Download as pdf or txt
Download as pdf or txt
You are on page 1of 79

Contents

1 CURVES IN SPACE 1
1.1 Arc Length and Natural parameter . . . . . . . . . . . . . . . 3
1.2 Contact of Curves . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Tangent to a curve . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Contact of a Curve and a Plane . . . . . . . . . . . . . . . . . 7
1.5 The Frenet Trihedron . . . . . . . . . . . . . . . . . . . . . . . 10
1.6 Curvature and Torsion . . . . . . . . . . . . . . . . . . . . . . 12
1.7 Serret-Frenet Formulas . . . . . . . . . . . . . . . . . . . . . . 16
1.8 Fundamental Theorem For Space Curves . . . . . . . . . . . . 18

2 SURFACES 20
2.1 The Tangent Plane . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2 The First Fundamental Form Of A Surface . . . . . . . . . . . 27
2.3 Length Of A Tangent Vector And Angle Between Two Tangent
Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4 Area Of A Surface . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5 Change Of Curvilinear Coordinates . . . . . . . . . . . . . . . 34

3 THE SECOND FUNDAMENTAL FORM 37


3.1 Second Fundamental Form . . . . . . . . . . . . . . . . . . . . 38
3.2 Gaussian Map and Gaussian Curvature
(Spherical Map) . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3 Formulas of Gauss and Weingarten . . . . . . . . . . . . . . . 43
3.4 Christoffel symbols . . . . . . . . . . . . . . . . . . . . . . . . 44
3.5 Codazzi Equations and Gauss Theorem . . . . . . . . . . . . . 46

4 CURVE OF SURFACES 50
4.1 Curvature Of A Curve On A Surface . . . . . . . . . . . . . . 50

1
4.2 Geodesic Curvature . . . . . . . . . . . . . . . . . . . . . . . . 52
4.3 Geodesic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.4 Normal Curvature . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.5 Principal Curvatures, Gaussian and Mean Curvatures . . . . . 60
4.6 Principal Directions and Lines of
Curvature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.7 Asymptotic Lines . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.8 Conjugate Direction . . . . . . . . . . . . . . . . . . . . . . . 67

5 DEVELOPABLE SURFACES 69
5.1 Envelopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.2 Developable Surfaces . . . . . . . . . . . . . . . . . . . . . . . 71
5.3 Ruled Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . 74

Index 76
Chapter 1

CURVES IN SPACE

Definition 1.0.1. Simple Arc.


A simple arc is a continuous one-to-one mapping of a closed interval [a, b]
of the real line into the Euclidean space of 3-dimensions E3 .
Definition 1.0.2. Curve.
A curve is a continuous locally one-to-one mapping of an interval into E3 .
A mapping of an interval (a, b) (or [a, b), (a, b], [a, b]) into E3 is called
locally 1-1 if ∀ t0 ∈ (a, b) ∃ δ > 0 such that the mapping restricted to
[t0 − δ, t0 + δ] is 1-1.
Thus, if an origin is chosen, a curve is the set of points whose position
vectors are the values of a continuous vector-valued function which is locally
one-to-one.
A curve is therefore represented by a continuous vector valued function

r̄ = r̄(t)

which is locally 1-1. This is called the parametric representation of the curve
and the variable t is called ”Parameter”. This vector equation is equivalent
to three scalar equations

x = x(t), y = y(t), z = z(t)

which express the components of the position vector r̄(t).


Definition 1.0.3. Regular curve.
A parametric curve r̄ = r̄(t) is called regular of class C n if r̄(t) is a function
of class C n (i.e., n derivatives exist and are continuous) and r̄(t)˙ 6= 0 ∀t.
Curves in space Arc length and natural parameter

Definition 1.0.4. Piece-wise regular curve.


A curve is called piece-wise regular of class C n if there is a parametric
representation r̄ = r̄(t), t ∈ (a, b), such that the interval (a, b) can be di-
vided by points a = t0 < t1 < t2 < · · · < tn = b so that r̄(t) is of class
˙ 6= 0. The
C n in every open interval (tk , tk+1 ) and satisfies the condition r̄(t)
points t1 , t2 , · · · , tn−1 for which this is not true are called singular points of
the curve r̄ = r̄(t).

Example 1.0.1. The graph of a continuous function y = f (x), defined


on a closed interval [a, b], in xy-plane is a simple arc.
This graph corresponds to the vector valued function

r̄(t) = tî + f (t)ĵ, a ≤ t ≤ b.

This vector valued function is continuous and obviously t1 6= t2 implies that

r̄(t1 ) = t1 î + f (t1 )ĵ 6= t2 î + f (t2 )ĵ = r̄(t2 ).

Example 1.0.2. The set of points of E3 which simultaneously satisfies


two equations
y = f (x), z = g(x),
where f (x) and g(x) are continuous functions defined on closed interval [a, b],
is a simpl arc.

Example 1.0.3. The circle is not a simple arc because every continuous
mapping of a closed interval onto the circle must map atleast two distinct
points of the interval to the same point, thus the mapping fails to be one-to-
one.

2
Curves in space Arc length and natural parameter

1.1 Arc Length and Natural parameter


It is a known result in calculus that the length of a piece-wise regular curve
r̄ = r̄(t), a ≤ t ≤ b is given by the formula

Zb
l= ˙
|r̄(t)|dt (1.1)
a

or, in coordinates

Z b " 2  2  2 #1/2
dx dy dz
l= + + dt. (1.2)
dt dt dt
a

If we now introduce a new parameter s by the formula


Zt
s= ˙
|r̄(t)|dt, (1.3)
t0

we can show that s is a monotonic increasing function of t, and hence we can


write the curve as r̄ = r̄(s).

The parameter s is called the natural parameter.

We can write
ds2 = dx2 + dy 2 + dz 2 = dr̄.dr̄.
Hence,  2
ds dr̄ dr̄
= · ˙ 2.
= |r̄|
dt dt dt
From (1.3), we have  
ds
˙
= |r̄| (1.4)
dt
at regular points of the representation r̄ = r̄(t).
Now,
dr̄ dr̄ dt r̄˙
r̄0 = = · = (1.5)
ds dt ds ˙
|r̄|

3
Curves in space Contact of curves

which shows that r̄0 = dr̄



ds
is a unit vector, i.e.,

r̄0 · r̄0 = 1 or |r̄0 | = 1. (1.6)

Consequently, for the natural parametrization, we can’t have r̄0 = 0 and the
derivative r̄0 doesn’t exist at an essentially singular points. Thus we have,

Theorem 1.1.1. The derivative of the position vector at a point of the


curve relative to the natural parameter, if it exists, is a unit vertor. At
essential singular points of a paremetric curve, this derivative doesn’t exist.

1.2 Contact of Curves


Comparison of functions which tend to zero:
Let α and β be functions. We say that α tends to zero faster than β or that
α is of higher order than β, in symbols

|α|
α = 0(β) iff → 0.
|β|

The symbol 0(β) does not denote any particular function, it stands for any
function α with the property that it is of higher order than β. Consequently,
the equality sign α = 0(β) means that α belongs to the class of functions
tending to zero faster than β and the equations α = 0(β), γ = 0(β) do not
imply that α = γ.
If the ratio |α|
|β|
is bounded in a neighbourhood of the limit values of the
arguments we say that α tends to zero at least as fast as β or that α is of
order not lower than β and we write,

α = 0(β).

If α = 0(β) and β = 0(α) simultaneously, we say that α and β are of the


same order.

Contact of a Curve with another curve:


Consider two curves C1 and C2 with natural parametrization, C1 : r̄ =
r̄1 (s) and C2 : r̄ = r̄2 (σ).

4
Curves in space Contact of curves

Let P0 be a common point on them which corresponds to the values s0 and


σ0 of the natural parameters s and σ of C1 and C2 i.e.,

P0 = r̄1 (s0 ) = r̄2 (σ0 ).

Take points P1 and P2 on C1 and C2 respectively at distances h from P0 so


that P1 = r̄1 (s0 + h) and P2 = r̄2 (σ0 + h).
Definition 1.2.1. The curves C1 and C2 have a contact of order n if
P1 P2 = 0(hn ) but P1 P2 6= 0(hn+1 ).
P1 P2 = 0(hn ) means lim Ph1 nP2 = 0.
h→0
|α|
In general, α = 0(β) if lim = 0.
β→0 |β|

We now have,
Theorem 1.2.1. The two curves, r̄ = r̄1 (s) and r̄ = r̄2 (σ), regular of
class C n+1 have a contact of order n at a nonsingular point p0 if and only if
at P0
(n) (n) (n+1) (n+1)
r̄1 = r̄2 , r̄10 = r̄20 , ....., r̄1 = r̄2 , r̄1 6= r̄2 .
Proof: As the two curves are regular of class C n+1 , by Taylors theorem, we
have
n (n) hn+1 (n+1)
r̄1 (s0 + h) = r̄1 (s0 ) − hr̄10 (s0 ) + · · · + hn! r̄1 (s0 ) + (n+1)! r̄1 (s0 ) + 0(hn+1 ),

hn (n) hn+1 (n+1)


r̄2 (σ0 + h) = r̄2 (σ0 ) − hr̄20 (σ0 ) + · · · + r̄ (σ0 )
n! 1
+ r̄
(n+1)! 2
(σ0 ) + 0(hn+1 ).

Now,
P1 P2 = r̄1 (s0 + h) − r̄2 (σ0 + h).
n (n) (n)
⇒ P1 P2 = [r̄1 (s0 ) − r̄2 (σ0 )] + h[r̄10 (s0 ) − r̄20 (σ0 )] + · · · + hn! [r̄1 (s0 ) − r̄2 (σ0 )] +
hn+1 (n+1) (n+1)
n+1! 1
[r̄ (s0 ) − r̄2 (σ0 )] + 0(hn+1 ).

If P1 P2 = 0(hn ) but P1 P2 6= 0(hn+1 ), the expression must start from the


n+1 (n+1) (n+1)
term hn+1! [r̄1 (s0 ) − r̄2 (σ0 )].

Hence the necessary and sufficient condition is


(n) (n) (n+1) (n+1)
r̄1 (s0 ) = r̄2 (σ0 ), r̄10 (s0 ) = r̄20 (σ0 ), · · · , r̄1 (s0 ) = r̄2 (σ0 ), r̄1 (s0 ) 6= r̄2 (σ0 ).

This completes the proof.

5
Curves in space Tangent to a curve

1.3 Tangent to a curve


Definition 1.3.1. A tangent to a curve at a point P is a straight line
which has a contact of order at least one with the curve at P .
Equation of the tangent:
Let the curve be r̄ = r̄(s) and the point p be given by r̄(s0 ). The equation
of a straight line through P is of the form
~ = σ~a + r̄(s0 ),
R
where ~a is the unit vector along the line and σ its natural parameter.

By theorem (1.2.1), this line has a contact of first order with the curve iff
~ ~ 0 (0) = r̄0 (s0 ). The first relation is obviously satisfied and
R(0) = r̄(s0 ) and R
the second relation gives ~a = r̄(s0 ). So, r̄0 (s0 ) is the directional unit vector
of the line.
Thus, the equation of tangent becomes

R̄ = r̄(s0 ) + σr̄0 (s0 ). (1.7)

The equation (1.7) can also be written as


0
(R̄ − r̄(s0 )) × r̄ (s0 ) = 0.

If a curve is given in a general parametrization r = r̄(t), then r̄˙ = dr


dt
is also
the directional vector of the tangent but not necessarily a unit vector and
the parametric equation of the tangent is
˙ 0 ).
R̄ = r̄(t0 ) + σ r̄(t (1.8)

The equation (1.8) is equivalent to


˙ = 0.
(R̄ − r̄(t0 )) × r̄(t)

If the equation of the curve is given in cartesian coordinates, x = x(t), y =


y(t), z = z(t), then the equation of the tangent at t = t0 will be
X − x(t0 ) Y − y(t0 ) Z − z(t0 )
= − . (1.9)
ẋ(t0 ) ẏ(t0 ) ż(t0 )

Exercise 1.3.1. Find the unit tangent vector and the equation of the
tangent to the following curves;

6
Curves in space Contact of a curve and a plane

1. Circle: x = a cos u, y = a sin u, z = 0.

2. Circular Helix: z = a cos u, y = a sin u, z = bu.

3. z = a(a + cos u), y = a sin u, z = 2a sin u2 .

1.4 Contact of a Curve and a Plane


Let P0 be a common point of a curve C and a plane S. Let P be a point
on the curve and the arc length P0 P be h, and let d be the perpendicular
distance of P from the plane S.

Definition 1.4.1. The curve C has a contact of order at least n at p0


with the plane if d = 0(hn ).

Definition 1.4.2. Osculating plane.


An osculating plane at a point P0 of the curve C is the plane which has a
contact of order at least 2 with the curve at P0 .

Equation of the osculating plane:


Let the curve C be r̄ = r̄(s) and let the point P0 be r̄(s0 ). The equation of
a plane through P0 is
(R̄ − r̄(s0 )) · m̄ = 0,
where m̄ is the unit normal to the plane. The distance of the point P with
position vector r̄(s0 + h) from the plane is

d = |(r̄(s0 + h) − r̄(s0 )) · m̄|.

By Taylor’s theorem, we get

h2 0
r̄(s0 + h) − r̄(s0 ) = hr̄0 (s0 ) + r̄ (s0 ) + 0(h2 ).
2
Hence,
h2 00
d = hr̄0 (s0 ) · m̄ +
r̄ (s0 ) · m̄ + 0(h2 ).
2
Therefore, the contact is of order 2 or more if and only if

r̄0 (s0 ) · m̄ = 0 and r̄00 (s0 ) · m̄ = 0.

