There is a minor confusion as shortest wavelength eigensolutions with n = (N-1) has a large eigenvalue whose modulus |lN-1| = l1 (lN-1 = -l1) is equal to that of "difficult" long wavelength case |
One would expect it to be damped strongly. This is an artifact of this iteration scheme as even (l = 0, 2, 4 ....) and odd grid points are decoupled |
Update of even l points only involves odd l |
Update of odd l points only involves even l |
We will return to this when we discuss relaxation |