Vardhaman Mahaveer Open University, Kota: Differential Geometry and Tensors-1
Uploaded by
Rahul NayakVardhaman Mahaveer Open University, Kota: Differential Geometry and Tensors-1
Uploaded by
Rahul NayakMA/MSc MT-04
Vardhaman Mahaveer Open University, Kota
Differential Geometry and Tensors-1
Course Development Committee
Chairman
Prof. (Dr.) Naresh Dadhich
Vice-Chancellor
Vardhaman Mahaveer Open University, Kota
Co-ordinator/Convener and Members
Subject Convener Co-ordinator
Prof. D.S. Chauhan Dr. Anuradha Sharma
Department of Mathematics Assistant Professor
University of Rajasthan, Jaipur Department of Botany, V.M.O.U., Kota
Members :
1. Prof. V.P. Saxena 4. Prof. S.P. Goyal 7. Dr. Paresh Vyas
Ex Vice-Chancellor Emeritus Scientist (CSIR) Assistant Professor
Jiwaji University, Deptt. of Mathematics Deptt. of Mathematics
Gwalior (MP) University of Rajasthan, Jaipur University of Rajasthan, Jaipur
2. Prof. S.C. Rajvanshi 5. Dr. A.K. Mathur 8. Dr. Vimlesh Soni
Deptt. of Mathematics Associate Prof. (Retired) Lecturer
Institute of Eng. & Tech. Deptt. of Mathematics Deptt. of Mathematics
Bhaddal, Ropar (Punjab) University of Rajasthan, Jaipur Govt. PG College, Kota (Raj.)
3. Prof. P.K. Banerjee 6. Dr. K.N. Singh 9. Dr. K.K. Mishra
Emeritus Fellow (UGC) Associate Prof. (Retired) Lecturer
Deptt. of Mathematics Deptt. of Mathematics Deptt. of Mathematics
J.N.V. University, Jodhpur University of Rajasthan, Jaipur M.S.J. Collage, Bharatpur (Raj.)
10. Dr. K.S. Shekhawat
Lecturer, Deptt. of Mathematics
Govt. Shri Kalyan College, Sikar (Raj.)
Editing and Course Writing
Editor Writers
Prof. P.R. Sharma 1. Dr. K.C. Sharma 2. Dr. Anil Sharma
Deptt. of Mathematics Lecturer Lecturer
University of Rajasthan, Jaipur Deptt. of Mathematics Deptt. of Mathematics
M.S.J. College, Bharatpur R.R. College, Alwar
3. Dr. G.R. Yadav 4. Dr. Paresh Vyas
Lecturer Assistant Professor
Deptt. of Mathematics Deptt. of Mathematics
R.R. College, Alwar University of Rajasthan, Jaipur
Academic and Administrative Management
Prof. (Dr.) Naresh Dadhich Prof. M.K. Ghadoliya Mr. Yogendra Goyal
Vice-Chancellor Director (Academic) Incharge
Vardhaman Mahveer Open University, Vardhaman Mahveer Open University, Course Material Production
Kota Kota and Distribution Department
Course Material Production
Mr. Yogendra Goyal
Assistant Production Officer
Vardhaman Mahaveer Open University, Kota
MA/MSc MT-04
Vardhaman Mahaveer Open University, Kota
Differential Geometry and Tensors-1
Unit No. Units Page No.
1. Space Curves, Tangent, Contact of Curve and Surface, Osculating Plane 1–22
2. Principal Normal and Binormal, Curvature, Torsion, Serret-Frenet’s
Formulae, Osculating Circle and Osculating Sphere 23–55
3. Existence and Uniqueness Theorems, Bertrand Curves, Involute,
Evolutes, Conoids, Inflexional Tangents, Singular Points, Indicatrix 56–79
4. Envelope, Edge of Regression, Ruled Surfaces, Developable Surface,
Tangent Plane to a Ruled Surface 80–102
5. Necessary and Sufficient Condition that a Surface = F (, ) should
Represent a Developable Surface, Metric of a Surface 103–113
6. First, Second and Third Fundamental Forms, Fundamental Magnitudes
of Some Important Surfaces, Orthogonal Trajectories, Normal Curvature 114–137
7. Meunier’s Theorem, Principal Direction and Principal Curvatures,
First Curvature, Mean Curvature, Gaussian Curvature, Umbilics,
Radius of Curvature of any Normal Section at an Umbilic on z = f (x, y).
Radius of Curvature of a given Section through any Point on
z = f (x, y), Lines of Curvature 138–173
8. Principal Radii, Relation between Fundamental Forms, Asymptotic
Lines, Differential Equation of an Asymptotic line, Curvature and
Torsion of an Asymptotic Line 174–197
9. Geodesics, Differential Equation of a Geodesic, Single Differential
Equation of a Geodesic, Geodesic on a Surface of Revolution,
Geodesic Curvature and Torsion, Gauss-Bonnet Theorem 198–239
10. Gauss’s Formulae, Gauss’s Characteristic Equation Weingarten
Equations, Mainardi-Codazzi Equations. Fundamental Existence
Theorem for Surfaces, Parallel Surfaces, Gaussian and Mean
Curvature for a Parallel Surface, Bonnet’s Theorem on Parallel Surfaces 240–253
Reference Books 254
PREFACE
The Present book entitled ‘‘Differential Geometry and Tensors-1’’ has been
designed so as to cover the unit-wise syllabus of Mathematics-Fourth paper for
M.A./M.Sc. (Previous) students of Vardhaman Mahaveer Open University, Kota. It
can also be used for competitive examinations. The basic principles and theory
have been explained in a simple, concise and lucid manner. Adequate number of
illustrative examples and exercises have also been included to enable the students
to grasp the subject easily. The units have been written by various experts in the
field. The unit writers have consulted various standard books on the subject and
they are thankful to the authors of these reference books.
MA/MSc MT-04
Vardhaman Mahaveer Open University, Kota
Differential Geometry and Tensors
Course Development Committee
Chairman
Prof. (Dr.) Naresh Dadhich
Vice-Chancellor
Vardhaman Mahaveer Open University, Kota
Co-ordinator/Convener and Members
Subject Convener Co-ordinator
Prof. D.S. Chauhan Dr. Anuradha Sharma
Department of Mathematics Assistant Professor
University of Rajasthan, Jaipur Department of Botany, V.M.O.U., Kota
Members :
1. Prof. V.P. Saxena 4. Prof. S.P. Goyal 7. Dr. Paresh Vyas
Ex Vice-Chancellor Emeritus Scientist (CSIR) Assistant Professor
Jiwaji University, Deptt. of Mathematics Deptt. of Mathematics
Gwalior (MP) University of Rajasthan, Jaipur University of Rajasthan, Jaipur
2. Prof. S.C. Rajvanshi 5. Dr. A.K. Mathur 8. Dr. Vimlesh Soni
Deptt. of Mathematics Associate Prof. (Retired) Lecturer
Institute of Eng. & Tech. Deptt. of Mathematics Deptt. of Mathematics
Bhaddal, Ropar (Punjab) University of Rajasthan, Jaipur Govt. PG College, Kota (Raj.)
3. Prof. P.K. Banerjee 6. Dr. K.N. Singh 9. Dr. K.K. Mishra
Emeritus Fellow (UGC) Associate Prof. (Retired) Lecturer
Deptt. of Mathematics Deptt. of Mathematics Deptt. of Mathematics
J.N.V. University, Jodhpur University of Rajasthan, Jaipur M.S.J. Collage, Bharatpur (Raj.)
10. Dr K.S. Shekhawat
Lecturer, Deptt. of Mathematics
Govt. Shri Kalyan College, Sikar (Raj.)
Editing and Course Writing
Editor Writers
Prof. P.R. Sharma 1. Dr. K.C. Sharma 2. Dr. Anil Sharma
Deptt. of Mathematics Lecturer Lecturer
University of Rajasthan, Jaipur Deptt. of Mathematics Deptt. of Mathematics
M.S.J. College, Bharatpur R.R. College, Alwar
3. Dr. G.R. Yadav 4. Dr. Paresh Vyas
Lecturer Assistant Professor
Deptt. of Mathematics Deptt. of Mathematics
R.R. College, Alwar University of Rajasthan, Jaipur
Academic and Administrative Management
Prof. (Dr.) Naresh Dadhich Prof. M.K. Ghadoliya Mr. Yogendra Goyal
Vice-Chancellor Director (Academic) Incharge
Vardhaman Mahveer Open University, V.M.O. University, Course Material Production
Kota Kota and Distribution Department
Course Material Production
Mr. Yogendra Goyal
Assistant Production Officer
Vardhaman Mahaveer Open University, Kota
MA/MSc MT-04
Vardhaman Mahaveer Open University, Kota
Differential Geometry and Tensors
Unit No. Units Page No.
1. Space Curves, Tangent, Contact of Curve and Surface, Osculating Plane 1–22
2. Principal Normal and Binormal, Curvature, Torsion, Serret-Frenet’s
Formulae, Osculating Circle and Osculating Sphere 23–55
3. Existence and Uniqueness Theorems, Bertrand Curves, Involute,
Evolutes, Conoids, Inflexional Tangents, Singular Points, Indicatrix 56–79
4. Envelope, Edge of Regression, Ruled Surfaces, Developable Surface,
Tangent Plane to a Ruled Surface 80–102
5. Necessary and Sufficient Condition that a Surface = F (, ) should
Represent a Developable Surface, Metric of a Surface 103–113
6. First, Second and Third Fundamental Forms, Fundamental Magnitudes
of Some Important Surfaces, Orthogonal Trajectories, Normal Curvature 114–137
7. Meunier’s Theorem, Principal Direction and Principal Curvatures,
First Curvature, Mean Curvature, Gaussian Curvature, Umbilics,
Radius of Curvature of any Normal Section at an Umbilic on z = f (x, y).
Radius of Curvature of a given Section through any Point on
z = f (x, y), Lines of Curvature 138–173
8. Principal Radii, Relation between Fundamental Forms, Asymptotic
Lines, Differential Equation of an Asymptotic line, Curvature and
Torsion of an Asymptotic Line 174–197
9. Geodesics, Differential Equation of a Geodesic, Single Differential
Equation of a Geodesic, Geodesic on a Surface of Revolution,
Geodesic Curvature and Torsion, Gauss-Bonnet Theorem 198–239
10. Gauss’s Formulae, Gauss’s Characteristic Equation Weingarten
Equations, Mainardi-Codazzi Equations. Fundamental Existence
Theorem for Surfaces, Parallel Surfaces, Gaussian and Mean
Curvature for a Parallel Surface, Bonnet’s Theorem on
Parallel Surfaces 240–253
11. Tensor Analysis, Kronecker Delta, Contravariant and Covariant
Tensors, Symmetric Tensors, Quotient law of Tensors,
Relative tensor 254–275
12. Riemannian Space, Metric Tensor, Indicator, Permutation Symbol
and Permutation Tensors, Christoffel Symbols and their Properties 276–296
13. Covariant Differentiation of Tensors, Ricci Theorem,
Intrinsic Derivative 297–313
14. Geodesics, Differential Equation of Geodesic, Geodesic
Coordinates, Field of Parallel Vectors 314–332
15. Riemannian-Christoffel Tensor and its Properties, Covariant
Curvature Tensor, Einstein Space, Bianchi’s Identity, Einstein
Tensor, Flat Space, Isotropic Point, Schur’s Theorem 333–348
Reference Books 349
PREFACE
The Present book entitled ‘‘Differential Geometry and Tensors’’ has been
designed so as to cover the unit-wise syllabus of Mathematics-Fourth paper for
M.A./M.Sc. (Previous) students of Vardhaman Mahaveer Open University, Kota. It
can also be used for competitive examinations. The basic principles and theory
have been explained in a simple, concise and lucid manner. Adequate number of
illustrative examples and exercises have also been included to enable the students
to grasp the subject easily. The units have been written by various experts in the
field. The unit writers have consulted various standard books on the subject and
they are thankful to the authors of these reference books.
UNIT 1 : Space Curves, Tangent, Contact of Curve and
Surface, Osculating Plane
Structure of the Unit
1.0 Objectives
1.1 Introduction
1.2 Space curves
1.2.1 Vector representation of a space curve
1.2.2 Unit tangent vector of a curve
1.2.3 The equation of tangent line to a curve at a given point
1.2.4 Equation of tangent line when the equation of the curve is given as the intersection of
two surfaces.
1.2.5 Direction cosines of the tangent line
1.2.6 Examples
1.2.7 Self-learning exercise-1
1.3 Contact of curve and surface
1.3.1 Definition
1.3.2 To find the condition that a curve and a surface have a contact of nth order
1.3.3 Inflexional tangent
1.3.4 Examples
1.3.5 Self-learning exercise-2
1.4 Osculating plane
1.4.1 To find the equation of the osculating plane
1.4.2 Equation of the osculating plane in terms of general parameter t
1.4.3 Equation of osculating plane in cartesian coordinates
1.4.4 Examples
1.4.5 Self-learning exercise-3
1.5 Answers to self-learning exercises
1.6 Exercises
1
1.0 Objectives
This unit provides a general overview of :
• Differential geometry
• Space curves
• Tangent
• Contact of curve and surface
• Osculating plane.
1.1 Introduction
Differential geometry is that part of geometry which is treated with the help of differential calcu-
lus. There are two branches of differential geometry :
Local differential geometry : In which we study the properties of curves and surfaces in the
neighbourhood of a point.
Global differential geometry : In which we study the properties of curves and surfaces as a
whole.
1.2 Space curves
A curve in space is defined as the locus of a point whose cartesian coordinates are the functions
of a single variable parameter u, say.
We can represent a space curve in the following two ways :
As intersection of two surfaces :
Let f1 (x, y, z) = 0, f2 (x, y, z) = 0 be two surfaces then these equations together represent the
curve of intersection of the above surfaces. If this curve lies in a plane then it is called a plane curve,
otherwise it is called to be skew, twisted or tortous.
For example, if f1 (x, y, z) = 0, represents a sphere and f2 (x, y, z) = 0 represents a plane then
these two equations together represent a circle.
Parametric representation :
If the coordinates of a point on a space curve be represented by the equations of the following
form
x = f1 (t), y = f2 (t), z = f3 (t) .....(1.2.1)
where f1, f2, f3 are real valued functions of a single real variable t ranging over a set of values
a t b.
The equation in (1.2.1) are called parametric equation of the space curve.
