POTW Minimizers of an Energy Functional

Click For Summary
In the discussion, it is established that a function u, which is continuous and satisfies the Laplace equation Δu = φ in a bounded open subset U of R^n, is a minimizer of the energy functional E(u). The energy functional is defined as the integral of a specific expression involving the gradient of u and the function φ. The boundary condition u|∂U = 0 is crucial for this minimization. The proof relies on variational methods and properties of Sobolev spaces. This result highlights the relationship between solutions to elliptic equations and minimization problems in calculus of variations.
Euge
Gold Member
MHB
POTW Director
Messages
2,072
Reaction score
245
Let ##U## be a bounded open subset of ##\mathbb{R}^n##. Given a continuous function ##\phi : \overline{U} \to \mathbb{R}##, show that any real-valued function ##u## of class ##C^2(\overline{U})## such that ##\Delta u = \phi## in ##U## and ##u|_{\partial U} = 0## is a minimizer of the energy functional $$\mathscr{E}(u) = \int_U d^n x\, \left(\frac{1}{2}|\nabla u(x)|^2 + \phi(x) u(x)\right)$$ over the class of functions ##\Sigma = \{v\in C^2(\overline{U}) : v|_{\partial U} = 0\}##.
 
  • Like
Likes jbergman and topsquark
Physics news on Phys.org
First, as

$$
\Delta u(x) = \phi (x)
$$

we have

$$
0 = \int_U d^nx \left( - \Delta u(x) + \phi (x) \right) \eta (x)
$$

for an arbitrary function ##\eta (x)##. If we take ##\eta \in \Sigma##, integrating by parts gives

$$
0 = \int_U d^nx \left( \nabla u(x) \cdot \nabla \eta (x)+ \phi (x) \eta (x) \right)
$$

Take an arbitrary ##v \in \Sigma##. We can write down an obvious inequality

$$
0 \leq \left( \| \nabla u \| - \| \nabla v \| \right)^2
$$

where ##\| \nabla u \| := \left( \sum_{i=1}^n (\partial_i u)^2 \right)^{1/2}## (##\| \nabla u \|^2 \equiv | \nabla u |^2##). Which implies

$$
0 \leq \frac{1}{2} | \nabla u |^2 + \frac{1}{2} | \nabla v |^2 - \| \nabla u \| \| \nabla v \|
$$

or

$$
\| \nabla u \| \| \nabla v \| \leq \frac{1}{2} | \nabla u |^2 + \frac{1}{2} | \nabla v |^2 .
$$

As ##\nabla u \cdot \nabla v \leq | \nabla u \cdot \nabla v | \leq \| \nabla u \| \| \nabla v \|## (Cauchy-Schwarz) we have

$$
\nabla u \cdot \nabla v \leq \frac{1}{2} | \nabla u |^2 + \frac{1}{2} | \nabla v |^2 .
$$

From which we have

$$
\int_U d^nx \left( \nabla u \cdot \nabla v + \phi v - \frac{1}{2} | \nabla u |^2 \right) \leq \int_U d^nx \left( \frac{1}{2} | \nabla v |^2 + \phi v \right)
$$

Using ##\int_U d^nx \left( \nabla u \cdot \nabla \eta+ \phi \eta \right) = 0## in this integral gives

\begin{align*}
\int_U d^nx \left( \nabla u \cdot \nabla [\eta + v] + \phi [\eta + v] - \frac{1}{2} | \nabla u |^2 \right)
\leq \int_U d^nx \left( \frac{1}{2} | \nabla v |^2 + \phi v \right)
\end{align*}

Putting ##\eta (x) = u(x) - v(x)## we have,

$$
\int_U d^nx \left( \frac{1}{2} | \nabla u (x) |^2 + \phi (x) u (x) \right) \leq \int_U d^nx \left( \frac{1}{2} | \nabla v (x) |^2 + \phi (x) v (x) \right)
$$

which is the desired result. Obviously, this choice of ##\eta (x)## means that ##\eta \in \Sigma##.
 
Last edited:
  • Like
Likes Office_Shredder and jbergman

Similar threads

Replies
3
Views
2K
Replies
2
Views
2K
Replies
1
Views
1K
Replies
1
Views
2K
Replies
4
Views
2K
Replies
2
Views
280
Replies
1
Views
143
Replies
8
Views
465