Professional Documents
Culture Documents
Full Ebook of Differential Geometry of Curves and Surfaces Third Edition Thomas F Banchoff Online PDF All Chapter
Full Ebook of Differential Geometry of Curves and Surfaces Third Edition Thomas F Banchoff Online PDF All Chapter
Full Ebook of Differential Geometry of Curves and Surfaces Third Edition Thomas F Banchoff Online PDF All Chapter
https://ebookmeta.com/product/topological-differential-and-
conformal-geometry-of-surfaces-universitext-norbert-acampo/
https://ebookmeta.com/product/tensor-algebra-and-analysis-for-
engineers-with-applications-to-differential-geometry-of-curves-
and-surfaces-1st-edition-paolo-vannucci/
https://ebookmeta.com/product/parametric-geometry-of-curves-and-
surfaces-architectural-form-finding-1st-edition-alberto-lastra/
https://ebookmeta.com/product/differential-geometry-of-plane-
curves-96th-edition-hilario-alencar/
A First Course in Differential Geometry Surfaces in
Euclidean Space Instructor Solution Manual Solutions
1st Edition Lyndon Woodward
https://ebookmeta.com/product/a-first-course-in-differential-
geometry-surfaces-in-euclidean-space-instructor-solution-manual-
solutions-1st-edition-lyndon-woodward/
https://ebookmeta.com/product/differential-geometry-1st-edition-
victor-v-prasolov/
https://ebookmeta.com/product/manifolds-vector-fields-and-
differential-forms-an-introduction-to-differential-geometry-1st-
edition-gal-gross/
https://ebookmeta.com/product/crc-standard-curves-and-surfaces-
with-mathematica-3rd-edition-von-seggern-david-h/
https://ebookmeta.com/product/differential-geometry-and-general-
relativity-volume-1-1st-edition-canbin-liang/
Differential Geometry
of Curves and Surfaces
Differential Geometry of
Curves and Surfaces
Third Edition
Thomas F. Banchoff
Stephen Lovett
Third edition published 2023
by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742
and by CRC Press
2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
© 2023 Stephen Lovett and Thomas F. Banchoff
First edition published by CRC Press 2010
Second edition published by CRC Press 2015
CRC Press is an imprint of Taylor & Francis Group, LLC
Reasonable efforts have been made to publish reliable data and information,
but the author and publisher cannot assume responsibility for the validity of
all materials or the consequences of their use. The authors and publishers have at-
tempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not
been obtained. If any copyright material has not been acknowledged please write
and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be
reprinted, reproduced, transmitted, or utilized in any form by any electronic,
mechanical, or other means, now known or hereafter invented, including photo-
copying, microfilming, and recording, or in any information storage or retrieval
system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, access
www.copyright.com or contact the Copyright Clearance Center, Inc. (CCC), 222
Rosewood Drive, Danvers, MA 01923, 978-750-8400. For works that are not
available on CCC please contact mpkbookspermissions@tandf.co.uk
Trademark notice: Product or corporate names may be trademarks or registered
trademarks and are used only for identification and explanation without intent to
infringe.
Preface vii
Acknowledgments xiii
Authors xv
v
vi Contents
Bibliography 359
Index 363
Preface
vii
viii Preface
Computer Applets
An integral part of this book is the access to on-line computer graph-
ics applets that illustrate many concepts and theorems introduced
in the text. Though one can explore the computer demos indepen-
dently of the text, the two are intended as complementary modes
of studying the same material: a visual/intuitive approach and an
analytical/theoretical approach. The text always motivates various
definitions and topics, but the graphical applets allow the reader to
explore examples further, and give a visual explanation for defini-
tions or complicated theorems. The ability to change the choice of
the parametric curve or the parametrized surface in an applet, or to
change other properties allows the reader to explore the concepts far
beyond what a static book permits.
Any element in the text (Example, Problem, Definition, Theo-
rem, etc.) that has an associated applet is indicated by the symbol
shown in this margin. Each demo comes with some explanation text.
The authors intended the applets to be intuitive enough so that after
using just one or two (and reading the supporting text) any reader
can quickly understand their functionality. At present the applets
run on a Java platform but we may add other formats possibly based
on standard computer algebra systems. As an interest to an instruc-
tor of this course, the Java applets are extensible in that they are
designed with considerable flexibility so that the reader can often
change whether certain elements are displayed or not. Often, there
are additional elements that one can display either by accessing the
Controls menu on the Demo window or the Plot/Add Plot menu on
any display window. Applets given in a computer algebra system are
naturally extensible through the capabilities of the given program.
