7

I find this problem challenging:

Use the method of characteristics to solve $u_t+u_x^2=t$ with $u(x,0)=0$.

I know I'm supposed to let $p=u_x$ and $q=u_t$. Then I get $F(x,t,u,p,q)=p^2+q-t=0$. But what to do from there eludes me.

Any help/hints to the solution process would be greatly appreciated.

Desperate Fluffy
  • 1,639
  • 2
  • 25
  • 47

2 Answers2

5

$u_t+u_x^2=t$

$u_{tx}+2u_xu_{xx}=0$

Let $v=u_x$ ,

Then $v_t+2vv_x=0$ with $v(x,0)=0$

Follow the method in http://en.wikipedia.org/wiki/Method_of_characteristics#Example:

$\dfrac{dt}{ds}=1$ , letting $t(0)=0$ , we have $t=s$

$\dfrac{dv}{ds}=0$ , letting $v(0)=v_0$ , we have $v=v_0$

$\dfrac{dx}{ds}=2v=2v_0$ , letting $x(0)=f(v_0)$ , we have $x=2v_0s+f(v_0)=2vt+f(v)$ , i.e. $v=F(x-2vt)$

$v(x,0)=0$ :

$F(x)=0$

$\therefore v=0$

$u_x=0$

$u(x,t)=g(t)$

$u_t=g_t(t)$

$\therefore g_t(t)=t$

$g(t)=\dfrac{t^2}{2}+C$

$\therefore u(x,t)=\dfrac{t^2}{2}+C$

$u(x,0)=0$ :

$C=0$

$\therefore u(x,t)=\dfrac{t^2}{2}$

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

Another way (probably better) way is to use the method of characteristic for non-linear first order PDE. We have:

$$ds = \frac{dx}{F_p} = \frac{dt}{F_t} = \frac{du}{pF_p + pF_p} = \frac{-dp}{F_x + pF_u} = \frac{-dq}{F_t + qF_u}$$

where $p = u_x$ and $q= u_y$

For our equation we obtain the following relations:

$$ds = \frac{dx}{2p} = \frac{dt}{1} = \frac{du}{2p^2+q} = \frac{-dp}{0} = \frac{dq}{1}$$

From this we obtain that $p = C_1$ is a constant and also $dq = dt \implies q=t+C_2$. Plugging into the PDE we obtain the relation $\boxed{C_2 = -C_1^2}$. Now as $q$ depends only on $t$ and $p$ only on $x$ we can integrate the following

$$du = pdx + qdt = C_1dx + (t-C_2^2)dt $$

$$\implies u = C_1x + \frac{t^2}{2} - C_1^2t + D$$

This gives you the complete integral of the PDE. Plugging in the initial values we have that $0 = C_1x + D \implies C_1 = D =0$, which yields the aprticular solution $u(x,t) = \frac{t^2}{2}$

Stefan4024
  • 35,843
  • Where can I find a derivation of the first equation? I want some intuitive explanation for it. – Prince Kumar Aug 22 '18 at 07:08
  • @PrinceKumar At the moment I don't have any resource at hands, so unfortunately I can't explicitly direct you to anything. However I think a quick Google search should yield some nice results. If I'm not mistaken this method is known as method of charactersitics – Stefan4024 Aug 22 '18 at 13:59
  • Yes. But I get many articles describing this for the case of 1st Order Linear PDE or at most Quasilinear, but not a general non-linear case. That's why I wanted to know any textbook sources as standard textbooks are much better at explaining such complex topics in simple manner. – Prince Kumar Aug 22 '18 at 19:24
  • @PrinceKumar The book by Evans on PDE covers the fully nonlinear case in $\S$3.2.2.c. $F$ fully nonlinear – JackpotWizard 180 Jan 06 '24 at 13:30