- #1
linda300
- 61
- 3
hello,
Let T be a open, connected and bounded subset of ℝ3 which has a smooth boundary bd(T).
Consider the equation Δu = -λu with either the Dirichlet condition (u=0 in bd(T)) or Neumann (where δu/δn = 0 on bd(T)).
Define:
[itex] \left\langle g,h \right\rangle =\iiint_{T}{g(x)h^{*}(x)}dx[/itex]
where * means conjugate
Show that for a Dirichlet problem the eigenvalues are positive and that for a Neumann problem they are not negative.
So the first greens identity is
[itex] \iint_{bd(T)}{v\frac{\partial u}{\partial n}}dS=\iiint_{T}{\nabla v\cdot \nabla u\,d\vec{x}}+\iiint_{T}{v\Delta u\,d\vec{x}} [/itex]
using this with v=u,
for a Dirichlet problem,
u=0 on the boundary so
[itex] 0=\iiint_{T}{\nabla u\cdot \nabla u\,d\vec{x}}+\iiint_{T}{u\Delta u\,d\vec{x}} [/itex]
then
[itex] \iiint_{T}{\nabla u\cdot \nabla u\,d\vec{x}}=\lambda \iiint_{T}{{{u}^{2}}\,d\vec{x}} [/itex]
is that correct?
and its the same result for the Neumann problem except the du/dn term is zero on the boundary so i can't really see how to get that the eigenvalues are positive, or non negative,
is this not the correct approach?
Let T be a open, connected and bounded subset of ℝ3 which has a smooth boundary bd(T).
Consider the equation Δu = -λu with either the Dirichlet condition (u=0 in bd(T)) or Neumann (where δu/δn = 0 on bd(T)).
Define:
[itex] \left\langle g,h \right\rangle =\iiint_{T}{g(x)h^{*}(x)}dx[/itex]
where * means conjugate
Show that for a Dirichlet problem the eigenvalues are positive and that for a Neumann problem they are not negative.
So the first greens identity is
[itex] \iint_{bd(T)}{v\frac{\partial u}{\partial n}}dS=\iiint_{T}{\nabla v\cdot \nabla u\,d\vec{x}}+\iiint_{T}{v\Delta u\,d\vec{x}} [/itex]
using this with v=u,
for a Dirichlet problem,
u=0 on the boundary so
[itex] 0=\iiint_{T}{\nabla u\cdot \nabla u\,d\vec{x}}+\iiint_{T}{u\Delta u\,d\vec{x}} [/itex]
then
[itex] \iiint_{T}{\nabla u\cdot \nabla u\,d\vec{x}}=\lambda \iiint_{T}{{{u}^{2}}\,d\vec{x}} [/itex]
is that correct?
and its the same result for the Neumann problem except the du/dn term is zero on the boundary so i can't really see how to get that the eigenvalues are positive, or non negative,
is this not the correct approach?