Geometry of the 2-sphere

October 28, 2010

The metric

The easiest way to find the metric of the 2-sphere (or the sphere in any dimension) is to picture it as embedded in one higher dimension of Euclidean space,
then restrict to constant radius.
The 3-dim Euclidean metric in spherical coordinates is
ds2 = dr2 + r2 d2 + r2 sin2 d2
so restricting to
r = R = const.
gives
ds2 = R2 d2 + R2 sin2 d2
This is the metric we will study. As a matrix,
 2

R
0
gij =
0 R2 sin2
with inverse
g ij =

1
R2

1
R2 sin2

There are more intrinsic ways to get this metric. One approach is to specify
the symmetries we require three independent rotations. There are techniques
for finding the most general metric with given symmetry, so we can derive this
form directly. Alternatively, we could ask for 2-dim spaces of constant curvature. Computing the metric for a general 2-geometry, then imposing constant
curvature gives a set of differential equations that will lead to this form.

Curvature: a plan

One definition of curvature starts by carrying a vector by parallel transport

around a closed loop. In general, the vector returns rotated from its original
direction. The difference between this angle and the angle expected in a flat
geometry is called the angular deficit. Next, calculate the area enclosed in the
1

loop. Then the curvature at point Q is the limit of the angular deficit per unit
area, as the loop shrinks to Q and the area to zero.
Explicitly, consider a closed curve C : R M2 , with tangent vector

u at each point. Let v0i be the components of an arbitrary vector at a point

P of the curve, and define the family of vectors v () around the curve as the
parallel transport of v0i along ui :
0

= ui Di v j


= ui i v j + v k jki

Since we have the metric, we can compute jki , so as soon as we specify the
curve, we can solve this equation for v i (). Then we can find the angle of
rotation, , by taking the inner product of v i (f inal ) with v i (initial ), where
we have
gij v0i v j (f inal )
cos =
gij v0i v0j
Then the angular deficit is
= 2
since transport around a closed loop in flat space will rotate the vector by 2.
For the area inside the loop, we integrate the 2-dim volume element.
This
p

is given by the square root of the determinant of the metric, g = det (gij ),
so that

g d d
A =

= R2
sin d d

3
3.1

Parallel transport on a non-geodesic circle

The curve

Now consider a circle around the sphere at constant 0 (e.g., constant latitude
on the surface of Earth). We can parameterize the curve by the angle , so the
curve is given by
xi = (0 , )
A vector tangent to the curve is
ti

dxi
d
(0, 1)

=
=

The length of this tangent vector is given by

l2

= gij ti tj
= R2 sin2 0
2

so the unit tangent is

ui =

3.2

1
(0, 1)
R sin 0

The connection

We also need the connection. We have

ijk
mjk

= g im mjk
1
(gmj,k + gmk,j gjk,m )
=
2

Since the metric only has one non-constant component, g , and that one depends only on , the only non-vanishing derivative of the metric is g, . This
means that the only non-vanishing ijk must have two s and one index. Using
the symmetry of the connection, we have

1
(g, + g, g, )
2

1
g,
2
= R2 sin cos
1
=
(g, + g, g, )
2
= R2 sin cos
=

Raising the first index is easy because the metric is diagonal. We have simply

=
=
=

=
=
=

3.3

g
1
R2 sin cos
R2 sin2
cos
sin
g
1
2 R2 sin cos
R
sin cos

Parallel transport

The parallel transport equation is

0

= ui Di v j


= ui i v j + v k jki


1
=
v j + v k jk
R sin 0
3

There are two components to check. For j = we have

0

1
v + v k k
R sin 0


1
v

+ v
R sin 0


v
1

v sin 0 cos 0
R sin 0

=
=
=

For j = ,
0



1
v + v k
k
R sin 0


v
1
cos 0
+v
R sin 0
sin 0

=
=

Therefore, we need to solve the coupled equations,

0

v
v sin 0 cos 0

v
cos 0
+ v

sin 0

Taking a second derivative of the first equation and substituting the second,
0

2 v
v

sin 0 cos 0
2

2 v
cos 0
+ v
sin 0 cos 0
2
sin 0
2 v
+ v cos2 0
2

=
=
=

Similarly, differentiating the second equation and substituting the first we have
0

=
=
=

2 v
v cos 0
+
2

sin 0
2 v
cos 0
+ v sin 0 cos 0
2

sin 0
2 v
+ v cos2 0
2

Each of these is just the equation for sinusoidal oscillation, so we may immediately write the solution,
v ()

v ()

A cos + B sin

C cos + D sin
4

where
= cos 0

Starting the curve at = 0, it will

close at = 2. Then for v we have the
initial condition v (0) = v0 , v0 , and from the original differential equations
we must have

v 
= v0 sin 0 cos 0
=0

cos 0
v 
= v0

=0
sin 0

These conditions determine the constants A, B, C, D to be

v ()

=
=

v ()

v0 sin 0 cos 0
sin

v0 cos + v0 sin 0 sin

sin
v0 cos v0
sin 0
v0 cos +

This gives the form of the transported vector at any point around the circle.

