Hyperbolic plane H2

The hyperbolic plane is given as follows. It is an abstract surface.

Let
H2 = {(x, y ) ∈ R2 | y > 0}
so that the first fundamental form is
1
gij (x, y ) = .
y2

That is to say, if α(t) = (x(t), y (t)) is a regular curve, then

1
hα0 , α0 i = g11 (x 0 )2 + 2g12 x 0 y 0 + g22 (y 0 )2 = (x 0 )2 + (y 0 )2 .


y 2 (t)
Gaussian curvature of H2

To compute the Gaussian curvature.

By previous result
 2
∂2

2 ∂
K =y + log y = −1.
∂x 2 ∂x 2
Completeness of H2

H2 is complete: every divergent path has infinite length.

Let α : [0, 1) be a divergent path. Let α(t) = (x(t), y (t)), then
the length of α is given by
Z 1
1
1
`(α) = (x 0 )2 (t) + (y 0 )2 (t) 2 dt.
0 y (t)

Since α is divergent, there exist tk → 1 so that (i) y (tk ) → 0; (ii)

y (tk ) → ∞; or (iii) b > y (tk ) > a > 0 for some a, b and
x(tk ) → ∞. In any case we will have `(α) = ∞.
Geodesics of H2

R
Consider the energy functional Ldt, with

1
L = y −2 (ẋ)2 + (ẏ )2 .


2
To find the E-L equations, first we have
∂L


 = 0;

 ∂x
∂L

= y −2 ẋ;




∂ ẋ
∂L
= −y −3 (ẋ)2 + (ẏ )2 ;





 ∂y
 ∂L

= y −2 ẏ .



∂ ẏ
Hence the E-L equations are:
∂L d ∂L d

 0=
 − ( ) = − (y −2 ẋ);
∂x dt ∂ ẋ dt
∂L d ∂L d
− ( ) = −y −3 (ẋ)2 + (ẏ )2 − (y −2 ẏ ).


 0=

∂y dt ∂ ẏ dt
Geodesics of H2 , cont.
That is:
ẍ − 2y −1 ẋ ẏ = 0;


ÿ + y −1 (ẋ)2 − (ẏ )2 = 0.

Obviously, x =constant is a geodesic. Next consider the

semi-circle: x = R cos t, y = R sin t, 0 < t < π. Then
cos t 0
x 00 − 2y −1 x 0 y 0 = − x
sin t
and
cos t 0
y 00 − y −1 (x 0 )2 + (y 0 )2 = −


y.
sin t
So it is a pre-geodesic.
Proposition
The geodesics are either {x = constant} or semicircles with
centers at the x-axis and orthogonal to the x-axis.
Isometry of H2

Let F : H2 → H2 defined as

F (x, y ) = (u, v ),

so that
az + b
u + iv =
cz + d
with a, b, c, d real and ac − bd = 1. Here z = x + iy .
Then
(az + b)(cz̄ + d)
u + iv =
|cz + d|2
Imaginary part of (az + b)(cz̄ + d) is

(ad − bc)y = y > 0.

So this is well-defined.
Isometry of H2 , cont.

Let α(t) = (x(t), y (t)) be a curve in H2 , t ∈ [0, 1]. Then the

length of α(t) is given by
Z 1 Z 1
1
1 1
`(α) = (x 0 )2 (t) + (y 0 )2 (t) 2 dt = |z 0 (t)|dt.
0 y (t) 0 y (t)

Let β(t) = F ◦ α. Write β(t) = (u(t), v (t)), then the length of β is

Z l
1
`(β) = |w 0 (t)|dt
0 v (t)

where w = u + iv .
Isometry of H2 , cont.

Now
y
v= .
|cz + d|2

az + b 0
w 0 =( )
cz + d
az 0 (cz + d) − cz 0 (az + b)
=
(cz + d)2
z 0
= .
(cz + d)2
So
1 1
|w 0 (t)| = |z 0 (t)|.
v (t) y (t)
Isometry of H2 , cont.

