2

This is an assignment question which I've been working on to solve the inhomogeneous wave equation $u_{tt} - c^{2}u_{xx} = f(x,t)$.

I separated the equation out into a system of two equations: $u_{t} + cu_{x} = v$ and $v_{t} - cv_{x} = f(x,t)$.

It says to solve the first differential equation to find that $u(x, t) = \int\limits_{0}^{t} v(x-ct+cs,s) ds$.

My idea is to use the linearity of the differential equation $u_{t} - cu_{x}$ to get that the solution to the homogenous equation is $f(x-ct)$ and then guessing that another solution is $g(x+ct)$ and using this to derive the solution given but didn't have any luck.

What's the general process to solving these kind of differential equations? I'm mostly attempting random ideas to solve it.

Nick R
  • 319

2 Answers2

0

Similar to how to solve this PDE:

Let $\begin{cases}p=x+ct\\q=x-ct\end{cases}$ ,

Then $\dfrac{\partial u}{\partial x}=\dfrac{\partial u}{\partial p}\dfrac{\partial p}{\partial x}+\dfrac{\partial u}{\partial q}\dfrac{\partial q}{\partial x}=\dfrac{\partial u}{\partial p}+\dfrac{\partial u}{\partial q}$

$\dfrac{\partial^2u}{\partial x^2}=\dfrac{\partial}{\partial x}\left(\dfrac{\partial u}{\partial p}+\dfrac{\partial u}{\partial q}\right)=\dfrac{\partial}{\partial p}\left(\dfrac{\partial u}{\partial p}+\dfrac{\partial u}{\partial q}\right)\dfrac{\partial p}{\partial x}+\dfrac{\partial}{\partial q}\left(\dfrac{\partial u}{\partial p}+\dfrac{\partial u}{\partial q}\right)\dfrac{\partial q}{\partial x}=\dfrac{\partial^2u}{\partial p^2}+\dfrac{\partial^2u}{\partial pq}+\dfrac{\partial^2u}{\partial pq}+\dfrac{\partial^2u}{\partial q^2}=\dfrac{\partial^2u}{\partial p^2}+2\dfrac{\partial^2u}{\partial pq}+\dfrac{\partial^2u}{\partial q^2}$

$\dfrac{\partial u}{\partial t}=\dfrac{\partial u}{\partial p}\dfrac{\partial p}{\partial t}+\dfrac{\partial u}{\partial q}\dfrac{\partial q}{\partial t}=c\dfrac{\partial u}{\partial p}-c\dfrac{\partial u}{\partial q}$

$\dfrac{\partial^2u}{\partial t^2}=\dfrac{\partial}{\partial t}\left(c\dfrac{\partial u}{\partial p}-c\dfrac{\partial u}{\partial q}\right)=\dfrac{\partial}{\partial p}\left(c\dfrac{\partial u}{\partial p}-c\dfrac{\partial u}{\partial q}\right)\dfrac{\partial p}{\partial t}+\dfrac{\partial}{\partial q}\left(c\dfrac{\partial u}{\partial p}-c\dfrac{\partial u}{\partial q}\right)\dfrac{\partial q}{\partial t}=c^2\dfrac{\partial^2u}{\partial p^2}-c^2\dfrac{\partial^2u}{\partial pq}-c^2\dfrac{\partial^2u}{\partial pq}+c^2\dfrac{\partial^2u}{\partial q^2}=c^2\dfrac{\partial^2u}{\partial p^2}-2c^2\dfrac{\partial^2u}{\partial pq}+c^2\dfrac{\partial^2u}{\partial q^2}$

$\therefore c^2\dfrac{\partial^2u}{\partial p^2}-2c^2\dfrac{\partial^2u}{\partial pq}+c^2\dfrac{\partial^2u}{\partial q^2}-c^2\dfrac{\partial^2u}{\partial p^2}-2c^2\dfrac{\partial^2u}{\partial pq}-c^2\dfrac{\partial^2u}{\partial q^2}=f\left(\dfrac{p+q}{2},\dfrac{p-q}{2c}\right)$

$-4c^2\dfrac{\partial^2u}{\partial pq}=f\left(\dfrac{p+q}{2},\dfrac{p-q}{2c}\right)$

$\dfrac{\partial^2u}{\partial pq}=-\dfrac{1}{4c^2}f\left(\dfrac{p+q}{2},\dfrac{p-q}{2c}\right)$