2
1.2.1 Vector representation of a space curve :
If r be the position vector of a current point A on the space curve whose cartesian coordianates
be x, y, z then we know that
r xiˆ yjˆ zkˆ
or r f1 (t ) iˆ f 2 (t ) ˆj f3 (t ) kˆ
or r = f (t)
or r = (f1 (t), f2 (t), f3 (t)) .....(1.2.2)
where f is a vector valued function of a single variable t. Thus space curve may be defined as :
A space curve is the locus of a point whose position vector r with respect to a fixed origin
may be expressed as a function of single parameter.
1.2.2 Unit tangent vector of a curve :
Consider two neighbouring points A (x, y, z) and B (x + x, y + y z + z) on a curve C whose
position vectors are r and r + r, respectively. We have
Fig. 1.1
AB OB OA r r r r .
Let s be length of arc AB measured along the curve and arc PA = s is measured from any
convenient point P on the curve.
AB r
Unit vector along chord AB
AB Chord AB
r Arc AB
.....(1.2.3)
s Chord AB
But as B tends to A, then the chord AB tends to be tangent at P.
Arc AB
Also we know that lim 1
B A Chord AB
3
r Arc AB dr
Hence, unit vector along tangent at A lim 1
B A s Chord AB ds
dr
r .....(1.2.4)
ds
Unit tangent vector at A is denoted by tˆ and is taken in the direction of s increasing
If r x, y , z i.e. r xiˆ yiˆ zkˆ
dr dx dy dz
then tˆ , ,
ds ds ds ds
dx dy ˆ dz ˆ
i.e. tˆ iˆ j k .....(1.2.5)
ds ds ds
Since tˆ is unit tangent vector, tˆ 1.
2 2 2
dx dy dz
1
ds ds ds
2 2 2
dx dt dy dt dz dt
or 1
dt ds dt ds dt ds
2 2 2 2
ds dx dy dz
or
dt dt dt dt
or s 2 x 2 y 2 z 2 ,
ds dx
where s , x etc. .....(1.2.6)
dt dt
and t is any parameter.
1.2.3 The equation of tangent line to a curve at a given point :
The tangent line to a curve at any point A is defined as the limiting position of a straight line
through the point A and a neighbouring point B on the curve as B tends to A along the curve.
t
P
B
A
O r
Fig. 1.2
4
Let r r s be the parametric equation of a curve and A be any point on it whose position
ˆ dr
vector is r and a unit tangent vector at A be denoted by t r .
ds
Let P be any point on the tangent line at A whose position vector is R (say).
Also AP w tˆ where AP w
But OP OA AP
R r w tˆ or R r w r .....(1.2.7)
Equation (1.2.2) gives us the equation of tangent line at A.
Tangent line in cartesian form :
We may write r xiˆ yjˆ zkˆ
r xiˆ y ˆj z kˆ
and R Xiˆ Y ˆj Zkˆ
Substituting these value in equation (1.2.2) of tangent line, we get
Xiˆ Yjˆ Zkˆ xiˆ yjˆ zkˆ c xiˆ y ˆj z kˆ
or Xiˆ Yjˆ Zkˆ x cx iˆ y cy ˆj z cz kˆ,
where c is a non-zero constant.
Equating coefficients of iˆ, ˆj , kˆ from both sides
X x cx, Y y cy, Z z cz
X x Y y Zz
i.e. c,
x y z
X x Yy Zz
i.e. .....(1.2.8)
x y z
This is the required equation of tangent line at (x, y, z) and direction cosines of the tangent line
are proportional to x, y, z.
1.2.4 Equation of tangent line when the equation of the curve is given as the intersec-
tion of two surfaces :
Let the equation of two surfaces are
F1 (x, y, z) = 0 and F2 (x, y, z) = 0 .....(1.2.9)
where x, y, z are functions of a parameter.
F1 dx F1 dy F1 dz
Now 0 .....(1.2.10)
x dt y dt z dt
F2 dx F2 dy F2 dz
0 .....(1.2.11)
x dt y dt z dt
5
Hence from equation (1.2.3) and (1.2.4)
x y z
F1 F2 F1 F2 F1 F2 F1 F2 F1 F2 F1 F2
y z z y z x x z x y y x
which are the direction ratios of the tangent and dot represents differentiation w.r. to ‘t’.
Therefore, the equation of tangent line at a point (x, y, z) on the curve of intersection of the two
given surfaces is given as
X x Yy Z z
.....(1.2.13)
F1 F2 F1 F2 F1 F2 F1 F2 F1 F2 F1 F2
y z z y z x x z x y y x
1.2.5 Direction-cosines of the tangent line :
Let A (x, y, z) and B (x + x, y + y, z + z) be adjacent points on a given curve in rectangular
coordinate axes. r the measure of chord AB is given by
2 2 2
r 2 = x + y + z
Let s be the length of the arc measure from some fixed point P to any point A on the curve.
If the measure of the arc AB of the curve be s then
2 2 2 2
r x y z
s s s s
Chord AB
Since lim 1
B A Arc AB
2 2 2
dx dy dz
1 ,
ds ds ds
2 2 2 2
ds dx dy dz 2
or r
dt dt dt dt
Hence s 2 x 2 y 2 z 2 .....(1.2.14)
dx
where x, y, z are functions of t and x etc.
dt
But x , y , z are direction ratios of a tangent line therefore the direction cosines of the tangent
line at A are
x y z dx dy dz
, , or , ,
s s s ds ds ds
dr dx ˆ dy ˆ dz ˆ
But i j k.
ds ds ds ds
The direction cosines of the tangent line are x, y, z which are the components of r where a
prime denotes differentiation with respect to s. Clearly r 1, i.e. r is unit vector along the tangent.
6
1.2.6 Examples :
Ex.1. Find the equation to the tangent at the point on the circular helix
x = a cos , y = a sin , z = C
Sol. The vector equation of the helix is given by
r a cos iˆ a sin ˆj C kˆ
r a sin iˆ a cos ˆj C kˆ
The equation of the tangent in given by
R r r
or
R a cos iˆ a sin ˆj C kˆ a sin iˆ a cos ˆj C kˆ
If R Xiˆ Yjˆ Zkˆ,
then Xiˆ Yjˆ Zkˆ a cos sin iˆ a sin cos ˆj C kˆ
X a cos Y a sin Z C
which gives .
a sin a cos C
It is the required equation of tangent line.
Ex.2. Show that the tangent at any point of the curve whose equations are
x = 3t, y = 3t2, z = 2t3
makes a constant angle with line
y = z – x = 0.
Sol. The direction-rations of the tangent at ‘t’ to the given curve are
3, 6t, 6t2 i.e., x, y, z
The direction ratios of the given line are
1, 0, 1.
If be the angle between the tangent and the given line, than
3 1 6t 0 6t 2 1
cos
9 36t 2 36t 4 1 0 1
3 1 2t 2
1
2 3 1 2t 2 2
which is independent of t, hence is constant.
Ex.3. Show that the tangent at a point of the curve of the intersection of the ellipsoid
x2 y 2 z2
1
a2 b2 c 2
7
and the confocal whose parameter is given by
x X x y Y y z Z z
.
a 2 b2 c2 a 2
b2 c2 a2 b2
c 2 a2 b2 c 2
Sol. The equation of a confocal to the ellipsoid
x2 y 2 z 2 .....(1)
F1 1 0
a 2 b2 c 2
is x2 y2 z2 .....(2)
F2 2 1 0
a b2 c 2
Equations to a tangent line are
X x Yy Zz
.....(3)
F1 F2 F1 F2 F1 F2 F1 F2 F1 F2 F1 F2
y z z y z x x z x y y x
F1 2 x F1 2 y F1 2 z F2 2x F 2 y F2 2z
Here 2, 2, 2, 2 , 2 2 , 2
x a y b 3 c x a y b z c
Putting these values in equation (3), we get
x X x y Y y z Z z
2
a b c 2 2
a 2
2
b c a 2 2
b 2
c 2
a 2
b2 c 2
which are the required equations of the tangent.
1.2.7 Self-learning exercise-1 :
1. Name the branches of differential Geometry.
2. If the curve lies in a plane then it is called ..... .
3. The intersection of two surfaces is called ........ .
4. Write the equation of a tangent line at a point.
5. Write the equation of tangent line when the equation of the curve is given as the intersection
of two surfaces.
1.3 Contact of curve and surface
We know that in a plane curve the tangent at A is the limiting position of the chord AB when B
coincides with A. In a similar manner if A1, A2, ..., An+1 be points on a given curve lying on a given
surface and if A2, A3, ..., An+1 all coincide with A1, than we say that a curve has a contact of nth order
with the surface at A1. We may also say that the curve and the surface has (n + 1) points of contact.
1.3.1 Definition :
If A, A1, A2,..., An points on a given curve lie on a given surface and A1, A2, ..., An coincide
with A, then curve and surface are said to have the contact of nth order at the point A.
8
1.3.2 To find the condition that a curve and a surface have a contact of nth order :
Let the equation of the curve C be given by
r = {x (t), y (t), z (t),} .....(1.3.1)
and the equation of the surface S be given by
f (x, y, z) = 0 .....(1.3.2)
The values of t corresponding to the points of intersection of the curve C and surface S are the
roots of the equation
F (t) = f {x (t), y (t), z (t)} = 0 .....(1.3.3)
Let t = t0 be a root of the equation F (t) = 0 so that
F (t0) = 0. .....(1.3.4)
Then t = t0 give as a point of intersection of C and S.
Put t = t0 + h so that
F (t) = F (t0 + h). .....(1.3.5)
Expanding F (t) about t0 by Taylor’s theorem, we get
2 3
h h
F t F t0 h F t0 F t0 F t0 ... . .....(1.3.6)
2 3
Since t0 is a solution of the equation (1.3.4) therefore F (t0) = 0, then we have
2 3
h h
F t h F t0 F t0 F t0 ... . .....(1.3.7)
2 3
We have the following cases :
(i) If F t0 0, then we say that the curve and the surface have a simple intersection at r
(t0).
(ii) If F t0 0, but F t0 0, then F (t) is of second order of h and we say that t0 is a
double zero of F (t) and in this case C and S have two points of contact (or contact of first
order) at r (t0).
(iii) If F t0 0, F t0 0, but F
t 0, then F (t) is of third order of h and we say that
0
t0 is a triple zero of F (t) and in this case we say that C and S have three point contact or
contact of second order.
(iv) In general if F t0 0, F t0 0,..., F n 1 t0 0, but f n (t0) 0, then F (t) is of nth
order of h and we say that C and S have a n point contact or contact of (n – 1)th order.
1.3.3 Inflexional tangent :
A straight line which meets the surface S in three coincident points i.e., it has a second order
point of contact is called inflexional tangent to the surface at that point.
9
1.3.4 Examples :
Ex.4. Find the plane that has three point contact at the origin with the curve
x = t4 –1, y = t3 –1, z = t2 –1.
Sol. Let the equation of the plane at the origin be
lx+my+nz=0 .....(1)
The equations of the given curve are
x = t4 –1, y = t3 –1, z = t2 –1 .....(2)
At origin,
t4 –1 = 0, t3 –1 = 0, t2 –1 = 0.
Clearly, t = 1 satisfies all of these three equations. Hence, at the origin, we have t = 1.
Now the points of intersection of the curve (2) and the surface (1) are given by the zeroes of the
function
F (t) = l (t4 –1) + m (t3 –1) + n (t2 –1)
or F (t) = l t4 + m t3 + n t2 – l – m – n .....(3)
For three point contact, we should have
F t 0, F t 0.
Now F = 4l t3 + 3m t2 + 2nt = 0 .....(4)
and F = 12l t2 + 6m t + 2n = 0 .....(5)
At the origin i.e. at t = 1, the equation (4) and (5) become
4l + 3m + 2n = 0, 12l + 6m + 2n = 0 .....(6)
Solving equation (6), we get
l m n
3 8 6
Hence the required equation of plane is
3x – 8y + 6z = 0.
Ex.5. Prove that if the circle
lx + my + nz = 0, x2 + y2 + z2 = 2cz
has three point contact at the origin with the paraboloid
ax2 + by2 = 2z,
then c = (l2 + m2)/(bl2 + am2)
Sol. Let the parametric equation of the circle be
x = f1 (t), y = f2 (t), z = f3 (t). .....(1)
Putting these values of x, y, z in the equation of the paraboloid, we get
F (t) = ax2 + by2 – 2z = 0 .....(2)
where x, y, z are functions of t.
10
For a three point contact at the origin we must have
F t F t F t 0, at the origin.
F (t) = ax2 + by2 – 2z = 0 .....(3)
F t 2axx 2byy 2 z 0 .....(4)
F t ax 2 by 2
z axx byy 0 .....(5)
2 2
At the origin we have z 0, ax by
z 0. .....(6)
Proceeding as above with the equations of the circle, we get
lx my nz 0 xx yy zz cz 0 .....(7)
lx my nz 0 x 2 y 2 z 2 xx yy zz cz 0 .....(8)
At the origin from (7) cz 0 i.e. z 0 so that lx my 0 .....(9)
At the origin from (8) and putting z 0, we have
x 2 y 2 cz 0 or x 2 y 2 c ax 2 by 2
x 2 y 2 x y
c , from (7) k , say
ax 2 by 2 m l
or c
l 2
m2 .
am 2
bl 2
Ex.6. Find the lines that have four point contact at (0, 0, 1) with the surface
x4 + 3xyz + x2 – y2 – z2 + 2yz – 3xy – 2y + 2z = 1
Sol. Any line through (0, 0, 1) is
x y z 1
k say .....(1)
l m n
Thus the parametric equations of the line are
x = lk, y = mk, z = nk + 1 .....(2)
Putting the values of x, y, z in the equation of the surface we get
F (k) = l4 k4 + 3lm k2 (nk + 1) + l2 k2 – m2 k2 – n2k2
+ 2mk (nk + 1) – 3lm k2 – 2mk + 2 (nk + 1) – 1 = 0
= l4 k4 + 3lmn k3 + (l2 – m2 – n2 + 2mn) k2 = 0 .....(3)
For four point contact we must have
F k 0, F k 0, F k 0, F
k 0 at k 0 for 0, 0,1
F k , F k are clearly zero at k = 0 .....(4)
F k 0 gives k 0
l2 – m2 – n2 + 2mn = 0. F gives lmn = 0.
Thus the direction ratios satisfy the above two relations.
11
Ex.7. Find the inflexional tangents at (x, y, z) on the surface
y2 z = 4ax.