7
Curves in space Contact of a curve and a plane

From these equations, we have

r̄0 (s0 ) × r̄00 (s0 )


m̄ = .
|r̄0 (s0 ) × r̄00 (s0 )|

Thus, the equation of the osculating plane becomes

R̄ − r̄(s0 ) · (r̄0 (s0 ) × r̄00 (s0 ))


= 0.
|r̄0 (s0 ) × r̄00 (s0 )|

Which can be written as

[(R̄ − r̄(s0 )), r̄0 (s0 ), r̄00 (s0 )] = 0. (1.10)

In cartesian coordinates, this can be written as

X − x0 Y − y0 Z − z0
x00 y00 z00 = 0. (1.11)
00 00 00
x0 y0 z0

Note: If the curve is given in generalized parameter r̄ = r̄(t), then


 
0 dr̄ dt dt
r̄ = · ˙
= r̄ ,
dt ds ds
2 (1.12)
dt d2 t
r̄00 = r̄ + r̄˙ 2 .
ds ds

From the set of equations (1.12), we obtain

dt 3
r̄0 × r̄00 = (r̄˙ × r̄¨)( ).
ds
Hence (1.10) becomes
 3
  dt
˙ 0 ), r̄¨(t0 )
(R̄ − r̄(t0 )), r̄(t = 0,
ds

i.e. the equation of osculating plane can be written as

˙ 0 ), r̄¨(t0 )] = 0.
[(R̄ − r̄(t0 )), r̄(t (1.13)

8
Curves in space Contact of a curve and a plane

In cartesian coordinates (1.13) becomes

X − x(t0 ) Y − y(t0 ) Z − z(t0 )


ẋ0 (t0 ) ẏ 0 (t0 ) ż 0 (t0 ) = 0.
00 00
ẍ (t0 ) ÿ (t0 ) z̈ 00 (t0 )

Here, R̄ denotes the position vector of a point of the osculating plane and
X, Y, Z denote its coordinates.
Note: The osculating plane is not determined when r̄00 (s0 ) = 0 or when
r̄0 (s0 ) is proportional to r̄00 (s0 ).

Exercise 1.4.1. Find the osculating plane of the following curves;


1. Circular helix: x = a cos u, y = a sin u, z = bu.

2. z = t, y = t − t2 , z = 2t.

3. z = y, x = 21 z 2 .

Answer: (1) az − abu + bx sin u − by cos u = 0.


(2) 2x − z = 0.
(3) x = y.

9
Curves in space The Frenet Trihedron

1.5 The Frenet Trihedron


We associate with every point of parametric curve of class C 2 an orthonormal
triplet of unit vectors; The tangent vector t̄, the principal normal
vector n̄ and the binormal vector b̄.
We know that unit tangent vector t̄ is

t̄ = r̄0 . (1.14)

Since r̄0 · r̄0 = 1, we get


r̄0 · r̄00 = 0,
i.e., r̄00 ⊥ t̄ and is therefore normal to the curve. Also, equation (1.10) shows
that r̄00 lies in the osculating plane.

Definition 1.5.1. Principal normal (t̄).


Principal normal n̄ is the unit normal which lies in the osculating plane.

Obviously,
r̄00
n̄ = . (1.15)
|r̄00 |

Definition 1.5.2. Binormal (b̄)


Binormal b̄ is the unit normal which is perpendicular to the osculating plane
such that the three vectors t̄, n̄, b̄ form a right handed system.

The Binormal b̄ is given by


r̄0 × r̄00
b̄ = t̄ × n̄ = . (1.16)
|r̄00 |

Some simple relations among t̄, n̄, b̄ are



t̄ · t̄ = 1 t̄ · n̄ = 0 t̄ · b̄ = 0 
n̄ · t̄ = 0 n̄ · n̄ = 1 n̄ · b̄ = 0 , (1.17)
b̄ · t̄ = 0 b̄ · n̄ = 0 b̄ · b̄ = 1


t̄ × t̄ = 0 t̄ × n̄ = b̄ t̄ × b̄ = −n̄ 
n̄ × t̄ = −b̄ n̄ × n̄ = 0 n̄ × b̄ = t̄ . (1.18)
b̄ × t̄ = n̄ b̄ × n̄ = −t̄ b̄ × b̄ = 0

10
Curves in space The Frenet Trihedron

Expressions for t̄, n̄, b̄ in generalized parameter:


We know that,
ds2 = dr̄ · dr̄,

 2
ds dr̄ dr̄
i.e., = · ˙ 2,
= |r̄|
dt dt dt
ds
or ˙
= kr̄k. (1.19)
dt
Hence, from (1.12), (1.17) and (1.19) we get

r̄˙ dt
t̄ = [∵ r0 = r̄˙ ]. (1.20)
˙
|r̄| ds

Also, from (1.12), we have

dt 3
r̄0 × r̄00 = (r̄˙ × r̄¨)( ).
ds
Since b̄ is a unit vector in the direction of r̄0 × r̄00 , i.e., in the direction of
r̄˙ × r̄¨, we get
r̄˙ × r̄¨
b̄ = . (1.21)
|r̄˙ × r̄¨|
Finally,
(r̄˙ × r̄¨) × r̄˙
n̄ = b̄ × t̄ = . (1.22)
|r̄˙ × r̄¨||r̄|
˙
The Three Planes:
The equation of the osculating plane can now be written as

(R̄ − r̄) · b̄ = 0.

The plane through n̄ and b̄ is called the normal plane, its equation be
written as
(R̄ − r̄) · t̄ = 0.
The plane through t̄ and b̄ is called the Rectifying plane, its equation can be
written as

(R̄ − r̄) · n̄ = 0.

11
Curves in space Curvature and Torsion

1.6 Curvature and Torsion


Definition 1.6.1. Curvature (κ):
Let ω denotes the angle between the tangent vectors at P : r̄(s) and at
Q : r̄(s + h). Then the limit of the ratio wh as h → 0 is called the Curvature
(κ) of the curve at P , i.e., κ is the arc rate turning of the tangent.

(a) (b)

Figure 1.1

Expression for k:
In figure 1.1(b), complete the ∆OAB by joining A and B, then draw
angle bisector OC of ∠AOB. it is easy to see that

|AC| = |BC|,

|AB|
or |BC| = .
2
Now, in ∆OCB,

ω |CB| |AB| 
sin( ) = = |CB| = , ∵ |OB| = 1, is a unit vector.
2 |OB| 2

ω
i.e., 2 sin( ) = |t̄(s + h) − t̄(s)|.
2
Hence,
ω ω |t̄(s + h) − t̄(s)|
κ = lim = lim ω × ,
h→0 |h| h→0 2 sin( ) |h|
2

i.e., κ = |t̄0 (s)| = |r̄00 (s)|. (1.23)

12
Curves in space Curvature and Torsion

From (1.15) we have,


r̄00 r̄00
r̄ = =
|r̄00 | κ
dt̄

= κn̄,
ds
dt̄
or κ = n̄ · . (1.24)
ds
If follows from (1.23) that in cartesian coordinates
s 2  2 2  2 2
d2 x dy dz
κ= 2
+ 2
+ . (1.25)
ds ds ds2

If the curve is given in generalized parameter, then

r̄˙
 
0 dt̄ dt d
κ = |t̄ | = | × | = | ˙ −1 ,
||r̄|
dt ds dt |r̄|˙

˙ −4 |(r̄˙ · r̄¨)r̄¨ − (r̄˙ · r̄¨)r̄|


i.e., κ = |r̄| ˙ −4 |(r̄˙ × r̄¨) × r̄|.
˙ = |r̄| ˙
Since r̄˙ ⊥ (r̄˙ × r̄¨), we have

|(r̄˙ × r̄¨) × r̄|


˙ = |r̄˙ × r̄¨||r̄|.
˙

Hence,
|r̄˙ × r̄¨|
κ= . (1.26)
˙3
|r̄|
Exercise 1.6.1. A curve r̄ = r̄(s) is a straight line iff κ = 0.
Exercise 1.6.2. The curvature of a circle is same at every point and
equals the reciprocal of its radius.

let r̄ = r̄(s) be a regular curve of class C3 . We know that b̄ = t̄ × n̄. Then

b̄0 = t̄0 × n̄ + t̄ × n̄0 = κn̄ × n̄ + t × n̄0 = t̄ × n̄0 . (1.27)

Since n̄ is a unit vector, n̄0 is orthogonal n̄ and is therefore parallel to the


rectifying plane. It follows that n̄0 is a linear combination of t̄ and b̄, say

n̄0 = µt̄ + τ b̄.

13
Curves in space Curvature and Torsion

Substituting it in (1.27), we get

b̄0 = t̄ × (µt̄ + τ b̄) = τ (t̄ × b̄) = −τ n̄,

because (t, n, b) is a neighbourhood orthogonal system.

Torsion (τ )
We know that b̄ · b̄ = 1, hence b̄ · b̄0 = 0, i.e., b̄0 ⊥ b̄. Again b̄ · t̄ = 0 gives
b̄0 · t̄ − b̄ · t̄0 = 0, or b̄0 · t̄ = −b̄ · (κn̄) = 0,
i.e., b̄0 is also perpendicular to t. Hence b̄0 is along the unit vector n̄ and we
can write
db̄
= −τ n̄. (1.28)
ds
We call τ the Torsion of the curve.

Expression for the Torsion (τ ):


From (1.28), we have
db̄
τ = −n̄ · = −n̄ · (t̄ × n̄)0 = [t̄, n̄, n̄0 ],
ds
r̄00 r̄000 r̄00 0 1
i.e., τ = [r̄0 , , − 2 κ ] = 2 [r̄0 , r̄00 , r̄000 ],
κ κ κ κ
0 00 000 0 00 000
[r̄ , r̄ , r̄ ] [r̄ , r̄ , r̄ ]
i.e., τ = = . (1.29)
κ 2 r̄00 · r̄00
In Cartesian coordinates this formula becomes
x0 y 0 z 0
x00 y 00 z 00
x000 y 000 z 000
τ = 00 2 . (1.30)
(x ) + (y 00 )2 + (z 00 )2
If the curve is given in generalized parameter, it is easily seen that
...
... dt 6 [r̄,
˙ r̄¨, r̄ ]
 
0 00 000 ˙ r̄¨, r̄ ]
[r̄ , r̄ , r̄ ] = [r̄, = .
ds ˙6
|r̄|
Hence, using (1.27) and (1.29), we get
...
˙ r̄¨, r̄ ]
[r̄,
τ= . (1.31)
(r̄˙ × r̄¨) · (r̄˙ × r̄¨)

14
Curves in space Curvature and Torsion

Exercise 1.6.3. Find κ and τ for the following curves;


1. z = a cos u, y = a sin u, z = bu;

2. z = u, y = u2 , z = u3 ;
1+u 1−u2
3. z = u, y= u
, z= u
;

4. y = f (x), z = g(x);

5. x = a(3u − u3 ), y = 3au2 , z = a(3u + u3 ).

15
Curves in space Serret-Frenet Formulas

1.7 Serret-Frenet Formulas


dt̄ db̄
In equations (1.25) and (1.28), we have found the values for ds and ds . To
dn̄
complete the Serret-Frenet formulas, we now find the value of ds . It is obvious
that dn̄
ds
⊥ n̄. Hence, we can write

dn̄
= αt̄ + β b̄. (a)
ds
dn̄
This gives α = t̄ · ds
. Differentiating the relation t̄ · n̄ = 0, we get

dr̄ dt̄
t̄ · + n̄ · = 0,
ds ds
dn̄ dt̄
i.e., α = t̄ · = −n̄ · = −κ. (b)
ds ds

Again from (a) we get


dn̄
. β = b̄ ·
ds
Differentiating the relation b̄ · n̄ = 0, we get

dn̄ db̄
b̄ · + n̄ · = 0,
ds ds
dn̄ db̄
i.e., β = b̄ ·
= −n̄ · = −n̄ · (−τ n̄) = τ. (c)
ds ds
Thus, the equations (a), (b) and (c) give the relation

dn̄
= τ b̄ − κt̄. (1.32)
ds
The relations (1.25), (1.28) and (1.32) together are called the Serret-Frenet
Formulas, viz.,
dt̄
= κn̄
ds
db̄ (1.33)
= −τ n̄
ds
dn̄
= τ b̄ − κt̄,
ds

16
Curves in space Serret-Frenet Formulas

t0
    
0 κ 0 t
or,  n0  =  −κ 0 τ   n  .
b0 0 −τ 0 b

Definition 1.7.1. Darpoux Vector:


The vector
d¯ = τ t̄ + κb̄
is called Darboux vector and it measures the angular velocity of the moving
tetrad t̄, n̄ and b̄ when the curve traverses at a constant speed.

Note: Serret-Frenet formulae can be written as

t̄0 = d¯ × t̄

n̄0 = d¯ × n̄
b̄0 = d¯ × b̄.

Exercise 1.7.1. Solve the following exercise;


1. Find the unit tangent t̄, the unit principal normal n̄ and the unit bi-
normal b̄ along the curve;

x = 3t − t3 , y = 3t2 , z = 3t + t3 .

2. Show that the binormal of a circular helix makes a constant angle with
the axis of the cylinder on which the helix lies.

3. Show that when all the tangent lines of a curve pass through a fixed
point, the curve in a straight line.

4. Show that when all the osculating planes of a curve pass through a
fixed point, the curve is plane curve (i.e. τ = 0).

17
Curves in space Fundamental Theorem For Space Curves

1.8 Fundamental Theorem For Space Curves


Theorem 1.8.1. If two single valued continuous functions κ(s) and
τ (s), s > 0, are given, then there exists one and only one space curve, deter-
mined except for its position in space, for which s is the arc length, κ the
curvature and τ the torsion.
Proof. We assume that the functions are analytic. Then we can write
in the neighbourhood of a point s = s0 .
h2 00
r̄(s) = r̄(s0 ) + hr̄0 (s0 ) + r̄ (s0 ) + · · · ,
2!
provided the series is convergent in a certain interval. Then substituting
for r̄0 , r̄00 etc., their values with respect to the three perpendicular vectors
t̄, n̄, b̄ at P (s0 ), we get
r̄0 = t, r̄00 = κn̄, r̄000 = −κ2 t̄00 + κ0 n̄ + κτ b̄, etc.,
so that,
h2 h3
r̄(s) = r̄(s0 ) + ht̄ + kn̄ + (−k 2 t̄ + k 0 n̄ − kτ b̄) + · · · , (1.34)
2! 3!
where all terms can be formed by differentiating the Frenet formulas as well
as t̄, n̄, b̄ at P (s0 ) are supposed to exist because of the analytic character
of functions. If, we now choose at an arbitrary point r̄(s0 ) and arbitrary set
of three perpendicular unit vectors and select them as t̄, n̄, b̄, then equation
(1.34) determines the curve uniquely (proved).
Note: The equations κ = κ(s), τ = τ (s) are called the Natural or Intrinsic
equations of the space curve.

