GEOMETRY HW 4

CLAY SHONKWILER
1.5.10
Consider the map
(t) =
_
_
_
(t, 0, e
1/t
2
), t > 0
(t, e
1/t
2
, 0), t < 0
(0, 0, 0), t = 0
(a) Prove that is a dierentiable curve.
Proof. If we denote (t) = (x(t), y(t), z(t0), then it is clear that x(t) is
dierentiable, with x

(t) = 1 for all t. Now, y(t) and z(t) are certainly

dierentiable for t ,= 0. To check that they are dierentiable at zero as well,
it suces to show that
lim
t0
d
dt
(e
1/t
2
) = lim
t0
2
t
3
e
1/t
2
= 0.
To see this, we re-write and then apply LHopitals Rule twice:
lim
t0
2
t
3
e
1/t
2
= lim
t0
2
t
3
e
1/t
2
= lim
t0
6
t
4
2
t
3
e
1/t
2
= lim
t0
3
t
e
1/t
2
= lim
t0
3
t
2
2
t
3
e
1/t
2
= lim
t0
3t
2e
1/t
2
= 0.
Hence, we have agreement at t = 0, so is dierentiable and

(t) =
_
(1, 0,
2
t
3
e
1/t
2
) t 0
(1,
2
t
3
e
1/t
2
, 0) t 0

(b) Prove that is regular for all t and that the curvature k(t) ,= 0, for
t ,= 0, t ,=
_
2/3, and k(0) = 0.
1
2 CLAY SHONKWILER
Proof. As we can see from the explicit solution for

(t) given above, [

(t)[
1 for all t, so is a regular curve. Furthermore, since
d
dt
_
2
t
3
e
1/t
2
_
=
_
4
t
6

6
t
4
_
e
1/t
2
,
we see that

(t) =
_
_
_
_
0, 0,
_
4
t
6

6
t
4
_
e
1/t
2
_
t 0
_
0,
_
4
t
6

6
t
4
_
e
1/t
2
, 0
_
t 0
Hence, so long as t ,= 0 and
4
t
6

6
t
4
,= 0
(i.e. t ,=
_
2/3), then we see that
k(t) = [

(t)[ =

__
4
t
6

6
t
4
_
e
1/t
2
_
2
=
_
4
t
6

6
t
4
_
e
1/t
2
,= 0.
Furthermore, a LHopital argument similar to that given in (a) above demon-
strates that k(0) = 0.
(c) Show that the limit of the osculating planes as t 0, t > 0, is the
plane y = 0 but that the limit of the osculating planes as t 0, t < 0, is
the plane z = 0.
Proof. The osculating plane is the plane spanned by

(t) and

(t)
k(t)
, so its
normal vector is given by

(t)

(t)
k(t)
.
Note that, as t 0,

(t) (1, 0, 0). When t > 0,

(t)
k(t)
=
1
k(t)
(0, 0, k(t)) = (0, 0, 1),
so

(t)

(t)
k(t)
(1, 0, 0) (0, 0, 1) = (0, 1, 0).
If (x, y, z) is a vector in the osculating plane, then
0 = (x, y, z), (0, 1, 0)) = y,
so we see that the limit of the osculating planes is given by y = 0.
On the other hand, when t < 0,

(t)
k(t)
=
1
k(t)
(0, k(t), 0) = (0, 1, 0),
so

(t)

(t)
k(t)
(1, 0, 0) (0, 0, 1) = (0, 0, 1).
GEOMETRY HW 4 3
If (x, y, z) is a vector in this osculating plane, then
0 = (x, y, z), (0, 0, 1)) = z
so we see that the limit of the osculating planes is given by z = 0.
(d) Show that can be dened so that 0, even though is not a
plane curve.
Proof. Recall from the denition that
n(t) =

(t)
[

(t)[
and
b

(t) = n.
Also, b is the normal vector to the osculating plane, so we see that, from
part (c) above, for 0 < t <
_
2/3 and t >
_
2/3,
b(t) = (0, 1, 0),
meaning that b

(t) = (0, 0, 0) on this interval. On the other hand, again

using our results from part (c), we know that, for
_
2/3 < t < 0 and
t <
_
2/3,
b(t) = (0, 0, 1),
meaning that b