In previous editions, the applets were delivered via Java embedded
into a browser. That is no longer the case. Now all the applets are
delivered via a single Maple workbook, using embedded components.
The user does not need the full Maple software, but only needs to
download the free MaplePlayer. The reader will find instructions to
download the MaplePlayer software and the workbook that accom-
panies this text at:
http://cs.wheaton.edu/∼slovett/diffgeo
Organization of Topics
Chapters 1 through 4 cover alternately the local and global theory
of plane and space curves. In the local theory, we introduce the
fundamental notions of curvature and torsion, construct various asso-
ciated objects (e.g., the evolute, osculating circle, osculating sphere),
and present the fundamental theorem of plane or space curves, which
x Preface
A Comment on Prerequisites
The mathematics or physics student often first encounters differential
geometry at the graduate level. Typically, at that point, one is imme-
diately exposed to the formalism of manifolds, thereby skipping the
intuitive and visual foundation that informs the deeper theory. The
Preface xi
Notation
As a comment on vector notation, this book consistently uses the fol-
lowing conventions. A vector or vector function in a Euclidean vector
⃗
space is denoted by ⃗v , X(t), ⃗
or X(u, v). Often γ indicates a curve
⃗ ⃗
parametrized by X(t) while writing X(t) = X(u(t),⃗ v(t)) indicates a
curve on a surface. The unit tangent and the binormal vectors of a
curve in space are written in the standard notation T⃗ (t) and B(t)
⃗ but
⃗ ⃗
the principal normal is written P (t), reserving N (t) to refer to the
unit normal vector to a curve on a surface. For a plane curve, U ⃗ (t)
⃗
is the vector obtained by rotating T (t) by a positive quarter turn.
Furthermore, we denote by κg (t) the curvature of a plane curve since
one identifies this curvature as the geodesic curvature in the theory
of curves on surfaces.
In this book, we often work with matrices of functions. The func-
tions themselves are denoted, for example, by aij , and we denote the
matrix by (aij ). Furthermore, it is essential to distinguish between
a linear transformation between vector spaces T : V → W and its
matrix with respect to given bases in V and W . Following common
notation in current linear algebra texts, if B is a basis in V and B ′ is a
B
basis in W , then we denote by T B′ the matrix of T with respect to
these bases. If the bases are understood by context, we simply write
[T ] for the matrix associated to T .
Occasionally, there arise irreconcilable discrepancies in definitions
or notations (e.g., the definition of a critical point for a function
Rn → Rm , or how one defines θ and ϕ in spherical coordinates).
In these instances the authors made a choice that best suits their
purposes and indicated commonly used alternatives.
xii Preface
Section Dependency
The topics in this book primarily oscillate back and forth between lo-
cal theory and global theory, first of plane curves, then space curves,
and then surfaces in R3 . In the authors’ perspective, the Gauss-
Bonnet Theorem (presented in Sections 8.2 and 8.3) serves as the
climax theorem of the textbook. Sections depend on each other ac-
cording to the following chart.
1.1–1.5
2.1–2.4 3.1–3.4
4.1–4.3 5.1–5.5
7.3 8.1–8.5
8.6–8.7 9.1–9.2
Thomas Banchoff
Our work at Brown University on computer visualizations in differen-
tial geometry goes back more than forty-five years, and I acknowledge
my collaborator computer scientist Charles Strauss for the first fifteen
years of our projects. Since 1982, an impressive collection of students
have been involved in the creation and development of the software
used to produce the applets for this book. The website for my 65th
birthday conference
http://tombanchoff.com
Stephen Lovett
I would first like to thank Thomas Banchoff my teacher, mentor,
and friend. After one class, he invited me to join his team of stu-
xiii
xiv Acknowledgments
xv
CHAPTER 1
1.1 Parametrizations
Borrowing from a physical understanding of motion in the plane, we
can think about plane curves by specifying the coordinates x and y
as functions of a time variable t, which give the position of a point
traveling along the curve. Thus we need two functions x(t) and y(t).