Sol. We know that the inflexional tangents are tangents which have three point contact with the
given surface.
Any line through (x1, y1, z1) is
x x1 y y1 z z1
k say .....(1)
l m n
x = x1 + lk, y = y1 + mk, z = z1 + nk .....(2)
Substituting the values of x, y, z in the given surface, we get
F (k) = (z1 + nk) (y1 + mk)2 – 4a (x1 + lk) .....(3)
F k = n (y1 + mk)2 + 2m (y1 + mk) (z1 + nk) – 4al .....(4)
F k = 2nm (y1 + mk) + 2m2 (z1 + nk) + mn (y1 + mk) .....(5)
For three point contact at (x1, y1, z1) i.e. where k = 0, we must have
F k 0, F k 0, F k 0
Hence from (3), (4) and (5), we get
y12 z1 4ax1 0 .....(6)
2my1z1 ny12 4al 0 .....(7)
2mny1 + 2m2 z1 = 0
or 2ny1 + mz1 = 0
or mz1 .....(8)
n
2 y1
Putting for n in (7), we get
2mz1 2
2my1 z1 y1 4al 0
2 y1
3my1 z1 .....(9)
l
8a
Putting for l and n in the equation (1), we get the required inflexional tangent as
x x1 y y1 z z1
3my1z1 / 8a m mz1 / 2 y1
x x1 y y1 z z1
put y12 z1 4ax1
or 3 y12 z1 / 4a 2 y1 z1
x x1 y y1 z z1
.
3 x1 2 y1 z1
12
Ex.8. Prove that the condition that four consecutive points of a curve should be coplanar
is
x y z
x y z 0 .
x y z
Sol. Let the parametric equation of the given curve be
x = f1 (t), y = f2 (t), z = f3 (t). .....(1)
Let the parameter for a given point A (x0, y0, z0) be t0 so that the equation of the plane through
A is
(x – x0) l + (y – y0) m + (z – z0) n = 0. .....(2)
Putting for x, y, z from (1) and (2), we get
F (t) = [x (t) – x (t0)] l + [y (t) – y (t0)] m + [z (t) – z (t0)] n = 0. .....(3)
The plane (2) passes through four consecutive points if it has four point contact, i.e., if
F t0 0, F t0 0, F t0 0, F
t 0 .
0 .....(4)
These conditions are equivalent to
ym
xl zn
0 .....(5)
xl
ym
zn 0 .....(6)
xl
ym
zn 0 and clearly F (t0) = 0. .....(7)
Eliminating l, m, n between the above equations, we have
x y z
x y z 0.
x y z
(Here dashes and dots represent derivative.)
1.3.5 Self-learning exercise-2 :
1. Write the condition for simple intersection of a curve and surface.
2. Write the condition for two point contact of a curve and a surface.
3. Write the condition for three point contact of a curve and a surface.
4. Write the condition for n point contact of a curve and surface.
5. Define an inflexional tangent.
1.4 Osculating plane
Definition : The osculating plane at a point P of a curve C of class greater then or equal to two
is the limiting position of the plane passing through the tangent line at P and a neighbouring point Q on
the curve C as QP. (or which contains the tangent line at P and is parallel to the tangent at Q as
Q P).
13
Alternative : Let P, Q, R be three points on a curve C, the limiting position of the plane PQR,
when Q and R tend to P, is called the osculating plane at the point P.
1.4.1 To find the equation of the osculating plane :
C
R P
R r (s)
O
Fig. 1.3
Let r r s be the given curve C of class 2, in terms of parameter s, where s is the length of
the arc of the curve measured from a fixed point on it. Let P and Q be two neighbouring points on the
curve C with r s and r s s be their position vectors. Let R be the position vector of current
point R on the plane containing the tangent line at P and the point Q.
Here OP r s , OQ r s s , OR R
Hence PQ OQ OP r s s r s
and PR OR OP R r s
Again if tˆ be the unit tangent vector at P,
then, dr
tˆ r s .
ds
Now the vectors PR , tˆ and PQ are coplanar lying in the plane PQR and hence their scalar triple
product is zero.
PR, tˆ, PQ 0 .....(1.4.1)
or R r s , r s , r s s r s 0 .....(1.4.2)
2
r s s r s r s s r s
s .....
but .....(1.4.3)
2
We know that [a b c] = a.(b c).
14
Equation (1.4.2) may be written as
R r s r s r s s r s 0 .....(1.4.4)
form (1.4.3) and (1.4.4)
R r s r s r s s r s
s 2 ... 0
2 .....(1.4.5)
2
s
or R r s r s r s terms of higher order of s 0
2
or R r s r s r s 0 s 0 .....(1.4.6)
The plane PQR tends to be the osculating plane when Q tends to P i.e. when s 0, and
hence the equation of the osculating plane is
R r s r s r s 0
or R r s , r s , r s 0 .....(1.4.7)
Equation (1.4.7) represents the equation of the osculating plane in terms of parameter s of the
point P.
1.4.2 Equation of the osculating plane in terms of general parameter t :
Let P (t) and Q (t + t) be the two neighbouring points on curve C. Let position vector of P and
Q be r t and r t t with respect to origin, respectively..
The tangents at P and Q will be parallel to the vectors r t and r t t , respectively..
Therefore the plane through the tangents at P (t) and Q (t + t) is perpendicular to the vector
r t r t t
or to the vector r t r t t r t r t r t 0
r t t r t
i.e. to the vector r t .....(1.4.8)
t
As Q P, t 0 in this unit the osculating plane is perpendicular to the vector r t r t .
If R be the position vector of any current point on the osculating plane, the equation of the
osculating plane may be written as
R r r r 0
or R r, r ,
r 0 .....(1.4.9)
1.4.3 Equation of osculating plane in cartesian coordinates :
Let the coordinates of a point P on a given curve C be (x, y, z) and coordinates of any current
point be (X, Y, Z), these are functions of a parameter t.
15
Then r xiˆ yjˆ zkˆ
and R Xiˆ Yjˆ Zkˆ
Substituting these values in (1.4.9) the equation of the osculating plane is given by
X x iˆ Y y ˆj Z z kˆ, xi
ˆ yj ˆ,
ˆ zk xiˆ zkˆ 0
yjˆ
X x Y y Z z
x y z 0
or .....(1.4.10)
x
y
z
which is the equation of the osculating plane at a point P (x, y, z).
Theorem : To show that when the curve is analytic, there exists a definite osculating plane
at a point of inflexion, provided the curve is not a straight line.
Proof : We know that r tˆ is a unit tangent vector, therefore r2 1. .....(1)
Differentiating w.r.t. ‘s’ we get
2r r 0 or r r 0 .....(2)
Again differentiating, we get
r r r r 0
.....(3)
r r 0
(At a point P where r 0, the tangent line is called inflexional and the point P is called the
point of inflexion.)
If r 0, then r is linearly independent of r . Differentiating successively (3) and applying, above
argument shall get
r r m 0, m 2 .....(4)
where r m is the first non-zero derivative of r at P.
Therefore if r m 0, from equation (1.4.3), we get
m
s r m s 0 s m1
r s s r s
m
.....(5)
Hence the equation of the osculating plane at P is
R r s , r s , r m s 0 .....(6)
Again if for all m 2 the derivative r m 0, we conclude r tˆ is constant (since the curve
under consideration is analytic) i.e. the tangent vector is same at each point of the curve and hence the
curve is a straight line.
Hence equation (6) is the equation of the osculating plane at a point of inflexion P when the
curve is not straight line.
16
1.4.4 To find the osculating plane at a point of a space curve given by the intersection
of two surfaces.
Let the equations of the surfaces be
f r 0 and g r 0 .....(1.4.11)
The equations of the tangent planes of these surfaces are given by
R r f 0 and R r g 0 .....(1.4.12)
where f and g are normal vectors to f r 0 and g r 0 respectively and R be the position
vector of current point on the plane.
The equation of the plane through the tangent line to the curve of intersection of the two sur-
faces is
F R r f R r g 0 .....(1.4.13)
If (1.4.13) be the equation of the osculating plane at P, it must have three point contact with the
curve at P. Therefore the required conditions are
F 0, F 0, F 0 ; .....(1.4.14)
when R r and dashes denote differentiation with respect to parameter ‘t’.
F 0 gives
R f R r f
R g R r g 0 .....(1.4.15)
At P , R r , condition (1.4.12) reduces to
r f r g 0 .....(1.4.16)
But we know that r is a tangent vector and f and g are normal vectors to f r 0 and
g r 0 and hence both
r f 0 and r g 0 .....(1.4.17)
Hence F 0 reduces to an identity..
Now consider the condition F 0 at P, R r , we have
r f 0
r g 0
r f
or .....(1.4.18)
r g
Now differentiating the equation (1.4.17), we get
r f r f 0, r g r g 0, .....(1.4.19)
or r f r f
and r g r g
17
r f r f
, from (1.4.18) .....(1.4.20)
r g r g
Putting the value of in (1.4.13), we get
R r f
r f
form (1.4.20)
R r g r g
R r f R r g
or .....(1.4.21)
r f r g
Above equation represents the equation of the osculating plane at P.
Cartesian form :
Let f r f x, y, z , g r g x, y , z
R Xi Yj Zk , r xi yj zk
f f f
f iˆ ˆj kˆ
x y z
f f x iˆ f y ˆj f z kˆ
f
f xx x f xy y f xz z iˆ
substituting in equation (1.4.21) of the osculating plane, we get
X x f x Y y f y Z z f z X x g x Y y g y Z z g z .....(1.4.22)
x 2 f xx ... 2 yz
f yz ... x 2 g xx ... 2 yz
g yz ...
1.4.4 Examples :
Ex.9. For the curve x = 3t, y = 3t2, z = 2t3, show that any plane meets it in three points
and deduce the equation to the osculating plane at t = t1.
Sol. Let the equation of the plane be
Ax + By + Cz + D = 0 .....(1)
F (t) = 3At + 3Bt2 + 2Ct3 + D = 0 .....(2)
which is cubic in t. Hence the plane meets the given curve in three points.
Also x 3, y 6t , z 6t 2
x 0,
y 6,
z 12t .....(3)
Hence the equation of osculating plane at the point t1 is
x 3t1 y 3t12 z 2t13
3 6t1 6t12 0
0 6 12t1
or 2t12 x 2t1 y z 2t13 is the required equation of the osculating plane at t = t1.
18
Ex.10. Find the osculating plane at the point ‘t’ on the helix.
x = a cos t, y = a sin t, z = ct
Sol. We know that the equation of the osculating plane is
R r , r , r 0
or R r r r 0
.....(1)
Here r a cos t , a sin t , c t .....(2)
r a sin t , a cos t , c .....(3)
r a cos t , a sin t , 0 .....(4)
i j k
r r a sin t a cos t c
a cos t a sin t 0
ca sin t i ca cos t j a 2 k
ca sin t , ca cos t , a 2 .....(5)
Also R r X a cos t , Y a sin t , Z ct .....(6)
Hence R r r r X a cos t , Y a sin t , Z ct ca sin t , ca cos t , a 2
or c (X sin t – Y cos t) + a Z – act = 0 .....(7)
Alternative method :
The equation of the osculating plane is
X x Y y Z z
x y z 0
x y z
X a cos t Y a sin t Z ct
a sin t a cos t c 0
a cos t a sin t 0
or c (X sin t – Y cos t) + a Z – act = 0 .....(8)
Ex.11. Prove that the osculating plane at (x1, y1, z1) on the curve of intersection of the
cylinders x2 + z2 = a2, y2 + z2 = b2 is given by
xx13 zz13 a 4 yy13 zz13 b 4
.
a2 b2
Sol. We know that osculating plane at (x, y, z) a point on the curve of intersection of two sur-
faces is given by
19
X x fx X x gx
2
2 .....(1)
x f xx 2 yz
f yz x g xx 2 yz
g yz
where f = x2 + z2 – a2, g = y2 + z2 – b2, x, y, z are functions of ‘t’.
x y z
xx zz 0, yy zz 0 or
(1/ x) (1/ y) (1/ z )
fx = 2x, fy = 0, fz = 2z, fxx = 2, fxy = 0, fxz = 0, fyy = 0, fyz = 0, fxz = 2, gx = 0, gy = 2y,
gz = 2z, gxx = 0, gxy = 0, gxz = 0, gyy = 2, gyz = 0, gzz = 2.
Hence the equation of the osculating plane at (x1, y1, z1) is
x x1 2 x1 z z1 2 z1 y y1 2 y1 z z1 2 z1
1 1 1 1
2 2 2 2 2 2
x1 z1 y1 z1
xx zz x
1 1
2
1
z12 x12 z12
yy zz y z y
1 1
2
1
2
1
2
1 z12
x12 z12 y z 2
1
2
1
Put x12 z12 a 2 and y12 z12 b 2
xx zz a x yy zz b y
1 1
2 2
1 1 1
2 2
1
a2 b2
or
xx13 zz1 a 2 a 2
z12 yy zz b b
3
1 1
2 2
z12
2 2
a b
or
xx13 zz13 a 4 a 2 z12 zz1 yy 3
1 zz13 b 4 b 2 z12 zz1
2 2
a b
xx13 zz13 a 4 yy13 zz13 b 4
or Hence proved.
a2 b2
Ex.12. Show that the osculating plane at (x, y, z) on the curve x2 + 2ax = y2 + 2by = z2 +
2cz has the equation
(b2 – c2) (x + a)2 (X – x) + (c2 – a2) (y + b)2 (Y – y) + (a2 – b2) (z + c)2 (Z – z) = 0
Sol. Let f = x2 – y2 + 2ax – 2by
g = x2 – z2 + 2ax – 2cz.
Let P (x, y, z) be any point on the curve of intersection of these surfaces and x, y, z are func-
tions of ‘t’.
xx yy ax by 0 or x x a y y b 0
xx zz ax cz 0 or x x a z z c 0
x x a y y b z z c
20
x y z
or .....(2)
y b z c z c x a x a y b
Again fx = 2(x + a), fy = – 2(y + b), fz = 0,
gx = 2(x + a), gy = 0, gz = – 2 (z + c)
fxx = 2, fyy = – 2, fzz = 0, fxy = fyz = fzx = 0
gxx = 2, gyy = 0, gzz = – 2, gxy = gyz = gzx = 0.
