MAT598 Intro to Geometric Structures - Homework Set 04

Joseph Wells
Arizona State University
October 15, 2015

[RM] 5.4. Prove the following proposition.

, g) is an isometry.
Proposition (Naturality of the Riemannian Connection). Suppose : (M, g) (M

a. takes the Riemannian connection of g to the Riemannian connection of g, in the sense that
X ( Y ) .
(X Y ) =

b. If is a curve in M and V is a vector field along , then

t ( V ) .
Dt V = D
c. takes geodesics to geodesics: if is the geodesic in M with initial point p and initial velocity V , then is
with initial point (p) and initial velocity V .
the geodesic in M

Proof.
a. Begin by defining the following map:
: X(M ) X(M ) X(M )




(X, Y ) 7 1

(
Y
)
.

is a linear connection (and in fact, well even have the audacity to name it something
We first prove that
crazy like the pullback connection). Let f, g C (M ) and V, W X(M ). Then





Y = 1

(
Y
)

(f
V
+gW
)

f V +gW


(f ) V +(g) W ( Y )
= 1



1
1
W ( Y )
= (f ) V ( Y ) + (g 1 )





V ( Y ) + 1
W ( Y )
= 1
(f 1 )
(g 1 )





V ( Y ) + g1
W ( Y )
= f 1






= f
Y + g
Y,
V

is C -linear in the first argument. For a, b R, we have

so




X ( (aV + bW ))

(aV + bW ) = 1

X



= 1

(a
V
+
b
W
)
X



X ( V ) + b
X ( W )
= 1
a

2




X ( W )
X ( V ) + b1

= a1






= a
(V ) + b
(W ),
X

is R-linear in the second argument. Lastly, for f C (M ),

so





X ( (f Y ))

(f Y ) = 1

X



1

= 1

(f

)
Y



1
X ( Y ) + X(f 1 ) Y
= (f 1 )





X ( Y ) + 1 X(f 1 ) Y

= f 1




= f
Y + 1
X(f 1 ) Y


X

=f
Y + 1
(Xf 1 ) Y


X

=f
Y + (Xf )Y,

is a connection on M . Let be the torsion tensor associated with

and the torsion tensor
and thus
We then have that
associated with .

X Y
Y X [X, Y ]
(X, Y ) =




X Y 1
Y X 1
= 1

([ X, Y ])




= 1

[
X,

Y
]

= 1
( X, Y )) = 1
(
(0) = 0,
is symmetric. Moreover, since
so




 
= 1
g = 1 (
1 ) g = 1
g 0,
g

is a symmetric linear connection that is compatible with the metric.

is compatible with g. Therefore

= . It follows then that X Y =
X ( Y ).
By uniqueness of Riemannian connections,

be the covariant derivative along , and define a map

b. Let be a curve in M , let D
t : X() X()
D


t V .
V 7 1
D

Then, for V X(),



t V = 1
t V
D
D

!
Pt1
V (t) V (t0 )
0t
=
lim
tt0
t t0



1  1 1
= lim
Pt0 t V (t) 1
( V (t0 ))
tt0 t t0
P 1 V (t) V (t0 )
= lim t0 t
= Dt V,
tt0
t t0
1

t ( V ).
so Dt V = D
c. Let be a geodesic in M . Then Dt = 0, and
t d ( ) = (Dt )
D
= 0,
dt
. Since (0) = p and (0)
so is a geodesic in M

= V , then it follows that ( )(0) = (p) and


d

= V.
dt t=0 ( ) = d(0) ((0))

