Homework 7.

Solutions
1 Show that great circles are geodesics on sphere. Do it
a) using the fact that for geodesic, acceleration is orthogonal to the surface.
b ) using straightforwardly equations for geodesics
c) using the fact that geodesic is shortest.
a) See Lecture notes, subsection 3.2.3
2

i dx
b) ) The straightforward equations for geodesic: ddtx2 + dx
dt km dt = 0 are just equation of motion for
free Lagrangian on the Riemannian surface. Hence in the case of sphere they are equations of motion of the
2 2
2
2
2
Lagrangian of free particle on the sphere is L = R +R2 sin . Its equations of motion are second order
differential equations

(
sin cos 2 = 0

+ 2cotan = 0

with intitial conditions

t=0 = a
(t)|t=0 = 0 , (t)|
(t)|t=0 = 0 , (t)|

t=0 = b

(1)

for geodesics (t), (t) starting at the initial point p = (0 , 0 ) with initial velocity v0 = a
+ b
. (All

Christoffel symbols vanish except = sin cos , and = = cotan .)

This differential equation is not very easy to solve in general case. On the other hand use the fact that
rotations are isometries of the sphere. Rotate the sphere in a way such that the initial point transforms
to the point the point 0 = 2 , 0 = 0 and then rotate the sphere with respect to the axis 0X such that
-component of velocity becomes zero. We come to the same differential equation but with changed initial
conditions:
(

t=0 = 0
(t)|t=0 = 2 , (t)|
sin cos 2 = 0
(2)
with intitial conditions

(t)|t=0 = 0, (t)|

t=0 = 0
+ 2cotan = 0

where we denote by 0 the magnitude of initial velocity. One can easy check that the functions

(t) = 2
(t) = 0 t

are the solution of the differential equations for geodesic with initial conditions (2). Hence this is geodesic

passing through the point 0 = 2 , = 0 with initial velocity 0

. We see that this geodesic is the equator
of the sphere. We proved that an arbitrary geodesic after applying the suitable rotation is the great-circle
equator. On the other hand an equator is the great circle (the intersection of the sphere x2 + y 2 + z 2 = R2
with the plane z = 0) and the rotation transforms the equator to the another great circle. Hence all arcs of
great circles are geodesics and all geodesics are arcs of great circles.
c) See the lecture notes the subsection 3.4.1 (Again on geodesics on sphere and on Lobachevsky plane.)
2 Consider in E3 a vector X =

attached at the point p: x =

R
2 ,y

= 0, z =

3R
.
2

Consider also a sphere x2 + y 2 + z 2 = R2 in E3 and the following two curves: a curve C1 which is the
intersections of this sphere with plane y = 0 and a curve C1 which is the intersections of this sphere with
the plane z = 23R . Both curves C1 , C2 pass through the point p.
Show that the vector X is tangent to the sphere and express this vector in spherical coordinates.

Describe the parallel transport of the vector X along these closed curves.
What will be the result of parallel transport of the vector X along these closed curves?
1

The vector X0 =

attached at the point p = x =

2

R
2 ,y

= 0, z =

3R
2

of the sphere x2 + y 2 + z 2 = R2

is tangent to this sphere since X0 (x + y + z R ) = 0:

X (x2 + y 2 + z 2 R2 )|p = 0 =
The vector X0 =

2
(x + y 2 + z 2 R2 )p = 2y|p = 0 .
y

attached at the point p is proportional to the vector

X0 = b0
. The spherical

coordinates of the point p are 0 = 6 (cos 0 = 23 ). Since the metric on the sphere is G = R2 (d+ sin2 d2 )
then calculating the length of the vector X0 in ambient space in Cartesian coordinates and on the sphere in
spherical coordinates we come to
s

