Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. 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

I need help with this exercise:

Given the PDE $$u_t=-10u_{xx}\tag{1},$$ with periodic boundary conditions in $[-1,1]$: $$u(-1,t)=u(1,t), \qquad u_x(-1,t)=u_x(1,t).$$

A) Obtain the solution $u(x,t)$ if $u(x,0)=u_0(x)=\sin(\frac{\pi}{10}x)$

B) If we consider as initial data $u_0(x)=x^2$ defined also in $[-1,1]$, does (1) have a solution in this case?

Thank you very much.

share|cite|improve this question
up vote 3 down vote accepted

First of all, periodic boundary conditions look like a pair of boundary conditions, e.g. like $$u(-1,t)=u(1,t), \quad u_x(-1,t)=u_x(1,t).$$

Start by separating variables, finding the eigenvalues $\lambda_n$, eigenfunctions $X_n(x)$, and temporal solutions $T_n(t)$.

Superimpose those to get something of the form $u(x,t)=\sum_n c_n X_n(x)T_n(t)$.

To obtain the coefficients in the series solution above, evaluate $u(x,t)$ at $t=0$, apply the initial data, and recognize this as an appropriate Fourier series. Use your knowledge of Fourier series and/or orthogonality on $[-1,1]$ to deduce formulas for the coefficients.

Edit based on progress in the comments:

The superimposed solution should look like $$u(x,t)=\sum_{n=1}^\infty [a_n\cos(n\pi x)+b_n\sin(n\pi x)]e^{10(n\pi)^2 t}.$$

Then, the initial condition $u(x,0)=f(x)$, $-1<x<1$, leads to $$u(x,0)=\underbrace{\sum_{n=1}^\infty a_n\cos(n\pi x)+b_n\sin(n\pi x)=f(x)}, \quad -1<x<1,$$ which can be found as standard Fourier coefficients.

share|cite|improve this answer
Thanks, John. I could do A without problems, I think. But, in B, I follow the same method as in A, and I got what it seems a solution too. ¿Should I find any problem with $u_0(x)=x^2$? – Mark_Hoffman Jan 21 '13 at 22:40
(For B, I got $u(x,t)=\sum_{n=1}^{\infty}\frac{4(-1)^n}{(n\pi)^2}cos(n\pi x)*e^{10n^2\pi^2t}$) This verifies the boundary and initial conditions, and I think that also verifies, $u_t=-10u_{xx}$. ¿It's ok, or are there any problems considering $x^2$ as an initial function in the pde? – Mark_Hoffman Jan 21 '13 at 23:07
Check your eigenfunction calculations again. You should get sines and cosines for periodic BCs. – JohnD Jan 21 '13 at 23:57
Sorry I misunderstood. Yes, you are correct for that specific initial data. – JohnD Jan 22 '13 at 0:34
What you have is fine: $b_n=0$ for all $n$, and $a_n=\int_{-1}^1 x^2\cos(n\pi x)\,dx$. There is what happens when you are computing the full Fourier series of an even function (such as $x^2$): the sine coefficients vanish leaving simply a Fourier cosine series! (Similarly, for an odd function, the cosine coefficients would vanish leaving a Fourier sine series. You should be able to prove it with just calculus.) – JohnD Jan 22 '13 at 0:58

I doubt that whether you can really solve it even for A).

Let $u(x,t)=X(x)T(t)$ ,

Then $X(x)T'(t)=-10X''(x)T(t)$



$\begin{cases}T(t)=c_3(n)e^{10n^2\pi^2t}\\X(x)=\begin{cases}c_1(n)\sin n\pi x+c_2(n)\cos n\pi x&\text{when}~n\neq0\\c_1x+c_2&\text{when}~n=0\end{cases}\end{cases}$

$\therefore u(x,t)=\sum\limits_{n=0}^\infty C_1(n)e^{10n^2\pi^2t}\sin n\pi x+\sum\limits_{n=0}^\infty C_2(n)e^{10n^2\pi^2t}\cos n\pi x$

$u(-1,t)=u(1,t)$ :

$\sum\limits_{n=0}^\infty C_1(n)e^{10n^2\pi^2t}\sin(-n\pi)+\sum\limits_{n=0}^\infty C_2(n)e^{10n^2\pi^2t}\cos(-n\pi)=\sum\limits_{n=0}^\infty C_1(n)e^{10n^2\pi^2t}\sin n\pi+\sum\limits_{n=0}^\infty C_2(n)e^{10n^2\pi^2t}\cos n\pi$

$-\sum\limits_{n=0}^\infty C_1(n)e^{10n^2\pi^2t}\sin n\pi+\sum\limits_{n=0}^\infty C_2(n)e^{10n^2\pi^2t}\cos n\pi=\sum\limits_{n=0}^\infty C_1(n)e^{10n^2\pi^2t}\sin n\pi+\sum\limits_{n=0}^\infty C_2(n)e^{10n^2\pi^2t}\cos n\pi$

$2\sum\limits_{n=0}^\infty C_1(n)e^{10n^2\pi^2t}\sin n\pi=0$


$\therefore u(x,t)=\sum\limits_{n=0}^\infty C_2(n)e^{10n^2\pi^2t}\cos n\pi x$

$u_x(x,t)=-\sum\limits_{n=0}^\infty n\pi C_2(n)e^{10n^2\pi^2t}\sin n\pi x$

But $u_x(-1,t)\neq u_x(1,t)$

share|cite|improve this answer

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.