What is the image of x =constant. That is the image of F (0, y ).

aiy + b
u + iv = F (0, y ) = .
ciy + d

It is a circle with centered at 21 (a/c + b/d) if c 6= 0, d 6= 0.

Geodesic equations of surfaces of revolution

Consider the surface of revolution given by

X(u, v ) = (f (v ) cos u, f (v ) sin u, g (v ))

with f > 0. In the following f 0 means dvdf

, etc. If there is come
confusion, we will write fv instead, etc.
Consider u 1 ↔ u, u 2 ↔ v .

 g11 = E = hXu , Xu i = f 2 , ;


g12 = g21 = F = hXu , Xv i = 0

g22 = G = hXv , Xv i = (f 0 )2 + (g 0 )2 .


So ( 0
Γ111 = 0, Γ112 = ff , Γ122 = 0;
0 f 0 f 00 +g 0 g 00
Γ211 = − (f 0 )2ff+(g 0 )2 , Γ212 = 0, Γ222 = (f 0 )2 +(g 0 )2
.
Geodesic equations of surfaces of revolution

Hence geodesic equations are:

0
(
ü + 2ff u̇ v̇ = 0;
0 f 0 f 00 +g 0 g 00
v̈ − (f 0 )2ff+(g 0 )2 (u̇)2 + (f 0 )2 +(g 0 )2
(v̇ )2 = 0.

Corollary
Any meridian is a geodesic. A parallel X(u, v0 ) is a geodesic if and
only if f 0 (v0 ) = 0.
General geodesics

To study the behavior of general geodesics, we begin with the

following lemma:
Lemma
Let a1 (t), a2 (t) be smooth functions on (T1 , T2 ) ⊂ R such that
a12 + a22 = 1. For any t0 ∈ (T1 , T2 ) and θ0 such that
a1 (t0 ) = cos θ0 , a2 (t0 ) = sin θ0 , there exists unique a smooth
function θ(t) with θ(t0 ) = θ0 such that a1 (t) = cos θ(t) and
a2 (t) = sin θ(t).
Proof of the lemma

Proof: Suppose θ satisfies the condition. Then a10 = −θ0 sin θ,

a20 = θ0 cos θ. Hence θ0 = a1 a20 − a2 a10 . From this we have
uniqueness. To prove existnce, fix t0 ∈ (T1 , T2 ) and let θ0 be such
that cos θ0 = a1 (0), sin θ0 = a2 (0). Let
Z t
θ(t) = θ0 + (a20 a1 − a10 a2 )dτ.
t0

Let f = (a1 − b1 )2 + (a2 − b2 )2 , where b1 = cos θ, b2 = sin θ.

Then f = 2 − 2a1 b1 − 2a2 b2 .
Proof of lemma, cont.

Then

1
− f 0 =a10 b1 + a1 b10 + a20 b2 + a2 b20
2
=a10 b1 − θ0 a1 b2 + a20 b2 + θ0 a2 b1
=(a20 a1 − a10 a2 )(−a1 b2 + a2 b1 ) + a10 b1 + a20 b2
= − a12 a20 b2 + a2 a20 a1 b1 + a1 a10 a2 b2 − a22 a10 b1 + a10 b1 + a20 b2
= − a12 a20 b2 − a1 a10 a1 b1 − a2 a20 a2 b2 − a22 a10 b1 + a10 b1 + a20 b2
=0

because a12 + a22 = 1 and a1 a10 + a2 a20 = 0.

General geodesics, cont.

Now let α(s) = X(u(s), v (s)) be a geodesic on M parametrized by

arc length. Let e1 = Xu /|Xu | and e2 = Xv /|Xv |. Then e1 , e2 are
orthonormal. Let
α0 = a1 e1 + a2 e2 .
By the lemma there exists smooth function θ(s) such that
a1 = sin θ, a2 = cos θ. Note that θ is the angle between α0 and the
meridian. That is:
sin θ = hα0 , e1 i = f u̇.
Clairaut’s Theorem
Proposition (Clairaut’s Theorem)
r (s) sin θ(s) is constant along α, where r (s) is the distance of α(s)
from the z-axis.

