1
$\begingroup$

I'm trying to prove that the following PDE has no solutions that are non-constant functions $f\in C^1(\mathbb{R}^2)$.

$$ (y+1)u_y + x u_x = 0 $$

I'm not sure what is the theory that would help start formulating a solution for this since the material provided in my class is relatively bad explained and/or I'm missing something.

I tried first solving it with the method of characteristics:

$$ \frac{dy}{dx} = \frac{y+1}{x} \implies \frac{y+1}{x} = K $$

Then I consider the variable change: $$ \xi = \frac{y+1}{x}\\ \eta = y $$ And I get of course: $$ (\eta+1)u_\eta = 0 \implies u=constant $$

But I suspect this doesn not constitute the proof requested. This would just proof there is constants solutions. I've tried also asumming it has such solution $u(x, y)$ non-constant to get to an absurd.

Then $u_x \neq 0$ and $u_y \neq 0$.

And Then it would be

$$(y+1)u_y = -xu_x \quad \text{with} \quad (y+1)u_y\neq 0; xu_x\neq 0; y\neq1; x\neq0$$

But I don't get anything.

$\endgroup$
3
$\begingroup$

$$ (y+1)u_y + x u_x = 0 $$ $\frac{dy}{y+1}=\frac{dx}{x}=\frac{du}{0}$

First characteristics, from $\quad\frac{dy}{y+1}=\frac{dx}{x}$ : $$\frac{y+1}{x}=c_1$$ Second characteristics, from $\frac{du}{0}=$finite, which implies $u=c_2$ :

General solution on the form of implicit equation : $$\Phi\left(\frac{y+1}{x}\:,\:u\right)=0$$ $\Phi$ is an arbitrary function of two variables, to be determined according to some boundary conditions.

Or equivalently, on explicit form : $$u(x,y)=f\left(\frac{y+1}{x}\right)$$ $f$ is an arbitrary function, to be determined according to some boundary conditions.

If $f$ is a non constant function, $u$ is necessarily a function of $\frac{y+1}{x}$, so is not $C^1$. Thus the PDE has no solutions that are non-constant functions $f\in C^1(\mathbb{R}^2)$.

$\endgroup$
  • $\begingroup$ Just a comment, your answer seems to imply that if the problem were, for example, defined for $C^1((0,\infty)\times(0,\infty))$, then the solution would not necessarily be constant, but this is not the case. The solution is constant also. $\endgroup$ – Enredanrestos May 15 '18 at 6:44
  • $\begingroup$ @Enredanrestos : So $u(x.y)=\frac{x}{1+y}$ is not a solution of the PDE in the first quadrant? $\endgroup$ – LutzL May 15 '18 at 15:47
  • $\begingroup$ Yes it is, I was wrong. The point apparently is that the characteristics dont cross 'inside' the domain.. $\endgroup$ – Enredanrestos May 15 '18 at 21:47
2
$\begingroup$

The method of characteristics says that $u$ is constant on each ray through $(x,y) = (0,-1)$. Thus the value at every point is equal to the value at $(0,-1)$, therefore is constant.

$\endgroup$

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.