[RM] 5-2. Let M R3 be a surface of revolution, parametrized as in Exercise 3.3. It will simplify computations if we
assume that the curve (called the generating curve for the surface) is unit speed.
a. Compute the Christoffel symbols of the induced metric in (, t) coordinates.
b. Show that each meridian { = 0 } is a geodesic on M .
c. Determine the necessary and sufficient conditions for a latitude circle {t = t0 } to be a geodesic.
Proof.
a. Given our parametrization (, t) = (a(t) cos , a(t) sin , b(t)), we have
dx2 = a 2 (t) cos2 dt2 + a2 (t) sin2 d2 a(t)a(t)
cos sin dt d,
dy 2 = a 2 (t) sin2 dt2 + a2 (t) cos2 d2 + a(t)a(t)
cos sin dt d,
2
2
2
dz = b (t) dt .
Supposing has unit speed, a 2 (t) + b 2 (t) = 1, so we have
g = dx2 + dy 2 + dz 2 = dt2 + a2 (t) d2 .
The components of the metric are then
g = dt2 ( , ) + a2 (t)2 ( , ) = a2 (t),
gt = gt = dt2 ( , t ) + a2 (t) d2 ( , t ) = 0,
gtt = dt2 (t , t ) + a2 (t) d2 (t , t ) = 1,
and the components of the inverse are
g = dt2 ( , ) + a2 (t)2 ( , ) =

1
,
a2 (t)

g t = g t = dt2 ( , t ) + a2 (t) d2 ( , t ) = 0,
g tt = dt2 (t , t ) + a2 (t) d2 (t , t ) = 1.
So we can finally compute the Christoffel symbols:
=

X 1
g k ( gk + gk k g ) = 0,
2

k{,t}

t =

X 1
g tk ( gk + gk k g ) = a(t)a(t),

k{,t}

t = t =

X 1
a(t)

g k ( gtk + t gk k gt ) =
,
2
a(t)

k{,t}

tt = tt =

X 1
g tk ( gtk + t gk k gt ) = 0,
2

k{,t}

tt =

X 1
g k (t gtk + t gtk k gtt ) = 0,
2

k{,t}

ttt =

X 1
g tk (t gtk + t gtk k gtt ) = 0.
2

k{,t}

b. Let : I M be a unit speed meridian curve in M given by (s) = ((s), t(s)) = (0 , t(s)). We show that
satisfies the geodesic equation, ie, that
00 (s) + 0 (s)0 (s) + 0 (s)t0 (s)t + t0 (s)0 (s)t + t0 (s)t0 (s)tt = 0
and that
t00 (s) + 0 (s)0 (s)t + 0 (s)t0 (s)tt + t0 (s)0 (s)tt + t0 (s)t0 (s)ttt = 0

Since the first coordinate is constant, 0 (s) = 00 (s) = 0, and by part (a), = tt = tt = ttt = 0. Thus we
need only show that t00 (s) = 0. Indeed,
2

1 = | 0 (s)|2 = (t0 (s)) ,

so t00 (s) = 0, as desired.

, g) are Riemannian manifolds, a map : M M

is a local isometry if each point p M
[RM] 5-7. If (M, g) and (M

has a neighborhood U such that |U is an isometry onto an open subset of M . Suppose M is connected, and suppose
are local isometries such that for some point p M , (p) = (p) and = at p. Show that .
, : M M
Thanks to Lindsey K. Gamard for the idea about proving openness of N .
Proof. Let N = {p M : (p) = (p)}. N is nonempty by assumption, and we claim furthermore that N is clopen.
To see that N is open, fix some p N and suppose dp = dp (such a points exists by assumption). Let U be a
neighborhood of p in M and V a neighborhood of the origin in Tp M such that expp : V U is a diffeomorphism.
be a neighborhood of (p) in M
and V a neighborhood of the origin in T (p)M
such that exp(p) : V U

Similarly, let U
1
is a diffeomorphism. Let X = U (U ) U , which is an open neighborhood of p. By naturality of the exponential
map, we have the following commutative diagram:
V Tp M

dp =dp

V T(p) M

expp

U M

exp(p) =exp(p)

It follows then that, for any x X,

1
(x) = exp(p) dp exp1
p (x) = exp(p) dp expp (x) = (x),

hence X is an open neighborhood of p in N . If we can show that this is true for every p N , we will have that N is
open, from which it follows that N = M . [Upon typing this, I realize that Im stuck and unsure how to continue from
here. In the problem statement, we only know that dp = dp for a particular p M , so its not clear how to extend
this procedure to all points in N .]
To see that N is closed, let pn be a sequence of points in N converging to p M . By smoothness of both and ,
lim (pn ) = (p)

pn p

and

lim (pn ) = (p).

pn p

By uniqueness of limits we thus have that (p) = (p), so p N whence N is closed.