$u(p,q)=F(p)+G(q)-\dfrac{1}{4c^2}\int_b^q\int_a^pf\left(\dfrac{r+s}{2},\dfrac{r-s}{2c}\right)dr~ds$

$u(x,t)=F(x+ct)+G(x-ct)-\dfrac{1}{4c^2}\int_b^{x-ct}\int_a^{x+ct}f\left(\dfrac{r+s}{2},\dfrac{r-s}{2c}\right)dr~ds$

doraemonpaul
  • 16,178
  • 3
  • 31
  • 75
0

The one dimensional wave equation is easy because as you did you can factor it into (setting $c=1$) $(\partial_t-\partial_x)(u_t+u_x)=f$, or calling $v\equiv u_t+u_x$ we have $v_t-v_x=f$, or $\frac{dv(s,x+t-s)}{ds}=f(s,x+t-s)$, so that integrating we get $v(t,x)-v(0,x+t)=\int_0^t f(s',x+t-s')\,ds'$. Now as for the $u_t+u_x=v$ equation, this is the same as $\frac{du(s,x-t+s)}{ds}=v(s,x-t+s)$ and integrating gives $u(t,x)-u(0,x-t)=\int_0^tv(s',x-t+s')\,ds$. Then you can throw them together.

In response to the comment:

Well $u_t+u_x$ is just the derivative of $u$ along the line $x-t=$ const., or in other words it is just $\nabla u\cdot(1,1)$ if you prefer. So we have $v(t_0+s,x_0+s)=u_t(t_0+s,x_0+s)+u_x(t_0+s,x_0+s)=\frac{du(t_0+s,x_0+s)}{ds}\,,$ and presumably you are given an initial condition $u(0,x)=g(x)$, and since we know that $v(s,x_0+s)=\frac{du(s,x_0+s)}{ds}$ then to find $u(x,t)$ we want to integrate $s$ from $0$ to $t$ to get $u(t,x_0+t)-u(0,x_0)=\int_o^t v(s,x_0+s)\,ds\,.$ Now this is ok but since we are free to choose $x_0$ it is simplest if we choose $x_0=x-t$ because then we get $u(t,x)-u(0,x-t)=\int_o^t v(s,x-t+s)\,ds$.

JLA
  • 6,442
  • 2
  • 24
  • 40
  • Where do you get from $u_{t} + u_{x} = v$ to $\frac{du(s, x-t+s)} {ds} = v(s, x-t+s)$. I can verify that statement, but it seems pulled out of nowhere. – Nick R Oct 11 '14 at 02:18
  • @Nick I edited the post, hopefully that answers your question. – JLA Oct 11 '14 at 02:49
  • Makes a ton of sense, great thanks. I completely forgot about trying to use the method of characteristics on the differential equation. However, one thing is that when you try to combine the integrals and solve for v you just get a double integral over time of f(x,s) ds dt which isn't the characteristic triangle. Do the initial conditions somehow need to be factored? Or the $\frac{1}{2c}$ term...doesn't seem quite right. – Nick R Oct 11 '14 at 10:58
  • Sorry to double comment, couldn't figure out how to edit. But when you combine the integrals you get that from your derivation $v(s, x-t+s) = \int\limits_{0}^{s} f(x, s') ds$ which isn't of the right form to be integrating over the characteristic triangle. – Nick R Oct 11 '14 at 11:15
  • @Nick Are you sure that's what you get? Check your variables by the way, they don't make sense. – JLA Oct 12 '14 at 03:33
  • Sorry, made a typo when I filled it out. Meant that you get that $v(s, x-t+s) = \int\limits_{0}^{s}f(x, s') ds'$ which when you plug into the equation for $u$ yields $u(x,t) = \int\limits_{0}^{t} \int\limits_{0}^{s} f(x,s) ds dt$ which isn't correct. – Nick R Oct 12 '14 at 17:25
  • @Nick Yeah that's what I figured, but I did a quick calculation and that's not what I got for $v(s,x-t+s),$ but I may be wrong. – JLA Oct 12 '14 at 18:07
  • Ah I see it, canceled out the variable that was being integrating over when you can't. Thanks for all of the help, really appreciate it. The derivation with the characteristics was spot on. – Nick R Oct 12 '14 at 18:14