Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
0% found this document useful (0 votes)
72 views

Lecture Notes 11: 2.4 Intrinsic Metric and Isometries of Surfaces

The document discusses intrinsic metrics and isometries of surfaces. It defines the intrinsic distance between two points on a surface as the infimum of the lengths of all curves connecting the two points. It proves that this turns the surface into a metric space and establishes properties of the intrinsic distance under isometries. It introduces the first and second fundamental forms and proves Gauss's Theorema Egregium, which states that the Gaussian curvature of a surface depends only on the first fundamental form and is thus intrinsic to the surface.

Uploaded by

Sanjeev Shukla
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
72 views

Lecture Notes 11: 2.4 Intrinsic Metric and Isometries of Surfaces

The document discusses intrinsic metrics and isometries of surfaces. It defines the intrinsic distance between two points on a surface as the infimum of the lengths of all curves connecting the two points. It proves that this turns the surface into a metric space and establishes properties of the intrinsic distance under isometries. It introduces the first and second fundamental forms and proves Gauss's Theorema Egregium, which states that the Gaussian curvature of a surface depends only on the first fundamental form and is thus intrinsic to the surface.

Uploaded by

Sanjeev Shukla
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Math 497C Mar 3, 20041

Curves and Surfaces


Fall 2004, PSU

Lecture Notes 11

2.4 Intrinsic Metric and Isometries of Surfaces


Let M ⊂ R3 be a regular embedded surface and p, q ∈ M , then we define

distM (p, q) := inf{Length[γ] | γ : [0, 1] → M, γ(0) = p, γ(1) = q}.

Exercise 1. Show that (M, distM ) is a metric space.

Lemma 2. Show that if M is a C 1 surface, and X ⊂ M is compact, then


for every  > 0, there exists δ > 0 such that
 
 distM (p, q) − p − q  ≤ p − q

for all p, q ∈ X, with distM (p, q) ≤ δ.

Proof. Define F : M × M → R by F (p, q) := distM (p, q)/p − q, if p = q


and F (p, q) := 1 otherwise. We claim that F is continuous. To see this let
pi be a sequnce of points of M which converge to a point p ∈ M. We may
assume that pi are contained in a Monge patch of M centered at p given by

X(u1 , u2 ) = (u1 , u2 , h(u1 , u2 )).

Let xi and yi be the x and y coorindates of pi . If pi is sufficiently close to


p = (0, 0), then, since ∇h(0, 0) = 0, we can make sure that

∇h(txi , tyi )2 ≤ ,

for all t ∈ [0, 1] and  > 0. Let γ : [0, 1] → R3 be the curve given by

γ(t) = (txi , tyi , h(txi , tyi )).


1
Last revised: November 8, 2004

1
Then, since γ is a curve on M ,
distM (p, pi ) ≤ Length[γ]
 1
= x2i + yi2 + ∇h(txi , tyi ), (xi , yi ) 2 dt
0 1 
≤ x2i + yi2 + (x2i + yi2 )2 dt

0

≤ 1 +  x2i + yi2
≤ (1 + )p − pi 
So, for any  > 0 we have
distM (p, pi )
1≤ ≤1+
p − pi 
provided that pi is sufficiently close to p. We conclude then that F is con-
tinuous. So U := F −1 ([1, 1 + ]) is an open subset of M × M which contains
the diagonal ∆M := {(p, p) | p ∈ M }. Since ∆X ⊂ ∆M is compact, we may
then choose δ so small that Vδ (∆X ) ⊂ U , where Vδ (∆X ) denotes the open
neighborhood of ∆X in M × M which consists of all pairs of points (p, q)
with distM (p, q) ≤ δ.
Exercise 3. Does the above lemma hold also for C 0 surfaces?
If γ : [a, b] → M is any curve then we may define

LengthM [γ] :=
 k 
 

sup distM (γ(ti ), γ(ti−1 ))  {t0 , . . . , tk } ∈ Partition[a, b] .
i=1

Lemma 4. LengthM [γ] = Length[γ].


Proof. Note that
distM (γ(ti ), γ(ti−1 )) ≥ γ(ti ) − γ(ti−1 ).
Thus LengthM [γ] ≥ Length[γ]. Further, by the previous lemma, we can make
sure that
distM (γ(ti ), γ(ti−1 )) ≤ (1 + )γ(ti ) − γ(ti−1 ),
which yields LengthM [γ] ≤ (1 + ) Length[γ], for any  > 0.

2
We say that f : M → M is an isometry provided that

distM (f (p), f (q)) = distM (p, q).


Lemma 5. f : M → M is an isometry, if and only if Length[γ] = Length[f ◦
γ] for all curves γ : [a, b] → M .
Proof. If f is an isometry, then, by the previous lemma,
Length[γ] = LengthM [γ] = LengthM [f ◦ γ] = LengthM [f ◦ γ].
The converse is clear.
Exercise 6. Justify the middle equality in the last espression displayed
above.
Theorem 7. f : M → M is an isometry if and only if for all p ∈ M , and
v, w ∈ Tp M ,  
dfp (v), dfp (w) = v, w .
Proof. Suppose that f is an isometry. Let γ : (−, ) → M be a curve with
γ(0) = p, and γ  (0) = v. Then, by the previous lemma
   

γ (t) dt = (f ◦ γ) (t) dt
− −

Taking the limit of both sides as  → 0 and applying the mean value theorem
for integrals, yields then that
v = γ  (0) = (f ◦ γ) (0) = dfp (v).
Thus df preserves the norm, which implies that it must preserve the inner-
product as well (see the following exercise).
Conversely, suppose that v = dfp (v). Then, if γ : [a, b] → M is any
curve, we have
 b  b  b
 
(f ◦ γ) (t)dt = dfγ(t) (γ (t))dt = γ  (t)dt.
a a a

So f preserves the length of all curves, which, by the previous Lemma, shows
that f is an isometry.
Exercise 8. Show that a function F : Rn → Rn preserves the norm  ·  if
and only if it preserves the inner product ·, · .

