Professional Documents
Culture Documents
1. The Catenoid
!
Consider the integrand F (x, y, y ! ) = y 1 + (y ! )2 in Eq. (1.5) when y is a function of x.
Solution
d ∂F ∂F d yy ! ! y ! y ! + yy !! yy ! y ! y !! !
( ! )− = (! )− 1 + (y ! )2 = ! −! 3 − 1 + (y ! )2 = 0
dx ∂y ∂y dx 1 + (y ! )2 1 + (y ! )2 1 + (y ! )2
yy !! − 1 − y !2 = 0.
∂F yy ! !
y! !
− F = y! ! − y 1 + (y ! )2 = C
∂y 1 + (y ! )2
dy dx x+D
# #
! =± =± .
2
y −C 2 C C
C sinh(α)dα x+D
#
=α=± .
C sinh(α) C
Thus
x+D
y = C cosh( )
C
2. Dido’s Problem
!
Consider the integrand F (s, y, y ! ) = y 1 − (y ! )2 in Eq. (1.10) when y is a function of s.
13
(e) What geometric curve does this parametric solution represent?
Solution
(a) Since F does not depend on s a first integral is
∂F −2y ! !
y! !
− F = y!y ! − y 1 − (y ! )2 = C
∂y 2 1 − (y ! )2
Rearranging gives !
y = −C 1 − (y ! )2
or !
! C 2 − y2
y =± .
C
(b) Separating variables gives
C dy
# #
! = ± ds.
C 2 − y2
The integral on the left is an inverse trig function
y
C sin−1 ( ) = ±(s + D)
C
Thus
s+D
y = ±C sin
C
(c) The boundary conditions give
D
0 = ±C sin
C
and
L+D
0 = ±C sin
C
L
So we should choose D = 0 and C = π . By the way, formally there are more general
solutions for L/C = nπ and D/C = mπ for integer n, m. However, for n > 1 we cannot
obtain a positive y throughout. The ± sign can be absorbed by choosing m even or odd.
Thus,
L sπ
y = sin
π L
(d) Now
dy sπ
= cos
ds L
Thus from
dx dy
( )2 + ( )2 = 1
ds ds
we can determine that
dx sπ
= ± sin
ds L
Thus, by integrating
L sπ
x(s) = ∓ cos + C2
π L
The boundary conditions (and choosing x to be positive) lead to
L sπ L
x(s) = −
cos + .
π L π
There is no endpoint condition on x but instead the x-coordinate of the endpoint of the
optimal curve is determined by x(L) = 2L/π.
14
(e) This is a semicircle of radius Lπ with centre at ( Lπ , 0). The area of the city is L2 /(2π). Notice
that the first integral can now be retrospectively interpreted as the radius of the city. First
integrals often have interesting interpretations.
3. The Catenary
(a) Determine F (x, y, y ! ) for the integrand in Eq. (1.12) when y is a function of x.
(b) Solve the corresponding Euler-Lagrange equations or the first integral if there is one.
(c) Apply the boundary conditions y = h at both x = a and x = b to determine the constants
of integration.
Solution
(a) The integral is
B B
λ
# #
(ρgy + λ)ds = ρg (y + )ds
A A ρg
λ
but since λ is arbitrary there is not much point in writing ρg instead of writing just λ so
one can do the calculation using (y + λ)ds.
When y is a function of x this gives
!
F (x, y, y ! ) = (y + λ) 1 + y !2 .
(b) Since F does not depend on x the appropriate first integral is
∂F (y + λ)y ! !
y! ! − F = y! ! − (y + λ) 1 + y !2 = C
∂y 1 + y !2
or
(y + λ)
−! = C.
1 + y !2
Thus,
(y + λ)2 − C 2
y !2 = .
C2
Solving by separation gives
dy dx
# #
! =±
2
(y + λ) − C 2 C
and a good substitution is y + λ = C cosh α or dy = C sinh α dα and so
$x %
α=± +D .
C
The solution is y = C cosh[(x/C) + D] − λ.
(c) Sorry the question was meant to say x = −a and x = a to make the boundary conditions
easier to solve. The boundary conditions x = −a and x = a can be used to eliminated D
and λ. Even though λ was introduced to control L at this stage we may use it to satisfy
the other boundary conditions. Hence we get
x a
y = C cosh[ ] − C cosh[ ] + h.
