Math 501: Differential Geometry

Homework 7
Lawrence Tyler Rush
<me@tylerlogic.com>

April 7, 2013
http://coursework.tylerlogic.com/courses/math501/homework07

1


Taken off the homework.

2


3 do Carmo pg 212 problem 11


Let X be a parametrization of a surface with normal N. Define Y to be
Y (u,v) = X (u,v)+ aN (u,v)
(3.1)

for some positive a.

(a)


We know the following to hold
Nu   =  a11Xu + a21Xv
Nv   =  a12Xu + a22Xv
where (aij) is the matrix representation of the differential of N. section 3.1 yields
Yu  =  Xu + aNu
Yv  =  Xv + aNv
and thus combining the two sets of equations above we are left with
Yu  =  Xu + a(a11Xu + a21Xv) = (1 + aa11)Xu + aa21Xv
Yv  =  Xv + a(a12Xu + a22Xv) = aa12Xu + (1+ aa22)Xv
With this, we can take the cross product of Y u and Y v revealing that
Yu ∧ Yv =  ((1+ aa11)Xu + aa21Xv) ∧(aa12Xu + (1 + aa22)Xv)
        =  (1+ aa  )aa X  ∧ X  + aa  aa X  ∧ X  + (1 + aa )(1+ aa22)X  ∧X  + aa  (1+ aa  )X  ∧ X
                 11   12  u    u    21  12  v   u         11           u    v    21      22  v   v
        =  aa21aa12Xv ∧ Xu + (1+ aa11)(1 +aa22)Xu ∧ Xv
        =  - aa21aa12Xu ∧ Xv + (1 + aa11)(1+ aa22)Xu ∧ Xv
        =  - aa21aa12Xu ∧ Xv + (1 + aa11 + aa22 +aa11aa22)Xu ∧ Xv
        =  (1+ a(a  + a  )+ a2(a  a  - a a  ))X  ∧X
                  11   22      11 22   21 12   u    v
Now since K = det([dN]) and H = -12tr([dN]) for Gaussian and mean curvatures of X, then
a11 + a22 = - 2H

and

a11a22 - a21a12 = K

resulting in

                      2
Yu ∧Yv = (1- 2Ha + Ka  )Xu ∧ Xv

(b)


Let F be the homeomorphism from S to the parallel surface defined by F(p) = p + aN(p). Thus we have
Fu  =   Xu + aNu = Xu + a(a11Xu + a21Xv) = (1+ aa11)Xu + aa21Xv
Fv  =   Xv + aNv = Xv + a(a12Xu + a22Xv ) = aa12Xu + (1 + aa22)Xv
indicating that
                                                       (                  )
[dFp]{Xu,Xv} = (dFp(Xu ) dFp(Xv )) = (Fu Fv) = ([Fu]X [Fv]X) =  1+ aa11   aa12
                                                           aa21    1+ aa22

which results in

det(dF ) = 1+ aa  + aa  + aa aa  - aa  aa  = 1 +a(a  + a  )+ a2(a  a  - a  a ) = 1 - 2Ha + Ka2
      p         11    22    11 22    12  21        11   22      11 22   12 21
(3.2)

Now because of part (a), we know that the normal field for the parallel surface, call it M, at F(p) is the same as the normal field for S at p, i.e.

N (p) = M (F(p))
(3.3)

We will make use of this to determine the Gaussian and mean curvatures.

Gaussian Curvature Using the chain rule with subsection 3.3 gives us that

     dNp  =   dMF( (p)dFp    )
det(dNp)  =   det dMF (p)dFp
det(dN )  =   det(dM    ) det(dF )
       p            F(p)       p
which leads to
   (      )   det(dNp)      K
det dMF (p) = det(dF-) = det(dF--)
                    p          p
(3.4)

The combination of subsection 3.2 and paragraph 3.4 leads to the parallel surface having a Gaussian curvature, K, of

