I am trying to solve the PDE $$x\frac{\partial u}{\partial y}+\frac{\partial u}{\partial x}=u,$$ subject to the conditions $u(x,0)=0$ and $u(0,y)=y$.

The characteristic equations are: $$\frac{dx}{dt}=1, \ \frac{dy}{dt}=x, \ \frac{du}{dt}=u.$$ Solving these ODEs gives \begin{align} x(t,s)&=t+A(s) \\ y(t,s)&=\frac{t^2}{2}+A(s)t+B(s) \\ u(t,s)&=C(s)e^t. \end{align}

Case $1$, $u(x,0)=0$: \begin{align} x(0,s)&=s\implies x(t,s)=t+s \\ y(0,s)&=0\implies y(t,s)=\frac{t^2}{2}+st \\ u(0,s)&=0\implies u(t,s)=0 \\ \therefore u(x,y)&=0 \end{align}

Case $2$, $u(0,y)=y$: \begin{align} x(0,s)&=s\implies x(t,s)=t \\ y(0,s)&=0\implies y(t,s)=\frac{t^2}{2}+s \\ u(0,s)&=0\implies u(t,s)=se^t \\ \therefore u(x,y)&=\left(y-\frac{x^2}{2}\right)e^x \end{align}

The solutions say that case $1$ is valid for $y\leq x^2/2$ and case $2$ is valid for $y>x^2/2$. Have absolutely no idea why this.

I solved this problem using a Laplace transform and I arrived at the same conditions. But how can this possibly be interpreted when using the method shown.


$$x\frac{\partial u}{\partial y}+\frac{\partial u}{\partial x}=u$$ I agree with your calculus. A different presentation of the Charpit-Lagrange equations : $$\frac{dy}{x}=\frac{dx}{1}=\frac{du}{u}=dt$$ A first characteristic equation comes from $\frac{dy}{x}=\frac{dx}{1}$ : $$y-\frac{x^2}{2}=c_1$$ A second characteristic equation comes from $\frac{dx}{1}=\frac{du}{u}$ : $$e^{-x}u=c_2$$ General solution of the PDE on implicit form $c_2=F(c_1)$ : $$e^{-x}u=F\left(y-\frac{x^2}{2}\right)$$ $$u(x,y)=e^xF\left(y-\frac{x^2}{2}\right)$$ $F(X)$ is an arbitrary fonction to be determined according to the boundary conditions.

First condition alone :

$u(x,0)=0=e^xF\left(0-\frac{x^2}{2}\right)\quad\implies\quad F(X)=0\quad\text{on}\quad X<0$

Putting $F=0$ into the above general solution leads to $$u(x,y)=0$$ Of course this trivial solution was obvious by inspection of the PDE.

Second condition alone :


The function $F$ is determined : $F(X)=X$. We put it into the above general solution where $X=y-\frac{x^2}{2}$ .

$$u(x,y)=e^x (y-\frac{x^2}{2})$$

Both conditions together :

In order to satisfy both conditions, necessarily the function $F(X)$ is a piecewise function now defined on two domains : $$F(X)=\begin{cases} 0 && X<0\\ X && X>0 \end{cases}$$ Since $X=y-\frac{x^2}{2}$ the boundary between the two domains is $y=\frac{x^2}{2}$. $$u(x,y)=\begin{cases} 0 && y<\frac{x^2}{2}\\ e^x (y-\frac{x^2}{2}) && y>\frac{x^2}{2} \end{cases}$$ This is in agreement with your own results.

  • $\begingroup$ (+1) I like this method presented, thanks! My main question is though, if I were to use the method that I have shown, how am I too see the correct domains for the piecewise function of $u(x,y)$? For instance, for case $2$, how is it obvious that $y>\frac{x^2}{2}$ given that I have shown $u(x,y)=e^x\left(y-\frac{x^2}{2}\right)$? Why can't $y<\frac{x^2}{2}$? $\endgroup$ – Stuart-James Burney Apr 28 at 14:28

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.