Helices (Curves of constant slope): They are defined by the property


that the tangent makes a constant angle α with a fixed line l in space (called
the axis).

The circular helix: r̄ = (a cos u, a sin u, bu) is a special case of such curve.
Definition 1.8.1. General helix or Cylindrical Helix:
A curve is called a general helix or cylindrical helix or a curve of constant
slope if there is a fixed vector (line) in space called the axis of the helix such
that the angle α between the tangent vectors and the axis is constant.

18
Curves in space Fundamental Theorem For Space Curves

Note: We exclude the case, α = 0, for which the tangent vector are all
parallel in which case the curve is a straight line.

Theorem 1.8.2. A necessary and sufficient condition that a curve be a


helix is that the ratio of curvature to torsion be constant.

Proof. Necessary Condition: Let a unit vector ā be placed in the


direction of l. Then for the helix

t̄ · ā = cos α (constant).

Differentiating, we get

ā · n̄ = 0.

Hence ā lies along the plane of t̄ and b̄ and makes an angle α with t̄,

i.e., ā = t̄ cos α + b̄ sin α.

Differentiating this, we get

0 = κn̄ cos α − τ n̄ sin α = (κ cos α − τ sin α)n̄,


κ
or, = tan α = Constant.
τ
k
Sufficient Condition: If is constant, we can always find a constant
τ
angle α such that,
n̄(k cos α − τ sin α) = 0,
d
i.e., (t̄ cos α − b̄ sin α) = 0,
ds
i.e., t̄ cos α − b̄ sin α) = ā (const. unit vect.).
Hence,
ā · t̄ = cos α (Proved).

19
Chapter 2

SURFACES

Definition of a surface:
Definition 2.0.1. Simple Sheet of a Surface:
A set of points of E 3 which is in a continuous one-to-one correspondence with
a closed rectangle in the plane is called a Simple Sheet of the surface, i.e., a
simple sheet is the set of points of E 3 whose position vectors are the values
of a 1-1 continuous vector-valued function
r̄ = r̄(u1 , u2 ),
defined on a closed rectangle a ≤ u1 ≤ b, c ≤ u2 ≤ d, u1 and u2 are the
parameters.

Example: The set of points which satisfies the equation


z = f (x, y),
where f is a continuous function of two variables defined in the rectangle
a ≤ x ≤ b, c ≤ y ≤ d, is an example of a simple sheet of a surface.

Indeed, in this case x and y play the role of parameters and the continuous
vector valued function is
r̄ = xî + y ĵ + f (x, y)k̂.
This function is obviously one to one since for two different points of the
plane xy, the values of the function must differ in at least one component.
Surfaces Definition of a surface

Note: Intuitively, we can say that a simple sheet of a surface is obtained


from a rectangle by streching, sqeezing and bending but without tearing or
gluing together.
Note: Instead of using a rectangle in this definition, we could use any
bounded closed simply connected domain in the plane, in particular a closed
disc.
The surface of a cylinder is not a simple sheet because it cann’t be obtained
from a rectangle without gluing it together. Similarly, neither a sphere nor
a flat annulus is a simple sheet.

is of class C n if every point P of


P P
We say that aPsurface has a
neighbourhood in which has a parametric representation

r̄ = r̄(u1 , u2 ),

where r̄ is a function of class C n .

Definition 2.0.2. Regular Point and Singular Point:


A point P of the surface corresponding to the values (u10 , u20 ) of the param-
eters is said to be a Regular Point if

r̄1 (u10 , u20 ) × r̄2 (u10 , u20 ) 6= 0, (2.1)


∂ r̄ ∂ r̄
where r̄1 = ∂u 1 and r̄2 = ∂u2 .

A point for which r̄1 × r̄2 = 0 in a given parameterization is called a


Singular Point of the parametrization.
P
In Cartesian coordinates the equation of becomes

x = x(u1 , u2 ), y = y(u1 , u2 ), z = z(u1 , u2 ),

∂x
and if we denote by x1 , etc., the components of the vector r̄1 × r̄2 are
∂u1
y1 z1 ∂(y, z) z1 x1 ∂(z, x) x1 y 1 ∂(x, y)
= , = , = .
y2 z2 ∂(u1 u2 ) z2 x2 ∂(u1 u2 ) x2 y 2 ∂(u1 u2 )

Hence (2.1) implies that for a regular point, at least one of the Jacobian’s
∂(y, z) ∂(z, x) ∂(x, y)
1 2
, 1 2
and should be non-zero.
∂(u , u ) ∂(u , u ) ∂(u1 , u2 )

21
Surfaces Definition of a surface

Implicit Equation of the Surface:


The equation
F (x, y, z) = 0 (2.2)
will represent a surface provided F (x, y, z) is a function of class C n and at
least one of the partial derivatives Fx , Fy , Fz is non-zero.
Let at a point P, Fz 6= 0, then by implicit function theorem, we can write
(2.2) as
z = f (x, y), (2.3)
which gives the parametric representation

z = u1 , y = u2 , z = f (u1 , u2 ).

Now, differentiating (2.2) and (2.3), we get

Fx dx + Fy dy + Fz dz = 0
∂z ∂z
· dx + · dy − dz = 0
∂x ∂y
whence, we obtain
∂z Fx ∂z Fy
=− , =− . (2.4)
∂x Fz ∂y Fz
The vectors r̄1 , r̄2 and r̄1 × r̄2 have components
     
∂z ∂z ∂z ∂z
1, 0, , 0, 1, and − , − , 1 . (2.5)
∂x ∂y ∂x ∂y
Thus r̄1 × r̄2 6= 0, and the parameterization is regular at P .

Definition 2.0.3. Parametric Curve:


When we keep u2 = constant in the equation r̄ = r̄(u1 , u2 ), r̄ depends
on one parameter u1 only and thus determines a curve on the surface called
the Parametric Curve, u2 = const. or the u1 −curve. Similarly putting u1 =
const. we get the parametric curve, u1 = const. or the u2 − curve.
We also call (u1 , u2 ) as the Curvilinear Coordinates on the surface and
the parametric curves as the Coordinate Curves.
Exercise 2.0.1. Write the following surfaces in the form F (x, y, z) = 0
and find the cartesian equation of the parametric curve;

22
Surfaces Definition of a surface

1. Sphere: x = a cos θ cos φ, y = a cos θ cos φ, z = a sin θ;

2. Circular cone: x = u sin α cos φ, y = u sin α cos φ, z = u cos α;

3. Ellipsoid: x = a sin u cos v, y = b sin u cos v, z = c cos u;

4. Hyperboloid of two sheets: x = a sinh u cos v, y = b sinh u sin v, z =


c cosh u.

23
Surfaces The Tangent Plane

2.1 The Tangent Plane


Definition 2.1.1. Tangent Plane: P
P Let P0 be a regular point of the surface and P a variable point of
. Consider all the planes passing through P0 and the distance of P from
those planes. For every fixed plane the distance will tend to zero as P → P0
remaining on the surface. The plane whose distance from P is of highest order
of smallness as P tends to P0 onPthe surface with respect to the distance P0 P
is called the Tangent Plane to at P0 .
The line through P0 perpendicular to the tangent plane at P0 is called
the Normal to theP surface at P0 . Any vector in this direction is called the
Normal Vector to .

Equation Of The Tangent Plane: P


Let P0 be the point r̄ = r̄(u10 , u20 ) on . The equation of a plane through
P0 is
(R̄ − r̄0 ) · N̄ = 0,
where N̄ is the normal to the plane at P0 .
The distance d of a point P : r̄ = r̄(u1 , u2 ) from the plane is

|(r̄ − r̄0 ) · N̄ |
d= .
|N̄ |

Putting u1 = u10 + h, u2 = u20 + k and using the notation r̄i,0 = r̄i (u10 , u20 ), we
get by using Taylor’s theorem
p
r̄ = r̄(u10 + h, u20 + k) = r̄0 + r̄1,0 h + r̄2,0 k + 0( (h)2 − (k)2 ).

Hence, p
[r̄1,0 h + r̄2,0 k + k + 0( (h)2 + (k)2 )] · N̄
d= ,
|N̄ |
r̄1,0 N r̄2.0 N̄ √
=h −k + 0( h2 + k 2 ),
|N̄ | |N̄ |
d
and |P0 P |
→ 0 if and only if

r̄1,0 · N̄ = 0 and r̄2,0 · N̄ = 0.

24
Surfaces The Tangent Plane

This condition is satisfied if we take

N̄ = r̄1,0 × r̄2,0 .

The equation of the tangent plane then becomes

(R − r̄(u10 , u20 )), r̄1 (u10 , u20 ), r̄2 (u10 , u20 ) = 0.


 
(2.6)

The unit normal to the surface at P0 will be


r̄1 × r̄2
(2.7)
|r̄1 × r̄2 |

evaluated at P0 .
If the surface is given in Cartesian form

x = x(u1 , u2 ), y = y(u1 , u2 ), z = z(u1 , u2 ),

the equation (2.6) of the tangent plane takes the form

X −x Y −y Z −z
x1 y1 z1 = 0, (2.8)
x2 y2 z2

where X, Y, Z are the coordinates of generic point of the plane and x, y, z; x1 , y1 , z1


are the values of the functions x, y, z and their derivatives at the point of
contact.

25
Surfaces The Tangent Plane

Tangent plane for the surface F(x, y, z) = 0:  


∂z ∂z
It follows from (2.4) and (2.5) that r̄1 × r̄2 has components − ∂x , − ∂y ,1 ,
 
Fx Fy
i.e., Fz , − Fz , 1 .
Hence, (Fx , Fy , Fz ) is a normal vector to the surface. The unit normal is
given by
(F , F , F )
p 2x y 2 z 2 . (2.9)
Fx − Fy + Fz
Since the equation of the tangent plane is (R̄ − r̄0 ) · N̄ = 0, in Cartesian
coordinates this will have the form

(X − x)Fx − (Y − y)Fy − (Z − z)Fz = 0, (2.10)

where X, Y, Z are the coordinates of the points of the plane and the deriva-
tives Fx , Fy , Fz are evaluated at the contact point (x, y, z).

Exercise 2.1.1. Solve the following Questions;


1. Find the tangent plane and normal vector for the circular cone x =
u sin α cos φ, y = u sin α sin φ, z = u cos α;

2. Tangent developable is the surface generated by the tangents to a curve.


Find its normal vector and tangent plane;

3. Prove that if all the normals to a surface pass through a fixed point,
then the surface is a sphere;

4. Find the tangent


√ √planes to the sphere x2 + y 2 + z 2 = 1 at the points
(1, 0, 0), ( 2/2, 2/2, 0), ( √13 , √13 , √13 ).

26
Surfaces The First Fundamental Form Of A Surface

2.2 The First Fundamental Form Of A Sur-


face
A curve on the surface r̄ = r̄(u1 , u2 ) is given by the equations u1 = u1 (t), u2 =
u2 (t), i.e., the position vector of any point on the curve is given by r̄ =
r̄(u1 (t), u2 (t)), a ≤ t ≤ b.
The tangent to the curve at a point P0 is

dr̄ dui
(t0 ) = r̄i (u1 (t0 ), u2 (t0 )) , (i = 1, 2).
dt dt
Replacing the derivatives by differentials which means multiplication by a
scalar dt, we obtain another vector tangent to the curve as

dr̄ = r̄i dui , (i = 1, 2). (2.11)

From equation (2.6), it is seen that r̄1 and r̄2 lie in the tangent plane, hence
dr̄ which is a linear combination of r̄1 , r̄2 also lies in the tangent plane, i.e.,
dr̄ is also a tangent vector to the surface.
The distance between two points P and Q, corresponding to the para-
metric vector tp and tQ , along the arc on the surface is found by integrating
|dr̄|,
ZtQ
i.e., s = |dr̄|.
ts

Hence,

ds2 = |dr̄|2 = dr̄ · dr̄ = (r̄i dui ) · (r̄j duj ) = (r̄i · r̄j )dui duj .

If we put
gij = r̄i · r̄j , (2.12)
We can write
ds2 = gij dui duj , (2.13)
or ds2 = g11 (du2 ) + 2g12 du1 du2 + g22 (du2 )2 .
This is called the first fundamental form or the metric form or the line element
on the surface.

27
Surfaces The First Fundamental Form Of A Surface

It is obvious that the distance between two points P and Q along the arc
is then given by the formula
ZtQ r
dui duj
s= gij dt. (2.14)
dt dt
tP

Since the scalar product r̄i · r̄ is symmetric, it is obvious from (2.12) that gij
is also symmetric, i.e., gij = gji .
The coefficients gij form a 2 × 2 symmetric matrix
 
g11 g12
[gij ] = , (2.15)
g12 g22
whose determinant, called the discriminant of the first fundamental form, is
denoted by g.
g g
g = 11 12 = g11 g22 − (g12 )2 . (2.16)
g12 g22
The elements of the inverse matrix will be denoted by g ij , i.e.,
 11 12   −1
ij g g g11 g12
[g ] = = .
g 21 g 22 g21 g22

Hence, we have
g22 g12 g11
g 11 = , g 12 = g 21 = − , g 22 = , (2.17)
g g g
which shows that g ij is also symmetric, i.e., g ij = g ji . Since
 11 12     
g g g11 g12 1 0
= ,
g 12 g 22 g12 g22 0 1

which can be written as


g ik gkj = δji (2.18)
where 
1 i = j,
δji =
0 i=6 j.
Finally, making use of the identity

(r̄1 × r̄2 ) · (r̄1 × r̄2 ) = (r̄1 · r̄1 )(r̄2 · r̄2 ) − (r̄1 · r̄2 )(r̄1 · r̄2 ),

28
Surfaces The First Fundamental Form Of A Surface

we can write
r̄1 · r̄1 r̄1 · r̄2 g11 g12
|r̄1 × r̄2 |2 = = = g, (2.19)
r̄1 · r̄2 r̄2 · r̄2 g12 g22

i.e., g = |r̄1 × r̄2 |2 .


