Can anyone help me with (b)(i)? I've done the first part of it. I've tried putting some boundary conditions in but cannot find $B_n$ What fact should I use?

An infinite slab of material, of constant thermoconductivity $\kappa$, occupies the region of $\mathbb{R}^3$ defined by $0 \leq x \leq a$. Initially the temperature $T(x,t)$ of the slab is given by $$T(x,0) = 0.$$ Subsequently the face at $x = 0$ is kept at zero temperature and the face at $x = a$ is kept at unit temperature.

(a) Write down the boundary conditions satisfied by $T$. Show that $$U(x,t) = T(x,t) - \frac{x}{a},$$ satisfies the heat equation $$\frac{\partial{U}}{\partial{t}} = \kappa\,\frac{\partial^2{U}}{\partial{x^2}},$$ and deduce the boundary conditions on $U$.

(b) (i) Show that, for $t > 0$, $$T(x,t) = \frac{x}{a} + \sum_{n=1}^\infty B_n \exp\left(-\frac{n^2 \pi^2 \kappa t}{a^2}\right) \sin\left(\frac{n\pi x}{a}\right)$$ and give an explicit expression for the constants $B_n$.


Note that the initial condition on $T$ is $T(x,0)=0$. But the equation you are really solving is for $U$, which has initial condition $U(x,0)=-x/a$. Thus, assuming you could derive

$$U(x,t) = \sum_{n=1}^{\infty} B_n e^{-n^2 \pi^2 \kappa t/a^2} \sin{\left ( \frac{n \pi x}{a}\right )}$$


$$U(x,0) = \sum_{n=1}^{\infty} B_n \sin{\left ( \frac{n \pi x}{a}\right )}$$

which means that $u$ is being represented by a Fourier sine series and the $B_n$ are given by corresponding Fourier sine coefficients. These are given by

$$B_n = \frac{2}{a} \int_0^a dx \, \left ( -\frac{x}{a} \right ) \sin{\left ( \frac{n \pi x}{a}\right )} = -\frac{2}{a^2} \int_0^a dx \; x \sin{\left ( \frac{n \pi x}{a}\right )}$$

I'll leave it to you to evaluate the integral.

  • $\begingroup$ That makes sense! I've been taught fourier series first and we just started learning about heat equation. I struggled to see the relationship between those two. Do you mind explaining how they're related? Other than cases like this? $\endgroup$ – user120227 Feb 14 '14 at 21:03
  • $\begingroup$ @user120227: The boundary values provide the eigenvalues and eigenfunctions which become the Fourier series expansion. Here, the zero BC's at either end imply a sine solution. The initial conditions therefore provide the coefficients of the Fourier series as I demonstrated. This of course works when you have finite BC's because you have a second-order equation in $x$. $\endgroup$ – Ron Gordon Feb 14 '14 at 21:32
  • $\begingroup$ THank you. This sounds stupid but why do I put in $U(x,0)$ in the place of where $f(x)$ goes for fourier series? $\endgroup$ – user120227 Feb 14 '14 at 21:49
  • $\begingroup$ @user120227: All I can say is, exactly. You can say that the initial condition is $U(x,0)=f(x)$ if that makes you more comfortable. $\endgroup$ – Ron Gordon Feb 14 '14 at 23:27

Fourier coefficients of the initial data  $U(x,0)=-x/a$  will be the desired constants $$ \begin{gather} B_n=\frac{2}{a}\int\limits_0^a U(x,0)\sin{\biggl(\frac{n{\pi}x}{a}\biggr)}dx= -\frac{2}{a^2}\int\limits_0^a x\sin{\biggl(\frac{n{\pi}x}{a}\biggr)}dx=\\ =\frac{2}{an\pi}\int\limits_0^a x\,d\cos{\biggl(\frac{n{\pi}x}{a}\biggr)}= \frac{2}{an\pi}x\cos{\biggl(\frac{n{\pi}x}{a}\biggr)}\biggr|_0^a-\frac{2}{an\pi} \int\limits_0^a \cos{\biggl(\frac{n{\pi}x}{a}\biggr)}dx=\\ =\frac{2}{n\pi}\cos{(n\pi)}-\frac{2}{n^2{\pi}^2}\sin{\biggl(\frac{n{\pi}x}{a}\biggr)} \biggr|_0^a=\frac{2(-1)^n}{n\pi}. \end{gather} $$

  • $\begingroup$ @Ron Gordon: Thanx Ron. This was a typical TeX-copy-paste slip of the keyboard. If I had noticed that you posted your answer I would not have posted mine. $\endgroup$ – mkl314 Feb 14 '14 at 23:09

Your Answer

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

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