MHB Why are there 2 cosines used in the general solution for this domain?

  • Thread starter Thread starter Dustinsfl
  • Start date Start date
  • Tags Tags
    Domain Laplace
Click For Summary
The discussion focuses on the derivation of the general solution for the function \(\phi(x, y)\) under the specified boundary conditions. The use of two cosine functions arises from the separation of variables approach, where \(\phi(x, y)\) is expressed as a product of functions dependent on \(x\) and \(y\). The periodic cosine function is not applied to \(y\) due to the nature of the boundary conditions, specifically that \(\phi(x, b) = 0\), which necessitates the use of hyperbolic functions for \(y\). The integration involving cosine functions leads to the appearance of the Dirac delta function, simplifying the solution. The query about the extra factor of 1/2 indicates a potential oversight in the integration process that needs clarification.
Dustinsfl
Messages
2,217
Reaction score
5
The domain is \(0\leq x < \infty\) and \(0\leq y\leq b\).
\begin{align*}
\phi(x, 0) &= f(x)\\
\phi(x, b) &= 0
\end{align*}
I want to show that
\[
\phi(x, y) = \frac{1}{\pi}\int_0^{\infty} \int_{-\infty}^{\infty}f(\xi) \frac{\sinh[u(b - y)]}{\sinh(ub)} \cos[u(\xi - x)]d\xi du
\]
Why isn't the periodic function cosine used on y? How should this be started then?
 
Physics news on Phys.org
Question is at the bottom in red.
Our boundary conditions on \(x\) are
\[
\lim_{x\to\pm\infty}\phi(x, y) = 0.
\]
Let \(\phi(x, y) = \varphi(x)\psi(y)\).
Then \(\frac{\varphi''}{\varphi} = - \frac{\psi''}{\psi} = -k^2\).
\begin{align}
\varphi(x) &\sim\left\{\cos(kx), \sin(kx)\right\}\\
\psi(y) &\sim\left\{\cosh(ky), \sinh(ky)\right\}
\end{align}
From the boundary conditions on \(y\), we see that we need to make a change of
variables.
That is, \(y\to b - y^*\); however, since the choice of variables are
arbitrary, let the change of variable be \(b - y\).
So we have is
\[
\psi(y) \sim\left\{\cosh(k(b - y)), \sinh(k(b - y))\right\}.
\]
Using the boundary condition \(\phi(x, b) = 0\), we have that
\[
\psi(y) \sim\sinh(k(b - y)).
\]
The general solution is then
\begin{alignat}{2}
\phi(x, y) &= \int_0^{\infty}\left[A(k)\cos(kx) + B(k)\sin(kx)\right]
\sinh(k(b - y))dk\\
\phi(x, 0) &= \int_0^{\infty}\left[A(k)\cos(kx) + B(k)\sin(kx)\right]
\sinh(kb)dk && ={} f(x)
\end{alignat}
Let \(A^*(k) = A(k)\sinh(kb)\) and \(B^*(k) = B(k)\sinh(kb)\),
and let's multiple through by \(\cos(k'x)\).
\begin{alignat}{2}
\phi(x, 0) &= \int_{-\infty}^{\infty}\int_0^{\infty}
\left[A^*(k)\cos(kx)\cos(k'x) + B^*(k)\sin(kx)\cos(k'x)\right]dkdx
&& ={} \int_{-\infty}^{\infty}f(x)\cos(k'x)dx\\
\phi(x, 0) &= \int_0^{\infty}\int_{-\infty}^{\infty}
A^*(k)\cos(kx)\cos(k'x)dxdk && ={} \int_{-\infty}^{\infty}f(x)\cos(k'x)dx
\end{alignat}
Note that \(\cos(kx)\cos(k'x) = \frac{1}{2}\left[\cos(x\Delta k) +
\cos[x(k + k')]\right]\).
From class, we know that the integral of \(\cos[x(k + k')]\) is zero.
\begin{align}
\int_{-\infty}^{\infty}\cos(kx)\cos(k'x)dx &=
\int_{0}^{\infty}\cos(x\Delta k)dx\\
&= \mathcal{R}e
\left\{\int_0^{\infty}\exp\left[(i\Delta k - \eta)x\right]dx\right\}\\
&= \mathcal{R}e\left\{\left.
\frac{\exp\left[(i\Delta k - \eta)x\right]}
{i\Delta k - \eta}\right|_0^{\infty}\right\}\\
&= \mathcal{R}e\left\{\frac{1}{-i\Delta k + \eta}\right\}\\
&= \frac{\eta}{\eta^2 + (\Delta k)^2}
\end{align}
In class, we have shown that
\(\frac{\eta}{\eta^2 + (\Delta k)^2} = \pi\delta(k - k')\).
Therefore, we have that
\[
\int_{-\infty}^{\infty}\cos(kx)\cos(k'x)dx = \pi\delta(k - k').
\]
Now we have that
\[
\phi(x, 0) = \pi\int_0^{\infty}A^*(k)\delta(k - k')dk = \int_{-\infty}^{\infty}f(x)\cos(k'x)dx
\]
The integral \(\int_0^{\infty}A^*(k)\delta(k - k')dk = A^*(k')\) so
\begin{align}
A^*(k') &= \frac{1}{\pi}\int_{-\infty}^{\infty}f(x)\cos(k'x)dx\\
A(k) &= \frac{1}{\pi\sinh(kb)}\int_{-\infty}^{\infty}f(x')\cos(kx')dx'
\end{align}
Now we can write the general solution \cref{lapgensoln} as
\[
\phi(x, y) = \frac{1}{\pi}\int_0^{\infty}\int_{-\infty}^{\infty}
\frac{\sinh(k(b - y))}{\sinh(kb)}f(x')\cos(kx)\cos(kx')dx'dk.
\]
Again we have that \(\cos(kx)\cos(kx') = \frac{1}{2}\left[\cos(k(x' - x)) +
\cos(k(x' + x))\right]\) and \(\cos(k(x' + x))\) integrates out to zero.
Let \(x' = \xi\) and \(k = u\).
Then the solution is
\[
\phi(x, y) = \frac{1}{2\pi}\int_0^{\infty}\int_{-\infty}^{\infty}
\frac{\sinh(u(b - y))}{\sinh(ub)}f(\xi)\cos(u(\xi - x))d\xi dk.
\]
I have an extra factor of 1/2. Where am I missing a factor of 2?
 
Last edited:

Similar threads

  • · Replies 5 ·
Replies
5
Views
3K
Replies
17
Views
6K
  • · Replies 17 ·
Replies
17
Views
3K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 25 ·
Replies
25
Views
4K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 1 ·
Replies
1
Views
3K
Replies
8
Views
4K