Why do we need boundary conditions for finite difference methods?
More specifically why do we need boundary conditions for finite difference methods for option pricing?