The equation of the osculating plane is
X x fx X x gx
x 2 f xx 2 yz
f yz x 2 g xx 2 yz
g yz
X x 2 x 2a Y y 2 g 2b Z z 0
or 2 2 2 2
2 y b z c 2 z c x a
X x 2 x 2a Y y 0 Z z 2 z x
2 2 2 2
2 y b z c 2 x a y b
or (X – x) (x + a)2 (b2 – c2) = 0.
1.4.5 Self-learning exercise-3 :
1. Define osculating plane.
2. Write the equation of osculating plane.
3. Write the equation of osculating plane in cartesian coordinates.
1.5 Answers to self-learning exercises
Self-learning exercise-1
1. (a) Local differential geometry (b) Global differential geometry
2. Plane curve
3. Curve
X x Yy Zz
4.
x y z
X x Yy Zz
5. F F F F F F F F F F F F
1 2
1 2 1 2
1 2 1 2
1 2
y z z y z x x z x y y x
Self-learning exercise-2
1. If F 0
2. If F 0 but F 0
21
3. If F 0, F 0 but F
0
4. If F 0, F 0,..., F n 1 0 but F n 0
5. A straight line which meets the surface S in the three coincident.
Self-learning exercise-3
1. Let P, Q, R be three points on a curve C, the limiting position of the plane PQR, when Q and R
tend to P, is called osculating plane at point R.
2. R r , r , r 0
X x Y y Zz
3. x y z 0
x
y
z
1.6 Exercises
1. Find the equation of the tangent line to the curve x = t, y = t2, z = t3 at the point t = 1.
x 1 y 1 z 1
[Ans. ]
1 2 3
2. Find the equation of the tangent line at the point t = 1 to the curve r = (1 + t, t2, 1 + t3).
x 2 y 1 z 2
[Ans. ]
1 2 3
3. Define a space curve and write its parametric equations.
4. Determine a, h, b so that the paraboloid 2z = ax2 + 2hxy + by2 may have the closest possible
contact at the origin with the curve x = t3 – 2t2 + 1, y = t3 – 1, z = t2 – 2t + 1. Find also the
a h b 1
order of contact. [Ans. ; Fourth ]
45 3 5 54
5. Show that the curve x = t, y = t2, z = t3 has six point contact with the paraboloid x2 + y2 = y at
the origin.
6. Find the equation of the osculating plane of the curve given by
r = (a sin t + b cos t, a cos t + b sin t, c sin 2t).
[Ans. 2cx {a cos t (2 – cos 2t) – b sin t (2 + cos 2t)}
+ 2cy {a sin t (2 – cos 2t) – b cos t (2 – cos 2t)}
+ 2 (b2 – a2) + 3c (b2 – a2) sin 2t = 0]
7. For the curve x = 3t, y = 3t2, z = 2t3 show that any plane meets it in three points and deduce
the equation of the osculating plane at t = t1. [Ans. 2t12x – 2t1y + z = 2t13]
22
Unit 2 : Principal Normal and Binormal, Curvature, Torsion,
Serret-Frenet’s Formulae, Osculating Circle and
Osculating Sphere
Structure of the Unit
2.0 Objectives
2.1 Introduction
2.2 Principal normal and binormal
2.2.1 Principal normal
2.2.2 Binormal
2.2.3 The fundamental unit vectors tˆ, nˆ , bˆ
2.2.4 Direction cosines of the tangent, principal normal and binormal
2.2.5 Self-learning exercise-1
2.3 The three fundamental planes
2.3.1 Osculating plane
2.3.2 Normal plane
2.3.3 Rectifying plane
2.3.4 Equations of principal normal and binormal
2.3.5 Self-learning exercise-2
2.4 Curvature and Torsion
2.4.1 Curvature
2.4.2 Torsion
2.4.3 Skew-curvature
2.4.4 Self-learning exercise-3
2.5 Serret-Frenet’s formula
2.5.1 Theorems on curvature and Torsion
2.5.2 Examples
2.6 Osculating circle and osculating sphere
2.6.1 Osculating circle
2.6.2 Properties of the locus of the centre of circle of curvature
2.6.3 Osculating sphere
2.6.4 Examples
2.7 Answers to self-learning exercises
2.8 Exercises
23
2.0 Objectives
This unit provides a general overview of
• Principal normal and Binormal
• Three fundamental planes
• Curative and Torsion
• Serret-Frenet’s formula
• Osculating circle and osculating sphere
2.1 Introduction
In this unit we shall study principal normal and Binormal. Equations of principal normal
and Binormal Curvature and Torsion. Formulae for radius of curvature and radius of Torsion, Serret-
Frenet’s formula. Theorem on curvature and Torsion. In the last of this unit detailed study is given
on osculating circle and osculating sphere.
2.2 Principal Normal and Binormal
All the normals to a given curve at any point lie in the normal plane. Two nromals namely
principal normal and binormal are significant and defined in this section.
n
N Principal normal
pla ne
lating
Oscu
P
t
T Tangent
Normal
plane
Rectifying plane
B Binormal
Fig. 2.1
24
2.2.1 Principal normal :
The principal normal at any point P of a given curve C is defined as the normal which lies
in the osculating plane at P.
From the above definition it is clear that the principal normal is the line of intersection of
the normal plane and osculating plane because being normal it must lie in normal plane and be-
ing principal normal it must lie in osculating plane.
The unit vector along principal normal shall be denoted by n̂ .
2.2.2 Binormal :
The binormal at any point P of a curve C is defined as the normal which is perpendicular
to the osculating plane.
From the above definition it in clear that binormal is perpendicular to principal normal
because the perpendicular to osculating plane and the latter lies in the osculating plane.
The unit vector along the binormal shall be denoted by b̂ .
2.2.3 The fundamental unit vectors tˆ , nˆ , bˆ :
We know that principal normal and binormal are perpendicular to each other and both these
normals are perpendicular to tˆ . Hence these three form a triad of mutually perpendicular unit
vectors such that tˆ, nˆ , bˆ form a right handed orthogonal system of axes.
Therefore tˆ nˆ bˆ, nˆ bˆ tˆ, bˆ tˆ nˆ
nˆ bˆ 0, bˆ tˆ 0, tˆ nˆ 0 .....(2.2.1)
O t
t
Frenet’s frame b
Fig. 2.2
2.2.4 Direction cosines of the tangent, principal normal and binormal :
We will denote the direction-cosines of tangent by (l1, m1, n1) of principal normal by
(l2, m2, n2) and of binormal by (l3, m3, n3).
25
(i) when the parameter is arc length ‘s’
We know that unit tangent vector tˆ is given by r, we have
r
d r dx ˆ dy ˆ dz ˆ
tˆ r i j k .....(2.2.2)
ds ds ds dz
dx dy dz
Hence l1 , m1 , n1 .....(2.2.3)
ds ds ds
The binormal is perpendicular to the osculating plane. Equation of osculating plane is given
by
r r
R rr , rr, rr 0 or ( R rr ) ( rr rr ) 0
r r
Therefor the vector r r is normal to the osculating plane. This implies that binormal is
r r
parallel to the vector r r .
r r
ˆ r r
Hence b r r .....(2.2.4)
r r
y z zy zx xz
l3 , m3 ,
( yz zy)2 ( yz zy)2
xy y x
n3 .....(2.2.5)
( yz zy )2
Further ˆ (r r) r (r r ) r (r r ) r r
n̂ b tˆ
r r r r r r
since r r 1 and r r 0
x y
Hence, l2 , m2 ,
( yz zy)2 ( y z zy)2
z
n2 , .....(2.2.6)
( yz z y)2
(ii) when the parameter is ‘t’.
Here r xiˆ yjˆ zkˆ
r xiˆ yˆj z kˆ xiˆ
r xiˆ
r r ( yz z y) iˆ .....(2.2.7)
Hence the principal normal being parallel to the vector
( r r ) r
i.e. ( y z z y) iˆ ( xiˆ)
i.e. {z (z x – x z) – y (x y – y x)} iˆ
26
The direction ratios of it are
z (z x – x z) – y (x y – y x);
x (x y – y x) – z (y z – z y);
and y (y z – z y) – x (z x – x z)
Since the binormal is parallel to the vector r r i.e., to ( y z z y ) iˆ, its direction
ratios are
y z – z y, z x – x z, x y – y x. .....(2.2.8)
2.2.5 Self-learning exercise-1
1. Principal normal lies in the ........................... plane
2. ........................... is perpendicular to the osculating plane.
3. Give the formula for bˆ and nˆ.
2.3 The three fundamental planes
At each point of the curve there is a triad of orthogonal unit vectors which determine three
fundamentals planes as shown in the figure (2.1) which contains two of these, the third being the
normal to that plane and which are mutually perpendicular.
2.3.1 Osculating plane :
The plane through P containing tˆ and nˆ whose normal is therefor b̂ is called osculating
plane whose equation is given as ( R r ) bˆ 0 .
In cartesian coordinates let (X, Y, Z) be a current point and (x, y, z) the point whose oscu-
lating plane is determined.
R Xiˆ Yjˆ Zkˆ, r xiˆ yjˆ zkˆ
Again let (lr, mr, nr) (r = 1, 2, 3) be the direction ratios of the tangent, principal normal
and binormal so that
tˆ l1iˆ m1 ˆj n1kˆ ,
n̂ l 2iˆ m 2 ˆj n 2 kˆ ,
bˆ l3iˆ m3 ˆj n3kˆ .
Substituting the values of R, r and bˆ in ( R r ) bˆ 0
X x iˆ Y y ˆj Z z kˆ l3iˆ m3 ˆj n3kˆ 0
or l3 (X – x) + m3 (Y – y) + n3 (Z – z) = 0 .....(2.3.1)
2.3.2 Normal plane :
The plane through P containing bˆ and nˆ whose normal is therefore tˆ is called normal plane
whose equation in given by R r tˆ 0 .
27
In cartesian coordinates : Substituting the values of R, r and tˆ in
R r tˆ 0
X x iˆ Y y ˆj Z z kˆ l1iˆ m1 ˆj n1kˆ 0
or l1 (X – x) + m1 (Y – y) + n1 (Z – z) = 0 .....(2.3.2)
2.3.3 Rectifying plane :
The plane through P containing bˆ and tˆ whose normal is therefore n̂ is called rectifying
plane whose equation is given by R r nˆ 0 .
In cartesian coordinates : Substituting the values of R , r and nˆ in R r nˆ 0
X x iˆ Y y ˆj Z z kˆ l2iˆ m2 ˆj n2 kˆ 0
or l2 (X – x) + m2 (Y – y) + n2 (Z – z) = 0 .....(2.3.3)
2.3.4 Equations of principal normal and binormal :
R Q
C
O t
r P
b
Fig. 2.3
Let r be the position vector of any point P on the curve C referred to O as origin. Also let
R be the position vector of a current point Q on the principal normal.
We have OP r , OQ R, PQ nˆ , since n̂ is the unit vector along the principal normal
and is some scalar.
Now OQ OP PQ
i.e. R r nˆ .....(2.3.4)
which is the required equation of the principal normal at the point P of the curve C.
Similarly, if R is the position vector of a current point R on binormal, then equation of
the binormal at the point P on the curve C is given by
R r bˆ, .....(2.3.5)
where is a scalar.
28
2.3.5 Self-learning exercise-2
1. Define osculating plane.
2. Define normal plane.
3. Define rectifying plane.
4. Write equations of principal normal.
5. Write equation of binormal.
2.4 Curvature and Torsion
2.4.1 Curvature :
The rate of change of the direction of tangent with respect to the arc length ‘s’ as the point
P ( r ) moves along the curve is called curvature vector of the curve whose magnitude is denoted
by k (kappa) called the curvature at P.
dtˆ
Hence k | tˆ | | r | .....(2.4.1)
ds
Radius of curvature : The reciprocal of the curvature is called the radius of curvature
and is denoted by .
1
.....(2.4.2)
k
Curvature at a point : Let P and Q be two neighbouring points on a curve such that PQ
= s, where OP = s and the unit tangents at these point be denoted by tˆ and tˆ tˆ which makes
angle and + with a fixed direction.
Fig. 2.4
Through Q draw vector QA parallel to tˆ . If the vectors QA and QB are respectively tˆ
and tˆ tˆ then | QA | | QB | 1 and the angle between tham is .
29
Also QB QA AB or AB QB QA tˆ
Now from isosceles triangle QAB.
AB 2QA sin
2
AB 2 QA sin
2
tˆ sin 2
| tˆ | 2 sin or
2 2
tˆ sin 2 d tˆ
lim lim or 1.
0 0 2 d
The curvature at point P
d d dtˆ dtˆ
k lim tˆ r . .....(2.4.3)
0 s ds dt ds ds
2.4.2 Torsion :
The rate of change of the direction of binormal with respect to arc length as the point P
moves along the curve is called the torsion vector of the curve whose magnitude is denoted by
called the torsion at P.
dbˆ
Hence bˆ .....(2.4.4)
ds
Radius of Torsion : The reciprocal of the torsion is called the radius of torsion and is
1
denoted by .
Torsion at a point :
Fig. 2.5
30
Let P and Q be two neighbouring points on a curve such that PQ = s where OP = s and
the unit binormals at these points be denoted by bˆ and bˆ bˆ and be the angle between these
vectors.
Average rate of change of direction of the osculating plane over the arc PQ .
s
The torsion of the curve at P
d
lim . .....(2.4.5)
s 0 s ds
Aliter :
Fig. 2.6
From isosceles triangle QRS, we have
RS RS 2 QR sin
2
bˆ 2sin
2
bˆ sin 2
2
bˆ sin 2
Hence lim 1
0 2
Torsion at P is
dbˆ dbˆ d
ds d ds
d 1 .....(2.4.6)
ds
31
2.4.3 Skew-curvature :
The rate of change of the direction of principal normal with respect to arc length as the
point P moves along the curve is called the skew curvature vector and its magnitude is given by
dnˆ
2
2 .
.....(2.4.7)
ds
2.4.4 Self learning exercise-3
1. Give the formula for curvature.
2. Give the formula for torsion.
2.5 Serret-Frenet Formulae
Arc derivative of three unit vectors tˆ, nˆ , bˆ are known as Serret-Frenet formulae as given
below
dtˆ
1. tˆ nˆ.
ds
dnˆ
2. nˆ bˆ tˆ.
ds
dbˆ
3. bˆ nˆ. .....(2.5.1)
ds
Proof : 1. Since tˆ tˆ t 2 1 .
Different with respect to arc length s
dtˆ
2tˆ 0
ds
dtˆ
tˆ 0
ds
dtˆ
is perpendicular to tˆ . .....(2.5.2)
ds
dtˆ ˆ
But t r
ds
The equation of osculating plane at a point P on a curve is
R r , r , r 0 .....(2.5.3)
ˆ
r t in the osculating plane.
