Solving a 2nd order PDE with boundary data

103 Views Asked by At

I have a feeling I may be making a trivial mistake here, but I would really appreciate it if someone could verify my method.

I have a 2nd order PDE:

$$u_{xx} - x^2 u_{yy} - \frac{1}{x} u_x$$

I can reduce this to normal form and find the general solution is given by:

$$u(x,y) = f(x^2 - 2y) + g(x^2 + 2y)$$

Now I have to impose the boundary data: $u(1,y) = 1$ and $u_x(1,y) = 2$.

So to solve this I plug in the data, the first condition gives:

$$1 = f(1-2y) + g(1+2y)$$

And the second:

$$2 = 2f'(1-2y) + 2g'(1+2y)$$

Now I differentiate the first equation and simultaneously solve with the second to give:

$f(1-2y) = \frac{1}{2} y + c_1$ and $g(1+2y) = \frac{1}{2}y + c_2$. Now this is where the difficulty is, when i substitute these into the first equation, it is impossible to determine the coefficients.

Thanks for any help

1

There are 1 best solutions below

0
On BEST ANSWER

Start with

$$f(1−2y)+g(1+2y)=1 \text{ (1)}$$,

$$f'(1−2y)+g'(1+2y)=1 \text{ (2)}$$.

@Dmoreno's comment reminded us to differentiate (1) against $y$ and obtain

$$0 = f'(1−2y)-g'(1+2y)= \text { (3)}$$.

From (2) and (3) we obtain:

$$f'(1−2y)=g'(1+2y)=1/2 $$ So $$ f(1-2y)=(1-2y)/2+c_1, g(1+2y)=(1+2y)/2+c_2$$

Substituting them into (1) leads to:

$$1=(1-2y)/2+c_1+(1+2y)/2+c_2=1+c_1+c_2 \implies c_2=-c_1$$