3
2.5 Gauss’s Theorema Egregium
Lemma 9. Let X : U → M be a proper regular chart. Then X := f ◦
X : U → M is a proper regular chart as well and gij = g ij on U .
Proof. Using the last theorem we have

g ij (u1 , u2 ) = Di X(u1 , u2 ), Dj X(u1 , u2 )


= Di (f ◦ X)(u1 , u2 ), Dj (f ◦ X)(u1 , u2 )
= dfX(u1 ,u2 ) (Di X(u1 , u2 )), dfX(u1 ,u2 ) (Dj X(u1 , u2 ))
= Di X(u1 , u2 ), Dj X(u1 , u2 )
= gij (u1 , u2 ).

Exercise 10. Justify the third equality in the last displayed expression above.
Let F denote the set of functions f : U → R where U ⊂ R2 is an open
neighborhood of the orgin.
Lemma 11. There exists a mapping Briochi : F × F × F → F such that for
any chart X : U → M centered at p ∈ M ,

K(p) = Briochi[g11 , g12 , g22 ](0, 0).

Proof. Recall that


det lij (0, 0)
K(p) = ,
det gij (0, 0)
and, by Lagrange’s identity,

X1 × X2 1
lij = Xij , = Xij , X1 × X2 ,
X1 × X2  det gij

where Xij := Dij X, and Xi := Di X. Thus

det(Xij (0, 0), X1 (0, 0) × X2 (0, 0) )


K(p) = .
(det gij (0, 0))2
Next note that

det(Xij , X1 × X2 ) = X11 , X1 × X2 X21 , X1 × X2 − X12 , X1 × X2 2

4
The right hand side of the last expression may be rewritten as
det(X11 , X1 , X2 ) det(X22 , X1 , X2 ) − (det(X12 , X1 , X2 ))2 ,
where (u, v, w) here denotes the matrix with columns u, v, and w. Recall
that if A is a square matrix with transpose AT , then det A = det AT . Thus
the last expression displayed above is equivalent to
det((X11 , X1 , X2 )T (X22 , X1 , X2 )) − det((X12 , X1 , X2 )T (X12 , X1 , X2 )),
which in turn can be written as
 
X11 , X22 X11 , X1 X11 , X2
det  X1 , X22 X1 , X1 X1 , X2 
X2 , X22 X2 , X1 X2 , X2
 
X12 , X12 X12 , X1 X12 , X2
− det  X1 , X12 X1 , X1 X1 , X2  .
X2 , X12 X2 , X1 X2 , X2
If we expand the above determinants along their first rows, then X11 , X22
and X12 , X22 will have the same coefficients. This implies that we can
rewrite the last expression as
 
X11 , X22 − X12 , X12 X11 , X1 X11 , X2
det  X1 , X22 X1 , X1 X1 , X2 
X2 , X22 X2 , X1 X2 , X2
 
0 X12 , X1 X12 , X2
− det X1 , X12 X1 , X1 X1 , X2  .
X2 , X12 X2 , X1 X2 , X2
Now note that each of the entries in the above matrices can be expressed
purely in terms of gij , since
1 1
Xii , Xj = Xi , Xi j = (gii )j ,
2 2
1
Xij , Xi = Xi , Xj i − Xi , Xji = (gij )i − (gii )j ,
2
and
X11 , X22 − X12 , X12 = X1 , X22 1 − X1 , X12 2
1 1
= (g21 )21 − (g11 )21 − (g11 )2 .
2 2
5
Substituting the above values in the previous matrices, we define

Briochi[g11 , g22 , g33 ] :=


 
 (g21 )21 − 12 (g11 )21 − 12 (g11 )2 21 (g11 )1 21 (g11 )2
1
2
det  (g21 )2 − 12 (g11 )2 g11 g12 
(det(gij )) 1
(g )
2 22 2
g21 g22
 1 1

0 2
(g11 )2 2
(g22 )1 
− det  21 (g11 )2 g11 g12  .
1
(g )
2 22 1
g21 g22

Evaluating the above expression at (0, 0) yields that Gaussian curvature


K(p).

Theorem 12. If f : M → M is an isometry, then K(f (p)) = K(p), where


K and K denote the Gaussian curvatures of M and M respectively.

Proof. Let X : U → M be a chart centered at p, then X := f ◦ X is a chart


of M centered at f (p). Let gij and g ij denote the coefficients of the first
fundemental form with respect to the chartst X and X respectively. Then,
using the previous two lemmas, we have

K(f (p)) = Briochi[g 11 , g 12 , g 22 ](0, 0)


= Briochi[g11 , g12 , g22 ](0, 0)
= K(p).

Exercise 13. Let M ⊂ R3 be a regular embedded surface and p ∈ M .


Suppose that K(p) = 0. Does there exist a chart X : U → M such that D1 X
and D2 X are orthonormal at all points of U .

You might also like