Therefore tˆ is perpendicular to b̂ and it is also perpendicular to tˆ .
Hence, tˆ is parallel to bˆ tˆ i.e. along the principal normal n̂ . Therefore tˆ is propor--
tional to n̂ , i.e.,
32
dtˆ ˆ
t nˆ
ds
tˆ nˆ .....(2.5.4)
(The direction of the principal normal is so chosen that curvature is always positive)
2. Since bˆ bˆ 1 .
Differentiating with respect to ‘s’, we have
dbˆ
bˆ 0 .....(2.5.6)
ds
dbˆ dbˆ
This implies is perpendicular to b̂ and thus lies in osculating plane.
ds ds
Also bˆ tˆ 0, on differentiating with respect to s, we get
dbˆ ˆ ˆ dtˆ
t b 0
ds ds
dbˆ ˆ ˆ
or t b nˆ 0
ds
dbˆ ˆ
or
ds
t 0 as bˆ nˆ
dbˆ
is perpendicular to tˆ . .....(2.5.7)
ds
ˆ dbˆ
Thus db is perpendicular to the vector b̂ . This implies that is parallel to bˆ tˆ i.e.
ds ds
dbˆ
is parallel to n̂ .
ds
dbˆ
Hence nˆ. .....(2.5.8)
ds
dbˆ
Taking nˆ .....(2.5.9)
ds
(In the right hand screw system, by convention, is negative).
3. Since n̂ bˆ tˆ
dnˆ dbˆ ˆ ˆ dtˆ
t b .....(2.5.10)
ds ds ds
dtˆ
Using nˆ (formula 1)
ds
dbˆ
and nˆ (formula 2),
ds
33
dnˆ
we get ˆ ˆ bˆ nˆ )
( nt
ds
bˆ tˆ ( nˆ tˆ bˆ and bˆ nˆ tˆ) .....(2.5.11)
The Serret-Frenet formulae can be put in the matrix form as follows :
0 0 tˆ tˆ
0
nˆ nˆ .....(2.5.12)
0 0 bˆ bˆ
2.5.1 Theorems on curvature and rorsion :
Theorem 1. The necessary and sufficient condition for the curve to be a straight line is
that = 0 at all points of the curve.
Proof : Necessary condition : Given the curve is a straight line.
To prove =0
We know that the vector equation of a straight line is
r a s c, .....(1)
where a and c are constant vectors and s be the measure of the length of the arc from the point
whose position vector is a .
Differentiating with respect to s, we get
.....(2)
r c r 0
dtˆ
Also by definition tˆ r 0
ds
Hence the condition is necessary.
Sufficient condition.
Given = 0.
To prove curve is a straight line.
Here =0
.....(3)
r 0
On integration, we have
r c (a constant vector) .....(4)
Again, integrating, we get
r a s c, .....(5)
where a is another constant vector..
Clearly r a s c denotes a straight line.
Hence the condition is sufficient.
34
Theorem 2. The necessary and sufficient condition that a given curve be a plane curve is
that = 0 at all point of the curve or in other words r r r 0
Proof : Necessary condition :
Given the given curve is a plane curve.
To prove = 0.
If the given curve be a plane curve then we know that tangent and normal at all points lie
in the plane of the curve. (It means that plane of the curve is the osculating plane at all points of
the curve.)
Hence the unit normal i.e. binormal b̂ is same at all points which means that b̂ is a con-
stant vector both in magnitude and direction and as such
dbˆ
bˆ 0
ds
or nˆ 0
or = 0 (by Serret-Frenet formula) .....(1)
Hence the condition is necessary.
Sufficient condition :
Given = 0.
To prove the curve is a plane curve.
If = 0
dbˆ
nˆ 0 .....(2)
ds
and hence b̂ is a constant vector i.e. the direction of binormal is same at all points of the curve.
This means the osculating plane is same at all points of the curve i.e. osculating plane
contains the curve. Hence the curve must be a plane curve.
Hence the condition is sufficient.
Theorem 3. If the tangent and the binormal at a point of a curve make angles , re-
spectively with a fixed direction, then
sin d
.
sin d
Proof : Let â denotes a unit vector in the fixed direction, then by the given condition, we
have
cos aˆ tˆ .....(1)
and cos aˆ bˆ .....(2)
d
sin aˆ tˆ
ds
aˆ nˆ aˆ nˆ .....(3)
35
d
and sin aˆ bˆ aˆ ( nˆ ) ( aˆ nˆ ) .....(2)
ds
sin d
from (1) and (2) we have .
sin d
Hence proved.
Theorem 4. The principal normals at consecutive points of a curve do not intersect un-
less = 0.
Proof : Let the position vectors of two consecutive points P and Q on a curve be
r and r d r and let the principal normals at these points be nˆ and nˆ d nˆ respectively..
Q
n + dn
n
dr
r+
r P
Fig. 2.7
In order to prove that the principal normals at these points may intersect, it is necessary
that the three vectors d r , nˆ , nˆ d nˆ are coplanar..
These vectors are coplanar if
d r , nˆ, nˆ dnˆ 0
or d rˆ, nˆ, d nˆ 0 .....(1)
or r , nˆ, n 0
or tˆ, nˆ, bˆ tˆ 0
or tˆ, nˆ, bˆ 0
or tˆ, nˆ, bˆ 0
= 0
since tˆ, nˆ, bˆ 1.
36
Theorem 5. Prove that
| r r |
(i) 3
| r |
| r r r |
(ii)
| r r |2
Proof : (i) We know that .....(1)
r tˆ
r tˆ nˆ (by Serret-Frenet formula) .....(2)
r r (tˆ n ) b̂ .....(3)
d r d r ds
Again r
sr .....(4)
dt ds dt
r
s r s r s
s r s 2 r .....(5)
from (4) and (5) s r s 2 r )
r r s r (
r r s 3 ( r r ) .....(6)
from (3) | r r | | bˆ | .....(7)
| r r |
| r r | | r r |
s 3 | r |3
because from (4) | r | s | r | s.1 s
or
r tˆ is unit vector..
Again from equation (3) r r bˆ
Differentiating again with respect to s, we have
r r r r bˆ bˆ
or 0 r r bˆ nˆ ( bˆ nˆ ) .....(8)
Now r r r r r r
r r r
nˆ bˆ nˆ [from 8]
r r r = 2 .....(9)
r r r
2
r r r
2
r r
37
r r r s6
2
r r 2 s3
r r r
or 2
r r
2.5.2 Examples :
Ex.1. For the curve x = 3u, y = 3u2, z = 2u3.
3 2
Prove that
2
1 2u2
Sol. Here r = (3u, 3u2, zu3)
r = (3, 6u, 6u2) = 3 (1, 2u, 2u2)
r = 3 (0, 2, 4u)
or r = 6 (0, 1, 2u)
r = 6 (0, 0, 2)
| r | 3 1 4u 2 4u 4 3 1 2u 2
2
r r = 18 [1, 2u, 2u ] × [0, 1, 2u]
= 18 [4u2 – 2u2, 0 – 2u, 1 – 0]
= 18 [2u2, – 2u, 1]
18 4u 2 4u 4 1 18 1 2u 2
| r
r|
1 2u 2u 2
3.6.6 0 1 2u 216
r r
r
0 0 2
r
3
r 18 1 2u 2
2 1
3 2
r 27 1 2u 2 3 1 1 2u 2
r
rr 216 2 1
2 2
2
r
r 18.18 1 2u 2 3 1 2u 2
1 3 2
2
1 2u 2 ;
1 3 2
1 2u 2
2
38
3 2
Now for a left hand system
2
1 2u 2
3 2
2
1 2u 2
Ex.2. For the curve x = a (3u – u3), y = 3au2, z = a (3u + u3)
show that the curvature and torsion are equal.
Sol. Here r xiˆ yjˆ zkˆ
or
r a 3u u iˆ 3au ˆj a 3u u kˆ
3 2 3
Differentiating with respect to s, we get
d r du
tˆ
ds
a 3 3u 3 iˆ 3aujˆ a 3 3u 3 kˆ
ds .....(1)
ˆ dr du
or t = a [(3 – 3u2), 6u, (3 + 3u2)]
ds ds
du
= 3a [(1 – u2), 2u, (1 + u2)]
ds
2
2 2 2 2 2 2 du
tˆ2 = 9a [(1 – u ) + 4u + (1 + u ) ]
ds
2
du
or 1 = 9a2 [2 (1 + u4) + 4u2]
ds
2
du
or 1= 18a2 (1 + u2)2
ds
du 1
ds 3 2 a 1 u 2
1 1 u2 2u
tˆ , , 1
2
2 1 u 1 u2
Differentiating again with respect to s, we get
1 4u
2 1 u2 , du
0
tˆ ,
2 2 2
ds
1 u 1 u 2 2
1
4u
,
2 1 u2 , 0
6a 3 3
1 u2 1 u 2
39
16u 2 4 1 u 2 2
2
k | tˆ |
2 1
2 6
36a
1 u 2
2
1 1 u
2
2
1 1
6 2 4
9a 2
1 u 9a 1 u 2
1 1
2
3a 1 u 2
= 3a (1 + u2)2
Now tˆ n̂
1 ˆ
or n̂ t
2 2
3a 1 u 2 4u ,
2 1 u2
, 0
6a 3 3
1 u
2
1 u 2
n̂
1 4u
,
2 1 u2
, 0
2 1 u2 1 u2
b̂ tˆ nˆ
1 1 u 2
,
2u 4u
, 1 ,
2 1 u2
, 0
2 2 1 u 2 1 u 2 1 u 2 1 u2
iˆ ˆj kˆ
1 1 u2 2u
1
2 2 1 u2 1 u2
4u
2 1 u2 0
1 u2 1 u2
2
1 2 1 u
0
2
, ,
4u 2 1 u
2
8u 2
2 2 1 u2 1 u2 2 2
1 u2 1 u2
40
or b̂
1 2 1 u
2
,
4u , 2
2 2 1 u2 1 u2
1 1 u 2 2u
2
, 2
, 1 .
2 1 u 1 u
Values of bˆ and tˆ are same except a change of sign.
Ex.3. Find the radii of curvature and torsion of the helix
x = a cos , y = a sin , z = a tan .
Sol. Here r = (a cos , a sin , a tan )
r = (– a sin , a cos , a tan )
r = (– a cos , – a sin , 0)
r = (a sin , – a cos , 0)
| r | a sin cos tan = a sec
2 2 2
r r = a2 (– sin , cos , tan ) × (– cos , – sin , 0)
= a2 (sin tan , – cos tan , sin2 + cos2 )
= a2 (sin tan , – cos tan , 1)
| r
r | a 2 sin 2 tan 2 cos 2 tan 2 1
a 2 tan 2 1 a 2 sec
sin cos tan
3
| r r r | a cos sin 0
sin cos 0
= a3 tan
| r r | a 2 sec 1
3 3 2 cos2
| r | a sec a
= a sec2
r
r r
2
|r r |
a3 tan 1
sin cos
a 4 sec2 a
= a sec cosec
41
Ex.4. Determine the function f () so that
x = a cos , y = a sin , z = f ();
shall be a plane curve.
Sol. Here r = (a cos , a sin , f ()),
r = (– a sin , a cos , f () ),
r = (– a cos , – a sin , f () ),
r = (a sin , – a cos , f () ),
The condition for a curve to be a plane curve is
r r
r 0
a sin a cos f ()
a cos a sin
f ( ) 0
i.e.
a sin
a cos f ()
Applying R1 + R3, we get
0 0 f ()
f ( )
a cos a sin
f ( ) 0
a sin a cos
f ( )
or a 2 f ()
f () 0
or f ()
f ( ) 0
or f ()
f ( ) A
or D 2 f ( ) f ( ) A
f () A B sin ( C )
Ex.5. Find the radii of curvature and torsion at a point of the curve
x2 + y2 = a2, x2 – y2 = a z,
Sol. The parametric equation of the curve may be given by
x = a cos , y = a sin , z = acos 2
Therefore r = (a cos , a sin , a cos 2)
r = a (– sin , cos , – 2 sin 2)
r = a (– cos , – sin , – 4 cos 2)
r = a (sin , – cos , 8 sin 2)
2
r r = a (– sin , cos , – 2 sin 2) × (– cos , – sin , – 4 cos 2)
2 3 3
or r r = a (– 4 cos 4 sin , 1)
42
sin cos 2sin 2
3
r r r a cos sin 4cos 2
sin cos 8cos 2
Applying R1 + R3, we get
0 0 6sin 2
3
r r r a cos sin 4cos 2
sin cos 8cos 2
= a3 (6 sin 2) 1 = 6 a3 sin 2
Also | r |2 = a2 (sin2 + cos2 + 4 sin2 2)
= a2 [1 + 4 (1 – cos2 2)]
= a2 [5 – 4 cos2 2]
| r r |2 = a4 (5 + 12 cos2 2)
2
1 | r
r|
Therefore 2 6
|r |
a 4 5 12 z 2 a 2
3
a2 5 4z 2
a2
3
2
5a 4 z
2 2
Hence
a 5a 12 z
2 2 2
1 r r
r 6a3 sin 2
and 2 4
| r
r| a 5 12 cos 2 2
a 5 12 z 2 a 2
6 1 z 2 a2
5a 2 12 z 2
or .
2 2
6a a z
2.6 Osculating circle and osculating sphere
2.6.1 Definitions of osculating circle and osculating sphere :
Osculating circle : If P, Q, R are three points on a curve, the circle, PQR in its limiting
position where Q, R tend to P is called the circle of curvature at P and radius of circle is the
radius of curvature and is denoted by .
or
43
The circle which has three point contact with the curve at P is called the osculating circle
at a point P on a curve.
To find the centre and radius of circle of curvature :
c1
C
c
t
P
O r
b
Fig. 2.8
Let c be the position vector of the centre C1 of the osculating circle at point P to the
curve C whose equation be r r ( s ). Let a be the radius of the circle
| r c | = a i.e. | r c |2 = a2 .....(1)
where r is the position vector of the point P. The osculating circle is the intersection of the sphere
(1) and the osculating plane at P.
