b)';x(1)=[];x(end)=[];%define diffusion matrix
Ad=2*diag(ones(N,1));Ad=Ad-diag(ones(N-1,1),-1);Ad=Ad-diag(ones(N-1,1),1);Ad=Ad/h^2;sigma=3;%define transport matrix
At=zeros(N,N);At=At+diag(ones(N-1,1),1);At=At-diag(ones(N-1,1),-1);At=At*sigma/(2*h);A=Ad+At;F=f(x);U=A\F;U=[0 ; U ; 0];x=[a ; x ; b];plot(x,U,'--');%plot approximation
hold on;plot(x,sol(x));%plot exact solution
Best Answer