Since |r̄1 × r̄2 |2 6= 0 at regular points, we have shown that at regular points

g > 0. (2.20)

Exercise 2.2.1. Find the first fundamental form and gij for the following
surfaces;

1. Sphere: x = a cos θ cos φ, y = a cos θ sin φ, z = a sin θ;

2. Surface of revolution: x = f (u) cos v, y = f (u) sin v, z = h(u);

3. Helicoid: x = u cos v, y = u sin v, z = av;

4. Ellipsoid: x = a cos u cos v, y = b cos u sin v, z = c sin u;

5. Torus: x = (b + a cos u) cos v; y = (b + a cos u) sin v, z = a sin u.

Answers:
1. ds2 = a2 θ2 + a2 cos2 θdφ2 ;

2. ds2 = (f 02 + h02 )du2 + f 2 dv 2 ;

3. ds2 = du2 + (a2 + u2 )dv 2 ;

4. ds2 = ((a2 cosv +b2 sin2 v) sin u2 + c2 cos2 u)du2 ;

5. ds2 = a2 du2 + (b + a cos u)2 dv 2 .

29
Length Of A Tangent Vector And
Surfaces Angle Between Two Tangent Vectors

2.3 Length Of A Tangent Vector And Angle


Between Two Tangent Vectors
A tangent vector ā at a point (u1 , u2 ) of the surface lies in the tangent plane
at that point, and therefore can always be written as a linear combination of
r̄1 and r̄2 , i.e.,
ā = ai r̄i , (i = 1, 2), (2.21)
where ai ’s are called the components of the tangent vector ā.
Hence the square of its length is given by

|ā|2 = ā · ā = (ai r̄i ) · (aj r̄j ) = ai aj (r̄i · r̄j ),

i.e., |ā|2 = gij ai aj . (2.22)


The scalar product of two tangent vectors ā = ai r̄i and b̄ = bj r̄j at the
same point is

ā · b̄ = (ai r̄i )(bj r̄j ) = ai bj (r̄i · r̄j ) = gij ai bj . (2.23)

If φ be angle between ā and b̄, we have

ā · b̄
cos φ = ,
|ā||b̄|

using (2.22) and (2.23), this can be written as

gij ai bj
cos φ = √ p . (2.24)
gnm am an gpq bp bq

We now consider two curves

ui = φi (t) and uj = ψ j (τ )

on the surface through a common point ui = φi (t0 ) = ψ i (τ0 ).


Let dui and δui denote the differentials in the direction of each of these
curves, i.e.,
dφi dψ i
dui = dt and δui = dτ
dt dτ

30
Length Of A Tangent Vector And
Surfaces Angle Between Two Tangent Vectors

Then the vectors dr̄ = r̄i dui and δr̄ = ri δui are tangents at P0 to the two
curves respectively, and hence the angle between them is given by

dr̄ · δr̄ gij dui δuj


cos φ = =√ p . (2.25)
|dr̄||δr̄| gmn dum dun gpq δup δuq

Angle Between Parametric Curves:


If we take the two curves as parametric curves u2 = const. and u1 = const.
we shall have du2 = 0 and δu1 = 0.
Hence if φ is the angle between these parametric curves, it follows from
(2.25) that
gij du1 δu2
cos φ = p p ,
g11 (du1 )2 g22 (δu2 )2
g12
i.e., cos φ = √ √ . (2.26)
g11 g22
Thus, the coordinate lines are orthogonal at a point if and only if at that
point g12 = 0.
If the coordinate lines are orthogonal at every point, we say that the
curvilinear coordinates are orthogonal, and then the first fundamental form
has the shape
ds2 = g11 (du1 )2 = g22 (du2 )2 . (2.27)

Exercise 2.3.1. Solve the following questions;


1. Prove that on a surface with first fundamental form

ds2 = (du1 )2 + G(u1 u2 )(du2 ),

the two curves defined by the differential equations


√ √
du1 + Gdu2 = 0 and du1 − Gdu2 = 0

are orthogonal;

2. Find a formula for the angle with which the two curves, φ(u1 , u2 ) =
const. and ψ(u1 , u2 ) = const., intersect;
Eφ2 ψ2 − F (φ2 ψ1 + φ1 ψ2 ) + Gφ1 ψ1
Ans: cos a = p p ;
E(φ2 )2 − 2F φ1 φ2 + G(φ1 )2 E(ψ2 )2 − 2F ψ1 ψ2 + G(ψ1 )2

31
Length Of A Tangent Vector And
Surfaces Angle Between Two Tangent Vectors

3. Show that the orthogonal trajectories of the family of curves given by

M du + N dv = 0

are given by

(EN − F M )du + (F N − GM )dv = 0;

4. Show that a necessary and sufficient condition that the curves

Adu2 + 2Bdudv + Cdv 2 = 0

form an orthogonal net is

EC − 2F B + GA = 0;

5. Assume the parametric curves to be orthogonal, show that the differen-


tial equation of the curves bisecting the angles of the parametric curves
is
du2 − Gdv 2 = 0.

Note: In the examples (2) to (5) the line element is taken as

ds2 = Edu2 + 2F dudv + Gdv 2 ,

i.e., g11 = E, g12 = F and g22 = G; u1 = u, v 2 = v.


Show that in this relation (2.25) can be written as

du δu du δv dv δu dv δv
cos φ = E + F( + )+G (10.5A)
ds δs ds δs ds δs ds δs
and the orthogonality condition is

Eduδu + F (duδv + dvδu) + Gdvδv = 0. (10.5B)

32
Surfaces Area Of A Surface

2.4 Area Of A Surface

Figure 2.1

Let δA be the area of a small patch P QRS bounded by the coordinate


curves such that the vertices are respectively (u1 , u2 ), (u1 + du1 , u2 ), (u1 +
du1 , u2 + du2 ) and (u1 u2 + du2 ). Consider the tangent vectors P Q0 and P S 0
to the coordinate curves at P such that P Q0 = r̄1 du1 and P S 0 = r̄2 du2 .
Then in the limit, the area of the parallelogram P Q0 R0 S 0 will be the same as
δA, i.e.,
δA = (r̄1 du1 ) × (r̄2 du2 ) = |r̄1 × r̄2 )| du1 du2 .
Hence, the area of the domain S on the surface r̄ = r̄(u1 , u2 ) which corre-
sponds to the rectangle Ω on the plane of parameters u1 , u2 is given by the
formula

ZZ ZZ
1 2
area of S = |r̄1 × r̄2 | du du = gdu1 du2 . (2.28)
Ω Ω

Exercise 2.4.1. Find the area of a sphere and a torus.


Answer: 4πa2 ; 8πa(a + bπ
2
).
Sphere: −Π ≤ θ ≤ Π, 0 ≤ φ ≤ Π, g = a4 cos2 a.
Torus: −Π ≤ v ≤ Π, − Π ≤ u ≤ Π, g = a2 (b + a cos u)2 ,
E = a2 , F = 0, G = (b + a cos u)2 .

33
Surfaces Change Of Curvilinear Coordinates

2.5 Change Of Curvilinear Coordinates


0 0
If we take as now curvilinear coordinates the functions u1 = u1 (u1 , u2 ) and
0 0 0 0
u2 = u2 (u1 , u2 ) such that, u1 and u2 are also independent, i.e.,
0 0
∂(u1 , u2 )
6= 0. (2.29)
∂(u1 , u2 )
The values of the new coordinates at any point P being such that
0 0
r̄(u1 , u2 ) = r̄(u1 , u2 ) at P.
We denote,
0 0
∂r̄(u1 , u2 )
r̄i0 = .
∂ui 0
Then,
∂r̄(u1 , u2 ) ∂u1 ∂r̄(u1 , u2 ) ∂u2
r̄i0 = + ,
∂u1 ∂ui 0 ∂u2 ∂ui 0
∂ui
i.e., r̄i0 = r̄i i 0 . (2.30)
∂u
The coefficient of the first fundamental form will then be transformed as
follows;
∂ui ∂uj ∂ui ∂uj
   
gi0 j 0 = r̄i0 · r̄j 0 = r̄i i0 · r̄j j 0 = r̄i · r̄j i 0 j 0 ,
∂u ∂u ∂u ∂u
∂ui ∂uj
i.e., . gi0 j 0 = gij (2.31)
∂ui 0 ∂uj 0
To find the transformation formula for g ij , we proceed as follows.
We know that
0 0 0
g i j gj 0 k0 = δki 0 ,
using (2.31), we have
∂uj ∂uk
 
i0 j 0 0
g gjk j 0 k0 = δki 0 .
∂u ∂u
0 0
∂uk ∂uk ∂uk k
Multiplying both sides by ∂um
and using the results ∂uk0 ∂um
= δm and
0 k0 i0
δki 0 ∂u
∂um
= ∂u
∂um
, we get,
0
i0 j 0 ∂uj ∂ui
g ggm j 0 = .
∂u ∂um

34
Surfaces Change Of Curvilinear Coordinates

Multiplying both sides by g mn , we get


0
i0 j 0 ∂un mn ∂u
i
g =g .
∂uj 0 ∂um
0
∂uk
Once again multiplying both sides by ∂un
, we get
0
i0 k 0 mn ∂un ∂ui
g =g ,
∂uj 0 ∂um
i0 0
∂uj
i0 j 0 ij ∂u
or g =g . (2.32)
∂ui ∂uj
Now, if ā is any tangent vector, we know that it can be written as
ā = ai r̄i .
0
If ai and ai are the components of ā in the old and new coordinate systems,
we have
i0
i i0 i0 ∂u
ā = a r̄i = a r̄i0 = a r̄i i
∂u
i
0 ∂u
i.e., (āi − ai i0 )r̄i = 0,
∂u
which gives
i i0
i i0 ∂u i0 i ∂u
a =a or a = a . (2.33)
∂ui0 ∂ui
Now, using (2.31) and (2.33), we get the following result for any two tangent
vectors ā and b̄;
∂ui ∂uj 0 0
ā · b̄ = gij ai bj = gij i0 j 0 ai bj , (2.34)
∂u ∂u
0 0
i.e., gij ai bj = gi0 j 0 ai bj , (2.35)
which shows that the scalar product of two vectors remain invariant under
coordinate transformations.
Again, using (2.31) and taking determinant, we get
∂ui ∂uj
|gi0 j 0 | = |gij | ,
∂ui0 ∂uj 0
Since,
∂ui ∂(u1 u2 )
= and g = |gij |,
∂ui0 ∂u10 u20

35
Surfaces Change Of Curvilinear Coordinates

we have 2
∂(u1 u2 )

0
g =g . (2.36)
∂(ui0 u20 )
Finally, we show that the area remains invariant under coordinate transfor-
mation.
∂(du1 , du2 )
 
√ √
ZZ p ZZ ZZ
0 10 20 10 20
A = 0
g du du = g 10 20 du du = gdu1 du2 .
∂(du , du )
Ω0 Ω0 Ω

The last step has been written by using the theorem of coordinate transfor-
mation on the double integrals.
Thus we have shown that
A0 = A.

36
Chapter 3

THE SECOND FUNDAMENTAL


FORM

Definition & Expression: Assume the surface r̄ = r̄(u1 , u2 ) of class C 2 ,


and let us calculate the deviation of the surface from the tangent plane at a
point P with coordinate (u1 , u2 ). Take another point Q : (u1 + h1 , u2 + h2 )
×r̄2
on the surface. If N̄ = |r̄r̄11 ×r̄ 2|
be the unit normal at P , the distance from Q
to the tangent plane is
d = [r̄(u1 + h1 , u2 + h2 ) − r̄(u1 , u2 )] · N̄ .
Using Taylor’s theorem, we have
1
r̄(u1 + h1 , u2 + h2 ) − r̄(u1 , u2 )] = r̄i hi + r̄ij hi hj + 0(h1 )2 + (h2 )2 ), i = 1, 2,
2
∂ 2 r̄
where r̄ij = ∂ui ∂uj
. Since N̄ · r̄i = 0 , we get
1
d = (r̄ij · N )hi hj + 0(h1 )2 + (h2 )2 ).
2
We now put
r̄ij · r̄1 × r̄2 [r̄ij · r̄1 , r̄2 ]
bij = r̄ij · N̄ = = √ . (3.1)
|r̄1 × r̄2 | g
The distance of the point (u1 + h1 , u2 + h2 ) on the surface from the tangent
plane at (u1 , u2 ) is thus given by
1
bij hi hj (3.2)
2
p
neglecting the terms of order higher than 2 in (h1 )2 + (h2 )2 .
The Second fundamental form The Second fundamental form

Theorem 3.0.1. The signed distance of the point (u2 + h, u2 + h2 ) of a


surface r̄ = r̄(u1 , u2 ) of class C 2 from the tangent plane at (u1 , u2 ) equals
1
δ = bij hi hj
2
p
with an error of order higher than 2 relative to (h1 )2 + (h2 )2 .

The coefficients bij ar given by the formula (3.1) and defined on the point
of contact only. For sufficiently small h1 and h2 , δ is positive if the point
(u1 + h1 h2 + h2 ) is an that side of the tangent plane into which the vector N
at (u1 , u2 ) is pointing and negative if it is on the other side.

3.1 Second Fundamental Form


The quadratic differential form

II = bij dui duj , (3.3)

where the coefficient bij are defined by the formula (3.1) is called the second
fundamental form of the surface. From (3.1) we get

bij = bji .

