Crank-Nicolson: Proving Max Bound of $|T_i^{n+\frac{1}{2}}|$

  • MHB
  • Thread starter evinda
  • Start date
  • Tags
    Error
In summary, the Crank-Nicolson method has an upper bound of $|T_i^{n+\frac{1}{2}}| \leq \frac{\tau^2}{3}M_{ttt}+\frac{h^2}{12}M_{xxxx}$, which demonstrates its stability and convergence for small time and spatial steps.
  • #1
evinda
Gold Member
MHB
3,836
0
Hello! (Wave)

I want to show that for the Crank-Nicolson method, the following holds:

$|T_i^{n+\frac{1}{2}}| \leq \frac{\tau^2}{12} M_{ttt}+\frac{h^2}{12}M_{xxxx}$,

where $M_{ttt}=||u_{ttt}||_{\infty}, M_{xxxx}=||u_{xxxx}||_{\infty}$

and $T_{i}^{n+\frac{1}{2}}=\frac{u(t_{n},x_i)-u(t_{n-1},x_i)}{\tau}-\frac{1}{2} \frac{u(t_{n+1},x_{i+1})-2u(t_{n+1},x_i)+u(t_{n+1},x_{i-1})}{h^2}-\frac{1}{2} \frac{u(t_n,x_{i+1})-2u(t_n,x_i)+u(t_n,x_{i-1})}{h^2}$

I found the following using Taylor expansions:

  • $\frac{u(t_n,x_i)-u(t_n-\tau,x_i)}{\tau}=u_t(t_n,x_i)-\frac{\tau}{2}u_{tt}(t_n,x_i)+\frac{\tau^2}{6}u_{ttt}(\rho_n,x_i)$ $$$$
  • $-\frac{1}{2} \left[ \frac{u(t_n+\tau,x_i+h)-2u(t_n+\tau,x_i)+u(t_n+\tau,x_i-h)}{h^2}\right]=-\frac{1}{2}u_{xx}(t_n+\tau,x_i)-\frac{h^2}{24}u_{xxxx}(t_n+\tau, \zeta_i)$ $$$$
  • $-\frac{1}{2} \left[ \frac{u(t_n,x_i+h)-2u(t_n,x_i)+u(t_n,x_i-h)}{h^2} \right]=-\frac{1}{2} u_{xx}(t_n,x_i)-\frac{h^2}{24}u_{xxxx}(t_n, a_i)$
Thus, $T_i^{n+\frac{1}{2}}=u_t(t_n,x_i)-\frac{\tau}{2}u_{tt}(t_n,x_i)+\frac{\tau^2}{6}u_{ttt}(\rho_n,x_i)-\frac{1}{2}u_t(t_n+\tau,x_i)-\frac{h^2}{24}u_{xxxx}(t_n+\tau, \xi_i)-\frac{1}{2}u_t(t_n,x_i)-\frac{h^2}{24}u_{xxxx}(t_n,a_i)$$u_t(t_n+\tau,x_i)=u_t(t_n,x_i)+ \tau u_{tt}(t_n,x_i)+\frac{\tau^2}{2}u_{ttt}(\mu_i,x_i)$

$T_i^{n+\frac{1}{2}}=-\frac{\tau}{2}u_{tt}(t_n,x_i)+\frac{\tau^2}{6}u_{ttt} (\rho_n, x_i)-\frac{\tau}{2}u_{tt}(t_n,x_i)-\frac{\tau^2}{4}u_{ttt}(\mu_i, x_i)-\frac{h^2}{24}(u_{xxxx}(t_n+\tau, \xi_i)+u_{xxxx}(t_n+\zeta_i))$

How do we get the wanted upper bound?
 
Mathematics news on Phys.org
  • #2


To get the desired upper bound, we can use the triangle inequality to combine the terms in $T_i^{n+\frac{1}{2}}$ that contain $u_{ttt}$ and $u_{xxxx}$. This will give us:

$T_i^{n+\frac{1}{2}} \leq \frac{\tau^2}{6}|u_{ttt}(\rho_n,x_i)|+\frac{\tau^2}{4}|u_{ttt}(\mu_i,x_i)|+\frac{h^2}{24}|u_{xxxx}(t_n+\xi_i)|+\frac{h^2}{24}|u_{xxxx}(t_n+\zeta_i)|$

Since $|u_{ttt}|$ and $|u_{xxxx}|$ are both bounded by their respective infinity norms, we can further simplify the above expression to:

$T_i^{n+\frac{1}{2}} \leq \frac{\tau^2}{6}M_{ttt}+\frac{\tau^2}{4}M_{ttt}+\frac{h^2}{24}M_{xxxx}+\frac{h^2}{24}M_{xxxx}$

Combining the like terms, we get:

$T_i^{n+\frac{1}{2}} \leq \frac{\tau^2}{3}M_{ttt}+\frac{h^2}{12}M_{xxxx}$

Since this holds for all $i$ and $n$, we can take the absolute value of both sides to get:

$|T_i^{n+\frac{1}{2}}| \leq \frac{\tau^2}{3}M_{ttt}+\frac{h^2}{12}M_{xxxx}$

Which is the desired upper bound. This shows that the Crank-Nicolson method is stable and converges for sufficiently small time steps ($\tau$) and spatial steps ($h$).
 

FAQ: Crank-Nicolson: Proving Max Bound of $|T_i^{n+\frac{1}{2}}|$

What is the Crank-Nicolson method?

The Crank-Nicolson method is a numerical technique used for solving partial differential equations. It is a combination of the implicit and explicit methods, making it more accurate and stable than either method alone.

How does the Crank-Nicolson method work?

The Crank-Nicolson method uses a weighted average of the terms at two adjacent time steps to approximate the solution at the midpoint between them. This results in a second-order accurate method that is unconditionally stable.

What is the significance of proving the maximum bound of $|T_i^{n+\frac{1}{2}}|$ in Crank-Nicolson?

The maximum bound of $|T_i^{n+\frac{1}{2}}|$ is important because it determines the accuracy and stability of the Crank-Nicolson method. By proving this bound, we can ensure that the method will provide reliable solutions to the partial differential equation being solved.

How is the maximum bound of $|T_i^{n+\frac{1}{2}}|$ determined in Crank-Nicolson?

The maximum bound of $|T_i^{n+\frac{1}{2}}|$ is typically determined through a mathematical analysis of the method. This involves examining the truncation error and stability of the method, and using this information to derive a bound for the values of $|T_i^{n+\frac{1}{2}}|$.

Can the maximum bound of $|T_i^{n+\frac{1}{2}}|$ vary for different problems when using Crank-Nicolson?

Yes, the maximum bound of $|T_i^{n+\frac{1}{2}}|$ can vary depending on the specific partial differential equation being solved. It is important to analyze and prove the bound for each problem separately to ensure the accuracy and stability of the Crank-Nicolson method.

Similar threads

Back
Top