Scribd
Upload
21 views

Lecture 8

This document provides definitions and concepts related to curves and surfaces in 3D space, including: - Curves are defined as smooth maps from an interval to R3 or a surface, and their arc length, curvature, and torsion are defined. - Curvature measures how much a curve is bent, while torsion measures twisting. - Frenet formulas relate the derivatives of the tangent, normal, and binormal vectors of a curve parametrized by arc length. - An example calculates the curvature and torsion of a circle, showing it has constant curvature but no torsion.

Uploaded by

ThetaOmega
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
21 views

Lecture 8

This document provides definitions and concepts related to curves and surfaces in 3D space, including: - Curves are defined as smooth maps from an interval to R3 or a surface, and their arc length, curvature, and torsion are defined. - Curvature measures how much a curve is bent, while torsion measures twisting. - Frenet formulas relate the derivatives of the tangent, normal, and binormal vectors of a curve parametrized by arc length. - An example calculates the curvature and torsion of a circle, showing it has constant curvature but no torsion.

Uploaded by

ThetaOmega
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

LECTURE 8

3 Curves and Surfaces in R3


In this chapter we will move away from global topological properties (such as degree) and
study in more detail the local metric properties of submanifolds of dimension 1 (curves) and 2
(surfaces) immersed or embedded in R2 and R3 , as well curves on a surface of R3 . By focusing
on these low-dimensional manifolds it will make it easier to build rigorously the theory of
curvature. The main goal of this chapter is the proof of the “remarkable” theorem of Gauss,
i.e. that the Gaussian curvature, i.e. the product of the eigenvalues of the shape operator, is
a “geometric invariant” in the sense that it is invariant under isometric diffeomorphism.

3.1 Arc-length, curvature and torsion of curves


Definition 71. Given X manifold and I ⊂ R a non-empty interval (bounded or not, open,
closed or half-open) then a (smooth) curve is a smooth map α : I → X, and a regular
curve is a smooth map α : I → X that is an immersion: the differential Dαt : R → Tα(t) X
is injective for all t ∈ I.

Remark 72. A regular curve α is thus locally diffeomorphic onto its image which is a 1-
submanifold of X (the “curve”). Note that the definition does not require α to be injective,
so the curve can self-intersect.
When X has a metric (in general: Riemannian manifold, for us: R2 or R3 ), we can
measure the length of the curve:

Definition 73. Given a regular curve α : I → R3 , we call the arc-length of the curve from
t0 ∈ I to t ∈ I the quantity Z t
s(t) := |α̇(τ )| dτ.
t0

If I has endpoints a < b ∈ [−∞, +∞], the length of the curve is


Z b
ℓ(α) := |α̇(τ )| dτ.
a

We say that the curve is parametrized by arc-length if |α̇(τ )| = 1 for all τ ∈ I.

Remark 74. It is always possible to work with arc-length parametrization for a regular curve,
by the following crucial proceedure: let α : I → R3 a regular curve, then since |α̇| is smooth
and does not vanish it remains uniformly positive on any compact interval of I and the arc-
length function t 7→ s(t) is strictly increasing and smooth by composition, so can be inverted
smoothly by some function s 7→ t(s) and β(s) := α(t(s)) is parametrized by arc-length,
and is a regular curve with the same image and length. Note that a curve parametrized by
arc-length is automatically regular.

18
Definition 75. Given α : I → R3 a curve parametrized by arc-length s, we define the
curvature at s by κ(s) := |α̈(s)| (amplitude of the acceleration) and the normal vector at
s by α̈(s) = κ(s)n(s). When κ(s) ̸= 0, we denote the unit speed vector at s by t(s) := α̇(s),
we call the plane determined by t(s) and n(s) the osculating plane at s, we define the
binormal vector at s by b(s) := t(s) × n(s), we define the torsion τ (s) at s by ḃ(s) =
τ (s)n(s), and finally we call the three orthonormal vectors (t(s), n(s), b(s)) the Frenet
frame or Frenet trihedron (this definition is justified by the following proposition).

Remark 76. Intuitively the curvature of the curve corresponds to how much it is bent, and
the torsion of the curve corresponds to how much it is twisted. Note also that we define the
curvature when it is non-zero and using the absolute value of the acceleration. Indeed, in
general we cannot choose an algebraic sign. However when the curve is in R2 , we can define
n(s) so that (t, n) has the same orientation as the basis (e1 , e2 ) of the plane, and then define
ṫ(s) = κ(s)n(s) which gives a sign to the curvature.

Example 77. Let us consider a curve α : [0, 2π] → R3 describing a circle with radius r in a
plane: α(t) = (r cos t, r sin t, H) (with H constant). Then |α̇(t)| = r and the length is s(t) = rt
(starting from t0 = 0). So t(s) = s/r and the new version of the curve parametrised by arc-
length is β(s) = (r cos(s/r), r sin(s/r), H). Then β̇(s) = (− sin(s/r), cos(s/r), 0) and β̈(s) =
−(1/r)(cos(s/r), sin(s/r), 0) so we deduce κ(s) = 1/r and t(s) = (− sin(s/r), cos(s/r), 0) and
n(s) = (cos(s/r), sin(s/r), 0). This means that b(s) = t(s) × n(s) = (0, 0, 1) the curve has
no torsion τ (s) = 0.

Proposition 78. The following relations, called Frenet equations or formulas, hold


ṫ(s) = κ(s)n(s)
ṅ(s) = −κ(s)t(s) − τ (s)b(s)


ḃ(s) = τ (s)n(s).

Proof. Since the curve is parametrised by arc-length we have |α̇(s)|2 = 1 which implies by
differentiating 2α̈(s) · α̇(s) = 0 so that t(s) and n(s) are orthogonal and do indeed determine
a plane, when α̈ ̸= 0. In this case α̈ ̸= 0, we can then define b(s) = t(s) ×n(s) and this vector
is the vectorial product of two orthogonal unit vector, so also has unit norm. Therefore by
differentiating 2ḃ(s) · b(s) = 0, which means that ḃ is orthogonal to b. We then compute
ḃ(s) = ṫ(s) × n(s) + t(s) × ṅ(s) = κ(s)n(s) × n(s) + t(s) × ṅ(s) = t(s) × ṅ(s) so ḃ(s) is
also orthogonal to t(s). Therefore we deduce that ḃ(s) is colinear to n(s). This allows to
define properly the scalar quantitites κ(s) and τ (s). In order to compute ṅ we observe that
b = t × n implies n = b × t so

ṅ(s) = ḃ × t + b × ṫ = τ n × t + κb × n = −τ b − κt

which proves the second relation.

19

You might also like