Hence (3.3) can be written as

II = b11 (du1 )2 + 2b12 du1 du2 + b22 (du2 )2 . (3.4)

Further differentiating the relation N̄ · r̄i = 0, we get N̄ · r̄ij + N̄j · r̄i = 0,

bij = N̄ · r̄ij = −N̄j · r̄i = N̄i · r̄j . (3.5)


We write
b11 b12
b = det(bij ) = ,
b12 b22
(3.6)
and finally, when the surface is given in the form

x = x(u1 , u2 ), y = y(u1 , u2 ), z = z(u1 , u2 ).

38
The Second fundamental form The Second fundamental form

We can write (3.1) in the form


xij yij zij
1
bij = √ = x1 y1 z1 . (3.7)
g
x2 y2 z2

Note: In view of (3.5) we can also write (3.3) as


II = −dN̄ · dr̄, (3.8)

for, −dN̄ · dr̄ = −(barNi dui ) · (r̄j duj ) = −(N̄i · r̄j )dui duj = bij dui duj .

Classification of Points on a Surface:


1) P is called an Elliptic Point if b > 0 at p. Then the quadratic form
(3.3) is positive definite, i.e., the tangent plane at P lies on the same
side of the surface.
2) P is called a Parabolic Point if b = 0 at P .In this case II = 0 for one
direction du1 : du2 , and is of the same sign for other directions. This
will be the case if the tangent plane and the surface have a curve in
common.
3) P is called a Hyperbolic Point if b < 0 at P .
Exercise 3.1.1. Solve the following exercise;
1) Show that a necessary and sufficient condition that a surface be a plane
is that the second fundamental form is identically zero;
2) Show that for a sphere of radius a

ds2 = aII;

3) Show that for the surface of revolution

x = f (u) cos v, y = f (u) sin v, z = h(u)


f 0 h00 − h0 f 00 2 f h0
Π= p 2 2
du + p
2 2
dv 2 ;
0
f +h 0 0
f +h 0

4) Find the first and second fundamental forms for the surface z = f (x, y).

39
Gaussian Map and Gaussian Curvature
The Second fundamental form (Spherical Map)

3.2 Gaussian Map and Gaussian Curvature


(Spherical Map)
P
Let be a surface of class C2 , and let the parametric representation of the
1 2
P
surface be r̄ = r̄(u
P, u ). Consider the following map G : → S of a region
Ω of the surface into the unit sphere S.

G : P → P ∗ 3 OP ∗ = N̄

i.e., to every point p ∈ . This map assigns a point P ∗ ∈ S whose position


P
vector equals the positive normal unit vector N̄ of the surface at P . This is
called the Gaussian Map or the Spherical Map of the surface.

be mapped onto the region Ω∗ ∈ S.


P
Let a region Ω ∈
Definition 3.2.1. Gaussian Curvature:
P
Gaussian Curvature κ of the surface at P is defined as
area of Ω∗
κ = lim . (3.9)
Ω→P area of Ω
The limit is taken as the area Ω shrinks to the point P .
Expression for κ:
We have,
b11 b12 N̄1 r̄1 N̄1 r̄2
b= =
b21 b22 N̄2 r̄1 N̄2 r̄2
and
(N̄1 × N̄2 ) · (r̄1 × r̄2 ) = [N̄1 × N̄2 × r̄1 ] · r̄2 ,
i.e., (N̄1 × N̄2 ) · (r̄1 × r̄2 ) = [(N̄1 · r̄1 )N̄2 − (N̄2 · r̄1 )N̄1 ] · r̄2 ,
i.e., (N̄1 × N̄2 ) · (r̄1 × r̄2 ) = (N̄1 · r̄1 )(N̄2 · r̄2 ) − (N̄2 · r̄1 )(N̄1 · r̄2 ) = b,
i.e., b = (N̄1 × N̄2 ) · (r̄1 × r̄2 ). (3.10)
Now,
(N̄1 × N̄2 ) · (r̄1 × r̄2 ) = |N̄1 × N̄2 ||r̄1 × r̄2 | cos φ.
Since

(N̄1 × N̄2 ) || N̄ and also (r̄1 × r̄2 ) = g N̄ ,

40
Gaussian Map and Gaussian Curvature
The Second fundamental form (Spherical Map)

we get φ = 0, i.e.,

b = |N̄1 × N̄2 ||r̄1 × r̄2 | = g|N̄1 × N̄2 |. (3.11)

Since the position vector for P ∗ is N̄ , we get from (3.9) and using (3.11)
RR
|N̄1 × N̄2 |du1 du2
area of Ω∗ ∗
κ = lim = lim ΩRR
Ω→P area of Ω Ω→P
Ω∗ →P
|r̄1 × r̄2 |du1 du2


|N̄1 × N̄2 | b/ g b
= = √ = .
|r̄1 × r̄2 | g g
Hence,
b
κ= . (3.12)
g
Note: Since g > 0 everywhere we get the following results from (3.12);
(a) At Elliptic points k > 0;

(b) At Parabolic points k = 0;

(c) At Hyperbolic points k < 0.


Change of Coordinates:
Consider the change of coordinates
0 0
(u1 , u2 ) → (u1 , u1 ).
∂u i
We know that r̄i0 = r̄i ∂u i0 .
Hence. 0 0
0 ∂(u1 , u2 )
N̄ = N̄ where  = sign
∂(ui , u2 )
∂ui
and N̄i0 = N̄i . (3.13)
∂ui 0
Therefore,
∂ui ∂uj
bi0 j 0 = −N̄i0 · r̄j 0 = −(N̄i0 · r̄j 0 ) = ,
∂ui 0 ∂uj 0
∂ui ∂uj
i.e., bi0 j 0 =∈ bij . (3.14)
∂ui 0 ∂uj 0

41
Gaussian Map and Gaussian Curvature
The Second fundamental form (Spherical Map)

Taking determinant, we obtain

∂ui ∂uj
|bi0 j 0 | = | ∈ bij | ,
∂ui 0 ∂uj 0
2
∂(u1 , u2 )

0
i.e., b = b , (3.15)
∂(u1 0 , u2 0 )
2
∂(u1 , u2 )

0
We know that, g =g . (3.16)
∂(u1 0 , u2 0 )
h 1 2 i2
∂(u ,u )
b 0 b ∂(u1 0 ,u2 0 ) b
Hence κ0 = 0 = h 2 = = κ,
g g
i
∂(u1 ,u2 )
g ∂(u 1 0 ,u2 0 )

i.e., κ0 = κ, (3.17)
i.e., κ remains invariant under coordinate transformations.

Exercise 3.2.1. Solve the following exercise;


i) For a plane,
κ = 0;

ii) For a sphere of radius a,


l
κ= ;
a2
iii) For a surface of revolution x = f (u) cos v, y = f (u) sin v, z = h(u)

h0 f 0 h00 − f 00 h0
κ= .
f (f 02 + h02 )2

42
The Second fundamental form Formulas of Gauss and Weingarten

3.3 Formulas of Gauss and Weingarten


I. Suppose the surface r̄ = r̄(u1 , u2 ) is of class C 2 . The partial derivatives
of second order r̄ij of the vector valued function r at a non-singular point
(u1 , u2 ), being a vector in E 3 , can be expressed as a linear combination of
the three mutually perpendicular vector r̄1 , r̄2 and N̄ . Say

r̄ij = Γ1ij r̄1 + Γ2ij r̄2 + βij N̄ .

Multiplying both sides by N , we get βij = bij .


Hence,
r̄ij = Γkij r̄k + bij N̄ . (3.18)
This is called the Gauss Formula, and the coefficient Γkij are called the
Christoffel Symbols of second kind, which we shall study later.

II Since N̄i ⊥ N̄ , it can be expressed in terms of r̄1 and r̄2 only.

Let us write
N̄i = −bji r̄j . (3.19)
Taking dot product with r¯k we get,

N̄i · r̄k = −bji r̄j · r̄k ,

or, bik = bji gjk .

Multiplying both sides by g km , we get

bm
i = g
km
bik , (3.20)

i.e., N̄i = −g jk bik r̄j . (3.21)


This is called Weingarten Formula.

43
The Second fundamental form Christoffel Symbols

3.4 Christoffel symbols


It is obvious from (3.18) that Γky is symmetric in i and j, i.e. Γky = Γky .
Taking dot product in (3.18) with r̄l ,we have

r̄ij · r̄l = gkl Γkij = Γijl , (3.22)


where Γijl are called Christoffel Symbols of first kind

Γijl = gkl Γkij . (3.23)

From this we get Γkij = g kl Γkijl . (3.24)


∂gij ∂
Now, ∂ul = ∂ul (r̄i · r̄j ) = r̄il · r̄j + r̄i · r̄il cyclically permuting the indices i, j, l,
and using (15.5),we get 3 equations

∂gij
= Γilj + Γjli . (a)
∂ul
∂gjl
= Γjil + Γlij . (b)
∂ui
∂gli
= Γlji + Γijl . (c)
∂uj
(b) + (c) − (a) implies
 
1 ∂gjl ∂gli ∂gij
Γijl = + j − . (3.25)
2 ∂ui ∂u ∂ul

Hence,  
1 ∂gjl ∂gli ∂gij
Γkij = g Γijlkl
= g kl + j − . (3.26)
2 ∂ui ∂u ∂ul
Transformation for Γkij :
∂u i
Differentiating r̄i0 = r̄i ∂ui0 , we get

∂ui ∂uj ∂ 2 ui
r̄ i0 j 0 = r̄ij i0 j 0 + r̄i i0 j 0 .
∂u ∂u ∂u ∂u
Hence,

∂ui ∂uj ∂ul ∂ 2 ui ∂ul


Γi0 j 0 l0 = r̄i0 j 0 · r̄l0 = Γijl 0 , 0 , 0 + gil . (3.27)
∂ui ∂uj ∂ul ∂ui0 ∂uj 0 ∂ul0

44
The Second fundamental form Christoffel Symbols

0 0 0 0 k0 0
∂ul
Multiplying by g k l and using g k l = g kl ∂u
∂uk ∂ul
, we get after simplification
0 0
0 ∂ui ∂uj ∂uk ∂ 2 uk ∂uk
Γki0 j 0 = Γkij + . (3.28)
∂ui0 ∂uj 0 ∂uk ∂ui0 ∂uj 0 ∂uk
Exercise 3.4.1. Solve the following questions;
1. Find Γijk and Γkij for the surface on which g12 = 0;

2. Find Γkij for the surface of revolution x = f (u) cos v, y = f (u) sin v, z =
h(u), hence deduce their values for a sphere;

3. Obtain the law of transformation for Γkij (15.8) by differentiating the


law of transformation for gij .

45
The Second fundamental form Codazzi Equations and Gauss Theorem

3.5 Codazzi Equations and Gauss Theorem


We have Gauss formula (15.1)

r̄ij = Γlij r̄l + bij N̄ ,


Differentiating this with respect to uk , we get

Γlij ∂bij
r̄ijk = k
r̄l + Γlij r̄lk + k N̄ + bij N̄k .
∂u ∂u
Using Gauss and Weingarten formulas (3.18), (3.19), we get

∂Γlij ∂bij
r̄ijk = k
r̄l + Γlij (Γm
lk r̄m + blk N̄ ) + N̄ − bij blk r̄l .
∂u ∂uk
∂Γlij ∂bij
i.e., r̄ijk = ( k
+ Γm l l
ij Γmk − bij bk )r̄l + ( + blk Γlij )N̄ . (3.29)
∂u ∂uk
Since r̄ijk = r̄ikj , we get on equating the coefficients of r̄l and N̄ respectively

∂Γlij m l l ∂Γlik
+ Γ Γ
ij mk − b b
ij k ) = − Γm l l
ik )Γmj − bik bj (3.30)
∂uk ∂uj
and
∂bij ∂bik
k
+ blk Γlij = + bij Γlik . (3.31)
∂u ∂uj
These are called Codazzi Equations.

For i = 1, j = 1, k = 2, we get
∂b11 ∂b12
2
+ bl2 Γl11 = + bl1 Γl12 . (3.32)
∂u ∂u1
For i = 2, j = 1, k = 2, we get
∂b21 l ∂b22
+ b l2 Γ21 = + bl1 Γl22 . (3.33)
∂u2 ∂u1
For all other values of i, j, k (3.31) reduces to an identity. Hence (3.32) and
(3.33) are the only two independent Codazzi Equations.

46
The Second fundamental form Codazzi Equations and Gauss Theorem

Gauss Equations:
Consider now the left hand side of (3.30). Multiplying it by gls , we get

∂Γlij
gls − Γm
ij Γmks − bij bks. (3.34)
∂uk
Now
∂Γijs ∂(gls Γlij ) ∂Γlij ∂gls l
= = gls Γ .
∂uk ∂uk ∂uk ∂uk ij
Hence (3.34) can be written as

∂Γijs ∂gls l
k
− k Γij − Γlij Γlks − bij bks ,
∂u ∂u
 
∂Γijs ∂gls l l 1 ∂gks ∂gls ∂glk
or, − k Γij + Γij + k − − bij bks ,
∂uk ∂u 2 ∂ul ∂u ∂us
∂Γijs
or, − Γlij Γksl − bij bks .
∂uk
Hence (3.30) can now be written as

∂Γijs l ∂Γiks
− Γ ij Γksl − b ij b ks = − Γlik Γjsl − bik bjs ,
∂uk ∂uj
 
∂Γijs ∂Γiks
− glm Γlij Γm l m

or, bij bks − bik bjs = k
− j ks − Γik Γjs .
∂u ∂u
Simplifying the terms in the first bracket we get
 2
∂ 2 gik ∂ 2 gij ∂ 2 gks

1 ∂ gjs
bij bks − bik bjs = − − −
2 ∂ui ∂uk ∂uj ∂us ∂uk ∂us ∂ui ∂uj
−glm Γlij Γm l m

ks − Γik Γjs . (3.35)
Taking i = j = 1, k = s = 2, this gives
 2
1 ∂ 2 g11 1 ∂ 2 g22

∂ g12 l m l m

b= − − − glm Γ11 Γ22 − Γ12 Γ 12 . (3.36)
∂u1 ∂u2 2 (∂u2 )2 2 (∂u1 )2

This is called the Gauss Formula.