(t) = (0, 0, 0) on these intervals. Hence, we see that, any-

where t ,= 0, t ,=
_
2/3,
(0, 0, 0) = b

(t) = (0, 1, 0) = (0, 0, 1),

which is to say that 0 except possibly at these three points. However,
since is continuous, we extend to these points by continuity to see that
0 for all t.
1.6.3
Show that the curvature k(t) ,= 0 of a regular parametrized curve : I
R
3
is the curvature at t of the plane curve , where is the normal
projection of over the osculating plane at t.
Proof. We know that k(s) = [

(s)[, so it suces to show that

k(s) = [

(s)[ = [( )

(s)[
for all t I. Now, the osculating plane P
s
0
at s
0
is simply the plane given
by the span of t(s
0
) =

(s
0
)
|

(s
0
)|
and n(s
0
) =

(s
0
)
|

(s
0
)|
. Since t and n are both
unit vectors and are necessarily orthogonal, we know that the projection of
onto this plane is given by
(s) = t(s
0
), (s))t(s
0
) +n(s
0
), (s))n(s
0
).
Hence,
( )

(s) = t,

(s))t +n,

(s))n = n,

(s))n =

(s
0
),

(s))
[

(s
0
)
n.
4 CLAY SHONKWILER
Hence, when s = s
0
,
( )

(s
0
) =

(s
0
),

(s))
[

(s
0
)
n =
[

(s
0
)[
2
[

(s
0
)[
n.
Therefore,
[( )

(s
0
)[ = [[

(s
0
)[n[ = [

(s
0
)[ = k(s
0
).
Hence, we see that the curvature of is the curvature of its projection onto
the osculating plane.
3.2.5
Show that the mean curvature H at p S is given by
H =
1

_

0
k
n
()d,
where k
n
() is the normal curvature at p along a direction making an angle
with a xed direction.
Proof. By denitions, H =
k
1
+k
2
2
, where k
1
and k
2
are the maximum and
minimum normal curvatures, respectively. Also, k
n
() = k
1
cos
2
+k
2
sin
2
.
Therefore,
1

0
k
n
()d =
1

0
k
1
cos
2
+ k
2
sin
w
d
=
1

0
k
1
(1 sin
2
) + k
2
sin
2
d
=
1

0
k
1
+ sin
2
(k
2
k
1
)d
=
1

_
[k
1
]

0
+ (k
2
k
1
)
_

0
sin
2
d

= k
1
+
k
2
k
1

0
1cos 2
2
d
= k
1
+
k
2
k
1

_
/2 +
sin 2
4

0
= k
1
+
k
2
k
1
2
=
k
1
+k
2
2
= H.

3.2.8
Describe the region of the unit sphere covered by the image of the Gauss
map of the following surfaces:
(a) Paraboloid of revolution z = x
2
+ y
2
.
(b) Hyperboloid of revolution x
2
+ y
2
z
2
= 1.
Answer: Consider the map f : R
3
R given by f(x, y, z) = x
2
+ y
2

z
2
. Then df
p
= (f
x
, f
y
, f
z
) = (2x, 2y, 2z), which is nonzero so long as
(x, y, z) ,= 0. Since f(0) ,= 1, we see that 1 is a regular value of f. Hence,
we see that the hyperboloid S is a regular surface. Now, note that
[[f[[ =
_
4x
2
+ 4y
2
+ 4z
2
= 2
_
x
2
+ y
2
+ z
2
= 2
_
(1 + z
2
) + z
2
= 2
_
(1 + 2z
2
.
GEOMETRY HW 4 5
Since the hyperboloid is a regular surface, we know that the normal vector
eld is given by
(x
0
, y
0
, z
0
) = N =
f(p)
[[f(p)[[
=
(2x, 2y, 2z)
2