The point of intersection of the curve C and sphere (1) are given by
2
F (s) r (s ) c a 2 0
The curve will have three point contact if
F (s) = 0, F (s) = 0, F (s) = 0
2
F (s) = 0 r c a 2 ,
F (s) = 0 r c r 0
or r c tˆ 0 .....(2)
F (s) = 0 r c r r r 0
or r c nˆ 1 0
1
or r c nˆ
or r c nˆ .....(3)
44
Result (2) shows that tˆ is orthogonal to ( r c ) and as such ( r c ) lies in the normal
plane at P. Also by definition ( r c ) lies in the osculating plane at P. Hence ( r c ) lies along
the intersection of these two points i.e. along the principal normal at P.
r c anˆ .....(4)
where a is a scalar.
Putting (4) in (3), we get
a nˆ nˆ a = –
and hence ( r c ) n̂
Squaring, we get ( r c ) 2 = 2
or a2 = 2
= a .....(5)
Above equation gives the radius of curvature.
Also from r c nˆ , we get
c r nˆ .....(6)
Above relation shows that the centre c lies on the principal normal at a distance from
the point P whose position vector in r .
Results (5) and (6) give the radius and position vector of the centre of the circle of curva-
ture. (It should also be noted the sign of is always positive.)
Cartesian form : Let (, , ) be the centre of the circle of curvature at a point (x, y, z) of
a given curve, and a be its radius.
The equation of the circle can be written as the intersection of
2 2 2
Sphere: a2 ,
.....(1)
Osculating plane : l3 m3 n3 0
where l3, m3, n3 being d.c.s of the binormal.
Since the circle (1) has three point contact at (x, y, z) therefore
(x – )2 + (y – )2 + (z – )2 = a2 .....(2)
(x – ) l3 + (y – ) m3 + (z – ) n3 = 0 .....(3)
On differentiating of (3) with respect to ‘s’, we get
dx
(x – ) =0
ds
or (x – ) l1 = 0 .....(4)
Again differentiating (4) with respect to ‘s’ we get
dl dx
x ds1 l1 ds 0 .
45
Using Serret-Frenet formulae
dl1 l 2
etc.
ds
in equation (4), we get (x – ) l2 = – ......(5)
Squaring and adding equation (3), (4) and (5), we get
(x – )2 = 2 .....(6)
From equation (2), and equation (6), we conclude that
a =
Now multiplying (3) by l3, (4) by l1 and (5) by l2 and adding, we have
(x – ) = – l2
Hence = x + l2,
similarly, = y + m2 and = n2 + z .....(7)
Thus, it is clear that the centre of curvature lies on the principal normal.
2.6.2 Properties of the locus of the centre of circle of curvature :
Property 1 : The tangent to the locus of the centre of curvature lies in the normal plane of
the original curve and is inclined to n̂ at an angle tan 1 .
Proof : Let r ( c ) be position vector of centre of c1 then
r1 r nˆ .....(1)
ˆ d r1 ds
t1 r nˆ nˆ
ds1 ds1
ˆ ds
or tˆ1 tˆ nˆ b tˆ ds (By Serret-Frenet formulae)
1
ˆ ds
or tˆ1 nˆ b ds 1 .....(2)
1
Above relation shows that tˆ1 lies in the plane containing bˆ and nˆ i.e. normal plane c.
If be the angle made by tˆ1 with nˆ , then
tˆ1 cos nˆ sin bˆ .....(3)
Hence by comparing (2) and (3)
ds ds
cos and sin ;
ds1 ds1
1
tan tan .
46
Property 2 : If the curvature of a curve c is constant, then the curvature 1 of c1 is also
constant and its torsion 1 varies inversely as of the curve c.
Proof : Now if is constant i.e. 1/ const., then = 0 and hence from (2)
ds
tˆ1 bˆ .....(4)
ds1
2
ds
squaring both sides 1 bˆ
ds1
bˆ bˆ 1
ds
or 1
ds1
Putting in (4) tˆ1 bˆ .
Differentiating with respect to s1,
ds
tˆ1 bˆ
ds1
1
or 1nˆ1 nˆ or 1nˆ1 nˆ .....(5)
This relation show that n1 is parallel to n and if we choose the direction of n1 opposite to
that of n i.e. n1 = – n then from (5) we get 1 = = constant as is given to be constant.
Thus the curvature of c1 is also constant.
Again, bˆ1 tˆ1 nˆ1
or
bˆ1 bˆ nˆ bˆ nˆ n̂ bˆ tˆ
bˆ1 tˆ.
Differentiating with respect to s1, we get
ds
bˆ1 tˆ
ds1
1
or 1 n1 n
2 1
or 1 n n , n1 n
2 constant
1
1 = constant.
Above shows that torsion of c1 is inversely proportional to torsion of c.
47
Property 3 : Principal normal to c is normal to c 1 at the points where curvature is
stationary.
Proof : We know that the position vector c of the center of curvature is given by
c r nˆ
Let the locus of c r1 be given by
r1 r nˆ
Differentiating r1 with respect to s1, we have
ds
tˆ1 tˆ nˆ nˆ
ds1
tˆ nˆ bˆ tˆ dsds
1
ds
tˆ nˆ bˆ tˆ
ds1
ds
nˆ bˆ
ds1
dsds
tˆ1 nˆ nˆ nˆ bˆ nˆ
1
ds
ds1 bˆ nˆ 0
ds
tˆ1 nˆ
ds1
If = const = 0.
Therefore tˆ1 nˆ 0.
which shows that principal normal is normal to the locus of center of curvature at those points
where the curvature is stationary.
2.6.3 Osculating sphere or sphere of curvature :
A sphere which has a four point contact with the curve at a point P is called the osculating
sphere at P.
Let c be the centre and R the radius of sphere so that its equation is
r c 2 R 2 .....(1)
where r is the position vector of point P on the curve.
The points of intersection of the sphere with the curve r r s are given by
2
F s r s c R 2 0 .....(2)
48
The sphere will have a four point contact with the curve if
F 0, F 0, F 0, F 0,
F = 0, r c 2 R 2
F = 0, r c r 0 or r c tˆ 0 .....(3)
F = 0, r c tˆ r tˆ 0
or r c nˆ tˆ tˆ 0
or r c nˆ 1 0
1
or r c nˆ .....(4)
F = 0, r c nˆ r nˆ
or r c bˆ tˆ tˆ nˆ
or r c bˆ r c tˆ tˆ nˆ
or r c bˆ 0 0 [by (3)]
or r c bˆ
.....(5)
Result (3) shows that tˆ is orthogonal to r c and as such it lies in the normal plane at
P which contains n̂ and b̂ and hence it can be expressed as a linear combination of nˆ and bˆ.
r c nˆ bˆ
r c nˆ or [by (4)]
r c bˆ or [by (5)]
r c nˆ bˆ
or c r nˆ bˆ .....(6)
Above relation gives us the position vector of the center c of the osculating sphere.
Again r c 2 R 2
2
or n̂ bˆ R2
or 2 2 2 R 2 .....(7)
Above relation gives us the radius R of the sphere of curvature.
49
Properties of the locus of the counter of sphere of curvature.
(i) tˆ1, nˆ1, bˆ1 of c1 are parallel respectively to bˆ, nˆ , tˆ of c.
Proof : Let c be the original curve, c1 the locus of the centre of spherical curvature. Let
the suffix unity denote quantities belonging to c1.
The position vector c r1 say of centre of spherical curvature is given by
r1 r nˆ bˆ .....(1)
differentiating with respect to s1
dr1 ds
ds1
r nˆ nˆ bˆ bˆ bˆ
ds1
ds
or
tˆ1 tˆ nˆ bˆ tˆ bˆ nˆ
ds1
ds
or tˆ1 tˆ bˆ t nˆ bˆ nˆ
ds
1
ds
or tˆ1 bˆ .....(2)
ds1
which shows that t1 (tangent to c1) is || to b.
Squaring (2), we get
2 2
ds1
ds
ds1
or .....(3)
ds
from (2) & (3) tˆ1 bˆ .....(4)
Differentiating equation (4) with respect to ‘s1’.
dtˆ1 ds
bˆ
ds1 ds1
ds
or 1n̂1 nˆ .....(5)
ds1
which shows that n̂1 is parallel to nˆ.
2
2 ds 2
Squaring (3), we get
ds1
ds 1
or .....(6)
ds1
50
from (5) and (6), we get nˆ1 nˆ .....(7)
Equation (7) shows that directions of n1 and n are opposite to each other.
Taking cross product of (4) and (7)
tˆ1 nˆ1 bˆ nˆ or bˆ1 tˆ .....(8)
which shows that b̂1 is parallel to c .
(ii) The product of the torsion of c1 at corresponding points is equal to the product of cur-
vatures at these points.
Proof : Differentiating equation (8) with respect to s1
ds
bˆ1 tˆ
ds1
ds
or 1nˆ1 nˆ .....(9)
ds1
but from equation (7) nˆ1 nˆ , hence from equation (9), we have
ds
1 .....(10)
ds1
1
or 1 [using (6)]
or 1 1 .....(11)
1 1
(iii) If the curvature of c is constant then curvature 1 of c1 is also constant.
Proof : The curvature of c is constant
i.e. const., 0, 0.
ds
equation (3) reduces to .
ds1
Hence from (6), we have
or 1 .....(12)
1
which shows that the curvature 1 of c1 is also constant.
2.6.4 Examples :
Ex.6. If a curve lies on a sphere show that and are related by
d
0
ds
show that a necessary and sufficient condition that a curve lies on a sphere is that
51
d
0
ds
at every point on the curve.
Sol. Necessary condition :
Let the curve lie on a sphere then we have to prove the given condition. Now the sphere
will be osculating sphere for every point. The radius R of the osculating sphere is given by
R2 = 2 + 2 2 .....(1)
Differentiating with respect to ‘s’, we get
0 = + 2 + 2
Dividing by , we get
0
d
or 0
ds
d
or 0
ds
d
Sufficient condition : If 0 to show that the curve lies on a sphere
ds
2 22 a 2 [by (1)]
showing that the radius of osculating sphere is independent of the point on the curve.
Again the centre of spherical curvature is given by
C r nˆ bˆ
dc
ds
tˆ nˆ bˆ tˆ bˆ bˆ nˆ
b̂
d
But or is zero.
ds
dc
0
ds
or c = constant vector
i.e., the centre of osculating sphere is independent of the point on the curve.
Ex.7. Prove that the curve given by
x = a sin u, y = 0, z = a cos u
lies on a sphere.
52
Sol. Here r = a (sin u, 0, cos u) .....(1)
du
tˆ r = a (cos u, 0, – sin u)
ds
2
du du 1
squaring 1 a2
ds ds a
Hence tˆ = (cos u, 0, – sin u) .....(2)
du
tˆ n sin u , 0, cos u
ds
1
or nˆ sin u, 0, cos u
a
1 1
squaring 2 2
a a
= a = constant .....(3)
Hence nˆ sin u ,0, cos u .....(4)
bˆ tˆ nˆ 0,1, 0
bˆ nˆ 0, 0, 0
= 0 (as n̂ 0) .....(5)
We know that curve will lie on a sphere if
d
0
ds
Here =a
= 0 and also =0.
d
Therefore, the relation 0 is clearly satisfied. Hence the given curve lies
ds
on a sphere.
1 1 2
Ex.8. Prove that x2 y2 z 2
2 2 4
where dashes denote differentiation with respect to ‘s’.
Sol. Here r xiˆ yjˆ zkˆ
r xiˆ yˆj z kˆ
r x iˆ y ˆj z kˆ
r x iˆ y ˆj z kˆ
r 2 x2 y2 z2 .....(1)
53
nˆ
Also r tˆ, r tˆ nˆ
nˆ ˆ ˆ
r 2 nˆ b t nˆ
2
1 ˆ 1 ˆ
t b 2 nˆ
2
1 1 2 .
r 2 ......(2)
4 22 4
Hence from (1) and (2)
2 2 21 1 2
x y z 2 2 .
4
2.7 Answers to self-learning exercises
Self-learning exercise-1
1. osculating plane.
2. Binormal
ˆ r r r
3. b , nˆ .
r r r r
Self-learning exercise-2
1. If P, Q, R be three points on a curve, the limiting position of the plane PQR when Q and R
tend to P, is called the osculating plane at P.
2. The plane through P and perpendicular to the tangent line at P is called the normal plane
at P of the curve.
3. The plane through P and containing tangent and binormal is called rectifying plane.
4. R r nˆ
5. R r bˆ
2.8 Exercises
1. Show that the tangent and binormal at any point of the curve
x t 3 1, y 3(t 2 1), z 2 t 1
x y z
make the same angle with the line and that the three directions are coplanar..
1 0 1
2. Establish the Serret-Frenet formulae at a point of a space curve.
54
3. Find the radii of curvature and torsion of a helix
x = a cos , y = a sin , z = a tan .
2 a
Ans. a sec , sin cos
4. For the curve x = a (3t – t3), y = 3at2, z = a (3t + t3)
Show that = = 3a (1 + t2)2.
5. Find the osculating plane, curvature and torsion at any point ‘’ of the curve
x = a cos 2, y = a sin 2, z = 2a sin
[Ans. (sin + sin 2 cos ) x – 2 cos3 y + 2z = 3a sin ,
3/ 2
a
5sec 3cos ,
2a 1 cos 2
3 1/ 2
5 3cos 2
6. For the curve x = 2abt, y = a2 log t, z = b2 t2.
Show that = = (a2 + 2b2 t2) / 2abt.
7. Find the equation of the osculating sphere and osculating circle at (1, 2, 3) on the curve
x = 2t + 1, y = 3t2 + 2, z = 4t3 + 3.
[Ans. 3 (x2 + y2 + z2) – 6x – 16y – 18z + 50 = 0,
3 (x2 + y2 + z2) – 6x – 16y – 18z + 50 = 0, z = 3]
8. Show that the radius of spherical curvature of a circular helix
x = a cos , y = a sin , z = a cot a
is equal to circular curvature.
9. If the radius of spherical curvature is constant show that the curve either lies on a sphere
or has a constant curvature R2 = 2 + ()2 where R is constant.
