2003 II 5

From QED

< PlasmaWiki | Generals(Link to this page as [[PlasmaWiki/Generals/2003 II 5]])
Jump to: navigation, search

This diagram is the same as in White, for the bound-state problem.

By the Bohr-Sommerfeld condition, if we want the solution that goes to zero at both z=\pm\infty, we must have:

\int_{a}^{b}Q^{1/2}dx=\left(n+1/2\right)\pi

Then:

\int_{-E^{1/2}}^{E^{1/2}}\left(E-x^{2}\right)^{1/2}dx=\frac{E\pi}{2}

So that E = 2n + 1.

The solutions in region 1 are (using b = E1 / 2):

\left(b,z\right)=\frac{1}{\left(E-z^{2}\right)^{1/4}}\exp\left[-i\int_{b}^{z}\left(E-x^{2}\right)^{1/2}dx\right]

It is clear that this is unbounded, and is the dominant solution for z > b. Then to get the solution that is zero at z = a:

y\left(z=a\right)=0=\frac{A}{\left(E-a^{2}\right)^{1/4}}\exp\left[-i\int_{b}^{a}\left(E-x^{2}\right)^{1/2}dx\right]+\frac{B}{\left(E-a^{2}\right)^{1/4}}\exp\left[i\int_{b}^{a}\left(E-x^{2}\right)^{1/2}dx\right]

This relation can be simplified:

\frac{A}{B}=-\exp\left[2i\int_{b}^{a}\left(E-x^{2}\right)^{1/2}dx\right]

Then:

y_{1}=A\left(b,z\right)_{d}+B\left(z,b\right)_{s}

Crossing to region 2, we hop off the stokes line and so we pick up half of the stokes constant:

y_{2}=A\left(b,z\right)_{d}+\left[B+\frac{i}{2}A\right]\left(z,b\right)_{s}

Then we cross an anti-stokes:

y_{3}=A\left(b,z\right)_{s}+\left[B+\frac{i}{2}A\right]\left(z,b\right)_{d}

Now we must cross a stokes:

y_{4}=\left[\frac{1}{2}A+iB\right]\left(b,z\right)_{s}+\left[B+\frac{i}{2}A\right]\left(z,b\right)_{d}

This will then be the solution on the axis. Now if we were to connect with the zero at b:

y_{5}=\left[\frac{1}{2}A+iB\right]\left[b,-b\right]\left(-b,z\right)_{s}+\left[B+\frac{i}{2}A\right]\left(z,-b\right)_{d}\left[-b,b\right]

Using \left[-b,b\right]=\exp\left(-i\int_{-b}^{b}Q^{1/2}dz\right)=e^{-iW}:

y_{5}=\left[\frac{1}{2}A+iB\right]e^{iW}\left(-b,z\right)_{s}+\left[B+\frac{i}{2}A\right]e^{-iW}\left(z,-b\right)_{d}

Now requiring that y\left(z\right)=y\left(-z\right):

\left[\frac{1}{2}A+iB\right]\left(b,z\right)_{s}+\left[B+\frac{i}{2}A\right]\left(z,b\right)_{d}=\left[\frac{1}{2}A+iB\right]e^{iW}\left(-b,-z\right)_{s}+\left[B+\frac{i}{2}A\right]e^{-iW}\left(-z,-b\right)_{d}

Since Q1 / 2 is even, \left(-b,-z\right)=\left(z,b\right):

\left[\frac{1}{2}A+iB\right]\left(b,z\right)+\left[B+\frac{i}{2}A\right]\left(z,b\right)=\left[\frac{1}{2}A+iB\right]e^{iW}\left(z,b\right)+\left[B+\frac{i}{2}A\right]e^{-iW}\left(b,z\right)

This gives an equation for W:

B+\frac{i}{2}A=\left[\frac{1}{2}A+iB\right]e^{iW}

Using the condition for A / B:

e^{iW}=\frac{2-ie^{2iF}}{2i-e^{2iF}}

Where:

\begin{array}{rcl} W  =  \int_{-E^{1/2}}^{E^{1/2}}\left(E-x^{2}\right)^{1/2}dz=\frac{E\pi}{2}\\ F  =  \int_{E^{1/2}}^{a}\left(E-x^{2}\right)^{1/2}dx\end{array}

which will then give a condition on E. If a\rightarrow\infty, F\rightarrow\infty, and:

e^{iW}=\frac{2-ie^{2iF}}{2i-e^{2iF}}\rightarrow i

This will be satisfied if:

\frac{E\pi}{2}=\left(2n+1/2\right)\pi

Or E = 4n + 1. This is different from part b because we have removed the odd solutions.

This page was recovered in October 2009 from the Plasmagicians page on Generals_2003_II_5 dated 01:13, 7 May 2007.

Personal tools