Using vector notation to locate a point on the curve, we often write
⃗
X(t) ⃗
= (x(t), y(t)) for this pair of coordinate functions and call X(t)
2
a vector function into R . From a mathematical standpoint, t does
not have to refer to time and is simply called the parameter of the
vector function.
DOI: 10.1201/9781003295341-1 1
2 1. Plane Curves: Local Properties
⃗v
p⃗ + 2⃗v
p⃗
is called a direction vector of the line because all vectors along this
line are multiples of ⃗v . Then every point on the line can be written
with a position vector p⃗1 + t⃗v for some t ∈ R. Figure 1.1 shows
an example with t = 2. Therefore, we find that a line can also be
described by providing a point and a direction vector.
Using the coordinates of vectors, given a point p⃗ = (x0 , y0 ) and a
direction vector ⃗v = (v1 , v2 ), a line through p⃗ in the direction of ⃗v is
the image of the following vector function:
⃗
X(t) = p⃗ + t⃗v = (x0 + v1 t, y0 + v2 t) for t ∈ R.
Note that just as the same line can be determined by two different
pairs of points, so the same line may be specified by different sets of
these equations. For example, using any point p⃗ on the line will ulti-
mately trace out the same line as t varies through all of R. Similarly,
if we replace ⃗v with any nonzero multiple of itself, the set of points
traced out as t varies through R is the same line in R2 plane.
traces out a circle of radius R > 0 centered at the point (a, b). To
see this, note that by the definition of the sin t and cos t functions,
(cos t, sin t) are the coordinates of the point on the unit circle that
is also on the ray out of the origin that makes an angle t with the
positive x-axis. Thus,
⃗ 1 (t) = (cos t, sin t),
X with t ∈ [0, 2π],
1.1. Parametrizations 3
if for all ε > 0 there exists a δ > 0 such that 0 < |t − a| < δ implies
⃗
∥X(t) − w∥
⃗ < ε.
⃗
lim X(t) ⃗
= X(a).
t→a
⃗
lim X(t) =w
⃗ if and only if lim x(t) = w1 and lim y(t) = w2 .
t→a t→a t→a
⃗
Proof : Suppose first that limt→a X(t) = w.
⃗ Let ε > 0 be arbitrary
and let δ > 0 satisfy the definition of the limit of the vector function.
⃗ − w∥
Note that |x(t) − w1 | ≤ ∥X(t) ⃗ − w∥.
⃗ and that |y(t) − w2 | ≤ ∥X(t) ⃗
Hence, 0 < |t − a| < δ implies |x(t) − w1 | < ε and |y(t) − w2 | < ε.
Thus, limt→a x(t) = w1 and limt→a y(t) = w2 .
Conversely, suppose that limt→a x(t) = w1 and limt→a y(t) = w2 .
Let ε > 0 be an arbitrary positive real number. By definition, there √
exist δ1 and δ2 such that 0 < |t − a| < δ1 √ implies |x(t) − w1 | < ε/ 2
and 0 < |t−a| < δ2 implies |y(t)−w2 | < ε/ 2. Taking δ = min(δ1 , δ2 )
we see that 0 < |t − a| < δ implies that
r
⃗
p
2 2
ε2 ε2
∥X(t) − w∥
⃗ = |x(t) − w1 | + |y(t) − w2 | < + = ε.
2 2
This finishes the proof of the proposition. □
It is easy to see that for all t ∈ R, x(t)2 + y(t)2 = 1. This means that
the locus of X ⃗ is on the unit circle. However, this parametrization
does not trace out the entire circle as it misses the point (−1, 0). We
leave it as an exercise to determine a geometric interpretation of the
parameter t and to show that
⃗
lim X(t) ⃗
= lim X(t) = (−1, 0).
t→∞ t→−∞
6 1. Plane Curves: Local Properties
x(t)2 y(t)2
2
+ 2 =1 for all t.
a b
where m and n are positive integers. See Figure 1.2 for an example
of a Lissajous figure with m = 5 and n = 3.
One can point out that reflectors on bicycle wheels are usually not
attached directly on the tire but on a spoke of the wheel. We can
easily modify the above discussion to obtain the relevant parametric
equations for when the point of light is located at a distance b from
the center of the rolling wheel. We obtain
⃗
X(t) = (at − b sin t, a − b cos t).
⃗
Whether or not b < a, the curve traced out by X(t) is called a
trochoid .