1 + 2z
2
=
_
x

1 + 2z
2
,
y

1 + 2z
2
,
z

1 + 2z
2
_
.
Hence, we see that when z is constant, the z-coordinate of N is also constant
and the x- and y-coordinates of N are proportional to x and y, respectively.
In other words, if a vector (x
0
, y
0
, z
0
) is in the image of the Gauss map, then
the entire latitudinal cirle given by
(x, y, z) S
2
: z = z
0

will also be in the image. Hence, to describe the image, we need only nd
the maximum and minimum possible values of z
0
, which will, in turn, give
the image of the Gauss map as a band around the equator of the sphere.
To do so, we can assume that x = 0. Then, since they are the coordinates
of a point lying on the hyperbola, y and z satisfy the following relation:
y
2
z
2
= 1.
Note that this implies that y
2
1. In other words, z =
_
y
2
1, so we
see that
z
0
=

_
y
2
1

_
2y
2
1
As y , we see that z
0

2/2 and that, furthermore, if we consider

a single branch of the above expression, the possible values for z
0
start at
zero and increase or decrease monotonically. Hence, z
0
[0,

2/2). Based
on our above argument that concluded that this solution holds for all x and
y, we see that, in fact, the image of the hyperboloid under the Gauss map
is given by
(x, y, z) S
2
: [z[ <

2/2.

(c) Catenoid x
2
+ y
2
= cosh
2
z.
Answer: Consider the function f : R
3
R given by
f(x, y, z) = x
2
+ y
2
cosh
2
z.
Then, recalling that cosh z =
1
2
(e
z
+ e
z
), we see that
f = (f
x
, f
y
, f
z
) = (2x, 2y,
1
2
(e
2
z e
2z
)).
Then we see that the only critical value of f occurs at the point (0, 0, 0).
Now, f(0, 0, 0) = 1/2, so 0 is a regular value of f. Hence, the catenoid S
that we are considering is f
1
(0), where 0 is a regular value. Therefore, S
is a regular surface.
6 CLAY SHONKWILER
Thus, the normal vector eld is given by
(x
0
, y
0
, z
0
) = N
p
=
f(p)
||f(p)||
=
(2x,2y,
1
2
(e
2z
e
2z
))
q
4x
2
+4y
2
+
1
4
(e
4z
+e
4z
2)
=
(2x,2y,
1
2
(e
2z
e
2z
))
q
4 cosh
2
z+
1
4
(e
4z
+e
4z
)
1
2
)
=
(2x,2y,
1
2
(e
2z
e
2z
))
q
4 cosh
2
z+
1
2
cosh 4z
1
2
.
Hence, as in part (b) above, when z is constant then z
0
is constant and x
0
and y
0
are proportional to x and y, respectively. In other words, as before,
if (x
0
, y
0
, z
0
) is in the image of the Gauss map, then the entire latitudinal
circle given by
(x, y, z) S
2
: z = z
0

will also be in the image. Therefore, we need only determine the possible
values of z
0
in order to completely describe the image of the Gauss map.
Now, let (x, y, z) S and dene :=
1
q
4 cosh
2
z+
1
2
cosh 4z
1
2
. Then
(x
0
, y
0
, z
0
) =
_
2x
t
,
2y
z
,

1
2
(e
2z
e
2z
)
t
_
.
Then
1 = x
2
0
+ y
2
0
+ z
2
0
=
4x
2
t
2
+
4y
2
t
2
+ z
2
0
= 4
cosh
2
z
t
2
+ z
2
0
.
In other words,
z
0
=
_
1
1
t
2
cosh
2
z
_1
2
.
Now, we see that, it must be the case that
[z
0
[ < lim
z
_
1
1
t
2
cosh
2
z
_1
2
= lim
z
_
1
cosh
2
z
4 cosh
2
z+
1
2
cosh 4z
1
2
_1
2
= lim
z
_
1
1
4
(e
2z
+e
2z
+2)
e
2z
+e
2z
+2+
1
4
(e
4z
e
4z
)
1
2
_
1
2
= lim
z
_
1
1
4
_
e
2z
+e
2z
+2
e
2z
+e
2z
+
3
2
+
1
4
(e
4z

4z
)
__1
2
= lim
z
_
1
1
4
_
1+e
4z
+2e
2z
1+e
4z
+
3
2
e
2z
+
1
4
(e
2z
e
6z
)
__1
2
= lim
z
_
1
1
4
1
1+
1
4
e
2z
_1
2
= (1 0)
1
2
= 1,
where we arrived at the fourth equality by multiplying numerator and de-
nominator by e
2z
. Now, since z can be any element of R, we see that z
0
GEOMETRY HW 4 7
achieves every value on the sphere except z
0
= 1 and z
0
= 1. Therefore,
the image of the Gauss map is all of S
2
except the north and south poles.