3
1
2 3

2 3
2
2
2

|X0 | = = 1 = R sin 0 d b
= Rb
b=
=
, X0 =
=
.
y

2
R sin 0
3R
R sin 0
3R
Using the formula X = dX(t)
+ x k ikm X m = 0 for parallel transport we see that for an arbitrary
dt
curve (t), (t) on the sphere starting at the initial point (t0 ) = 0 , (t0 ) = 0 the parallel transport
X(t) = a(t) + b(t) of the initial vector X|t0 = a0 + b0 over the curve C is defined by the first order
differential equation:
(

a +
b = a sin cos b = 0
a(t0 ) = a0
with initial conditions
(3)
b = b + (a

b(t0 ) = b0
b +
a +
+ b)cotan
=0

(We use here the information about Christoffel symbols of sphere.) Apply this equation for both curves (For
the first curve there is another much more clear solution: see below.)
First consider the curve C1 . For the curve C1 in spherical coordinates: (t) = 0 + t, (t) = 0,
= 1, = 0. The differential equation (3) becomes:

a sin cos b = a = 0

b + (a
+ b)cotan
= b + bcotan 6 + t = 0

The solution of this equation is a(t) = 0, b(t) =

with initial conditions

1
R sin (t) ,

X =

R sin (t)

a(t0 ) = 0
b(t0 ) = b0 =

.
R sin(t+
6)

2 3
3R

(3a)

This answer

is evident without any calculation: Namely the vector X = y during transport along the curve C1 in the

ambient space E3 , remains tangent to the sphere. Hence X(t) = y

at any point of this circle. The vector
1

y at point (t) equals to R sin (t) . We come to the answer.

Note that during parallel transport along the curve C1 the final vector coincides with initial one.
For the curve C2 . This curve in spherical coordinates is (t) = 0 = 6 , (t) = t, = 0, = 1. Hence
the equation (3) for parallel transport becomes:

a sin cos b = a(t)

sin 0 cos 0 b(t) = 0
a(t0 ) = 0
with initial conditions
(3)
+ a(t)cotan 0 = 0
b(t)
b(t0 ) = b0
Here we have two first order differential equations. Introduce b0 (t) = sin 0 b(t). We have
(

or in other way

da(t)
0
dt cos 0 b (t)
0
db (t)
dt + cos 0 a(t)

=0
=0

i.e.

d
dt

a(t)
b0 (t)

= cos 0

d
(a + ib0 ) = i cos 0 (a + ib0 )
dt
2

0 1
1 0

a(t)
b0 (t)

The solution is
a(t) + ib0 (t) = eit cos 0 (a0 + ib00 ) = (cos(t cos 0 ) i sin(t cos 0 ))(a0 + ib00 )
In particular as a result of parallel transport along the closed latitude

a(t) + ib0 (t)

= e2i cos 0 (a0 + ib00 )
t=2

that is the vector rotates on the angle = 2 cos 0 .

We see that both answers for curves C1 and C2 are in coincidence with the formula of rotation of the
R
vector during parallel transport along the closed curve: = Kd. Indeed for the sphere K = R12 .
R
R
For the first curve C1 the first case = Kd = R12 2R2 = 2 and for the curve C2 = Kd =
1
2
R2 2R (1 cos 0 ) = 2(1 cos 0 )
3 Show that vertical lines x = a are geodesics (non-parameterised) on Lobachevsky plane.
We consider here the realisation of Lobachevsky plane (hyperbolic plane) as upper half of Euclidean plane
2
2
{(x, y): y > 0} with the metric G = dx y+dy
.
2
Consider second order differential equations defining geodesics with initial conditions such that horisontal velocity equals to zero: (we uses the information from Homework 6 or from Lecture notes about
Christoffels for Lobachevsky plane: xxx = 0, xxy = xyx = y1 , xyy = 0, yxx = y1 , yxy = yyx = 0, yyy = y1 .)

x 2xy y = 0

2
x 2
y + y yy = 0

x(t)t=t0 = x0 , x(t)
t=t0 = 0

y(t)
= y0 , y(t)
t=t = y 0
t=t
0

This equation has a solution and it is unique. One can see that if we put x(t) 0, i.e. curve is vertical then
2
2

we come to the equation y yy = 0. Solution of these equation gives curve x = x0 , y = y(t): y yy = 0.

The image of this curve clearly is vertical ray x = x0 , y > 0.
4 Show that the following transformations are isometries of Lobachevsky plane:
a) horizontal translation r r + a where a = (a, 0),
b) homothety: r r ( > 0),

c) inversion with the centre at the points of the line x = 0 :

(
xa
x0 = a + (xa)
ra
2 +y 2
ra+
where
a
=
(a,
0):
.
y
0
2
y = (xa)2 +y2
|r a|
2

remains invariant under these transformations.