⃗
X(t) = (f (t) cos t, f (t) sin t).
Problems
In Problems 1.1.1 through 1.1.4, sketch the curve with the given parametric
equations. Indicate with an arrow the direction in which t increases.
⃗
1. X(t) = (t3 , t2 ).
⃗
2. X(t) = (t2 , cos t).
⃗
3. X(t) = (t cos t, t sin t).
⃗
4. X(t) = (t2 − 1, t3 − t).
1 − t2 2t
⃗ = (x(t), y(t)) =
X , , for t ∈ R.
1+t 2 1 + t2
6. Let C be the circle of center (0, 2) and radius 1. Let S be the set of
points in R2 that are the same distance from the outside of circle C
as they are from the x-axis. Show the following.
⃗ 3 sin t 3 cos t
(a) S has a parametrization X(t) = ,2 − .
1 + cos t 1 + cos t
[Hint: Use the parameter t as the angle as shown below.]
(b) S is a parabola. Find the Cartesian equation for it.
10 1. Plane Curves: Local Properties
(0, 2)
t
⃗
X(t)
14. Consider the parametrized curve X⃗ : [0, 2π] → R2 that gives a defor-
⃗
mation of a cardioid X(t) = ((a + cos t) cos t, (a + cos t) sin t), where
a is a real number. Show that the curve intersects itself for |a| < 1.
Describe what happens for a = 0.
15. Let Γ be a circle and let O a point on Γ. Set L to be the line tangent
to Γ such that its point of tangency is diametrically opposite from O.
The cissoid of Diocles relative to Γ and O is the set of points P such
←→
that OP = AB, where A is the intersection of OP and Γ and where
←→
B is the intersection of OP and L. Using the setup in the following
diagram, find parametric equations for the cissoid of Diocles.
y
B
A
O a x
Γ L
⃗
form the standard reference frame. We call the vector function X(t)
the position vector. When one uses the standard reference frame, it
is not uncommon to write
⃗
X(t) = x(t)⃗ı + y(t)⃗ȷ.
⃗ 0)
factor of t1 − t0 . Hence, the vectorial rate of change between X(t
⃗ 1 ) is
and X(t
1 ⃗
⃗ 0) .
X(t1 ) − X(t
t1 − t0
To define an instantaneous rate of change along the curve at the
point t0 , we need to calculate (if it exists and if it has meaning) the
limit
⃗ ′ (t0 ) = lim 1 X(t
X ⃗ 0 + h) − X(t
⃗ 0) . (1.2)
h→0 h
⃗
According to Proposition 1.1.6, if X(t) = (x(t), y(t)), then the limit in
Equation (1.2) exists if and only if x(t) and y(t) are both differentiable
⃗ ′ (t0 ) = (x′ (t0 ), y ′ (t0 )).
at t0 . Furthermore, if this limit exists, then X
This leads us to the following definition.
C 0 ⊃ C 1 ⊃ C 2 ⊃ · · · ⊃ C ∞.
⃗
3. If X(t) = ⃗v (t) + w(t),
⃗ then
⃗ ′ (t) = ⃗v ′ (t) + w
X ⃗ ′ (t).
⃗
4. If X(t) = ⃗v f (t) is a vector function and f : J → I is a real
function into I, then
⃗ ′ (t) = f ′ (t)⃗v ′ f (t) .
X
⃗
Thus, the angle θ(t) between X(t) ⃗ ′ (t) is defined for all t ̸= 0
and X
and is equal to
!
⃗
X(t) ·X⃗ ′ (t)
t
−1
θ(t) = cos = cos−1 √ .
⃗
∥X(t)∥ ⃗ ′ (t)∥
∥X |t| 1 + t2
⃗ ′ (t0 )
X
⃗
∆X
⃗ 1)
X(t
⃗ 0)
X(t
O
Figure 1.5: Arc length segment.
⃗
Let C be the locus of a vector function X(t) for t ∈ [a, b]. Then
we approximate the length of the arc l(C) with the Riemann sum
n
X n
X
l(C) ≈ ∥X(t ⃗ i−1 )∥ ≈
⃗ i ) − X(t ⃗ ′ (ti )∥∆t.
