Solving conservation law IVP when characteristics intersect at $t=0$

characteristicspartial differential equations

This is my first post here so I apologize if there is anything wrong with it. If so, please let me know and I can edit the post.

I just tried to solve the following question:

Problem 2.

  1. Give the explicit solution for all $t\leqslant 1/2$, to the following equation:
    $$
    \left\lbrace\begin{aligned}
    &u_t+f(u)_x=0,\\
    &u(x,0)=\begin{cases}
    1, & x\leqslant 1\\
    1/2, &1<x\leqslant 3\\
    3/2, &x>3
    \end{cases}
    \end{aligned}\right.
    $$

    where $f(u)=4u(2-u)$.
  2. Represent the solution at times $t=0$, $t=1/4$ and $t=1/2$.
  3. Determine the solution beyond $t=1/2$?

What I have so far:

  • $f'(u) = 8 – 8u$, and for a solution $u(x,t), u$ is constant along the characteristic curves defined by $x'(t) = 8 – 8u, x(0) = a$ ($a$ real number), ie $u(x(t), t) = u(x(0), 0) = u(a, 0)$.
  • Then $u$ does not depend on $t$, so the characteristic curves are $x(t) = (8 – 8\cdot u(a,0)) \cdot t + a$.
  • ie $x(t)$ is given by $$\begin{cases}
    0t + a & \text{if $a<=1$} \\
    4t + a & \text{if $1 <= a <= 3$} \\
    -4t + a & \text{if $a>3$}
    \end{cases}$$

The question asks for the solution before $t = 1/2$, however I'm struggling with the following:

  1. Based on the above, it seems the characteristic curves already intersect at $t = 0$ (and $x = 3$), long before $t = 1/2$. Am I supposed to construct a shock wave solution using the Rankine-Hugoniot jump condition? I actually think there may be 2 shock wave solutions needed, because the characteristic curves do not cover a space around $(1, 0)$, it seems, since the curve $x = -4t + 3$ intersects the t-axis at $t=4/3$… However if I do this, wouldn't it be a solution for some range of t that isn't $t \leq 1/2$?
  2. On the other hand, is my reasoning above (before bullet #1) correct? This is, I think, analogous to what I have seen in class and in some other resources I looked at, however they all used the function $f(u) = u^{2} / 2$ and while I don't think this would change much, maybe I'm missing something vital.

Thank you in advance for any help.

Best Answer

The situation is very similar to this post and related ones, where piecewise constant initial data is considered (i.e, a series of Riemann problems).

  • At $x=1$, we have the characteristic speeds $f'(1) = 0$ on the left side and $f'(1/2) = 4$ on the right side of the discontinuity. Since the right slope $4>0$ is larger than that on the left, characteristic lines separate: a rarefaction wave forms over $1\leq x\leq 1+4t$, with a linear fan solution.
  • At $x=3$, we have the characteristic speeds $f'(1/2) = 4$ on the left side and $f'(3/2) = -4$ on the right side of the discontinuity. Since the right slope $-4<4$ is smaller than that on the left, characteristic lines intersect: a shock wave forms, which speed $s=0$ follows from the Rankine-Hugoniot condition (it's a static shock located at $x=3$).
  • These two waves interact at the time $t^*$ such that $1+4t^* = 3$, i.e $t^*=1/2$. Thus, one may distinguish the case $t<1/2$ from the case $t\geq 1/2$ (cf. linked post).

To link with the Burgers equation, one may consider the change of variable $v=8(1-u)$ such that $$ v_t + (\tfrac12 v^2)_x = 0\, , $$ $$ v(x,0) = \begin{cases} 0, & x\leq 1\\ 4, & 1<x\leq 3\\ -4, & x>3 \end{cases} $$