We have to show that Riemannian metric G = dx y+dy
2
a) horizontal translation. If x x + a, y y then dx and dy do not change. Hence G is invariant
under horisontal translations.
2
2
2
2
2
2
2
dy 2
dx2+
= dx y+dy
b) Homothety. If x x, y y where > 0 is a constant then dx y+dy
2
2
y2
does not change too.
c) inversion with the centre at the points of the line y = 0. Since we proved that horisontal translation
is isometry it suffices to consider inversion with centre at the point x = y = 0:
0
x
x = x2 +y
r
2
r 2:
y
0
y = x2 +y2
|r|

This line is called absolute.

3

02

02

Now by straightforward calculations one can show that dx y+dy

= dx y+dy
.
2
02
To avoid the straightforward calculations consider coordinates r, : x = r cos , y = r sin then
dx2 + dy 2
dr2 + r2 d2
d2
1 dr2
1
d2
2
=
=
+
=
(d
log
r)
+
.
2
2
2
2
y2
r2 sin
sin r2
sin
sin
sin2
Under transformation of inversion in these coordinates does not change, r 1r , log r log 1r . It is
evident that in coordinates u = log r, metric does not change. Hence inversion (with centre at the point
x = y = 0) is isometry. Hence an inversion with a centre at the arbitrary point (a, 0) and with an arbitrary
radius is isometry, since horisontal translation and homothety are isometries.
(In complex cooridinates this is so called Mobius transformation transformation z z1 which you
learned in the course Hyperbolic Geometry.)
5 Show that upper arcs of semicircles (x a)2 + y 2 = R2 , y > 0 are (non-parametersied) geodesics.
You may do this exercise solving explicitly differential equations for geodesics, but it is much more nice
to use inversion (Mobius ) transformation studied in the previous exercise: Consider the inversion of the
Lobachevsky plane with the centre at the point x = a R, y = 0 (see the exercise above). This inversion
does not change Riemannian metric, it is isometry. Isometry transforms geodesics to geodesics. On the other
1
hand it transforms the semicircle (x a)2 + y 2 = R2 , y > 0 to the vertical ray x = a R + 2R
, y > 0. This
can be checked directly. On the other hand the vertical ray is geodesic. Hence the initial curve was the
geodesic too.
6 Let ABC be triangle formed by geodesics on the sphere of the radius R. Express the area of this
triangle via its angles.
Do the previous exercise for the triangle on the Lobachevsky plane.
We use
1) the Theorem from Lecture notes about the rotation of the vector during parallel transport along
R
closed curve C = D: = D Kd a
2) the fact that velocity vector is covariantly constant along geodesic
3) during parallel transport scalar product and hence the angle between two vectors does not change.
Let ABC be the triangle formed by arcs of geodesics. We denote by , , angles at the vertices A, B, C.
Let vector A0 be velocity vector at the vertex A along the arc AB. Under parallel transport it becomes
the vector A1 which is the velocity vector of the arc AB attached at the vertex B. Take the vector B0 which
is velocity vector at the vertex B along the arc BC. The vector B0 has the angle with the vector A1 .
Do the parallel transport of two vectors A1 and B0 along the arc of geodesic BC. We arrive at the vertex
C with the vector B1 which is the velocity vector along the curve BC at the vertex C and the vector A2
which has the angle with the vector B1 since angles do not change during parallel transport. Now take
the vector C0 which is velocity vector at the vertex C along the arc CA. The vector C0 has the angle
with the vector B1 and it has the angle + with the vector A2 . Now make the final parallel transport
along the arc of geodesic CA of three vectors: vector C0 , the vector B1 and the vector A2 . We come to
three vectors: vectors C1 , vector B2 and the vector A3 at the vertex A. The vector C0 will transform to
the vector C1 which is the velocity vector of the curve CA at the vertex A. The vector B2 has the angle
with the vector C2 and the vector A3 has the angle with the vector C1 . Hence the vector A3 has the
angle + with the vector C1 . The vector C1 has the angle with initial velocity vector A0 . We see
that after parallel transport along the geodesics the vector A0 becomes the vector A3 which has the angle
+ + with the initial vector A0 . Now using the Theorem we come to
Z
++ =
Kd
4ABC