∥X
i=1 i=1
⃗ i) −
(See Figure 1.5 as an illustration for the approximation ∥X(t
⃗ ⃗ ′
X(ti−1 )∥ ≈ ∥X (ti )∥∆t.) Taking the limit of this Riemann sum, we
obtain the following formula for the arc length of C:
Z bp
l= (x′ (t))2 + (y ′ (t))2 dt. (1.3)
a
A rigorous justification for the arc length formula, using the Mean
Value Theorem, can be found in most calculus textbooks.
In light of Equation (1.3), we define the arc length function s :
⃗
[a, b] → R related to X(t) as the arc length along C over the interval
[a, t]. Thus,
Z t Z tp
s(t) = ⃗ ′ (u)∥ du =
∥X (x′ (u))2 + (y ′ (u))2 du. (1.4)
a a
⃗
Proposition 1.2.8 If X(t) is a regular parametrized curve, then there
is a regular reparametrization of X⃗ by arc length. Furthermore, if X ⃗
1.2. Position, Velocity, and Acceleration 17
The Inverse Function Theorem as used in the above proof (and its
more general multivariable counterpart) appear in most introductory
texts on analysis. (See [7] for example.) However, the key derivative
in Equation (1.7) follows from the chain rule for invertible functions
and is a standard differentiation formula.
The habit of reparametrizing by arc length has benefits and draw-
backs. The great benefit of this approach is that along a curve para-
metrized by arc length, the speed function s′ is identically 1 and the
velocity vector is exactly the tangent vector
⃗ ′ (s) = T⃗ (s).
X
This formulation simplifies proofs and difficult calculations. The main
drawback is that, in practice, most curves do not admit a simple for-
mula for their arc length function, let alone a formula that can be
written using elementary functions. For example, given an explicit
⃗
parametrized curve X(t), it is often very challenging to find s(t) as de-
fined in Equation (1.4). Furthermore, to reparametrize by arc length,
it would be necessary to find the inverse function t(s), representing
the original parameter t as a function of s. Even if it is possible to
find s(t), determining this inverse function usually cannot be written
with elementary functions. Then the parametrization by arc length
⃗
is X(s) ⃗
= X(t(s)). Even using a computer algebra systems (CAS),
reparametrizing by arc length remains an intractable problem.
18 1. Plane Curves: Local Properties
⃗ t (t) = X(t
L ⃗ 0 ) + (t − t0 )X
⃗ ′ (t0 ), with t ∈ R,
0
or alternatively
⃗ t (u) = X(t
L ⃗ 0 ) + uT⃗ (t0 ), with u ∈ R, (1.8)
0
Consequently, we can determine the angle the curve makes with itself
at the corner t0 as
α0 = cos−1 T⃗ (t+ ⃗ −
0 ) · T (t0 ) .
Example 1.2.10 Let X(t) ⃗ = (t, | tan t|). We calculate that X⃗ ′ (t) =
2
(1, sign(t) sec t) and deduce that t = 0 is a critical point because
sign(t) is not defined at 0, and hence X ⃗ is not differentiable there.
However, we also find that
1 1
T⃗ (0− ) = √ (1, −1) and T⃗ (0+ ) = √ (1, 1).
2 2
Thus,
T⃗ (0− ) · T⃗ (0+ ) = 0,
which shows that X ⃗ makes a right angle with itself at t = 0. However,
one can tell from the explicit values for T⃗ (t− ⃗ +
0 ) and T (t0 ) that the
π
exterior angle of the curve at t = 0 is 2 .
If t0 is a critical point, it may still happen that
lim T⃗ (t)
t→t0
When this limit exists, even though t0 is a critical point, one can
use Equation (1.8) as parametric equations for the tangent line at t0 ,
replacing T⃗ (t0 ) in Equation (1.8) with limt→t0 T⃗ (t).
Though we will not make use of the following comments until
Chapter 2, we finish this chapter on calculus concepts with a quick
review of notation for integration.
In geometry, physics, and other applications, we must sometimes
integrate a function along a curve. To this end, we use path and
line integrals of scalar functions or vector functions, depending on
the particular problem. Since we will make use of them, we remind
the reader of the notation for such integrals. Let C be a curve para-
⃗
metrized by X(t) over the interval [a, b]. Let f : R2 → R be a real
(scalar) function and F⃗ : R2 → R2 a vector field in the plane. Then
the path integral of f and the line integral of F⃗ over the curve C are
respectively
Z Z b
def p
f ds = f (x(t), y(t)) (x′ (t))2 + (y ′ (t))2 dt,
C a
Z Z b
def
F⃗ · d⃗s = F⃗ (x(t), y(t)) · T⃗ (t) dt.
