solving heat equation under Neumann boundary conditions

$\begingroup$

I am having trouble solving following equation \begin{align*} u_t - u_{xx} = 0 & \; ;0<x<L, t>0 \\ u(0, x)=x &\; ; 0<x<L\\ u_x(t, 0) = u_x(t, L) = 0&\; ; t >0 \end{align*}

I got the solution of the form $\displaystyle u(x, t) = \sum_{n=1}^\infty A_n e^{-\frac{-n^2 \pi ^2}{L^2}t} \cos ( \frac{2 n \pi}{L}x) $ but Fourier expansion of $u(0, x) = x$ has no cosine terms. Is the boundary condition properly formulated am I making mistake somethere?

$\endgroup$

1 Answer

$\begingroup$

If I understand correctly, the solution should be

$$\displaystyle u(x,t) = A_0+\sum_{n=1}^\infty A_n\; exp\left({\lambda_n \;t}\right) \cos ( \frac{ n \pi}{L}x)$$

with $$\lambda_n=-\frac{-n^2 \pi ^2}{L^2}$$

for $$u_t(x,t) - u_{xx}(x,t) = 0 \qquad \; ;0<x<L,\; t>0$$

and boundaries/initials (please check against your formulation):

\begin{align*} u(x,0)=\phi(x) &\qquad \; ; 0<x<L\\ u_x(0,t) =0,\;\; u_x(L,t) = 0&\qquad \; ; t >0 \end{align*}

You need to pick $A_n$ such that $\phi(x)=u(x,0)$

To do this we consider what we learned from Fourier series. In particular we look for $u$ as an infinite sum

$$\displaystyle u(x,t) = A_0+\sum_{n=1}^\infty A_n\; exp\left({\lambda_n \;t}\right) \cos ( \frac{ n \pi}{L}x)$$

and find $\{A_n\}$ while satisfying:

$$\phi(x)=u(x,0)=A_0+\sum_{n=1}^\infty A_n\; \cos ( \frac{ n \pi}{L}x)$$

This requires a $cos$ expansion of $\phi(x)$ in the interval $[0,L]$, then we gain for your case $\phi(x)=x$:

$$A_0=(1/L) \int_0^L \phi(x)dx= (1/L) \int_0^L x dx$$

and

$$A_n=(2/L)\int_0^L \phi(x)\cos ( \frac{ n \pi}{L}x)dx=(2/L)\int_0^L x\cos ( \frac{ n \pi}{L}x)dx$$

$\endgroup$ 2

Your Answer

Sign up or log in

Sign up using Google Sign up using Facebook Sign up using Email and Password

Post as a guest

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

You Might Also Like