Mysteries of Geometric Optics In MTW Chapter 22

TerryW
Gold Member
Messages
222
Reaction score
20
Homework Statement
I really don't understand what is going on here - can anyone shed any light please?
Relevant Equations
Contained in attachments
At the start of this section §22.5 (Geometric Optics in curved Spacetime), the amplitude of the vector potential is given as:

A = ##\mathfrak R\{Amplitude \ X \ e^{i\theta}\} ##

The Amplitude is then re=expressed a "two-length-scale" expansion (fine!) but it then is modified further to (22.25) by
introducing a "useful" parameter ε...
Introducing a parameter.png
But this amplitude is now varying at a quite different rate (if ε is anything other than unity) so it isn't equivalent to the original vector potential.

When we get into the calculations, (22.7) is readily derived but I have a problem with the gathering terms O(##\frac {1}{ε}##) or O(ε) etc because, if ε reverts to its eventual value unity thereby recovering the original rate of variation, justification for (22.28) and (22.9) looks a bit thin.

Using parameter to derive realtionships.png


Can anyone explain what is going on here and why this is all reasonable?

Regards
TerryW
 
Physics news on Phys.org
The introduction of the parameter ##\varepsilon## is just a convenient way to "tag" the various terms so that the power of ##\varepsilon## immediately tells you the order of the term in λ/L. A term with ##\varepsilon^2## is second order in λ/L even when ##\varepsilon = 1##. If you want, you could forgo introducing ##\varepsilon##. But, then, you would need to remember that ##a^\mu## is zero order in λ/L, ##b^\mu## is first order, ##c^\mu## is second order, ##\theta## is order -1,etc. When MTW "collect terms of order ##\epsilon^n##", they are just collecting terms of order (λ/L)n, which you could do even if you didn't introduce ##\varepsilon##.
 
It's not different from the usual eikonal expansion in standard Maxwell theory in flat space. Perhaps it helps to read about this first, e.g., in A. Sommerfeld, Lectures on Theoretical Physics, vol. 4 (optics).

In quantum mechanics a mathematically equivalent procedure is known as WKB approximation. There it's an expansion in powers of ##\hbar##.

The mathematical procedure is known as "singular perturbation theory".
 
TSny said:
The introduction of the parameter ##\varepsilon## is just a convenient way to "tag" the various terms so that the power of ##\varepsilon## immediately tells you the order of the term in λ/L. A term with ##\varepsilon^2## is second order in λ/L even when ##\varepsilon = 1##. If you want, you could forgo introducing ##\varepsilon##. But, then, you would need to remember that ##a^\mu## is zero order in λ/L, ##b^\mu## is first order, ##c^\mu## is second order, ##\theta## is order -1,etc. When MTW "collect terms of order ##\epsilon^n##", they are just collecting terms of order (λ/L)n, which you could do even if you didn't introduce ##\varepsilon##.
Thanks TSny,

I'd already decided that I didn't need ##\varepsilon## and that I could get to all the results by "remembering" the order of the terms, but your reply is helpful in reassuring me that I wan't missing something crucial.

CheersTerry
 
vanhees71 said:
It's not different from the usual eikonal expansion in standard Maxwell theory in flat space. Perhaps it helps to read about this first, e.g., in A. Sommerfeld, Lectures on Theoretical Physics, vol. 4 (optics).

In quantum mechanics a mathematically equivalent procedure is known as WKB approximation. There it's an expansion in powers of ##\hbar##.

The mathematical procedure is known as "singular perturbation theory".
Thanks for replying vanhees71,

I'll see if I can find something about the WKB approximation or singular perturbation theory - it doesn't look like I can get a free download of Sommerfield Vol IV. Maybe I can find something in my ancient copy of the Feynman lectures.RegardsTerry W
 
##|\Psi|^2=\frac{1}{\sqrt{\pi b^2}}\exp(\frac{-(x-x_0)^2}{b^2}).## ##\braket{x}=\frac{1}{\sqrt{\pi b^2}}\int_{-\infty}^{\infty}dx\,x\exp(-\frac{(x-x_0)^2}{b^2}).## ##y=x-x_0 \quad x=y+x_0 \quad dy=dx.## The boundaries remain infinite, I believe. ##\frac{1}{\sqrt{\pi b^2}}\int_{-\infty}^{\infty}dy(y+x_0)\exp(\frac{-y^2}{b^2}).## ##\frac{2}{\sqrt{\pi b^2}}\int_0^{\infty}dy\,y\exp(\frac{-y^2}{b^2})+\frac{2x_0}{\sqrt{\pi b^2}}\int_0^{\infty}dy\,\exp(-\frac{y^2}{b^2}).## I then resolved the two...
It's given a gas of particles all identical which has T fixed and spin S. Let's ##g(\epsilon)## the density of orbital states and ##g(\epsilon) = g_0## for ##\forall \epsilon \in [\epsilon_0, \epsilon_1]##, zero otherwise. How to compute the number of accessible quantum states of one particle? This is my attempt, and I suspect that is not good. Let S=0 and then bosons in a system. Simply, if we have the density of orbitals we have to integrate ##g(\epsilon)## and we have...
Back
Top