C a
20 1. Plane Curves: Local Properties
Problems
which gives the distance between a point X(t) ⃗ on the line and
the point (x0 , y0 ).
(c) Show that f (t0 ) simplifies to the distance formula given above
for d.
⃗ be a fixed point, and let ⃗l(t) = ⃗at+⃗b be the parametric equations
6. Let p
of a line. Prove that the distance between p ⃗ and the line is
s
(⃗a · (⃗b − p
⃗))2 ∥⃗a × (⃗b − p
⃗)∥
∥⃗b − p
⃗∥2 − = ,
∥⃗a∥2 ∥⃗a∥
19. (*) Consider the cycloid introduced in Example 1.1.13 given by X(t) ⃗ =
(t − sin t, 1 − cos t). Prove that the path taken by a point on the edge
of a rolling wheel of radius 1 during one rotation has length 8.
20. Calculate the arc length function of the curve X(t) ⃗ = (t2 , ln t), de-
fined for t > 0.
21. Let X⃗ : I → R2 be a regular parametrized curve, and let p ⃗ be a fixed
point. Suppose that the closest point on the curve X ⃗ to p
⃗ occurs at
t = t0 , which is neither of the ends of I. Prove that the line between
p
⃗ and the point closest to p ⃗ 0 ), is perpendicular to the
⃗, namely X(t
curve X ⃗ at t = t0 .
22. Prove the differentiation formulas in Proposition 1.2.3.
⃗
23. Prove that if X(t) ⃗ ·X
is a curve that satisfies X ⃗ ′ = 0 for all values of
⃗ is a circle. [Hint: Use ∥X∥
t, then X ⃗ =X
2 ⃗ ·X
⃗ and apply Proposition
⃗ 2 .]
1.2.3 to calculate the derivative of ||X||
22 1. Plane Curves: Local Properties
⃗
24. Consider the ellipse given by X(t) = (a cos t, b sin t). Find the ex-
tremum values of the speed function.
25. Consider the linear spiral of Example 1.2.4. Let n ≥ 0 be a non-
negative integer. Prove that the length of the nth derivative vector
function is given by
p
⃗ (n) (t)∥ =
∥X n2 + t 2 .
⃗
x(t) = (x(t), y(t)) = (r(t) cos(θ(t)), r(t) sin(θ(t))).
1.3 Curvature
Let X⃗ : I → R2 be a twice-differentiable parametrization of a curve
C. As we saw in the previous section, the decomposition of the ve-
⃗ ′ = s′ T⃗ into magnitude and unit tangent direction
locity vector X
separates the geometric invariant (the unit tangent T⃗ ) from the dy-
namical aspect (the speed s′ (t)) of the parametrization. Taking one
more derivative, we obtain the decomposition
⃗ ′′ = s′′ T⃗ + s′ T⃗ ′ .
X (1.9)
(T⃗ · T⃗ )′ = 0 =⇒ 2T⃗ · T⃗ ′ = 0,
=⇒ T⃗ · T⃗ ′ = 0.
Thus, T⃗ ′ is perpendicular to T⃗ .
Just as there are two unit tangent vectors at a regular point of the
curve, there are two unit normal vectors as well. Given a particular
parametrization, there is no naturally preferred way to define “the”
unit normal vector, so we make a choice.
⃗ ′′ = s′′ T⃗ + (s′ )2 κg U
X ⃗. (1.11)
⃗ = ⃗k.
Viewing the plane as the xy-plane in three-space, we have T⃗ × U
Thus,
⃗ ′×X
X ⃗ ′′ = (s′ T⃗ ) × (s′′ T⃗ + (s′ )2 κg U
⃗)
= s′ s′′ T⃗ × T⃗ + (s′ )3 κg T⃗ × U⃗
= (s′ )3 κg⃗k.
⃗ ′ (t) × X
(X ⃗ ′′ (t)) · ⃗k x′ (t)y ′′ (t) − x′′ (t)y ′ (t)
κg (t) = = . (1.12)
∥X ⃗ ′ (t)∥3 (x′ (t)2 + y ′ (t)2 )3/2
X⃗ ′ (s) = T⃗ (s),
⃗ ′′ (s) = κg (s)U
X ⃗ (s).
