Solving a Boundary Value Problem: Non-Uniform vs. Uniform Partitioning

Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
1 reply · 2K views
evinda
Gold Member
MHB
Messages
3,741
Reaction score
0
Hello! (Wave)Consider the boundary value problem
$\left\{\begin{matrix}
- \epsilon u''+u'=1 &, x \in [0,1] \\
u(0)=u(1)=0 &
\end{matrix}\right.$
where $\epsilon$ is a positive given constant.
I have to express a finite difference method for its numerical solution.
How can we know whether it is better to use non-uniform partition or uniform partition?
 
Physics news on Phys.org
Well, you can always do the experiment and see. This BVP has an exact answer that's not too difficult to obtain. If you do the substitution $v=u'$, you get a first-order linear DE, $v'-v/\varepsilon=-1/\varepsilon$, with solution
$$v=1+C_1 e^{x/\varepsilon}.$$
Integrating once yields
$$u=x+\varepsilon C_1 e^{x/\varepsilon}+C_2.$$
Applying the boundary conditions yields the system
\begin{align*}
\varepsilon C_1+C_2&=0 \\
\varepsilon C_1 e^{1/\varepsilon}+C_2&=-1,
\end{align*}
with solution
\begin{align*}
C_1&=\frac{1}{\varepsilon(1-e^{1/\varepsilon})} \\
C_2&=\frac{1}{e^{1/\varepsilon}-1}.
\end{align*}
Hence, the exact solution is
$$u(x)=x+\frac{e^{x/\varepsilon}-1}{1-e^{1/\varepsilon}}.$$
You can use this to compare how good the two numerical solutions are.