--                     K
K  = det(dMF (p)) = 1--2Ha-+-Ka2-

Mean Curvature Again using the chain rule with subsection 3.3 and employing Lemma A.1 and paragraph 3.4 we find that

       dN   =  dM    dF
        -p1        F-(1p) -p1
     dN p   =  dF p dM F(p)
  dFpdN -p1  =  dM - 1
      - 1         F(p) - 1
 tr(dM F(p))  =  tr(dFpdNp )
 tr(dMF (p))                       -1
det(dMF-(p)  =  tr([dFp]{Xu,Xv}[dNp]{Xu,Xv})
                            ((                  )         (             ))
 tr(dMF (p))  =  det(dMF (p))tr   1 + aa11   aa12    ---1----   a22   - a12
                                 a(a2(1   1 + aa22   det(dN)p()   - a21 a11))
                det(dNp)----1----     1 + aa11    aa12        a22   - a12
 tr(dMF (p))  =   det(dFp) det(dNp) tr      aa21   1+ aa22     - a21  a11
                      1
 tr(dMF (p))  =   1--2Ha-+-Ka2-(a22 +aa11a22 - aa12a21 - aa12a21 + a11 + aa11a22)
                      1
 tr(dMF (p))  =   1--2Ha-+-Ka2-((a11 + a22)+ 2a(a11a22 - a12a21))
                      1
 tr(dMF (p))  =   1--2Ha-+-Ka2-(- 2H + 2aK )

 tr(dMF (p))  =   -- 2(H---aK-)2
                1- 2Ha + Ka
Hence the mean curvature, H, is
H-= - 1 tr(dM    ) = ---H---aK----
      2      F(p)    1- 2Ha + Ka2

(c)


4 do Carmo pg 229 problem 9


Let S1 and S2 be regular surfaces with a conformal maps φ : S1 S2 and ψ : S2 S3.

(a) Inverses of isometries are isometries


The proof in Problem section 8 part (a) holds for this when λ is the constant function of 1.

(b) Composition of isometries is an isometry


The proof in Problem section 8 part (b) holds for this when λφ and λψ are both the constant functions of 1.

5 do Carmo pg 229 problem 10


Let φ : S S be a rotation about the axis of a surface of revolution, S. Because it is simply a rotation, φ is the restriction of some linear map of rotation, R : 3 3, to S. Hence for v S, φ(p) = Ap for some matrix of rotation A. Note that rotational matrices such as A are orthogonal. Thus we have the following for p S and v TpS with some curve α such that α(0) = p and α(0) = v
          |              |
dφp(v) =-d||   φ(α(t)) = d-||  A α(t)
        dt|t=0          dt|t=0

but because A is not dependent on t

 d||
--||   Aα(t) = A α′(t)|t=0 = A α′(0) = Av
dt t=0

resulting in p(v) = Av. Thus since A is orthogonal then for any v,w TpS

⟨dφ (v),dφ (w)⟩ = ⟨Av,Aw ⟩ = ⟨v,w⟩
   p     p

thereby giving us that φ is an isometry.

6


(a)


The drawing in Figure 1 has two points p and q in a planar surface for which there is no curve between them with length equal to the intrinsic distance between the two points. A curve that would potentially have a length of the intrinsic distance would need to go through the hole in the middle of the surface, but it obviously cannot while remaining a curve of the surface.

PIC

Figure 1: A planar surface with two points for no curve between the two has a length equal to the intrinsic distance.


(b)


From our first homework assignment, we know that for any curve α in S with α(a) = p and α(b) = q, L(α)ab ≥|p-q|. Thus because d(p,q) is the infimum a set of the lengths (from a to b) of curves in S which pass through p and q at a and b, respectively, then d(p,q) ≥|p - q|.

(c)


7


(a) Coefficients of the first fundamental form.


X (ϕ,θ)  =  (cosθcosϕ,cosθsin ϕ,sin θ)
   X    =  (- cosθsin ϕ,cos θcosϕ,0)
     ϕ
    Xθ  =  (- sin θcosϕ,- sinθsinϕ,cosθ)
     E  =  ⟨X ϕ,Xϕ⟩ = cos2θsin2ϕ+ cos2θcos2ϕ = cos2θ
     F  =  ⟨X ϕ,Xθ⟩ = cosθsinϕsinθcosϕ - cosθcosϕsin θsin ϕ = 0
     G  =  ⟨X  ,X  ⟩ = sin2 θcos2 ϕ+ sin2 θsin2ϕ + cos2 θ = sin2 θ+ cos2 θ = 1
              θ  θ

(b) Relation of M, M˜, X, and Y


Since M˜ is the map from the domain of X to the domain of Y , then we have
M (X(ϕ,θ) = Y ( ˜M (ϕ,θ))
(7.5)

for (ϕ,θ) in the domain of X.