Proof.
Denote the dα
ds by α0 etc. Since r (s) = f (v (s)),

r 0 = fv v 0 .

Also sin θ = hα0 , e1 i = u 0 f , so (sin θ)0 = u 00 f + u 0 v 0 fv .

(r sin θ)0 =fv v 0 u 0 f + u 00 f + fv u 0 v 0

 
00 2fv 0 0
=f u + uv
f
=0.
Nöether’s first theorem, a digression

Consider the action:

Z b
S= L(φ, φ̇, t)dt.
a

here φ = (φk ). To be precise, we denote L(u, v , t) so that in the

above, u k = φk , v k = φ˙k . Consider the transformation:

et = et (φ, t, )
φe = φ(φ,
e t, ).

e φe0 , et ). 0 means derivative w.r.t. et . Assume

Let Le = L(φ,
et = t, φ(φ, t, ) = φ at  = 0.
e
Nöether’s first theorem, cont.

The action is said to be invariant if

d et
L − Le = O(2 ).
dt
Hence we have  
d d et
L − Le = 0.
d dt =0

Let (
∂e
t
τ= ∂ =0 ;
ξ k = ∂∂φ =0 .
e

Proof is given at the end of the note for your reference.

Theorem (Nöether)
If L is invariant in the above sense, then
d  
τ H − pk ξ k = (ξ k − τ φ˙k )Ek
dt

where pk = ∂L
, H = pk φ˙k − L and Ek is the E-L expression
∂ φ˙k
!
∂L d ∂L
− .
∂ φ˙k
∂φk dt

In particular, if Ek = 0, then τ H − pk ξ k is conserved, i.e. constant

along the solution.
Consider m particles in three space with coordinates (x j , y j , z j )
with mass Mj . Then
1X
Mj (ẋ j )2 + (ẏ j )2 + (ż j )2 − V (t, x, y , z).
 
L=
2
j

Then
∂L
pk =
∂ φ˙k
which is Mj ẋ j , etc.
1X
H =Mj ẋ j ẋ j − Mj (ẋ j )2 + (ẏ j )2 + (ż j )2 − V (t, x, y , z)
 
2
j
1X
Mj (ẋ j )2 + (ẏ j )2 + (ż j )2 + V (t, x, y , z).
 
=
2
j

Suppose V does not depend on t (i.e. there is only one

independent variable). Then the transformation t → t + , with
x, y , x fixed is invariant. We have τ = 1, ξ = 0.
Hence we have,

 
d  X d
Mj (ẋ j )2 + (ẏ j )2 + (ż j )2  = (K + V ).


0= −L +
dt dt
j

Here K is the kinetic energy. So the total energy is preserved.

Geodesics of surfaces of revolution, cont.
Clairaut’s Theorem revisited: In this case for the energy functional,
1
L = (f 2 (u̇)2 + (fv2 + gv2 )(v̇ )2 ).
2
If we let u
e = u + , ve = v , et = t, then
L = Le
Here we have τ = 0, ξ 1 = 1, ξ 2 = 0.
p1 = f 2 u̇
So, we have
d 2
(f u̇) = 0
dt
along the geodesic.
In fact, this is just one of the E-L equations. Note that
Xu
sin θ = hα0 , e1 i = hXu u̇ + Xv v̇ , i = f u̇.
|Xu |
So r (s) sin θ(s) = f (α(s)) sin θ(s) = f 2 u̇.
Geodesics of surfaces of revolution, cont.

