Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top



$$\frac{\partial z}{\partial x}+2\frac{\partial z}{\partial y}+3z=0$$

Find the characteristics associated with this PDE and find an explicit solution $z(x,y)$ which satisfies the initial condition $z=e^{x}$ on the line segment $\{(x,y):0 \leq x \leq 1, y=0 \}$.

State the region on the $(x,y)$-plane in which $z(x,y)$ is uniquely determined by the initial condition.


To find the characteristics:

$$\frac{dx}{ds}=1, \quad \frac{dy}{ds}= 2, \quad \frac{dz}{ds}=-3z$$ which yields that $$x=s+A, \quad y=2s+B, \quad z=Ce^{-3s}$$

Are we now OK to parametrise the line segment as $L=(t,0,e^t)$ using our initial condition, giving rise to explicit parametric solutions

$$\begin{cases} x=s+t \\ y=2s \\ z=e^{t-3s} \end{cases}$$ for $0 \leq t \leq 1$.

I'm not convinced that having $z=e^x$ allows us to parametrise the line segment in the way. Also, to establish where the solution is uniquely determined:

At $t=0$,

$$x=s, \quad y=2s, \quad \text{so } y=2x$$ and at $t=1$,

$$ x=s+1, \quad, y=2s, \quad \text{so } y=2x-2$$

Not sure how much guidance this offers.

Any help with the various parts of this would be much appreciated. Not sure what's right/wrong really. Regards as always, MM.

share|cite|improve this question
up vote 2 down vote accepted

The solution using the method of characteristic is a way to reduce a partial differential equation to a set of ordinary differential equations. Originally, you wanted to find a function $z(x,y)$ which satisfies $$\frac{\partial z}{\partial x}+2\frac{\partial z}{\partial y}+3z=0\qquad\qquad(1)$$ as well as the boundary condition.

We make the ansatz, that $x=x(s)$ and $y=y(s)$ where $s$ is a parameter and $(x,y)$ is a curve in $\mathbb{R}^2$. Furthermore, let $z(s)=z(x(s),y(s))$. Then $$z'(s) = \frac{\partial z}{\partial x} x'(s) + \frac{\partial z}{\partial y} x'(s).$$ Note that this equation is equivalent to (1) if $$x'(s)=1, \qquad y'(s)=2, \qquad z'(s)=-3z.$$ (this is the same set of equations as you have in your post)

The solutions to the equations for $x(s)$ and $y(s)$ give us a family of curves $$ x(s)= x_0 +s \qquad y(s) = 2s,$$ parameterized by $x_0$ (note that the additive constant in $y$ can be absorbed in a reparameterization of the curve). These curves go through the point $(x_0,0)$ at $s=0$ for which we have the boundary value $z= e^{x_0}$ specified.

Along each of the curves (characteristics), the partial differential equation (1) becomes an ordinary differential equation $$z'(s)=z$$ as noted above. The solution is given by $z(s)=Ce^{-3s}$. The boundary value read $z(0)=C =e^{x_0}$. Thus we have $$z(s)= e^{x_0-3s}$$ as a solution along the characteristic.

It you want to obtain the (original) function $z(x,y)$, you have to find $x_0(x,y)$ and $s(x,y)$. In this case this is rather trivial and we obtain $x_0=x-y/2$ and $s=y/2$ such that $$ z(x,y) = e^{x-2y}.$$

share|cite|improve this answer
Many thanks, that's a great explanation. How can I determine the region of the $(x,y)$-plane in which $z(x,y)$ is uniquely determined? – Mathmo Jan 13 '12 at 10:39
@MiamiMaths: All $\mathbb{R}^2$ (as both $x(s)$ and $y(s)$ are invertible. – Fabian Jan 13 '12 at 16:33

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.