Q: Solve the following non-homogeneous problem: \begin{align*} \frac{\partial u}{\partial t} &= k \frac{\partial^2 u}{\partial x^2} + e^{-t} + e^{-2t} \cos \frac{3\pi x}{L} \\ \end{align*} With the following boundary and initial value conditions: \begin{align*} \frac{\partial u}{\partial x}(0,t) &= 0 \\ \frac{\partial u}{\partial x}(L,t) &= 0 \\ u(x,0) &= f(x) \\ \end{align*} Assume that $2 \neq k(3\pi/L)^2$. Use the method of eigenfunction expansion. Look for the solution as a Fourier cosine series. Assume appropriate continuity.

TEXTBOOK GIVEN ANSWER: \begin{align*} u &= \sum\limits_{n=0}^\infty A_n(t) \cos \frac{n \pi x}{L} \\ A_{n,n \neq 0, n\neq 3}(t) &= A_n(0) e^{-k \left( \frac{n \pi}{L} \right)^2 t} \\ A_0(t) &= A_0(0) + 1 - e^{-t} \\ A_3(t) &= A_3(0) e^{-k \left( \frac{3 \pi}{L} \right)^2 t} + \frac{e^{2t} - e^{-k \left( \frac{3 \pi}{L} \right)^2 t}}{k \left( \frac{3 \pi}{L} \right)^2 - 2} \\ A_0(0) &= \frac{1}{L} \int_0^L f(x) \, dx \\ A_{n \ge 1}(0) &= \frac{2}{L} \int_0^L f(x) \cos \frac{n \pi x}{L} \, dx \\ \end{align*}

My work: I fully understand how to get most of the answer, but I am stuck on the rest. Specifically, I don't understand the $n=0, n=3$ cases.

First, the part I fully understand:

We look for the solution as a Fourier cosine series. The right hand side is an even extended and periodized version of $u(x,t)$ along the $x$ variable.

\begin{align*} u(x,t) &\sim A_0 + \sum\limits_{n=1}^\infty A_n(t) \cos \frac{n \pi x}{L} \\ \end{align*}

The problem says we can assume appropriate continuity which means that we can assume the original, non-periodized $u(x,t)$ is continuous. We can apply term by term partial differentiation to both $x$ and $t$. For $t$, we are not periodized by $t$, the given function is continuous, so that's all we need. For $x$, we are dealing with the even extended, periodized version of a continuous function, which must be continuous, so we can use term by term partial differentiation there as well.

\begin{align*} \frac{\partial u}{\partial x} &\sim - \sum\limits_{n=1}^\infty \frac{n \pi}{L} A_n(t) \sin \frac{n \pi x}{L} \\ \frac{\partial u}{\partial t} &\sim \sum\limits_{n=1}^\infty A'_n(t) \cos \frac{n \pi x}{L} \\ \end{align*}

We also assume that $\frac{\partial u}{\partial x}$ is continuous. Since we are given that $\frac{\partial u}{\partial x}(0,t) = 0 = \frac{\partial u}{\partial x}(L,t)$, then we can see that the periodized version along the $x$ variable must also be fully continuous so that we can again use term by term differentiation to get:

\begin{align*} \frac{\partial^2 u}{\partial x^2} &\sim - \sum\limits_{n=1}^\infty \left(\frac{n \pi}{L}\right)^2 A_n(t) \cos \frac{n \pi x}{L} \\ \end{align*}

Now plugging that back in to the PDE:

\begin{align*} \frac{\partial u}{\partial t} &= k \frac{\partial^2 u}{\partial x^2} + e^{-t} + e^{-2t} \cos \frac{3\pi x}{L} \\ \sum\limits_{n=1}^\infty A'_n(t) \cos \frac{n \pi x}{L} &= k \left( - \sum\limits_{n=1}^\infty \left(\frac{n \pi}{L}\right)^2 A_n(t) \cos \frac{n \pi x}{L} \right) + e^{-t} + e^{-2t} \cos \frac{3\pi x}{L} \\ \end{align*}

That will be satisfied if the summation terms on the left/right side are all equal and the term on the right hand side outside the summation is zero:

\begin{align*} A'_n(t) &= -k \left(\frac{n \pi}{L}\right)^2 A_n(t) \\ A_n(t) &= A_n(0) e^{-k \left( \frac{n \pi}{L} \right)^2 t} \\ \end{align*}