3.4

Norm of v

We have claimed that the norm of a vector is not changed by parallel transport.
We can check this in the current example. The initial squared norm of v (0) is
2

|~v0 | = R2 v0

2

+ R2 sin2 0 (v0 )

while the norm of

|~v |

=
=

=
=

2


2

2
2
sin
R
cos + sin 0 sin + R sin 0 v0 cos v0
sin 0

2

2
R2 v0 cos2 + 2v0 v0 sin 0 cos sin + (v0 ) sin2 0 sin2



2 sin2
sin
2
+ v0
+R2 sin2 0 (v0 ) cos2 v0 v0 cos
sin 0
sin2 0


2
2
R2 v0 cos2 + 2R2 v0 v0 sin 0 cos sin + R2 (v0 ) sin2 0 sin2

2
2
+R2 (v0 ) sin2 0 cos2 R2 v0 v0 sin 0 cos sin + R2 v0 sin2



2


2
R2 v0
cos2 + sin2 + R2 (v0 ) sin2 0 sin2 + cos2

2
2
R2 v0 + R2 (v0 ) sin2 0
2

v0

v0

= |~v0 |

which, as claimed, is independent of .

Now we turn to the calculation of the curvature.
5

3.5

Curvature of the 2-sphere

We need the angular deficit and the area of the sphere enclosed by the circular
path.
3.5.1

The angular deficit

The angular deficit is given by

= 2
where the angle of rotatoin, , is given by
cos

gij v0i v j (f inal )

gij v0i v j (2)

gij v0i v0j

gij v0i v0j

The inner product in the numerator is

gij v0i v j (2)

=
=
=
=

R2 v0 v2
+ v0 v2
sin2 0





2
2


sin
sin 0
R
v0 v0 cos + v0 v0 sin 0 sin + v0 v0 cos v0
sin 0


2
R2 v0 v0 cos + v0 v0 sin 0 sin + (v0 ) sin2 0 cos v0 v0 sin 0 sin

2

2
R2 v0 + (v0 ) sin2 0 cos
2

= |~v0 | cos
The angle of rotation is therefore = = 2 cos 0 . Therefore, the angular
deficit is

3.5.2

2 (1 cos 0 )

The area enclosde by the loop

The area enclosed by the loop is

= R

=
=

d
0

d g
0

sin d

2
2R cos |00
2R2 (cos 0 1)
2R2 (1 cos 0 )

3.5.3

The curvature

The curvature is now given by the limit as we shrink the loop to a point,

A
2 (1 cos 0 )
= lim
0 0 2R2 (1 cos 0 )
1
=
R2
This increases as the sphere shrinks, which indeed makes the curvature greater.
Curvature =

lim

0 0

Comparison with the Riemann curvature tensor

We can also compute the curvature using the Riemann curvature tensor. We
already have the connection,

cos
sin
sin cos

so it is straightforward to compute the curvature using

i
Rjkm
= ijm,k ijk,m + ink njm inm njk

Since Rijkm = Rjikm = Rijmk , there is only one independent component.

All of the rest follow from the symmetries of the curvature tensor. We can
compute any one non-vanishing component. Choose

, , + n n n n

( sin cos ), 0 + 0 n





cos

2
2
= cos + sin ( sin cos )
sin

sin2

The full curvature tensor may be written in terms of the metric and Kronecker
delta by including all the necessary symmetries,


1
i
i
gjk
Rjkm
= 2 ki gjm m
R
Then check that


1

R
=
g g
2
R
1
=
g
R2
= sin2
7

Since the expression in terms of the metric has all the right symmetries, and the
value of the one independent component is correct, it gives the full curvature
tensor.
We may find the Ricci tensor by contraction:
Rjm

=
=
=

i
Rjim


1
i
ii gjm m
gji
2
R
1
(2gjm gjm )
R2
1
gjm
R2

The Ricci scalar is the contraction of this. Using the inverse metric,
R

=
=

g jm Rjm
1 jm
g gjm
R2
2
R2

which differs from our angular deficit formula only by an overall constant.