5

Solving the quaslinear PDE by the method of characteristics is a bit tricky for me.

I was trying to obtain the solution $u$ for the PDE

$$u_{t} - u^2 u_{x} = 0$$ The initial condition is given by:

$$u(x,0) = g(x) = \begin{cases} -0.5, & x \leq 0 \\ 1, & 0<x<1 \\ 0.5, & x \geq 1 \end{cases}$$

Using the method of characteristics I have tried to write the characteristic equation as

$$\frac{dt}{1} = \frac{dx}{-u^2}$$

Trying to obtain an expression for $u$ from the above equation seems tricky as there is no $u$ term.

Dylan
  • 16,575
BAYMAX
  • 4,972

1 Answers1

5

It is important to note that the flux $f(u)=-\frac13 u^3$ in the conservation law $u_t + f(u)_x = 0$ is nonconvex. The problem may be viewed as two neighbor Riemann problems, for which one can compute the waves, and their potential interactions. A plot of the base characteristic lines in the $x$-$t$ plane is given below:

characteristics

One observes that the method of characteristics provides a unique solution $u = g(x + u^2 t)$ in some specific parts of the plane only. In facts, there may be zero, one or two characteristics passing at a given point $(x,t)$. The previous plot suggest that a shock wave is generated at $x=0$, and that a rarefaction wave is generated at $x=1$. This claim is verified by convex hull constructions, a graphical method related to the Oleinik entropy condition for shock waves (see this post):

hull

Here the Rankine-Hugoniot condition gives the shock speed $s = -\frac{1}{4}$. Hence, the following solution for times $t< \frac{4}{3}$ is obtained: $$ u(x,t) = \left\lbrace \begin{aligned} &{-\tfrac12} & &\text{if}\quad x< {-\tfrac{1}{4} t}\\ &1 & &\text{if}\quad {-\tfrac{1}{4} t} < x \leq 1-t\\ &\sqrt{(1-x)/t} & &\text{if}\quad 1- t \leq x \leq 1-\tfrac14 t\\ &{\tfrac12} & &\text{if}\quad 1- \tfrac14 t \leq x \end{aligned} \right. $$ For larger times, the interaction between the shock and the rarefaction must be computed.

EditPiAf
  • 20,898