Computational Science Stack Exchange is a question and answer site for scientists using computers to solve scientific problems. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Suppose I have the parabolic system $$u_t=\nabla\cdot(k(x)\nabla u)+f,\quad (x,t)\in\Omega\times I$$ with Dirichlet boundary conditions $$u=g, \quad x\in\partial\Omega$$ and initial condition $$u(x,t)= h,\quad t=0.$$

Often times in engineering, we are more interested in the asymptotic (steady state) behavior of this PDE rather than the transient behavior. So, we sometimes neglect the time derivative term and solve the elliptic system $$-\nabla\cdot(k(x)\nabla u)=f,\quad (x,t)\in\Omega\times I$$ instead. The assumption is that over infinite time, $$\lim_{t\rightarrow \infty}u_{\mathrm{parabolic}}(x,t)= u_{\mathrm{elliptic}}(x,t)$$.

I've observed that when $f\equiv 0$, this limit is true, but I'm not sure if this is the case for arbitrary $f$ or if there are other necessary conditions to guarantee this limit to be true. Do the boundary conditions have to converge asymptotically to a constant value in order for the parabolic solution to converge to the elliptic solution?

Though I'm my question is formulated in the continuous case, I'm also curious if the same conditions are true for the discrete case. That is, assuming I use a stable and consistent finite difference scheme to approximate $u_{\mathrm{parabolic}}$ and $u_{\mathrm{elliptic}}$, should I expect $$\lim_{t\rightarrow \infty}u_{\mathrm{parabolic}}^\mathrm{fdm}(x,t)= u_{\mathrm{elliptic}}^\mathrm{fdm}(x,t)$$ if $u_{\mathrm{elliptic}}^\mathrm{fdm}$ and $u_{\mathrm{parabolic}}^\mathrm{fdm}$ are discretized on the same spatial grid and $\Delta t\rightarrow\infty$?

share|improve this question
up vote 3 down vote accepted

Yes, with Dirichlet boundary conditions you always have exponential convergence to the staedy state. Any PDE book will have a proof. For a nice explanation from a numerical perspective, see chapter 2 of LeVeque's FDM book.

share|improve this answer
The same should be true if there were mixed conditions (neumann and dirichlet), right? – Paul Jan 24 '13 at 15:51
@Paul Yes, the only sticky case in 1D is when both boundaries are Neumann. – David Ketcheson Jan 27 '13 at 8:37

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.