C C
The final remaining unknown C is determined by the length of the catenary
# a " # a
2 x x x a
#
L = ds = 1 + sinh dx = cosh dx = [C sinh ]ax=−a = 2C sinh .
x=−a C x=−a C C C
Unfortunately, the final result cannot be in a form more explicit than that.
15
4. Calculating ds in a different coordinate system
Cylindrical polar coordinates are defined by
x = ρ cos φ
y = ρ sin φ
z = z
Having warmed up with that calculation, repeat with spherical polar coordinates which are
defined by
x = r sin θ cos φ
y = r sin θ sin φ
z = r cos θ
(a) This is a simple application of the product rule dx = dρ cos φ − ρ sin φdφ.
(b) dy = dρ sin φ + ρ cos φdφ.
(c) Now
dx2 + dy 2 = (dρ cos φ − ρ sin φdφ)2 + (dρ sin φ + ρ cos φdφ)2
The cross terms cancel so
ds2 = (dr sin θ + r cos θdθ)2 + r 2 sin2 θdφ2 + (dr cos θ − r sin θdθ)2
16
5. Geodesics on a Sphere
The equation of a sphere in spherical polar coordinates is particularly simple: it is r = a, where
a is a constant.
(a) Starting with ds in spherical polar coordinates, write down the simplified form of ds when
r = a is a constant.
(b) Use this expression for ds to write down an integral that represents the distance between
two points connected by a path that lies on the surface of a sphere. Write the integral in
the form where φ is a function of θ.
(c) Write down a first integral for this integrand.
(d) Show by substitution that
φ − φ0 = sin−1 [α cot θ]
satisfies the first integral, where φ0 and α are two independent constants.
(e) The equation of a plane through the origin is Ax + By + Cz = 0. Rewrite this equation in
spherical polar coordinates. Rearrange the equation to make it look like the solution above
and find α and φ0 in terms of A, B and C.
(f) Thus give a simple geometric description and method of finding geodesics on a sphere.
Solution
sin2 θ φ!
! =C
1 + sin2 θ φ! 2
or ' (
C
φ! = ± ! .
sin θ sin2 θ − C 2
(d) Direct differentiation and some algebra, yields the result.
(e) In spherical coordinates this becomes
The r cancels, the cos θ can be moved to the other side and both sides divided by sin θ to
give
A cos φ + B sin φ = −C cot θ
Trig identities can be used to rewrite the left hand side as
!
A2 + B 2 sin(φ − φ0 ) = −C cot θ
√
where φ0 = − tan−1 (A/B) and α = −C/ A2 + B 2 .
17
(f) In other words, the curve with the shortest distance lies simultaneously on the surface of a
sphere AND on a plane through the origin. The intersection of such a plane and a sphere
is called a great circle.
Solution
(a)
ds2 = dx2 + dy 2 + dz 2
(b) !
F (x, y, z, y ! , z ! ) = 1 + y !2 + z !2 .
y!
! = C1 .
1 + y !2 + z !2
z!
! = C2 .
1 + y !2 + z !2
y !2 z !2 !
! +! − 1 + y !2 + z !2 = C3
1 + y !2 + z !2 1 + y !2 + z !2
(a) Determine the integrand F (t, x, y, ẋ, ẏ) for Eq. (1.6) when both x and y are functions of
some parameter t.
(b) Write down all the first integrals. Are any surprising?
(c) Show that the parametric form of the cycloid given in the text solves these equations.
(d) Find a different pair of functions x(t) and y(t) that also solve the same equations.
(e) Why should this be regarded as the same solution?
18
Solution
(a) !
ẋ2 + ẏ 2
F (t, x, y, ẋ, ẏ) = √
h−y
(b) The integrand is independent of x so
∂F ẋ
=√ ! =C
∂ ẋ h − y ẋ2 + ẏ 2
α2 − α2 cos(2t)
! =C
α sin(t) 2α4 − 2α4 cos(2t)
which simplifies to
1 − cos(2t)
! =C
α sin(t) 2 − 2 cos(2t)
or
1
√ =C
α 2
(d) You can let θ equal any function of t and the equations will still be satisfied. This is because
there are an equal number of dots in the top and bottom of the LHS of the first integral.
(e) The parametric form of a curve is not unique, but they all represent the same curve.
19