Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

A Brief Survey of Differential Geometry: Adrian Down August 29, 2006

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

A Brief Survey of Differential Geometry

Adrian Down
August 29, 2006

1 Introduction
Differential geometry is a branch of mathematics that uses the tools of cal-
culus and linear algebra to study geometry. In this course, we will write
equations describing geometric objects such as curves and surfaces. We will
analyze these equations using the tools of math 53 and 54 gain geometric
insight from these equations.
Note. The notation in this field is very precise and should be followed exactly.
The formulas will quickly become complicated and it will be useful to follow
the standardized notation.

2 Curvature
2.1 Motivation
Imagine a curve in a plane. Let x1 and x2 refer to the coordinates in this plane
(superscripts refer to coordinates). We would like to define the curvature κ
of the curve at a particular point on the plane.
This definition of curvature should meet at least two basic requirements:

1. The curvature of a straight line should be identically equal to 0 every-


where along the curve.

2. The curvature of a circle should be constant and nonzero at every point


on the circle.

1
Following the general method above, we should write a formula describing
the curve and manipulate it using calculus. The simplest way to write an
equation parameterizing a plane curve is by y = f (x) for some function f (x).
All of the information about the curve must be contained in the function f .

2.2 Possible (incorrect) definitions


It is instructive to try a few possible definitions for the curvature and see
how they fail to meet the requirements above.
One possibility is κ = f 0 (x). This definition fails to give κ ≡ 0 for a
straight line with nonzero slope.
We could also try κ = f 00 (x). This definition does ensure that κ ≡ 0 for
any straight line. We can verify if the second condition is also satisfied. The
unit circle can be parameterized,

S = x, y ∈ R x2 + y 2 = 1

1
⇒ y = f (x) = (1 − x2 ) 2
Taking the derivative,
−x
f 0 (x) = 1
(1 − x2 ) 2
x2 1
f 00 (x) = − 3 − 5
(1 − x2 ) 2 (1 − x2 ) 2
f 00 (x) is not constant on the unit circle, and so κ = f 00 (x) cannot be the
correct definition of curvature.
Note. Another argument against the choice κ = f 00 (x) is that this definition
is not invariant under rotation of the coordinate system.

2.3 Correct definition of curvature


The correct definition of curvature is,
|f 00 (x)|
κ= 3
(1 + f 0 (x)) 2
In can be shown that this formula is invariant to coordinate transforma-
tion. The denominator can be thought of as the necessary factor to make
the definition invariant.

2
3 Curves in space
3.1 Parameterization
Imagine a curve C in space, and let the coordinates in this space be denoted
by (x1 , x2 , x3 ). Our first objective is to describe this curve with an equation,
and then we would like to extract geometric information from this description
by manipulating the equation.
It will be convenient to describe the curve parametrically as a vector-
valued function α (boldface is used to denote vector quantities) that maps
from the real line to R3 . The behavior of the curve is determined by the
point on the real line, denoted as s, from which the function is mapping. For
each s, the vector α points to one point on the curve. The curve can be
written in component notation,
α(s) = (α1 (s), α2 (s), α3 (s))
The curve is thus completely described by the three functions α1 (s), α2 (s),
and α3 (s) for a ≤ s ≤ b.

3.2 Geometric interpretation


We can now use linear algebra and calculus to extract geometric information
from the equation describing the curve. Using linear algebra, we can express
α at any point in terms of a basis that spans R3 . We will see that the most
convenient basis is one constructed from α itself. At each point, we can build
orthonormal unit vectors T(s), N(s), and B(s). T is tangent to the curve,
N is the normal to the curve, and B, called the binormal, is perpendicular
to both T and N. This basis changes at each point on α, and describes a
comoving coordinate system that depends on the parameter s.
We will show that the derivatives of these orthonormal basis vectors can
be expressed using linear algebra,
 0    
T (s) 0 κ(s) 0 T (s)
N 0 (s) = −κ(s) 0 τ (s) N (s)
0
B (s) 0 −τ (s) 0 B(s)
d
where primes denote ds , κ is the curvature, and τ is the torsion.
This equation is called the Frenet-Serret formula and results from cal-
culus. We will use it to define the torsion τ . We will see that the torsion
represents the tendency for the curve to stay in a plane.

3
4 Surfaces in space
4.1 Parameterization
Now we consider two-dimensional surfaces lying in a three-dimensional space.
By analogy with the mapping from the real line to R3 considered in the case
of curves, surfaces are parameterized by a function x(u1 , u2 ) that maps from
a two dimensional space U with coordinates (u1 , u2 ) to R3 . The function x
can be written in component form,

x(u1 , u2 ) = (x1 (u1 , u2 ), x2 (u1 , u2 ), x3 (u1 , u2 ))

The surface is completely described by these three functions xi .

4.2 Geometric interpretation


We can now attempt to extract geometric information from this parameter-
ization of the surface. In analogy with the orthonormal basis constructed
before, we will use the function x to construct the most convenient comoving
basis in which to examine the surface. We will construct a comoving frame
at any point from the two vectors describing the plane tangent at that point,
denoted by x1 and x2 , and the normal to this tangent plane, denoted by n.
We will see that the tangent vectors are formed using partial differentiation,
 1
∂x ∂x2 ∂x3

∂x
x1 = = , ,
∂u1 ∂u1 ∂u1 ∂u1
 1
∂x ∂x2 ∂x3

∂x
x2 = = , ,
∂u2 ∂u2 ∂u2 ∂u2

Note. The comoving basis constructed to analyze surfaces is not in gen-


eral orthonormal. This is different than the basis used to analyze curves,
{ T, N, B }, which was orthonormal.
We would like an analog of the Frenet-Serret formula. Since the basis
chosen to analyze the surface is not orthonormal, the form of this formula is

4
more complicated,
3
X
xij = Lij n + Γkij xk 1 ≤ i, j, ≤ 2
k=1
2
X
nj = − Lkj xk 1≤j≤2
k=1

where Lij is called the coefficient of the second fundamental form and γijk is
called the Christoffel symbol of the second kind.
The first of these equations is called Gauss’s formula. The second is
Weingarten’s formula and represents how the surface is changing as the co-
ordinates xi change.

4.3 Intrinsic verses extrinsic properties


We would like to have a geometric interpretation of the quantities that we
have just discovered. In particular, we would like to identify which of these
quantities is intrinsic, meaning they can be measured at any particular point
on the surface without knowledge of the rest of the surface. We will see that
the Christoffel symbols are intrinsic, whereas the coefficients of the second
fundamental form are not.
One way to think of intrinsic quantities is as those that could be measured
by a bunch of mathematically-inclined bugs living on the surface that are
unaware of the larger three-dimensional space in which the curve lies.

You might also like