R2 ω 3 sin2 t + R2 ω 3 cos2 t R2 ω 3 1
κg (t) = 2 2 2 2 2 2 3/2
= 3 3
= .
(R ω sin t + R ω cos t) R ω R
the speed function, one recovers the common formula for centripetal
acceleration of
v2
(s′ )2 κg = .
R
In introductory physics courses, this formula is presented only in the
context of circular motion. With differential geometry at our disposal,
we see that centripetal acceleration is always equal to v 2 /R at all
points on a curve where κg ̸= 0, where by R one means the radius of
curvature.
In contrast to physics where one tends to refer to the radius of cur-
vature, the curvature function is a more geometrically natural quan-
tity to study. Indeed, the radius of curvature of a line segment is
undefined even though line segments are such useful geometric ob-
jects. On the other hand, a radius of curvature is 0 (and hence the
curvature is undefined) at degenerate curves that are points or at
critical points.
Example 1.3.4 Let I be a real interval and consider the plane curve
given as the graph y = f (x) of a twice-differentiable function f : I →
R. We can parametrize this as X(t)⃗ = (t, f (t)) for t ∈ I. A quick
application of (1.12) gives a curvature function of
f ′′ (t)
κg (t) = .
(1 + f ′ (t)2 )3/2
B H
B
C A C
F G
A
E
D E
G
H D F
p
precisely when f ′′ (x) = 0. Recalling from calculus that 1 + f ′ (x)2 is
the speed function s′ (x) along the curve, we see from yet another angle
that the curvature function of a function graph is κg = f ′′ (x)/s′ (x)3 .
X⃗ ′ (t) = φ′ (t)⃗b,
⃗ ′′ (t) = φ′′ (t)⃗b.
X
because ⃗b × ⃗b = ⃗0.
Conversely, if X(t)⃗ = (x(t), y(t)) is a curve such that κg (t) = 0,
then
x′ (t)y ′′ (t) − x′′ (t)y ′ (t) = 0.
We need to find solutions to this differential equation or determine
how solutions are related. Since X ⃗ is regular, X
⃗ ′ ̸= ⃗0 for all t ∈ I.
′
Let I1 be an interval where y (t) ̸= 0. Over I1 , we have
x′ y ′′ − x′′ y ′ d x′ x′
= = 0 =⇒ = C,
(y ′ )2 dt y ′ y′
where C is a constant. Thus, x′ = Cy ′ for all t. Integrating with
respect to t we deduce x(t) = Cy(t)+D. Similarly, over an interval I2
where x′ (t) ̸= 0, we deduce that y(t) = Ax(t) + B for some constants
A and B. Since I can be covered with intervals where x′ (t) ̸= 0 or
⃗ is a piecewise linear curve.
y ′ (t) ̸= 0, we deduce that the locus of X
⃗
However, since X is regular, it has no corners, and hence its locus is
a line segment. □
⃗ ′ = (U
U ⃗ ′ · T⃗ )T⃗ + (U
⃗ ′·U
⃗ )U
⃗.
28 1. Plane Curves: Local Properties
Consequently,
⃗ ′ (t) = −s′ (t)κg (t)T⃗ (t).
U
Since {T⃗ , U
⃗ } forms an orthonormal basis of R2 for all t, every
⃗
higher derivative of X(t), if it exists, can be expressed as a linear
combination of T⃗ and U ⃗ . The components of X ⃗ (n) (t) in terms of
T⃗ and U ⃗ involve sums of products of derivatives of s(t) and κg (t).
Furthermore, if X ⃗ is parametrized by arc length, then the coefficients
⃗ (n)
of X (s) only involve sums of powers of derivatives of κg (s).
We note that if a curve X ⃗ : [−a, a] → R2 is reparametrized by
⃗
X(−t), then the modified curvature function is −κg (t). Thus, the sign
of the curvature depends on what one might call the “orientation”
of the curve. Excluding this technicality, curvature has a physical
interpretation that one can eyeball on particular curves. If a curve
is almost a straight line, then the curvature is close to 0, but if a
regular curve bends tightly along a certain section, then the curvature
is high (in absolute value). Of particular interest are points where the
curvature reaches a local extremum.
Problems