47
The Second fundamental form Codazzi Equations and Gauss Theorem

Gauss Theorem: The Gauss curvature κ is a function of gij and its


derivatives of first and second order, and hence a bending invariant.
Proof.
b
κ=
g
 2
1 ∂ 2 g11 1 ∂ 2 g22
 
1 ∂ g12 l m l m

i.e., κ = − − − glm Γ11 Γ22 − Γ12 Γ12 .
g ∂u1 ∂u2 2 (∂u2 )2 2 (∂u1 )2
(3.37)
which is a function of gij and its first and second derivatives only and hence
a bending invariant.

Example 3.5.1. hSolve the followin questions;


∂Γl ∂Γl i
(1) Show that κ = g1 ∂u112 − ∂u121 + Γm Γ
11 m2
l
− Γ m l
Γ
12 m1 gl2 .
Hint: Use (3.34) in (3.30) and get
!
∂Γlij ∂Γlik m l m l

bij bks − bik bjs = gls − − g ls Γ Γ
ij mk − Γ Γ
ik mj
∂uk ∂uj

putting i = j = 1, k = s = 2, get the result.


2
(2) Show that κ = [r̄11 ,r̄1 ,r̄2 ][r̄22 ,r̄1g,r̄2 2 ]−[r̄12 ,r̄1 ,r̄2 ,] .
Hint: κ = (b11 b22 − b212 ) /g and
1 ×r̄2 )
b11 = N̄ · r̄11 = r̄11 · (r̄
|r̄1 ×r̄2 |
] = [r̄11 , r̄1 , r̄2 ]/g.

r̄11 · r̄22 r̄1 · r̄22 r̄2 · r̄22


1
(3) Show that κ = 2 r̄11 · r̄1 r̄1 · r̄1 r̄2 · r̄1
g
r̄11 · r̄2 r̄1 · r̄2 r̄2 · r̄2

r̄12 · r̄12 r̄1 · r̄12 r̄2 · r̄12


1
− r̄12 · r̄1 r̄1 · r̄1 r̄2 · r̄1 .
g2
r̄12 · r̄2 r̄1 · r̄2 r̄2 · r̄2

Hint : Use the result of Ex 2 and the fact that


ā · d¯ b̄ · d¯ c̄ · d¯
¯ ē, f¯] =
[ā, b̄, c̄][d, ā · ē b̄ · ē c̄ · ē .
ā · f¯ b̄ · f¯ c̄ · f¯

48
The Second fundamental form Codazzi Equations and Gauss Theorem

∂ 2 g12 1 ∂ 2 g11 1 ∂ 2 g22 1 ∂g11 ∂g12


∂u1 ∂u2
− 2 ∂u2 ∂u2
− 2 ∂u1 ∂u1 2 ∂u1 ∂u1
− 3 ∂g
∂u2
11

1 ∂g12 1 ∂g22
(4) Show that κ = ∂u2
− 2 ∂u1
g11 g12
g2
1 ∂g22
2 ∂u2
g12 g22
1 ∂g11 1 ∂g22
0 2 ∂u2 2 ∂u1

1 1 ∂g11
− 2 ∂u2
g11 g12 .
g2
1 ∂g22
2 ∂u1
g12 g22

Hint: Use the result of Ex 3.

(5) For orthogonal coordinates, show that


  √   √ 
−1 ∂ 1 ∂ g22 ∂ 1 ∂ g11
κ= √ √ + 2 √ .
g11 g22 ∂u1 g11 ∂u1 ∂u g22 ∂u2
Hint: Use the expression for κ as given in (3.37) and simplify.

Fundamental Theorem of Surface Theory(Bonnet The-


orem):
If gij and bij are given as functions of u1 and u2 which satisfy the Gauss
and Coddazi Equations and g 6= 0, then there exists a surface whose first
and second fundamental forms are ds2 = gij dui duj and

II = bij dui duj .

Illustration: Given ds2 = (du)2 +cos2 u(du)2 and II = (du)2 +(cos2 v)(dv)2 ,
find the surface.

49
Chapter 4

CURVE OF SURFACES

4.1 Curvature Of A Curve On A Surface


Let r̄ = r̄(u1 , u2 ) be a surface and let ui = ui (s) be a curve C on it. The
equation of C can be written as

r̄ = r̄(u1 (s), u2 (s)).

The curvature vector κ̄ of the curve C is


 
dt̄ 00
κ̄ = κn̄ = = r̄ , (4.1)
ds

where t̄ is the tangent and n̄ the principal normal of the curve C.


Consider now a unit vector ū = N̄ × t̄ which is perpendicular to both N̄ and
t̄. Since n̄ is also perpendicular to t̄, it will lie in the plane of N̄ and ū. we
resolve κ̄ along N̄ and ū and write

κ̄ = κ̄g + κ̄n , (4.2)

where κ̄g = κg ū and κ̄n = κn N̄ .


Definition 4.1.1. Normal Curvature Vector and Geodesic
Curvature Vector:
(1) The resolved part of κ̄ along N̄ is called the Normal Curvature vector κ̄n .

(2) The resolved part of κ̄ perpendicular to N̄ (along ū) is called the


geodesic Curvature vector κ̄g .
Curve of Surfaces Curvature Of A Curve On A Surface

Expressions for κg and κn :

We have
dui
r̄0 = r̄i ,
ds
and hence
dui duj d2 ui
κ̄ = r̄00 = r̄ij + r̄i 2 .
ds ds ds
Using the Gauss formula r̄ij = Γkij r̄k + bij N̄ , we get

 dui duj d2 uk
κ̄ = Γkij r̄k
+ bij · N̄ + r̄k 2 ,
ds ds ds
 2 k i j
dui duj
  
du k du du
or κ̄ = + Γij r̄k + bij N̄ . (4.3)
ds2 ds ds ds ds
Compairing this with (4.2), we get
 2 k i j

du k du du
κ̄g = + Γij r̄k , (4.4)
ds2 ds ds

dui duj
 
and κ̄n = bij N̄ . (4.5)
ds ds
Since κ̄n = κn N̄ , we get from (4.5)

bij dui duj


κn = , (4.6)
gij dui duj

κn is called the Normal Curvature.

51
Curve of Surfaces Geodesic Curvature

4.2 Geodesic Curvature


We have,
κ̄g = κg ū, hence κg = κ̄g · ū.

Theorem 4.2.1. The value of Geodesic Curvature is given by


du1 du2
√ ds ds
κ̄g = g (4.7)
d2 u1 i
1 du du
j
d2 u2 i
2 du du
j
+ Γij + Γij
ds2 ds ds ds2 ds ds
Proof. From (17.4),

du2 uk i j
 
k du du
κg = κ̄g · ū = + Γij r̄k · ū
ds2 ds ds

Now,

dui du2
r̄1 · ū = r̄i · (N̄ × t̄) = t̄ × r̄1 · N =
(r̄i × r̄1 ) · N = (r̄2 × r̄1 ) · N̄ ,
ds ds
√ du2
 
r̄1 × r̄2 √
⇒ r̄1 · ū = − g . since = N̄ and |r̄1 × r̄2 | = g
ds |r̄1 × r̄2 |
Similarly we show that
√ du1
r̄2 · ū = g .
ds
Hence,
 2 1 i j
√ du2
  2 2 i j
√ du1
  
du u 1 du du du u 2 du du
κg = + Γij g + + Γij g ,
ds2 ds ds ds ds2 ds ds ds

which is the result (4.7).

Corollary 4.2.1.
κg = [N̄ , r̄0 , r̄00 ]. (4.8)
κg = κ̄g · ū = κ̄ · ū = r̄00 · N̄ × t̄ = r̄00 · N̄ × r̄0 = [N̄ , r̄0 , r̄00 ].

52
Curve of Surfaces Geodesic Curvature

Corollary 4.2.2. The geodesic curvature of parametric curves are:


√ √
g g
(κg )1 = Γ211 √ , (κg )2 = −Γ122 √ . (4.9)
g11 g11 g22 g22
du2
Proof: For the curve u2 =const., ds
= 0 and ds2 = g11 (du1 )2 , i.e.,
du1
ds
= √g111 .
Putting these values in (4.7), we get
2 √
du1 du1
  
√ g
(κg )1 = g Γ211 = Γ211 .
ds ds (g11 )3/2

Similarly, we get the other part for u1 = constant.

Corollary 4.2.3. If the parametric curves are perpendicular, then


√ √
1 ∂ log g 11 1 ∂ log g 22
(κg )1 = − √ , (κg )2 = − √ . (4.10)
g 22 ∂u2 g 11 ∂u1

Proof: Given that g12 = 0, hence g = g11 g22 and


1 ∂g11 1
Γ211 = g 22 Γ112 = − g 22 2 and g 22 = .
2 ∂u g22
Hence, from Corollary 2
√ √
1 22 ∂g11 g11 g22 1 1 ∂g11 1 ∂ log g11
(κg )1 = − g √ =− =− .
2 ∂u2 g11 g11 2g22 g 11 ∂u2 g22 ∂u2

Similarly, we can prove the other part too.

Liouville’s Formula: The parametric curves are perpendicular, and


(κg )1 and (κg )2 be their geodesic curvatures, then the geodesic curvature of
a curve making an angle θ with u1 -curve is given by

κg = + (κg )1 cos θ + (κg )2 sin θ (4.11)
ds
Proof: Take the unit vectors ī1 , ī2 tangent to the perpendicular parametric
curves.

53
Curve of Surfaces Geodesic Curvature

Then, by definition

dī1 ˙ dī2 ˙
(κg )1 = · ī2 and (κg )2 = − · ī1 ,



ds1 √ ds√2 

where ds1 = g11 du1 and ds2 = g22 du2 . (4.12)
Also, we have t̄ = ī1 cos θ + ī2 sin θ 



and ū = −ī1 sin θ + ī2 cos θ.

Now,
dī1 dī1 ds2 dī1 ds2 dī1 dī1
= + = cos θ + sin θ.
ds ds1 ds ds2 ds ds1 ds2
Similarly,
dī2 dī2 dī2
= cos θ + sin θ.
ds ds1 ds2
Differentiating t̄ and using these results, we get

dt̄ dī1 dī1 dī2 dī2 dθ


= cos2 θ + cos θ sin θ + cos θ sin θ + sin2 θ + ū .
ds ds1 ds2 ds1 ds2 ds
Then
dt̄ dθ dī2 dī2 dī1 dī1
κg = ·ū = −ī1 cos θ sin2 θ−ī1 sin3 θ+ī2 cos3 θ+ī2 cos2 θ sin θ.
ds ds ds1 ds2 ds1 ds2
Using the expression for (κg )1 and (κg )2 and simplifying


κg = + (κg )1 cos θ + (κg )2 sin θ.
ds

54
Curve of Surfaces Geodesic

4.3 Geodesic
Definition 4.3.1. Geodesic:
A Geodesic on a surface is a curve whose geodesic curvature Kg = 0 (This is
also the curve of shortest distance between two points on the surface).
We have seen in (4.3) that
 2 k i j
dui duj
  
du k du du
κ̄ = + Γij r̄k + bij N̄ .
ds2 ds ds ds ds
Hence, the curve is a geodesic if and only if
d2 u k i
k du du
j
+ Γ ij = 0. (4.13)
ds2 ds ds
This is, therefore, the equation of the geodesic.
These are in fact two equations for k = 1, 2. However, if we eliminate
ds between (4.13) and ds2 = gij dui duj , we get only one relation.
This can also be achieved by using the expression (4.7) for κg putting
i 1 du2
it zero and using du ds
= du
du2 ds
.
We get for the single equation of geodesic:
 1 3
d2 u 1  du1 2  du1
 
2 du 1 2 1 2
− Γ11 + Γ 11 − 2Γ12 + 2Γ 12 − Γ 22 + Γ112 = 0.
(du2 )2 du2 du2 du2
(4.14)
1 2 du1
From (4.14), we see that when at a point P (u , u ) if direction du2 is given
d2 u1
then (du 2 )2 is determined, i.e., the geodesic is determined. Hence the following

theorem.
Theorem 4.3.1. At every point on a surface of class C 2 there exists a
unique geodesic in every direction.
Theorem 4.3.2. A curve C on a surface
P
is a geodesic if and only if
its osculating plane contains the normal N̄ to the surface.
Proof: We have κ̄ = κ̄g − κ̄n , hence C is geodesic if and only if κ̄g = 0, i.e.,
if and only if κ̄ = κ̄n . Since κ̄n is along N̄ and κ̄ along the principal normal
n̄. C is a geodesic if and only if N̄ = n̄. Since the osculating planes of C
passes through t̄ and n̄, it passes
P through N̄ , i.e., the osculating plane of C
must be a normal plane of .

55
Curve of Surfaces Geodesic

Exercise 4.3.1. Solve the following questions;


1. Show that on a plane the geodesics are straight lines;

2. Show that on a sphere the geodesics are great circles.;

3. Find the geodesics on a surface of revolution;

4. For a curve φ(u1 , u2 ) = 0 on a surface. Show that


 
1 g12 φ2 − g22 φ1 ∂ g12 φ1 − g11 φ2
(κg ) = − √ + .
g g11 (φ2 )2 − 2g12 φ1 φ2 + g22 (φ2 )2 ∂u2 g11 (φ2 )2 − 2g12 φ1 φ2 + g22 (φ1 )2

56
Curve of Surfaces Normal Curvature

4.4 Normal Curvature


Consider a curve on the surface passing through a given point P . The pro-
jection of the curvature vector κ̄ of the curve at P on the normal vector N̄
of the surface at P is called the Normal Curvature at P of the curve on the
surface.