10. Find the equation of the osculating sphere at origin of the curve
x = a1t3 + 3b1t2 + 3c1t, y = a2t3 + 3b2t2 + 3c2t, z = a3t3 + 3b3t2 + 3c3t.
x2 y 2 z 2 2x 2 y 2z
9 b1c1 b2c2 b3c3 a1 a2 a3
Ans. 0
3 c12 c2 2 c32 2b1 2b2 2b3
0 c1 c2 c3
55
UNIT 3 : Existence and Uniqueness Theorems, Bertrand
Curves, Involute, Evolutes, Conoids, Inflexional
Tangents, Singular Points, Indicatrix
Structure of the Unit
3.0 Objectives
3.1 Existence and uniqueness theorems
3.1.1 Existence theorem
3.1.2 Uniqueness theorem
3.2 Bertrand curves
3.2.1 Definitions
3.2.2 Properties of Bertrand curves
3.2.3 Theorem based on Bertrand curves
3.2.4 Self-learning exercise-1
3.3 Involute
3.3.1 Definition
3.3.2 General equation of the involute of a given space curve
3.3.3 To find curvature of the involute
3.3.4 To find torsion of the involute
3.3.5 Examples
3.3.6 Self-learning exercise-2
3.4 Evolute
3.4.1 Definition
3.4.2 General equation of the evolute of a given space curve
3.4.3 To find curvature of the evolute
3.4.4 To find torsion of the evolute
3.4.5 Examples
3.4.6 Self-learning exercise-3
3.5 Conoids
3.5.1 Definition
3.5.2 Equation of a conoid
3.5.3 Examples
3.5.4 Self-learning exercise-4
56
3.6 Inflexional tangents
3.6.1 Definition
3.6.2 Equation of the Inflexional tangents at a point of a given surface
3.6.3 Examples
3.7 Singular points
3.7.1 Definition
3.7.2 Singular tangent planes
3.7.3 Examples
3.7.4 Self-learning exercise-5
3.8 Indicatrix
3.8.1 Definition
3.8.2 Examples
3.9 Answers to self-learning exercise
3.10 Exercises
3.0 Objectives
This unit provides a general overview of :
• Existence and uniqueness theorems
• Bertrand curves
• Involute
• Evolute
• Conoids
• Inflexional tangents
• Singular points
• Indicatrix
3.1 Existence and uniqueness theorems
Existence and uniqueness theorem for space curves is also called fundamental theorem on space
curves
3.1.1 Existence theorem :
If k (s) and (s) are continuous functions of a real variable s (s 0) then there exists a space
curve for which k is the curvature, is the torsion and s is the arc-lengths measured from some suitable
base point.
57
Proof : The proof of this theorem depends on the existence theorem of the solution of differen-
tial equations which states that the linear differential equations
dx dy dz
ky , z kx, y .....(3.1.1)
ds ds ds
where k and are continuous functions of s in the interval 0 s a.
Equation (1) admits a unique set of solutions for a given set of values of x, y, z at s = 0.
In particular, there exists a unique set (x1, y1, z1), (x2, y2, z2) and (x3, y3, z3) which have values
(1, 0, 0), (0, 1, 0), (0, 0, 1) at s = 0, respectively.
Now,
dx1 dy dz
d 2
ds
x1 y12 z12 2 x
1
ds
y1 z1 1
ds ds
2 x1 ky1 y1 z1 kx1 z1 y1 0 [using (1)]
Hence,
x12 y12 z12 = constant = C1 (say)
Since at s = 0, x1 (0) = 1, y1 (0) = 0, z1 (0) = 0, therefore C1 = 1.
Thus, we get x12 y12 z12 1, s. .....(3.1.2)
x22 y22 z22 1,
Similarly, we get .....(3.1.3)
x32 y32 z32 1, for s
d
Further, x1 x2 y1 y2 z1z2
ds
dx dx dy dy dz dz
x1 2 x2 1 y1 2 y2 1 z1 2 z2 1
ds ds ds ds ds ds
= x1(ky2) + x2(ky1) + y1(z2 – kx2) + y2(z1 – kx1) + z1(– y2) + z2(– y1) = 0.
Hence on integration,
x1 x2 + y1 y2 + z1 z2 = const. = d1 (say)
The value of the constant d1, determined by the initial conditions and we get d1 = 0.
Thus we get x1 x2 + y1 y2 + z1 z2 = 0, s. .....(3.1.4)
x2 x3 y2 y3 z 2 z3 0
Similarly, we get for s .....(3.1.5)
x3 x1 x3 x1 x3 x1 0
Hence, we have six relations given by (3.1.2) to (3.1.5) in the elements of three sets namely
(x1, y1, z1), (x2, y2, z2) and (x3, y3, z3) defined for each value of s.
s
If r t ds , then r = r (s)
0
58
is the position vector of a point on the curve with (k, , s) curvature, torsion and arc-length
respectively and tˆ, nˆ, bˆ as unit tangent vector, unit principal normal vector and unit binormal
respectively
Hence the existence of curve is proved.
3.1.2 Uniqueness theorem :
A curve is uniquely determined, except as to position in space, when its curvature and torsion
are given functions of its arc-length.
Proof : If possible let there be two curves c1 and c having equal curvature k and equal torsion
for the same value of s. Let the suffix unity be used for quantities belonging to c1. Now, if c1 is moved
(without deformation) so that the two points on c and c1 corresponding to the same value of ‘s’
coincide. We have
d ˆ ˆ
ds
t t1 tˆ k1nˆ1 knˆ tˆ1
d ˆ ˆ
or
ds
t t1 tˆ knˆ1 knˆ tˆ1 [ k1 = k given] .....(3.1.6)
d
ds
nˆ nˆ1 nˆ bˆ1 ktˆ1 bˆ ktˆ nˆ1 .....(3.1.7)
d ˆ ˆ
ds
b b1 bˆ nˆ1 knˆ bˆ1 .....(3.1.8)
Adding equations (1), (2) and (3), we get
d ˆ ˆ
ds
t t1 nˆ nˆ1 bˆ bˆ1 0 .....(3.1.9)
which on integrating gives tˆ tˆ1 nˆ nˆ1 bˆ bˆ1 constant . .....(3.1.10)
If c1 is moved in such a manner that at s = 0 the two triads tˆ, nˆ, bˆ and tˆ1 , nˆ1, bˆ1 coincide.
Then at the point tˆ tˆ1 , nˆ nˆ1 , bˆ bˆ1 and then the value of constant in equation (3.1.10) becomes 3.
Thus, tˆ tˆ1 nˆ nˆ1 bˆ bˆ1 3 .....(3.1.11)
But the sum of three cosines is equal to 3 if each angle is zero or in an integral multiple of 2.
Thus for each pair of corresponding points tˆ tˆ1, nˆ nˆ1, bˆ bˆ1 .
Also, tˆ tˆ1 gives r r1
d
i.e.
ds
r r1 0 i.e., r r1 a (constant vector)
But when s = 0, r r1 0 or r r1 at all corresponding points and hence the two curves
coincide or the two curves are congruent.
Hence the uniqueness theorem is proved.
59
3.2 Bertrand curves
3.2.1 Definitions :
Two curves c and c1 are said to be Bertrand curves or conjugate if the principal normals to c
are also principal normals to c1.
3.2.2 Properties of Bertrand curves :
Property I. The distance between corresponding points of the two curve is constant.
C1
n1
t1
C
P1(r1)
n t
b1 P (r)
b
Fig 3.1
Proof : Let P and P1 be the corresponding points on the Bertrand curves c and c1. nˆ and nˆ1
be principal normals at P and P1 on curves c and c1.
Let the corresponding quantities for the curve c1 be denoted by the suffix unity.
Let PP1 = . Then the position vector r1 related to r as,
r1 r nˆ .....(3.2.1)
where is function of ‘s’.
Differentiating (1) with respect to ‘s’, we get
dr1 ds1
tˆ nˆ nˆ
ds1 ds
ds
or
tˆ1 1 tˆ bˆ k tˆ nˆ
ds
1 k tˆ nˆ bˆ .....(3.2.2)
By definition nˆ1 nˆ .....(3.2.3)
Taking the dot product of (3.2.2) and (3.2.3) and noting that tˆ nˆ bˆ nˆ 0, nˆ nˆ 1
0 =
which on integration gives
= constant.
i.e. PP1 = constant.
60
Property II. The tangents at the corresponding points of the associate Bertrand curves
are inclined at a constant angle.
Proof : If be the angle between tˆ and tˆ1 then we have to prove that is constant.
t1
t
O b1
n = n1 b
Fig 3.2
But tˆ tˆ1 cos
d ˆ ˆ ds
and
ds
t t1 tˆ tˆ1 tˆ tˆ1 1
ds
ds1
K nˆ tˆ1 tˆ K1 nˆ1
ds
ds1 ˆ
K nˆ1 tˆ1 K1 t nˆ. nˆ nˆ1
ds
=0 .....(3.2.4)
nˆ1 tˆ1 0, tˆ nˆ 0
Integrating we get tˆ tˆ1 constant cos , say .....(3.2.5)
= constant. .....(3.2.6)
Further, as the principal normals of the two curves coincide, it follows from the above that the
binormals of the two curves are also inclined at the same constant angle.
Property III. The curvature and torsion of either associate Bertrand curves are connected
by a linear relation.
Proof : From property I, we have = 0. .....(3.2.7)
Therefore,
ds
tˆ1 1 1 k tˆ bˆ .....(3.2.8)
ds
61
This implies that tˆ1, tˆ and bˆ are coplanar. On taking dot product with bˆ1 , we get
0 1 k tˆ bˆ1 bˆ bˆ1
But, tˆ bˆ1 cos 90 sin and bˆ . bˆ1 cos .....(3.2.9)
Hence, (1 – k) sin + cos = 0
1
or k tan for the curve c. .....(3.2.10)
This shows that and k are linearly related.
Again from (3.2.1), r1 r nˆ
r r1 nˆ
Above shows that the point P r is at a distance – along the normal at P1 r1 and tˆ is
inclined at an angle (– ) with tˆ1. Hence (3.2.10) takes the form
1 1
1 K1 tan or c1 K1 tan .....(3.2.11)
which gives the linear relationship between 1 and K1.
3.2.3 Theorem based on Bertrand curves :
Theorem. The torsion of the two Bertrand curves have the same sign and their product is
constant.
Proof : From property (III), we have
ds
tˆ1 1 1 K tˆ bˆ .....(3.2.12)
ds
and tˆ1 tˆ cos bˆ sin .....(3.2.13)
On comparing the coefficients in (3.2.12) and (3.2.13), we get
( ds1 / ds ) 1 K
.....(3.2.14)
1 cos sin
ds1 ds
then, we have sin , 1 K cos 1 .....(3.2.15)
ds ds
Replacing by – , by – and s by – s1 and s1 by s, we have
ds ds
– 1 sin ds , 1 K1 cos ds .....(3.2.16)
1 1
On multiplying, we have 1 = 1/2 sin2 K = constant as both and are constants and
(1 – K) (1 + K1) = cos2 . .....(3.2.17)
62
3.4.4 Self-learning exercise-1.
1. How many properties of Bertrand curves are ?
2. In which relation the torsion and curvature of Bertrand curves are connected ?
3. The angle between tangents of Bertrand curve are ......... .
3.3 Involute
3.3.1 Definition :
If tangents to a give space curve c are normals to another curve c1, then the curve c1 is called
involute of the curve c.
3.3.2 General equation of the involute of a given space curve :
Let c1 be an involute of c and let equation of c be r r s . Let the quantities belonging to c1
be distinguished by the suffix unity.
Any point P1 on c1 is given by
OP1 = OP + PP1 r1 r tˆ .....(3.3.1)
Where is to be determined. Differentiating equation (3.3.1)
ds
tˆ1 tˆ tˆ K nˆ .....(3.3.2)
ds1
Involute
Evolute P t
P1
C
r1
r
t
O C1
n
Fig 3.3
But tˆ is perpendicular to t1 for an involute, hence taking dot product of both sides of equation
(3.3.2) with tˆ and using tˆ tˆ1 0, we get
ds
1 0 i.e. 1 + = 0 i.e. ds + d = 0
ds1
hence on integration, we get s + = c or = c – s .....(3.3.3)
where c is constant of integration
r1 r c s tˆ .....(3.3.4)
This is the required equation of involute c1 of the curve c.
63
3.3.3 Curvature of the involute :
From (3.3.2) and (3.3.3), we have
ds
tˆ1 1 k nˆ where = c – s. .....(3.3.5)
ds
This shows that tˆ1 is parallel to nˆ. Taking the direction of tˆ1 same as that of nˆ, we get
ds1
K , tˆ1 nˆ .....(3.3.6)
ds
Differentiating equation (3.3.6), we have
dtˆ1 dnˆ ds 1
ds1 ds ds1
K1nˆ1 bˆ Ktˆ
K
.....(3.3.7)
1/ 2
On squaring K12
2
K
K1
2
2
K2 where = c – s. .....(3.3.8)
2 K 2 K
Hence, equation (3.3.8) determine curvature of the involute.
3.3.4 Torsion of the involute :
From (3.3.7), we have
bˆ Ktˆ
nˆ1 .....(3.3.9)
K K1
Therefore using (3.3.6) and (3.3.8), we have
tˆ Kbˆ tˆ Kbˆ
bˆ1 tˆ1 nˆ1 nˆ nˆ1 or bˆ 12
.....(3.3.10)
KK1
2 K 2
Differentiating (3.3.10) with respect to ‘s’ and using Serret-Frenet formulae, we find
dbˆ1 ds1 1 dtˆ dbˆ ˆ
tˆ Kbˆ
32
12 K tˆ K b KK ,
ds1 ds ds ds
2 K 2 2
K 2
1 tˆ Kbˆ
or 1n̂1 K
2 1/ 2
Knˆ K nˆ tˆ K bˆ K 32 KK ,
2
K 2 2
or 1n̂1 K
2
K 2 tˆ K bˆ tˆ K bˆ KK ,
32
2
K 2
K K K tˆ bˆ
or 1Knˆ1 32
.....(3.3.11)
2
K 2
64
Squaring both sides, we get
2
12 2 K 2
K K
, 2
2
K2
1
K K where = c – s. .....(3.3.12)
K 2 K 2
Hence equation (3.3.12) determines the torsion of the involute.
3.3.5 Example :
Ex.1. Find the involute of a circular helix given by.
x = a cos , y = a sin , z = a tan .
Sol. Here r = (a cos , a sin , a tan ).
Diff. with respect to
ds
r a sin , a cos , a tan
d
ds
tˆ a sin , a cos , a tan .....(1)
d
On squaring, we have
2
ds
2 2
2 2 2 2 2
a sin cos a tan a sec
d
ds
a sec
d
or s a sec d a sec
0
ds
Putting for in (1), we get
d
tˆ a sec a sin , cos , tan
tˆ cos sin , cos , tan
Now the equation of involute is,
r1 r c s tˆ
or r1 a cos , a sin , a tan c a sec cos sin , cos , tan
or x = a cos – (c – a sec ) cos sin
y = a cos – (c – a sec ) cos cos
z = a tan + (c – a sec ) sin
65
Ex.2. Show that the distance between corresponding points of two involutes is constant.
