(I am not sure why the system keeps deleting my reply)
Thank you for your reply Iajze. I am aware of the discussion surrounding
the vibrating rectangular plates, but I still think it can be helpful to
lay out my readings here.
In the
StackExchange discussion, they mentioned that the solution to the
biharmonic EV problem is not separable, but according to Everitt (2009),
the solution is separable for k=2. This kind of explains why the 1D
solver works for k=0,2 but not the rest. However, I would expect the 2D
solver to be able to find the correct eigenvalues for the rest of the
modes.
Another common method in the
literature for solving the biharmonic EV problem is to factorise it into
two Laplace EV problems. Unfortunately, this would not work for the
problem I am trying to solve which is a thin-film equation
https://en.wikipedia.org/wiki/Thin-film_equation.
So, to summarise my questions:
1.
Is it possible to use the chebfun (or diskfun) package to solve the
biharmonic EV problem, and more generally, any fourth-order linear
eigenvalue problem?
2. If it is not possible to use an
out-of-the-box solution, then I will have to implement my own. Am I
missing something causing the 2D solver to not find the correct
eigenvalues?
- W. N. Everitt, B. T.
Johansson, L. L. Littlejohn, C. Markett, Quasi-separation of the
biharmonic partial differential equation, IMA Journal of Applied
Mathematics, Volume 74, Issue 5, October 2009, Pages 685–709.