This demo illustrates how to: * Solve a linear partial differential equation with Neumann boundary conditions * Use mixed finite element spaces The solution for :math:`u` in this demo will look as follows: .. image:: ../neumann-poisson_u.png :scale: 75 % Equation and problem definition: -------------------------------------- The Poisson equation is the canonical elliptic partial differential equation. For a domain :math:`\Omega \subset \mathbb{R}^n` with boundary :math:`\partial \Omega`, the Poisson equation with particular boundary conditions reads: .. math:: - \nabla^{2} u &= f \quad {\rm in} \ \Omega, \\ \nabla u \cdot n &= g \quad {\rm on} \ \partial \Omega. Here, :math:`f` and :math:`g` are input data and :math:`n` denotes the outward directed boundary normal. Since only Neumann conditions are applied, :math:`u` is only determined up to a constant :math:`c` by the above equations. An additional constraint is thus required, for instance: .. math:: \int_{\Omega} u \, {\rm d} x = 0. This can be accomplished by introducing the constant :math:`c` as an additional unknown (to be sought in :math:`\mathbb{R}`) and the above constraint expressed via a Lagrange multiplier. We further note that a necessary condition for the existence of a solution to the Neumann problem is that the right-hand side :math:`f` satisfies .. math:: \int_{\Omega} f \, {\rm d} x = - \int_{\partial\Omega} g \, {\rm d} s. This can be seen by multiplying by :math:`1` and integrating by parts: .. math:: \int_{\Omega} f \, {\rm d} x = - \int_{\Omega} 1 \cdot \nabla^{2} u \, {\rm d} x = - \int_{\partial\Omega} 1 \cdot \partial_n u \, {\rm d} s + \int_{\Omega} \nabla 1 \cdot \nabla u \, {\rm d} x = - \int_{\partial\Omega} g \, {\rm d} s. This condition is not satisfied by the specific right-hand side chosen for this test problem, which means that the partial differential equation is not well-posed. However, the variational problem expressed below is well-posed as the Lagrange multiplier introduced to satisfy the condition :math:`\int_{\Omega} u \, {\rm d} x = 0` *effectively redefines the right-hand side such that it safisfies the necessary condition* :math:`\int_{\Omega} f \, {\rm d} x = - \int_{\partial\Omega} g \, {\rm d} s`. Our variational form reads: Find :math:`(u, c) \in V \times R` such that .. math:: a((u, c), (v, d)) = L((v, d)) \quad \forall \ (v, d) \in V \times R, .. math:: a((u, c), (v, d)) &= \int_{\Omega} \nabla u \cdot \nabla v \, {\rm d} x + \int_{\Omega} cv \, {\rm d} x + \int_{\Omega} ud \, {\rm d} x, \\ L(v) &= \int_{\Omega} f v \, {\rm d} x + \int_{\partial\Omega} g v \, {\rm d} s. :math:`V` is a suitable function space containing :math:`u` and :math:`v`, and :math:`R` is the space of real numbers. The expression :math:`a(\cdot, \cdot)` is the bilinear form and :math:`L(\cdot)` is the linear form. Note that the above variational problem may alternatively be expressed in terms of the modified (and consistent) right-hand side :math:`\tilde{f} = f - c`. In this demo we shall consider the following definitions of the domain and input functions: * :math:`\Omega = [0, 1] \times [0, 1]` (a unit square) * :math:`g = - \sin(5x)` (normal derivative) * :math:`f = 10\exp(-((x - 0.5)^2 + (y - 0.5)^2) / 0.02)` (source term)