1
Suppose F : R
n
R
k
is a C

-function and r R
k
is a regular value of
F; that is, for each p F
1
(r) w have that dF
p
: T
p
R
n
T
F(p)
R
k
is onto.
Show that for each p S = F
1
(r) there exists V S open neighborhood
of p in S and : U R
nk
V such that
(a) : U V R
n
is dierentiable.
(b) : U V is a homeomorphism.
(c) d
q
: T
q
R
nk
T
(q)
R
n
is injective for any q U.
Proof. Denote r = (r
1
, . . . , r
k
) and let p F
1
(r). Since r is a regular value,
the matrix
dF
p
=
_
_
_
_
_
_
F
1
x
1
F
1
x
2

F
1
xn
F
2
x
1
F
2
x
2

F
2
xn
.
.
.
.
.
.
.
.
.
F
k
x
1
F
k
x
2

F
k
xn
_
_
_
_
_
_
has a minor with nonzero determinant. Assume, without loss of generality,
that the rightmost k columns are such a minor. Now, dene H : R
n
R
n
by
H(x
1
, x
2
, . . . , x
n
) = (x
1
, x
2
, . . . , x
nk
, F
1
(x
1
, . . . , x
n
), . . . , F
k
(x
1
, . . . , x
n
)).
We will denote by (u
1
, . . . , u
n
) the elements in R
3
where H takes its values.
Now,
dH
p
=
_
_
_
_
_
_
_
_
_
_
_
_
1 0 0 . . . 0 0 . . . 0
0 1 0 . . . 0 0 . . . 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 0 . . . 1 0 . . . 0
F
1
p
1
F
1
p
2
. . . . . .
F
1
pn
.
.
.
.
.
.
F
k
p
1
F
k
p
2
. . . . . .
F
k
pn
_
_
_
_
_
_
_
_
_
_
_
_
.
Then
det dH
p
=

F
1
p
nk+1
. . .
F
1
pn
.
.
.
.
.
.
.
.
.
F
k
p
nk+1
. . .
F
k
pn

,= 0.
By the Inverse Function Theorem, then, there exists a neighborhood V

of
p and W of F(p) such that H : V

W is a dieomorphism. Hence, the

coordinate functions
x
1
= u
1
, . . . , x
nk
= u
nk
, x
nk+1
= g
1
(u
1
, . . . , u
n
), . . . , x
n
= g
k
(u
1
, . . . , u
n
)
8 CLAY SHONKWILER
of H
1
are dierentiable. In particular, for i = 1, . . . , k,
x
nk+i
= g
i
(u
1
, . . . , u
nk
, r
1
, . . . , r
k
) = h
i
(u
1
, . . . , u
nk
)
are dierentiable functions dened on

(V

), where

is the projection of
F
1
(r) onto the set
U

= W (u
1
, . . . , u
nk
, r
1
, . . . , r
k
).
Note that
H(F
1
(r) V ) = U

,
so, in fact,
H[
U
: U

F
1
(r) V

= V
is a dieomorphism. Now, if we let : R
n
R
n
be the projection onto the
set
A = (x
1
, . . . , x
nk
, 0, 0, . . . , 0,
then it is clear that [
U
is a dieomorphism of U

with [
U
(U

). Dene
U := [
U
(U

). Then : U V given by
(u
1
, . . . , u
nk
) = (u
1
, . . . , u
nk
, h
1
(u
1
, . . . , u
nk
), . . . , h
k
(u
1
, . . . , u
nk
))
is simply the composition = H[
U

1
U

. Since both of the functions in

the composition are dieomorphisms, so is . Hence, : U V precisely
fullls the desired requirements.
DRL 3E3A, University of Pennsylvania
E-mail address: shonkwil@math.upenn.edu