how do i evaluate the convergence

supercali

got fourier series as a result of solving a PDE. how do i evaluate the converg. using average error in order to determine the # of terms needed for it to converge to less than X%?

ChaseRLewis

Well brute force is Solve with n vars then solve with n+1 vars and compare |n-(n+1)|/|n+1| once that is less than X% difference you can do that. Since each individual component should have less value towards the total solution than the previous one.

ChaseRLewis

Quote by ChaseRLewis

Well brute force is Solve with n vars then solve with n+1 vars and compare |n-(n+1)|/|n+1| once that is less than X% difference you can do that. Since each individual component should have less value towards the total solution than the previous one.

oops was thinking about this I mistyped it's |f(n)-f(n+1)|/|f(n+1)| where f is your pde function and n equivalent to the number of components. meant that before but was rushing.

supercali	how do i evaluate the convergence got fourier series as a result of solving a PDE. how do i evaluate the converg. using average error in order to determine the # of terms needed for it to converge to less than X%?

ChaseRLewis	Well brute force is Solve with n vars then solve with n+1 vars and compare \|n-(n+1)\|/\|n+1\| once that is less than X% difference you can do that. Since each individual component should have less value towards the total solution than the previous one.