For the sphere we have K =

1
R2 ,

hence:
Z

++ =

Kd =
4ABC

For the Lobachevsky plane K = 1 and

Z
++ =

Area of spheric triangle ABC

R2

Kd = (Area of spheric triangle ABC)

4ABC

7 Let X(t) be parallel transport of the vector X along the curve on the surface M embedded in E3 , i.e.
v X = 0, where v is a velocity vector of the curve C and Levi-Civita connection (induced connection) on
the surface. Compare this condition v X = 0 (for internal observer) with the condition for external observe
that for the vector X(t) dX(t)
is orthogonal to the surface 2) .
dt
We know that Levi-Civita connection equals to the connection : X Y = (can.flat
Y)tangent . Show
X
that the definitions of External and Internal Observers are equivalent.
Denote by v the velocity vector of curve C on the surface M . For external observer parallel transport
means that dX
dt = v X is colinear to the normal vector to the surface. This is equivalent to the fact that
tangential component equals to zero: v Xtangent = 0. According to definition of induced (Levi-Civita)
connection this means that v X = 0. Thus we prove that both observers have equivalent definitions.
8 Let r = r(t) be an arbitrary geodesic on the Riemannian manifold. Show that magnitudes I = gik x i x k
is preserved along geodesic.
Let r = r(t) be an arbitrary geodesic on the sphere. Show that magnitudes I = sin2 and E =
sin2 2 + 2
are preserved along geodesics.
2
Let r = r(t) be an arbitrary geodesic on Lobachevsky plane. Show that magnitudes I =
are preserved along geodesics.
There are two fundamental facts:
1. For Lagrangian L = L(xi , x i ) the magnitude

E=

L k
x L
x k

2
vx
y2

and E =

2
vx
+vy2
2y 2

is preserved on equations of motions. This magnitude is called energy. More in details this means that for
L
d
L
any xi = xi (t) which is the solution of Euler-Lagrange differential equations dt
= x
k the magnitude
x k

E(t) = E|xi (t) =

L k
x L |xi (t)
x k

is a constant. The proof is immediate: using Euler-Lagrange equations of motions we come

dE
d
=
dt
dt

L k
d
L
L k dL
L k
L k dL
dL dL
k
x

L
=
x

+
=
x

+
=

= 0.
x

x
k
k
k
k
k
x
dt x
x
dt
x
x
dt
dt
dt

if xi = xi (t) are solutions of Euler-Lagrange equations.

L
The second fundamental fact is that if Lagrangian L = L(xi , x i ) does no depend on variable xm ( x
m =
0) then the magnitude
L
pm =
xm
2)

We defined parallel transport in Geometry course using the second condition

5

is preserved on equations of motions, i.e. for any xi = xi (t) which is the solution of Euler-Lagrange differential
L
d
L
equations dt
= x
k the magnitude
x k
pm (t) = pm |xi (t) =

L
| i
x m x (t)

is a constant. The proof is even simpler than for the first fact: using Euler-Lagrange equations of motions
we come

L
L
d
dpm (t)
=
|xi (t) =
= 0.
m
dt
dt x
xm
The magnitude pm is called generalised momentum (corresponding to the coordinate xm )
Now it is very easy to solve the problems for sphere and Lobachevsky plane.
g x p x q
Geodesics are solutions of Euler-Lagrange equations of motion for Lagrangian L = pq 2 . The energy

E=

L k
x L
x k

= 2L L = L

is preserved. Hence the Lagrangian itself preserves on geodesics. We see that for sphere the Lagrangian
2
2
g x p x q
g x p x q
L = pq 2
= 21 (2 + sin2 2 ) is preserved and for Lobachevsky plane the Lagrangian L = pq 2
= x 2y+2y
is preserved.
The Lagrangian of the sphere L = 12 (2 + sin2 2 ) does not depend explicitly on , hence I = L
=
2
sin is preserved.
2
2
The Lagrangian of the Lobachevsky plane L = x 2y+2y does not depend explicitly on x, hence I = yx2 is
preserved.