Let us analyse a geodesic α(s), 0 ≤ s < L ≤ ∞, on the surface of

revolution parametrized by arc length. Let us assume that
g (v ) is increasing, i.e. gv > 0.
Let r (s) and θ(s) be as in Clairaut’s Theorem. Let θ0 = θ(0). We
may assume that
0 ≤ θ0 ≤ π2 .
By Clairaut’s Theorem,
r (s) sin θ(s) = R for some constant R ≥ 0.
Note that r (s) ≥ R.
Case 1: R = 0, then θ = π/2 along α. So it is a meridian.
Case 2: R > 0. Then sin θ0 = R/r (0) < 1. So 0 ≤ θ0 < π/2. So
α is going up. Then we must have fv (α(0)) 6= 0. Let us assume
that fv (α(0)) > 0. Hence initially, r (s) = f (α(s)) > R and it
continue to go up. We consider two cases:
Case 2(i) There is no parallel above α(0) with radius R. Then α
will go up all the way.
Case 2(ii) There is a parallel above α(0) so that the radius is R.
Let C be the first one above α(0). Then we have two more
subcases:
(ii)(a) C is a geodesic. Then α will approach to C but never
intersect C .
(ii)(b) C is not a geodesic, then α will touch C and bounces away.
To summarize, in the above settings, we have:
Proposition
(i) If R = 0, then α is a meridian.
(ii) R > 0. Then geodesic will go up for all s, as long as r > R,
i.e. the z coordinate of α is increasing in s. Either α does not
come close to any parallel of radius R, and α will go up for all
s, or α will be close to a parallel C of radius R. Let C be the
first such parallel above α. Then we have the following cases:
(a) C is a geodesic. Then α will not meet C and α will come
arbitrarily close to C without intersecting C .
(b) C is not a geodesic. Then there is α(s0 ) ∈ C for some s0 and
α will bounce off from C and will turn downward.
Proof of Nöether’s theorem

Proof: At  = 0,
 
d d et
0= L−L e
d dt
∂ ∂ d et
= − Le − L ( )
∂ ∂ dt
˙k
∂L k ∂L ∂ φe ∂L
=− ξ − − τ − Lτ̇
∂φ k ˙
∂ φk ∂ ∂t

Now
˙k ˙k
∂ φe ∂ ∂ φe dt
= ( · )
∂ ∂ dt d et
=ξ˙k − φ˙k τ̇ .
So we have

∂L k ∂L  ˙k ˙k τ̇ + ∂L τ + Lτ̇

0= ξ + ξ − φ
∂φk ∂ φ˙k ∂t
Also,
∂ d ∂L ∂L ¨k
τ L =τ L − τ k φ˙k − τ φ
∂t dt ∂φ ∂ φ˙k
! !
d ∂L d ∂L ˙k d ∂L
=τ L − τ k φ˙k − τ φ + τ φ˙k
dt ∂φ dt ∂ φ˙k dt ∂ φ˙k
So
!
d ∂L ˙k
− τL − τ φ
dt ∂ φ˙k
!
∂L k ∂L  ˙k ˙
 ∂L ˙k d ∂L
= kξ + k
ξ − φ τ̇ − τ k φ + τ φ˙k
∂φ ∂φ ˙k ∂φ dt ∂φ ˙k
!
˙ ∂L ∂L 
˙ ˙
 d ∂L
k
=(ξ − τ φk ) k + ξ k − φk τ̇ + τ φ˙k
∂φ ∂φ˙k dt ∂φ ˙k

∂L ∂L ˙k d ∂L
=(ξ k − τ φ˙k ) k + ξ + τ φ˙k
∂φ ∂φ˙k dt ∂ φ˙k
! !
˙ ∂L d ∂L d ∂L d ∂L
k
=(ξ − τ φk ) k + ξ k
−ξ k
+ τ φ˙k
∂φ dt ∂ φ˙k dt ∂ φ˙k dt ∂ φ˙k
! !
∂L d ∂L d ∂L k
=(ξ k − τ φ˙k ) − + ξ .
∂φk dt ∂ φ˙k dt ∂ φ˙k
So we have
!
d   ∂L d ∂L
τ H − pk ξ = (ξ k − τ φ˙k )
k
−
dt ∂φk dt ∂ φ˙k

where
∂L
pk = , H = pk φ˙k − L
∂ φ˙k