I need to solve the following partial differential equation: $$F(x, y, u, p, q) = u_y + (1-2u)\cdot u_x = 0$$ with $$u(x, 0) = \left\{\begin{array}{cc} \frac{1}{4} & x < 0 \\ \frac{3}{4} & x>0 \end{array}\right.$$ Here we have $p = u_x$ and $q = u_y$. $u(x,y)$ is a continously differentiable function and $-\infty < x < \infty$ and $0 < y < \infty$.

I use the characteristic differential equations to get a solution. These are $$ \left\{\begin{array}{ll} \frac{dx}{ds} = F_p = 1-2u \\ \frac{dy}{ds} = F_q = 1\\ \frac{du}{ds} = pF_p+qF_q = p(1-2u) + q\\ \frac{dp}{ds} = -F_x - pF_u = 2p^2\\ \frac{dq}{ds} = -F_y - qF_u = 2pq \end{array}\right. $$

The initial conditions are, if $s=0$, that $x = \lambda, y = 0,$ $$ u = \left\{\begin{array}{cc} \frac{1}{4} & \lambda < 0\\ \frac{3}{4} & \lambda > 0\end{array}\right. $$ and, using that $$0 = \frac{du}{d\lambda} = p(\lambda)\frac{dx}{d\lambda} + q(\lambda)\frac{du}{d\lambda} = p(\lambda)\cdot 1 + 0$$ and $$ 0 = q + (1-2u)p = q + (1-2)\cdot 0 = q$$ we get $p=q=0$.

From this, it is easily seen that $y = s$. But from here, I'm stuck. The general solution of $\frac{dp}{ds} = 2p^2$ is $p(s) = \frac{1}{c_1 - 2s}$. It impossible for this function to sattisfy the initial condition, so does that mean that this partial differential equation doesn't have a solution?

Another thing is that the initial condition for $u(x, 0)$ isn't continous, so how can the solution of this problem be continously differentiable?


The PDE $u_y + (1-2u)u_x = 0$ is the Lighthill-Whitham-Richards traffic flow equation. The proposed initial-value problem may be interpreted as a one-way road with the car density $u=1/4$ before the abscissa $x=0$ (fast cars), and the car density $u=3/4$ after the abscissa $x=0$ (slow cars). In mathematical words, this is a Riemann problem, which cannot be solved by using the method of characteristics alone: we should consider weak solutions. The Lax entropy condition tells that the solution is a shock wave. Its speed is given by the Rankine-Hugoniot condition $s = 1- \frac{1}{4} - \frac{3}{4} = 0$. Therefore, the solution is a static shock wave $$ u(x,y) = \left\lbrace \begin{aligned} &1/4 &&\text{if}\quad x<0\\ &3/4 &&\text{if}\quad x>0 \end{aligned}\right. $$


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

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

$\dfrac{du}{dt}=0$ , letting $u(0)=u_0$ , we have $u=u_0$

$\dfrac{dx}{dt}=1-2u=1-2u_0$ , letting $x(0)=f(u_0)$ , we have $x=(1-2u_0)t+f(u_0)=(1-2u)y+f(u)$ , i.e. $u=F(x+(2u-1)y)$

$u(x,0)=\begin{cases}\dfrac{1}{4}&x<0\\\dfrac{3}{4}&x>0\end{cases}$ :


$\therefore u=\begin{cases}\dfrac{1}{4}&x+(2u-1)y<0\\\dfrac{3}{4}&x+(2u-1)y>0\end{cases}=\begin{cases}\dfrac{1}{4}&x-\dfrac{y}{2}<0\\\dfrac{3}{4}&x+\dfrac{y}{2}>0\end{cases}$

Hence $u(x,y)=\begin{cases}\dfrac{1}{4}&x<\dfrac{y}{2}\\\dfrac{3}{4}&x>-\dfrac{y}{2}\\c&\text{otherwise}\end{cases}$

  • $\begingroup$ I'm not sure I follow this. What is $f$ in $x(0) = f(u_0)$. It seems to me like it is an initial condition, but how exactly is it defined? Is it the inverse of $F$? If so, if $u = F(x + (2u-1)y)$, shouldn't $u(x,y)$ be a function of $x+(2u-1)y$ instead of $x$, like in your last expression for $u(x,y)$? $\endgroup$ – Pierre Jun 27 '15 at 23:04
  • $\begingroup$ @ImPact Although the general solution is of the form $u=F(x+(2u-1)y)$ , the constant-type condition leads the function only possible to be the constant-type function. $\endgroup$ – doraemonpaul Jun 28 '15 at 0:42

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.