Theorem 4.4.1. The nomral curvature at P of a curve on a given surface


depends on the direction of the curve and equals the ratio of the values of
the second fundamental form to the fundamental form for this direction;

bij dui duj


κn = . (4.15)
gij dui duj

Proof: We know that,


κn = κ̄n · N̄ = κ̄ · N̄ .
As we have seen that κ̄ = κ̄g +κ̄n , since κ̄g is tangent to the surface, κ̄g · N̄ = 0;
therefore, κ̄n = κ̄ · N̄ = κn N̄ , which implies that

bij dui duj


κn = ,
dsds
where s is the natural parameter on the curve. But for the natural parameter
we have the identity  2
ds dui duj
= gij = 1,
ds ds ds
which implies that
i duj
bij du
ds ds
κn = dui duj
.
gij ds ds
Multiplying the numerator and denominator by ds2 , we obtain (4.15). In
this form, the formula does not depend in the parameterization. Moreover,
the formula involves only the first directive of the coordinates of points of
the curve in the natural parameter. Thus it depends only on the direction
and also the sense of the curve. But the change of sign of all differentials
involved does not affect the ratio. Therefore, the normal curvature depends
on the directions only.

57
Curve of Surfaces Normal Curvature

We have defined normal curvature in section (4.1) and seen in (4.6) that
κn = κ̄n · N̄ = κ̄ · N , i.e.,
bij dui duj
κn = . (4.16)
gij dui duj
du1
Thus, κn for a curve depends on the ratio du2
only, i.e, on the tangent line.
Hence we have the following theorem.

Theorem 4.4.2. All curves on a surface which have the same line as
tangent have the same Normal Curvature κr .

58
Curve of Surfaces Normal Curvature

Meusnier Theorem: Let α be the angle between N̄ and n̄, so that


N̄ · n̄ = cos α.

Then,  
dt̄
κn = κ̄n · N̄ = κ̄ · N̄ = .N = κn̄N̄ = κ cos α,
dS
i.e., κn = κ cos α. (4.17)
Consequences of Meusnier Theorem:
κn depends on t̄ only (Th.1), and cos α depends on n̄ only since N̄ is fixed for
the surface. Hence from (4.17). κ depends on the osculating plane (Through
t̄ and n̄) if cos α 6= 0.
If, however, cos α = 0 then N̄ ⊥ n̄, i.e., the osculating plane is the tangent
plane to the surface. Hence;

Theorem 4.4.3. All curves on a surface through a point P which have


the same osculating plane through P have the same curvature κ at P , pro-
vided the osculating plane is not tangent to the surface.

Definition 4.4.1. Normal Section:


Normal Section at P is the curve along which a normal plane through P
cuts the surface.
For Normal Section N̄ = n̄, hence cos α = N̄ · n̄ = 1, i.e., κn = κ. Hence the
theorem;

Theorem 4.4.4. The curvature κ of a normal section at a point is equal


to the Normal Curvature κn of the section at P .

59
Principal Curvatures,
Curve of Surfaces Gaussian and Mean Curvatures

4.5 Principal Curvatures, Gaussian and Mean


Curvatures
du2
The expression for κn (4.6), by putting λ = du1
can be written as

b11 + 2b12 λ + b22 λ2


κn = . (4.18)
g11 + 2g12 λ + g22 λ2
Thus κn is a function of λ and hence will have maximum and minimum
values κ1 and κ2 obtained by putting dκ n

= 0. These are called the Principal
Curvatures k1 , k2 of the curve at P .
To obtain the principal curvatures, write (4.18) as

(g11 + 2g12 λ + g22 λ2 )κn = (b11 + 2b12 λ − b22 λ2 ).


dκn
Differentiating w.r.to λ and putting dλ
= 0, we get

(2g12 + 2λg22 )κn = 2b12 + 2b22 λ,


b12 + b22 λ
i.e., κn = . (a)
g12 + g22 λ
Further (4.18) can also be written as
b11 + b12 λ + λ(b12 + b22 λ)
κn = . (b)
(g11 + g12 λ) + λ(g12 + g22 λ)
Hence, from (a) and (b), we get for pricipal curvature
b11 + b22 λ b12 + b22 λ
κn = = . (4.19)
g11 + g12 λ g12 + g22 λ
This can also be written as

(b12 − κn g12 ) + λ(b22 − κn g22 ) = 0,
(4.20)
(b11 − κn g11 ) + λ(b12 − κn g12 ) = 0.

Eliminating λ, we get

(b11 − κn g11 )(b22 − κn g22 ) = (b12 − κn g12 )2 ,


i.e., (b11 b22 − b212 − (g11 b22 − 2g12 b12 + g22 b11 )κn + (g11 g22 − g12
2
)κ2n = 0,

60
Curve of Surfaces Principal Directions and Lines of Curvature

i.e., b − g(g 22 b22 + 2g 12 b12 + g 11 b11 )κn + gκ2n = 0,


b
i.e., κ2n − b1i kn + = 0. (4.21)
g
The roots of this equation are the principal curvatures κ1 and κ2 .

Mean Curvature: M = κ1 + κ2 = bii .

Gaussian Curvature(or Total Curvature): K = κ1 κ2 = gb .

4.6 Principal Directions and Lines of


Curvature
Definition 4.6.1. Principal Directions:
The two directions at a point for which κn is maximum or minimum are
called the Principal Directions.

Definition 4.6.2. Lines of Curvature:


The curves whose tangent at every point is along the principal directions at
that point are called the Lines of Curvature.

From (4.19) we get

(g11 b12 − g12 b11 ) + (g11 b22 − g22 b11 )λ + (g12 b22 − g22 b12 )λ2 = 0,

where λ = du2 /du1 . Hence,

(g11 b12 − g12 b11 )(du1 )2 + (g11 b22 − g22 b11 )dudu2 + (g12 b22 − g22 b12 )(du2 )2 = 0,
(4.22)
which can also be written as
(du2 )2 −du1 du2 (du1 )2
g11 g12 g22 = 0. (4.23)
b11 b12 b22

Equation (22.1) is the equation for the Principal directions. This is, therefore,
also the differential equation for the lines of curvature.

61
Curve of Surfaces Principal Directions and Lines of Curvature

du1 δu1
Let these directions be du2
and δu2
. Then it can be verified that they
satisfy the relation

du1 δu1 du1 δu1


 
g11 2 2 + g12 + + g22 = 0, (4.24)
du δu du2 δu2
which is a necessary and sufficient condition for them to be orthogonal. Hence
the result
Theorem 4.6.1. From any point of the surface there pass two line of
curvature which are perpendicular.
Theorem 4.6.2. The lines of curvature are parametric curves if and only
if

g12 = 0 = b12 . (4.25)


Proof: If we take the lines of curvature (which are perpendicular) as para-
metric curves then g12 = 0. Now, if du2 = 0 satifies (4.22), we get

g11 b12 = 0. (a)

If du1 = 0 satisfies (4.22), we get

g22 b12 = 0 (b)

Since, g = g11 g22 − g12 > 0, and g12 = 0, g11 , g22 cannot be zero. Hence
(a) & (b) imply that b12 = 0.
To prove the sufficiency if we put g12 = 0 = b12 in (4.22), we get du1 du2 = 0
which are parametric curves.

Euler’s Theorem: The Normal Curvature κn in any direction C at P


is given by
κn = κ1 cos2 α + κ2 sin2 α, (4.26)
where κ1 , κ2 are the principal curvatures at P , and α the angle between C
and the principal direction corresponding to κ1 .
Proof: Take the lines of curvature as parametric curves. Then g12 = 0 = b12 .
Hence,
b11 b22
κ1 = and κ2 = .
g11 g22

62
Curve of Surfaces Principal Directions and Lines of Curvature

Let ds1 be the length element along u1 -curve, so that



ds1 = g11 du1 .

Then,
ds1 √ du1 √ du2
cos α = = g11 , and sin α = g22
ds ds ds
Now,
2 2
b11 (du1 )2 + b22 ∗ du2 )2 √ du1 du2
 
b11 b22 √
κn = 2
= g11 + g21 ,
ds g11 ds g22 ds

i.e., κn = κ1 cos2 α + κ2 sin2 α.


Rodrigue’s Formula: (du1 ) is a principal direction if and only if
dN̄ + kdr̄ = 0 (4.27)

and then k is the principle curvature in this direction.


Proof: Let (du1 ) be the principal direction, then the expressions giving the
principal curvature κ (4.20) can be written as:

(κg11 − b11 )du1 + (κg12 − b12 )du2 = 0

and (κg12 − b12 )du1 + (κg22 − b22 )du2 = 0,


or, (κr̄1 · r̄1 + N̄1 · r̄1 )du1 + (κr̄1 · r̄2 + N̄2 · r̄1 )du2 = 0
and (κr̄1 · r̄2 + N̄1 · r̄2 )du1 + (κr̄2 · r̄2 + N̄2 · r̄2 )du2 = 0,
or, (dN̄ + κdr̄) · r̄1 = 0 and (dN̄ + κdr̄) · r̄2 ) = 0.
Showing that (dN̄ + κdr̄) ⊥ r̄1 and r̄2 . Since (dN̄ + κdr̄) ⊥ N̄ also, it
follows that
dN̄ + kdr̄ = 0.
To prove the sufficiency we proceed in the reverse direction.

Example 4.6.1. (use of Rodrigue’s formula):


Show that,
dN̄ · dN̄ + M dN̄ · dr̄ + Kdr̄ · dr̄ = 0. (4.28)

63
Curve of Surfaces Principal Directions and Lines of Curvature

Solution: Take the lines of curvature as parameteric curves, then g12 = 0 =


b12 .
Since the parametric curves have to satisfy Rodrigue’s formula, we get
N̄1 du1 + κ1 r̄1 du1 = 0 and N̄2 du2 + κ2 r̄2 du2 = 0.
i.e., N̄1 = −κ1 r̄1 and N̄2 = −κ2 r̄2 .
Hence,
dN̄ = dN̄1 du1 + N̄2 du2 = −κ1 r̄1 du1 − κ2 r̄2 du2 ,
i.e., dN̄ + κ1 dr̄ = (κ1 − K2 )r̄2 du2 .
Similarly,
dN̄ + κ2 dr̄ = (κ2 − κ1 )r̄1 du1 .
Hence,
(dN̄ + K1 dr̄) · (dN̄ + κ2 dr̄) = −(κ1 − κ2 )r̄1 r̄2 du1 du2 = 0,
i.e., dN̄ · dN̄ + (K1 + K2 )dN̄ · dr̄ + K1 K2 dr̄ · dr̄ = 0.
Exercise 4.6.1. Solve the following questions;
1. For th surface r̄ = (u, v, u2 +v 2 ) find the lines of curvature and principal
curvature;
2. Find the lines of curvature and principal curvature for the surface of
revolution;
3. Find the lines of curvature for z = f (x, y);
4. Prove that the sum of normal curvatures at a point in any pair of ⊥
directions as constant;
5. If two surfaces intersect at a constant angle and if the curve of inter-
section is a line of curvature on one surface, it is a line of curvature on
the other also;
6. For the surface F (x, y, z) = c, show that the principal directions are
given by
Fx dFx dx
Fy dFy dy = 0 and Fx dx + Fy dy + Fz dz = 0.
Fz dFz dz

64
Curve of Surfaces Asymptotic Lines

4.7 Asymptotic Lines


A direction (dui ) is called Asymptotic if

II = bij dui duj = 0. (4.29)

A curve whose direction at every point satisfies (4.29) is called Asymptotic


Line.
The differential equation of the Asymptotic line is

b11 (du1 )2 + 2b12 du1 du2 + b22 (du2 )2 = 0. (4.30)

Now,

dN̄ · dr̄ = (N̄i dui ) · (r̄j duj ) = (N̄i · r̄j )dui duj = −bij dui duj .

Hence for Asymptotic lines, we have

dN̄ · dr̄ = 0. (4.31)

Theorem 4.7.1. The normal curvature κn for Asymptotic line is zero.


Proof: From the equation (4.29), we have

bij dui duj


κn = = 0.
gij dui duj

Theorem 4.7.2. Parametric curves are Asymptotic lines iff b11 = 0 = b22 .
Proof: If we put b11 = 0 = b22 in (4.30), we get du1 du2 = 0, i.e., the
parametric curves. On the other hand if du1 = 0 and du2 = 0 are to satisfy
(4.30) we must have b11 = 0 = b22 .

Theorem 4.7.3. Either the Asymptotic line is a straight line or its os-
culating plane is the tangent plane of the surface.

Proof: For asymptotic line κn = κ̄ · N̄ = 0 = κn̄ · N̄ . Hence either κ = 0


which means it is a straight line or n̄ · N̄ = 0, i.e., the principal normal lies
in the tangent plane (so does t̄). Hence the osculating plane, which passes
through n̄ and t̄ is the tangent plane.

65
Curve of Surfaces Asymptotic Lines

Theorem 4.7.4. If the Asymptotic line is not a straight line, its torsion
τ 2 = −K (Gaussian Curvature).

Proof: Since for asymptotic line N̄ · n̄ = 0 (Theorem 3), and also N̄ · t̄ = 0,


db̄
hence N̄ =∈ b̄ (binormal). Now ds = τ n̄, i.e.,

db̄ db̄ dN̄ dN̄


τ2 = · = · . (a)
ds ds ds ds
We use the formula dN̄ · dN̄ + M dN̄ · dr̄ + Kdr̄ · dr̄ = 0 (from 4.28). Since
dr̄ · dr̄ = ds2 and dN̄ · dr̄ = 0 (from 4.31). We get −K = ddsN̄ · ddsN̄ = τ 2 .

Exercise 4.7.1. Solve the following questions;


1. Find the asymptotic lines on the surface r̄ = (r cos θ, r sin θ, log r);

2. For the surface F (x, y, z) = C. The asymptotic lines are given by


Fx dx + fY dy + Fz dz = 0 and dFx dx + dFY dy + dFz dz = 0;

3. Find the asymptotic lines for Z = x4 + y 4 .

66
Curve of Surfaces Conjugate Direction

4.8 Conjugate Direction


Definition 4.8.1. Given a direction (dui ), the direction (δui ) is called
Conjugate to it if
dN̄ · δr̄ = 0, (4.32)
i.e., (N1 du1 + N2 du2 ) · (r̄1 δu1 + r̄2 δu2 ) = 0,
i.e., b11 du1 δu1 + b12 (du1 δu2 + du2 δu1 ) + b22 du2 δu2 = 0. (4.33)

This relation is symmetric in α and δ. Hence we simply say that (dui )


and δui ) are conjugate directions.