(c) Coefficients of the first fundamental form of X


By subsection 7.5 we have
--
X(ϕ,θ) = M (X (ϕ,θ) = Y(M˜(ϕ,θ)) = Y(ϕ,z(θ)) = (cosϕ,sinϕ,z(θ))

which results in

--
X-ϕ  =  (- sin ϕ,cos ϕ,0)
 Xθ  =  (0,0,z′(θ))
  E- =  ⟨X-ϕ,X-ϕ⟩ = sin2 ϕ+ cos2ϕ = 1
  --    ⟨--  -- ⟩
  F- =  ⟨X-ϕ,X-θ⟩ = 0
  G  =   X θ,X θ = (z′(θ))2

(d)


In order to have M be a conformal map, we must satisfy
E- =  λ2E
       2   2
1  =  λ  cos θ
F-  =  λ2F

 0  =  0
and
     G- =  λ2G
  ′  2       2
(z(θ))  =  λ
indicating that z(θ) must be secθ.

(e)


From the result of the previous part of the problem we know that
      ∫
z(θ) =   secθ = ln|secθ + tan θ|

assuming z(0) = 0 and z(θ) > 0 for θ (0,π∕2).

8


Let S1 and S2 be regular surfaces with a conformal maps φ : S1 S2 and ψ : S2 S3.

(a) Inverses of conformal maps are conformal


Since S1 and S2 are diffeomorphic, then TpS1 = Tφ(p)S2 and p = φ(p)-1 for p S2. Thus for vectors v1,v2 TpS2 = Tφ-1(p)S1, the vectors p-1(v1) and p-1(v2) are vectors in Tp(S1). So we have
⟨         - 1              - 1   ⟩    2   ⟨  -1      -1   ⟩
dφ φ-1(p)(dφp  (v1)),dφ φ-1(p)(dφp (v2)) = λ (p) dφp (v1),dφp (v2)

which in turn implies

⟨      -1           - 1    ⟩    2  ⟨   -1      -1   ⟩
 d(φ ∘φ   )p(v1),d(φ ∘φ  )p(v2) = λ (p) dφ p (v1),dφp (v2)

however, φ φ-1 is the identity map implying that the above equation simplifies to

          2  ⟨  -1       -1   ⟩
⟨v1,v2⟩ = λ (p) dφp (v1),dφ p (v2)

which gives us what we’re looking for

⟨  -1      -1    ⟩  --1--
 dφp (v1),dφp (v2) = λ2(p) ⟨v1,v2⟩

(b) Composition of conformal maps is conformal


Let p S1 and v1,v2 TpS. Then we have the following
                              ⟨                           ⟩
⟨d(ψ ∘ φ)p(v1),d(ψ ∘φ )p(v2)⟩ =   dψ φ(p)(dφp(v1)),dψ φ(p)(dφp(v1))
                          =   λ2(φ(p))⟨dφ (v ),dφ (v )⟩
                               ψ        p  1    p 1
                          =   λ2ψ(φ(p))λ2φ(p)⟨v1,v2⟩
so for λ(p) = λψ(φ(p))λφ(p) we have
                            2
⟨d(ψ ∘φ )p(v1),d(ψ ∘φ)p(v2)⟩ = λ (p)⟨v1,v2⟩

giving us the fact that φ ψ is conformal since it is a diffeomorphism.

(c)


Appendix

A Helpful Lemmas

Lemma A.1 The trace of the inverse of a two-dimensional matrix A is

          tr(A)
tr(A -1) =------
         det(A)

Proof. Let A be the matrix denoted by ( a  b )
  c  d then

            (         )
A-1 = --1---   d   - b
      det(A )  - c  a

which in turn leads to

    -1     (   1   (  d  - b ))      1    (  d   - b )   d+ a    tr(A)
tr(A  ) = tr det(A-)  - c  a     = det(A) tr  - c a    = det(A) = det(A-)

__