16—Calculus of Variations 404
In the integral forT, where the starting point and the ending point are the source and image points,
the second order variation will be zero for these long, gradual changes in the path. The straight-line
path through the center of the lens takesleasttime if its starting point and ending point are closer
than this source and image. The same path will be a saddle (neither maximum nor minimum) if the
points are farther apart than this. This sort of observation led to the development of the mathematical
subject called “Morse Theory,” a topic that has had applications in studying such diverse subjects as
nuclear fission and the gravitational lensing of light from quasars.
Thin Lens
This provides a simple way to understand the basic equation for a thin lens. Let its thickness betand
its radiusr.
p q
t
n r
Light that passes through this lens along the straight line through the center moves more slowly as it
passes through the thickness of the lens, and takes a time
T 1 =
1
c
(p+q−t) +
n
c
t
Light that take a longer path through the edge of the lens encounters no glass along the way, and it
takes a time
T 2 =
1
c
[√
p^2 +r^2 +
√
q^2 +r^2
]
Ifpandqrepresent the positions of a source and the position of its image at a focus, these two times
should be equal. At least they should be equal in the approximation that the lens is thin and when you
keep terms only to the second order in the variation of the path.
T 2 =
1
c
[
p
√
1 +r^2 /p^2 +q
√
1 +r^2 /q^2
]
=
1
c
[
p
(
1 +r^2 / 2 p^2
)
+q
(
1 +r^2 / 2 q^2
)]
EquateT 1 andT 2.
(p+q−t) +nt=
[
p
(
1 +r^2 / 2 p^2
)
+q
(
1 +r^2 / 2 q^2
)]
(n−1)t=
r^2
2 p
+
r^2
2 q
1
p
+
1
q
=
2(n−1)t
r^2
=
1
f
(16.49)
This is the standard equation describing the focusing properties of thin lenses as described in every
elementary physics text that even mentions lenses. The focal length of the lens is thenf=r^2 /2(n−1)t.
That isnotthe expression you usually see, but it is the same. See problem16.21. Notice that this
equation for the focus applies whether the lens is double convex or plano-convex or meniscus:. If you
allow the thicknesstto be negative (equivalent to saying that there’s an extra time delay at the edges
instead of in the center), then this result still works for a diverging lens, though the analysis leading up
to it requires more thought.