Theorem 4.8.1. The parametric curves are conjugate iff b12 = 0.


Proof: If (4.33) is to be satisfied by du1 = 0 and δu2 = 0, we must have
b12 = 0.
If b12 = 0, (4.33) becomes

b11 du1 δu1 + b22 du2 δu2 = 0

which is satisfied by du1 = 0 and δu2 = 0.

Corollary 4.8.1. If parametric curves are conjugate and also orthogonal


then they are lines of curvature.

This follows from the following tables:



(1) parametric curves are orthogonal ⇐⇒ g12 = 0; 

(2) parametric curves are lines of curvature ⇐⇒ g12 = 0 = b12 ;

(4.34)
(3) parametric curves are asymptotic ⇐⇒ b11 = 0 = b22 ; 

(4) parametric curves are conjugate ⇐⇒ b12 = 0.

Note: The Asymptotic lines are self conjugate.

Theorem 4.8.2. At Elliptic and Hyperbolic points every direction has


a conjugate direction.

Proof: Given a direction (dui ), let (δui ) be the conjugate direction to it.
Then (4.33) can be written as

(b11 du1 + b12 du2 )δu1 + (b12 du1 + b22 du2 )δu2 = 0.

67
Curve of Surfaces Conjugate Direction

This will always have a root (δui ) iff

(b11 du1 + b12 du2 )2 + (b12 du1 + b22 du2 )2 > 0,

for all values of du1 , du2 , i.e.,

(b211 + b212 )(du1 )2 + 2(b11 b12 + b12 b22 )du1 du2 + (b212 + b222 )(du2 )2 > 0

which is true if its discriminant is negative,

i.e., b212 (b11 + b22 )2 − (b211 + b212 )(b212 + b222 ) < 0,

i.e., − (b11 b22 − b212 )2 < 0,


i.e., b11 b22 − b212 6= 0,
i.e., point is not parabolic.

Exercise 4.8.1. Solve the following exercise;


(1) On the surface r̄ = (u, v, u2 + v 2 ), find the conjugate family of the curve
u 2 + v 2 = c2 ;
(2) Prove that the principal directions bisect the asymptotic direction;
(3) Show that the mean curvature is zero on a surface whose asymptotic line
are perpendicular.

68
Chapter 5

DEVELOPABLE SURFACES

5.1 Envelopes
Consider the one parametric family of surfaces

F (x, y, z, c) = 0.

Two neighbouring surfaces are F (x, y, z, c) = 0 and F (x, y, z, c + δc) = c,


which can also be written as
F (x, y, z, c + δc) − F (x, y, z, c)
F (x, y, z, c) = 0 and = 0, δc 6= 0.
δc
These two surfaces will in general intersect on a curve. As δc → 0, the curve
of intersection approaches a limiting position called the Characteristic Curve,
given by

F (x, y, z, c) = 0 and F (x, y, z, c) = 0. (5.1)
∂c
Each C determines a characteristic curve. The locus of all these curves,
obtained by illuminating C from (5.1), gives a surface called the envelope of
the one-parameter family.
Now consider two neighbouring characteristic curves

F (x, y, z, c) = 0, F (x, y, z, c) = 0
∂c

and F (x, y, z, c + δc) = 0, F (x, y, z, c + δc) = 0,
∂c
Developable surfaces Envelopes

which in general intersect at a point called the characteristic point. The locus
of the points is the envelope of the characteristics and is called the Edge of
Regression, given by 
∂F (x, y, z, c) = 0,  
∂ 

F (x, y, z, c) = 0, (5.2)
∂c2
∂ 

F (x, y, z, c) = 0 

∂c2
and illuminating c from them.

Exercise 5.1.1. Solve the following exercise;


(1) Find the envelope, characteristics and edge of regression for the osculating
plane
(R̄ − r̄(s)) · b̄(s) = 0;
(2) Find the envelope and edge of regression for

x sin C − y cos C + z tan θ = C, θ is constant;

(3) Find the envelope and edge of regression for

x2 y 2 z62
c2 ( 2
+ 2 ) + 2 = 1.
a b c

70
Developable surfaces Developable surfaces

5.2 Developable Surfaces


A Developable Surface is the envelope of a one-parametric family of planes.

Developables associated with the space curves:

(1) Tangent Developable (Osculating Developable): It is the envelope


of osculating planes. Osculating planes is

F3 ≡ (R̄ − r̄(s)) · F2 (s) = 0. (a)


∂F
= 0 =⇒ −t̄ · b̄ + (R̄ − r̄)(−τ n̄) = 0. (b)
∂s
∂F
= 0 =⇒ (R̄ − r̄)(−k t̄ + τ b̄) = 0,
∂s2
or, (R̄ − r̄) · t̄ = 0. (c)
The edge of regression given by (a), (b) & (c) is the curve itself. i.e.,
(R̄ − r̄) = 0.
The characteristic lines given by (a), (b) & (c) are the tangent lines (R̄− r̄) =
ct̄.

(2) Polar Developables: It is the envelope of the Normal Planes

(R̄ − r̄) · t̄ = 0. (a)


∂F
= 0 =⇒ (R̄ − r̄) · n̄ = k −1 = ρ. (b)
∂c
∂ 2F
= 0 =⇒ (R̄ − r̄) · b̄ = τ −1 ρ0 . (c)
∂c2
Hence the edge of regression is

R̄ = r̄ + ρn̄ + τ −1 ρ0 b̄.

(3) Rectifying Develpables: It is the envelope of the Rectifying plane

(R̄ − r̄) · n̄ = 0. (a)


∂F
= 0 =⇒ (R̄ − r̄) · (−k t̄ + τ b̄) = 0. (b)
∂s

71
Developable surfaces Developable surfaces

The envelope is the locus of lines in the rectifying plane passing through the
point P of the curve and making an angle tan−1 κτ with tangent.
Now,
∂ 2F
= 0 =⇒ (R̄ − r̄) · (−k 0 t̄ − τ 0 b̄) = −κ (c)
∂s2
κ2 κτ
(b)+(c) ⇒ (R̄ − r̄) · b̄ = τ κ0 −τ 0 κ
, (R̄ − r̄) · t̄ = τ κ0 −τ 0 κ
.
Hence the edge of regression is

κb̄ − τ t̄
R̄ − r̄ = κ .
τ κ0 − τ 0 κ
Theorem 5.2.1. Any developable is either a cylinder or a cone or may
be regarded as the tangent developable of its edge of regression.

Proof: Consider the one parameter family of planes

X̄ · ā + p = 0, ā = ā(u), p = p(u). (a)

The characteristic line is given by

X̄ · ā + p = 0, X̄ · ā0 + p0 = 0. (b)

The characteristic point (whose locus is the edge of regression) is given by


the equation

X̄ · ā + p = 0, X̄ · ā0 + p0 = 0, X̄ · a00 + p00 = 0. (c)

The point does not exist when [ā, ā0 , ā00 ] = 0, in which case the vector field
ā is plane. We note that the vector ā is perpendicular to the plane (a). If
the vectors ā(u) are parallel to a plane π, the planes (a) are parallel to the
direction parallel to π.
In this case the envelope of the plane (a) is a cylinder with generating
lines perpendicular to the planeπ.
When the characteristic point is same for all planes, the envelope is a
cone generated by the characteristic lines.
In the general case, the locus of characteristic points is a curve C, the
edge of regression. Its equation is given by (c) where X̄ is also considered as
a function of the parameter u. Differentiating (c), we get

X̄ 0 · ā + X̄ · ā0 + p0 = 0, X̄ 0 · a0 + X̄ · ā00 + p00 = 0,

72
Developable surfaces Developable surfaces

i.e., X̄ 0 · ā = 0, and X̄ 0 · a0 = 0.
Differentiating again, we get

X̄ 00 · ā = 0,

which shows that ā is parallel to X̄ 0 × X̄ 00 .


The osculating planes of the curve C at the characteristic point is there-
fore identical with (a), i.e., the envelope is, therefore, a tangent developable.

Theorem 5.2.2. A necessary and sufficient condition for a surface to be


developable is that its Gaussian Curvature K be zero.
Proof: If the developable is a cylinder or a cone, the Gaussian Curvature is
evidently zero.
If these cases are excluded, it must be a tangent developable and its
equation can be written as

R̄ = r̄(s) + v t̄(s).

R̄1 = t + vκr̄, R̄2 = t̄, R̄11 = κn̄ − v(tr̄)0 , R22 = 0, R12 = κn̄.
R̄1 × R̄2 vκn̄ × t̄
N̄ = = = n̄ × t̄ = −b̄.
|R̄1 × R̄2 | vκ
Hence,
b11 b22 − b212
b11 = N̄ · R̄11 = −vκτ, b12 = 0, b22 = 0 and κ = = 0.
g
To prove the sufficiency, consider

b11 b22 −b212 = (r̄1 N̄1 )(r̄2 N̄2 )−(r̄1 N̄2 )(r̄2 N̄1 ) = (r̄1 ×r̄2 )·(N̄1 ×N̄2 ) = |r̄1 ×r̄2 |[N̄ N̄1 N̄2 ].

Since κ = 0, we get [N̄ N̄1 N̄2 ] = 0, hence N̄ , N̄1 , N̄2 are coplanar. Now
from N̄ · N̄1 = 0 = N̄ · N̄2 , it follows that N̄1 = 0 or N̄2 = 0 or N̄1 = κN̄2 .
Let N̄2 = 0. The tangent plane has the equation

(R̄ − r̄) · N̄ = 0.

Since [(R̄ − r̄) · N̄2 ] = −r̄N̄2 = 0,


the equation involves only one parameter, the surface is an envelope of one

73
Developable surfaces Ruled Surfaces

parameter family of planes and hence a developable.


If N̄1 = κN̄2 , we apply the transformation
u = u0 + v 0 , v 0 = u0 − κv 0
Then N̄10 = N̄1 + N̄2 and N̄20 = N̄1 −κN̄2 = 0, which brings it to the previous
case.

5.3 Ruled Surfaces


A Ruled Surface is a surface generated by the motion of a straight line, which
is called its generator.

We know that a developable surface is also generated by a straight


line (characteristic line), and hence is a ruled surface. But all ruled surfaces
are not developables. We shall now find a condition when a ruled surface
becomes a developable.

Let C be a curve r̄ = r̄(u) on the surface which meets each generator


precisely once. This is called a base curve or directrix.
Let ī = ī(u) be a unit vector in the direction of a generator passing
through a point A(u) of C. A general point P on the ruled surface is given
by
R̄ = r̄(u) + v̄ ī(u), ī = R̄ = R(· ·).
where ī=const. the surface is a cylinder and when r̄=const., it is a cone.
Now,
R̄1 = r̄0 + v ī0 , R̄2 = ī, R̄11 = r̄00 + v ī00 , R̄12 = ī0 , R̄22 = 0.
(r̄0 + v ī0 ) × ī [r̄0 , ī, ī]
N̄ = √ , b22 = 0, b12 = √ .
g g
Hence,
[r̄, ī, ī0 ]2
κ= .
g2
[r̄0 , ī, ī0 ] is called Parametric of Distribution. Hence κ = 0 is a necessary
and sufficient condition for surface to be developable. Hence we have the
following result;

74
Developable surfaces Ruled Surfaces

Theorem 5.3.1. A necessary and sufficient condition that a ruled surface


be a developable is that the parametric of distribution vanishes.

Note: In the case N̄2 = 0, i.e., the tangent plane is same all along the
generator, or, the whole generator become the tangent plane.

75
Index

Angle Between Two Tangent Vectors, First Fundamental Form Of A Sur-


30 face, 27
Arc Length, 3 Formulas of Gauss and Weingarten,
Area Of A Surface, 33 43
Asymptotic Lines, 65 Fundamental Theorem For Space
Curves, 18
Binormal vector, 10 Fundamental Theorem of Surface The-
Bonnet Theorem, 49 ory, 49
Change Of Curvilinear Coordinates, Gauss Equations, 47
34 Gauss Formula, 43
Characteristic Curve, 69 Gauss formula, 47
Christoffel Symbols, 43, 44 Gauss Theorem, 48
Codazzi Equations, 46 Gaussian Curvature, 40
Conjugate Direction, 67 Gaussian Curvatures, 61
Contact of a curve and a plane, 7 Gaussian Map, 40
Contact of Curves, 4 Geodesic, 55
Contact of order at least n, 7 Geodesic Curvature, 52
Contact of order n, 5 Geodesic Curvature Vector, 50
Coordinate Curves, 22
Curvature, 12 Helix, 18
Curve, 1 Hyperbolic Point, 39
Curvilinear Coordinates, 22
Length Of A Tangent Vector, 30
Darpoux Vector, 17 Lines of Curvature, 61
Liouville’s Formula, 53
Edge of Regression, 70 Locally one-one, 1
Elliptic Point, 39
Envelopes, 69 Mean Curvatures, 61
Equation of the osculating plane, 7 Meusnier Theorem, 59
Euler’s Theorem, 62 Natural or Intrinsic Equations, 18

76
Index Index

Natural parameter, 3 The Frenet Trihedron, 10


Normal, 24 The Three Planes, 11
Normal Curvature, 57 Torsion, 14
Normal Curvature Vector, 50
Normal plane, 11 Weingarten Formula, 43
Normal Section, 59
Normal Vector, 24

Osculating Developable, 71
Osculating plane, 7

Parabolic Point, 39
Parametric Curve, 22
Parametric of Distribution, 74
Piece-wise regular curve, 2
Polar Developables, 71
Principal Curvatures, 60
Principal Directions, 61
Principal normal vector, 10

Rectifying Develpables, 71
Rectifying plane, 11
Regular curve, 1
Regular Point, 21
Rodrigue’s Formula, 63
Ruled Surfaces, 74

Second Fundamental Form, 38


Serret-Frenet Formulas, 16
Simple Arc, 1
Simple Sheet of a surface, 20
Singular Point, 21
Singular point, 2
Spherical Map, 40

Tangent Developable, 71
Tangent Plane, 24
Tangent to a curve, 6
Tangent vector, 10

77

You might also like