Solving for the coefficients with the initial condition:

\begin{align*} u(x,t) &= \sum\limits_{n=0}^\infty A_n(t) \cos \frac{n \pi x}{L} \\ u(x,0) = f(x) &= \sum\limits_{n=0}^\infty A_n(0) \cos \frac{n \pi x}{L} \\ A_0(0) &= \frac{1}{L} \int_0^L f(x) \, dx \\ A_{n \ge 1}(0) &= \frac{2}{L} \int_0^L f(x) \cos \frac{n \pi x}{L} \, dx \\ \end{align*}

ok, so far so good. All of that lines up perfectly with the given answer. But I am unsure about how to get the rest.

From earlier, I said if the terms on the right hand side out side the summation came to zero, it would satisfy the PDE. That would mean that:

\begin{align*} e^{-t} + e^{-2t} \cos \frac{3\pi x}{L} &= 0 \\ e^{t} &= -\cos \frac{3\pi x}{L} \\ \end{align*}

I am not sure how to use this. I also don't see what to do with the problem given that $2 \neq k(3\pi/L)^2$



$$u \sim \sum_{n \ge 0} A_{n}(t) \cos \left( \frac{n \pi x}{L} \right)$$

then substituting into the PDE (and noting that $1 \equiv \cos(0 \pi x/L)$) yields

$$\sum_{n \ge 0} A_{n}'(t) \cos \left( \frac{n \pi x}{L} \right) = -k \sum_{n \ge 0} \left( \frac{n \pi}{L} \right)^{2} A_{n}(t) \cos \left( \frac{n \pi x}{L} \right) + e^{-t} \color{red}{\cos \left( \frac{0 \pi x}{L} \right)} + e^{-2t}\cos \left( \frac{3 \pi x}{L} \right)$$

Equating terms in $\cos$ for $n = 0, 1, \dots$ and letting $C_{n}$ be the constants of integration, we have

\begin{align} A_{0}'(t) = -k(0)+e^{-t} &\implies A_{0}(t) = -e^{-t} + C_{0} \\ &\implies A_{0}(0) = -1+C_{0} \\ &\implies A_{0}(t) = A_{0}(0) + 1 - e^{-t} \\\\ A_{1}'(t) = -k \left( \frac{\pi}{L} \right)^{2} A_{1}(t) &\implies A_{1}(t) = A_{1}(0) e^{-k \left( \frac{\pi}{L} \right)^{2} t} \\\\ A_{2}'(t) = -k \left( \frac{2 \pi}{L} \right)^{2} A_{2}(t) &\implies A_{2}(t) = A_{2}(0) e^{-k \left( \frac{2 \pi}{L} \right)^{2} t} \\\\ A_{3}'(t) = -k \left( \frac{3 \pi}{L} \right)^{2} A_{3}(t) + e^{-2t} &\implies A_{3}(t) = \frac{e^{-2t}}{-2 + k \left( \frac{3 \pi}{L} \right)^{2}} + C_{3} e^{-k \left( \frac{3 \pi}{L} \right)^{2} t} \\ &\implies A_{3}(0) = \frac{1}{-2 + k \left( \frac{3 \pi}{L} \right)^{2}} + C_{3} \\ &\implies A_{3}(t) = \frac{e^{-2t} - e^{-k \left( \frac{3 \pi}{L} \right)^{2} t}}{-2 + k \left( \frac{3 \pi}{L} \right)^{2}} + A_{3}(0) e^{-k \left( \frac{3 \pi}{L} \right)^{2} t} \end{align}

  • $\begingroup$ thank you!!! Interesting that the book says $A_3(t) = A_3(0) e^{-k \left(\frac{3 \pi}{L}\right)^2 t} + \frac{e^{2t} - e^{-k \left(\frac{3 \pi}{L}\right)^2 t}}{-2 + k \left(\frac{3 \pi}{L}\right)^2}$ where we get $A_3(t) = A_3(0) e^{-k \left(\frac{3 \pi}{L}\right)^2 t} + \frac{e^{-2t} - e^{-k \left(\frac{3 \pi}{L}\right)^2 t}}{-2 + k \left(\frac{3 \pi}{L}\right)^2}$. I presume the textbook just made a mistake. $\endgroup$ – clay Oct 14 '19 at 2:40

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.