Sol. The equation of the involute is,
r1 r c s tˆ
where ‘c’ is arbitrary constant.
Let c = c1 and c = c2 be the values of constant for the two point P and Q on the involute whose
position vectors are r1 and r2 say, so that
r1 r c1 s tˆ , r2 r c2 s tˆ
PQ PQ r2 r1 c1 c2 t c1 c2 constant.
3.3.6 Self-learning exercise-2.
1. Write the formulae of curvature of an involute
2. Write the formulae of torsion of an involute.
3. Write the equation of involute.
3.4 Evolute
3.4.1 Definition :
If the tangents to a curve c are normals to another curve c1, then c is called an evolute of c1.
3.4.2 General equation of the evolute of a given space curve :
In other words, we are given the equation of the involute c and are required to find its
evolute c1.
Let r r s be a given curve c.
Let r1 be the position vector of any point Q on c1 and that of the corresponding point P on c
be r.
Now, since the tangent to c1 are normals to c, the point Q must lie in the normal plane to the
curve c at P.
t1 Q
P
n evo
lute
C1
r
r1
n1
C
O
Fig 3.4
66
Thus, r1 r nˆ bˆ .....(3.4.1)
where and are to be found out.
Differentiating with respect to ‘s’, as PQ r1 r nˆ bˆ
ds
tˆ1 tˆ bˆ Ktˆ nˆ bˆ nˆ
ds1
ds
1 K tˆ nˆ tˆ bˆ .....(3.4.2)
ds
1
As tˆ1 lies in the normal plane of c at P, therefore it must be parallel to nˆ bˆ, hence compar--
ing this with the relation in (3.4.2), we obtain
1–K=0 i.e. =
d 1
and i.e., tan
2 2 ds
d
or tan 1 .....(3.4.3)
ds
Integrating equation (3.4.3), we get
a ds tan 1 [as and a is constant]
cot 1
or cot ds a
Substituting values of and in equation (3.4.1), we get
r1 r nˆ cot ds a bˆ. .....(3.4.4)
This is the required equation of evolute c1 of the curve c. As, we give different value of a, we
get infinite system of evolutes of the given curve, one evolute arising from each choice of a.
1
If we assume ds and a c
2
.....(3.4.5)
d
so that
ds
Hence, equation (3.4.5) of the evolute becomes,
r1 r nˆ tan c bˆ . .....(3.4.6)
3.4.3 Curvature of the evolute :
Differentiating equation (3.4.6) with respect to ‘s1’
67
ds
t1 r nˆ tan c bˆ
ds
1
t nˆ bˆ Ktˆ tan c bˆ
2 ds ds
+ nˆ tan c sec c b
ds1 ds1
ds
tan c nˆ tan c b
ds1
K sin c K cos c ds K
2 2 nˆ cos c bˆ sin c K 2
K cos c ds1
Hence the unit tangent to the evolute is,
t1 nˆ cos c bˆ sin c .....(3.4.7)
ds1 K sin c K cos c
where
ds
K 2 cos 2 c
Differentiating equation (3.4.7) w.r.t. ‘s1’
ds
K1nˆ1 bˆ Ktˆ cos c nˆ sin c nˆ sin c bˆ cos c
ds
1
ds
K cos c tˆ .....(3.4.8)
ds1
This equation shows that the principal normal to the evolute is parallel to tˆ.
We may choose the direction such that
n̂1 tˆ .....(3.4.9)
ds
Therefore K1 K cos c
ds1
K 3 cos3 c
or K1 .....(3.4.10)
K sin c K cos c
Hence equation (3.4.10) determines the curvature of evolute.
3.4.4 Torsion of the evolute :
We know bˆ1 tˆ1 nˆ1 nˆ cos c bˆ sin c tˆ
or bˆ1 cos c bˆ sin c nˆ .....(3.4.11)
68
Differentiating this relation with respect to ‘s’
ds
1nˆ1
ds1
sin c bˆ cos c nˆ cos c nˆ sin c bˆ Ktˆ
ds
or 1nˆ1 K sin c tˆ
ds1
K 3 sin c cos2 c
or 1 n1 t from (3.4.9) .....(3.4.12)
Kt sin c K cos c
Hence equation (3.4.12) determines the torsion of the evolute.
Now the relation between curvature and torsion is given by
1
= – tan ( + c) .....(3.4.13)
K1
3.4.5 Example :
Ex. 1. Find the evolutes of the circular helix
x = a cos , y = a sin , z = a tan
Sol. Here r a cos , a sin , a tan .....(1)
Equation of evolute of space curve r r s is given by,,
r1 r nˆ tan c bˆ .....(2)
where ds .....(3)
Differentiating, (1) gives
d
tˆ r a sin , cos , tan .....(4)
ds
Taking module of both sides
d d
1 a 2 sec 2 a sec .....(5)
ds ds
Using (5), (4) gives
tˆ cos sin , cos , tan
cos 2
tˆ Knˆ cos , sin , 0 [using (5)]
a
cos 2
which gives K i.e. a sec 2
a
and nˆ cos , sin , 0
bˆ tˆ nˆ cos sin tan , cos tan ,1
69
cos 2
bˆ t nˆ tan cos , sin , 0
a
cos
which gives, sin
a
1 1
sin cos ds s sin cos [Using (5)]
a a
= ( sin ).
Hence the equation of evolute is given by
x a cos tan 2 a tan sec sin tan c sin
y a sin tan 2 a tan sec cos tan c sin .....(6)
z a tan a tan sec tan c sin
Equation (6) gives required evolutes.
Ex. 2. Prove that the locus of the centre of curvature is an evolute and is given only when
the curve is plane.
Sol. The equation of evolute of space curve r r s is given by
r1 r nˆ tan c bˆ .....(1)
where ‘c’ is arbitrary constant.
The locus of the centre of curvature is given by the equation as
r1 r nˆ .....(2)
If equations (1) and (2) represent the same curves, then on comparison, we get
tan c 0 c n ; n is an integer .....(3)
On differentiating with respect to ‘s’, we get
d
0 0 .....(4)
ds
Hence the curve should be a plane curve.
3.4.6 Self-learning exercise-3.
1. Write down the equation of the evolute.
2. Write the formulae of the curvature of the evolute.
3. Write down the formulae of the torsion of the evolute.
4. Given relation between curvature and torsion of the evolute.
3.5 Conoids
3.5.1 Definition :
The surfaces generated by a moving straight line under certain conditions are called ruled
surfaces.
70
Cone and cylinder are examples of the ruled surface.
A conoid, is defined as the locus of a line which always intersects a fixed line (a given line) and a
given curve and is parallel to a given plane.
Right conoid : If the given line is at right angles to the given plane, the locus is a right conoid.
3.5.2 Equation of a conoid :
Let the coordinate axes be so chosen as fixed line be z-axis and xy-plane be the given plane.
In such a case, the generators of the conoid will project the given curve on the plane x = 1 in a
curve, whose equation be taken as,
x = 1, z = f (y) .....(3.5.1)
Let P (1, y, z) be any point on this curve, therefore
z1 = f (y1) .....(3.5.2)
Let Q (0, 0, z1) be the corresponding point on the fixed line. The generator of the conoid through
P is the line joining P and Q, whose equation is,
x 0 y 0 z z1
.....(3.5.3)
1 y1 0
Eliminating y1 and z1 between the eqn (3.5.2) and (3.5.3), we obtain the required equation of
the conoid i.e.,
z = f (y/x) .....(3.5.4)
3.5.3 Examples :
Ex.1. Find the equation to the conoid generated by lines parallel to the plane XOY, are
drawn to intersect OZ and the curve
x2 y2 2z
x2 y 2 r 2 , .
a 2 b2 c
Sol. The generators of the conoid are parallel to the plane XOY and intersect OZ, therefore their
equations may be written as
Z= and x = y .....(1)
say (x, y, ) is a point lying on the curve through which the generator of the conoid passes, then the
other point will be (0, 0, ).
Therefore the equation to the generators are,
x y z
.....(2)
x1 y1 0
x12 y12 2
Also, x12 y12 r2, 2
2
.....(3)
a b c
2 1
or cr 2 2 2 2 1 2
x1 y1 .....(4)
a b
71
Eliminating unknown constants and between (1) and (4), the required locus is
2 x2 y 2
cr 2 2 2z x 2 y 2 .
a b
Ex.2. Find the equation to the right conoid generated by lines which meet OZ, are paral-
lel to the plane XOY and intersect the circle
x = a, y2 + z2 = r2.
Sol. The generators of the conoid will project the given curve on the plane x = a is the circle
y2 + z2 = r2.
Let (x, y, z) or (a, y, z) is point on the circle through which the generators of the conoid pass.
Since the lines meet oz, therefore the other point will be (0, 0, z).
Therefore, the equations to the generating lines are,
x y z z1
.....(1)
a y1 0
Also, y12 + z12 = r2
Eliminating y1, z1, between (1) and (2), we have
x2 (z2 – r2) + a2 y2 = 0
which is the required equation of the right conoid.
3.5.6 Self-learning exercise-4.
1. Define right conoid.
2. Write equation of conoid.
3. How can we obtain the equation of conoid ?
3.6 Inflexional tangents
3.6.1 Definition :
Let the equation to the line through a point (x1, y1, z1) on a given surface be
x x1 y y1 z z1
(u ) .....(3.6.1)
l m n
The inflexional tangents are the lines which have three point contact inside the given surface where
u = 0.
Another definition : At a point P where r 0, the tangent line is called inflexional and the
point P is called point of inflexion.
3.6.2 The equation of the inflexional tangents at a point on given surface :
Let = f (,) be the equation to the surface, the point of intersection of the line
x y z
.....(3.6.2)
l m n
72
are given by z + = f (x + l , y + m )
2
2
f x, y l m f l m f ...
x y 2! x y
2 2
or f x, y pl qm
2!
rl 2 slm m 2 k +...
z z 2 z 2 z 2 z
where p , q ,r , s , t
x y x 2 xy y 2
Therefore the equation of the tangent plane at (x, y, z) is
( – x) p + ( – y)q = – z .....(3.6.3)
and the inflexional tangents are the lines of intersection of the tangent plane and the pair of planes given
by
r ( – x)2 + 2s ( – x) ( – y) + t ( – y)2 = 0. .....(3.6.4)
3.6.3 Examples :
Ex.1. Find the inflexional tangent at (x1, y1, z1) on the surface y2z = 4ax
Sol. The equation to a line through (x1, y1, z1) is
x x1 y y1 z z1
u say .....(1)
l m n
The inflexional tangents are the lines which have three point contact inside the surface where
u = 0.
From equation (1) substituting the values of x, y, z in the equation of surface y2z = 4ax, we get
F (u) = (mu + y1)2 (nu + z1) – 4a (lu + x1) = 0 .....(2)
For three point contact, we have
F (u) = (mu + y1) 2m (nu + z1) + (mu + y1)2 n – 4al = 0 .....(3)
F (u) = 2m2 (nu + z1) + 2mn (mu + y1) + 2mn (mu + y1) = 0 .....(4)
At u = 0, the above equations (2), (3) and (4) are reduced to
y12 z1 4ax1 0 ....(5)
2my1z1 ny12 4al 0 .....(6)
2m2z1 + 2mny1 + 2mny1 = 0
or mz1 + 2ny1 = 0 .....(7)
Using (7), (6) become
mz1 2 3my1 z1
2my1 z1 y1 4al 0 or l
2 y1 8a
Substituting value of l and n in (1), we get
x x1 y y1 z z1
3my1z1 / 8a m mz1 / 2 y1
73
x x1 y y1 z z1
or 3 y12 z1 / 4a 2 y1 z1
x x1 y y1 z z1
or [using (5)]
3x1 2 y1 z1
which is the required equation of the inflexional tangent. Answer
3.7 Singular points
3.7.1 Definition :
If at a point P (x, y, z) of the surface F (x, y, z) = 0
F F F
0 .....(3.7.1)
x y z
then every straight line through P (x, y, z) will meet surface in two coincident points, such a point is
called a singular point or the first order on the surface.
3.7.2 Tangents at the singular point :
The straight lines through P (x, y, z) whose direction ratio satisfy the equation
2
l m n F 0 .....(3.7.2)
x y z
meet the surface in three coincident points at P (x, y, z) and are called the tangents at the singular point.
Eliminating l, m, n from the equations of the straight lines and (2), we get the locus of the system
of tangents through P (x, y, z) as the surface
2 2 F 2 F
x ... 2 y z ... 0 .....(3.7.3)
x 2 yz
Singular points are classified according to the nature of the locus of the tangent lines represented
by (3.7.3) :
(i) if this locus is a proper cone, then the point P is called a conical point or conic node.
(ii) when it is a pair of distinct planes, then the point P is called a biplaner node or binode.
(iii) when the pairs of planes coincide, then the point P is called uniplanar node or unode.
3.7.3 Examples :
Ex.1. Find and classify the singular points of the surface
xyz – a2 (x + y + z) + 2a3 = 0
Sol. The equation of the surface can be written as,
F (x, y, z) = xyz – a2 (x + y + z) + 2a3 = 0 .....(1)
74
Differentiating (1) with respect to x, y and z respectively, we get
F
0 yz a 2 0 or yz a 2 , .....(2)
x
F
2 2, .....(3)
y 0 xz a 0 or xz a
F
0 xy a 2 0 or xy a 2 . .....(4)
z
From (2), (3) and (4), we get (a, a, a) which is a singular point.
Now, shifting the origin at (a, a, a) by substituting x = X + a, y = Y + a, z = Z + a the equation
of the surface reduce to
(X + a) (Y + a) (Z + a) – a2 (X + a + Y + a + Z + a) + 2a3 = 0
= XYZ + a (XY + YZ + ZX) = 0.
The locus of the inflexional tangents are
a (XY + YZ + ZX) = 0
which is an equation of a cone therefore (a, a, a) is a conic node.
Ex.2. Prove that the z-axis is a nodal line with unodes at the points (0, 0, – 2) and
(0, 0, 2/3) for the surface
2xy + x3 – 3x2y – 3xy2 + y3 + z (x2 – xy + y2) = 0.
Sol. The origin is singular point and the locus of the inflexional tangent is
2xy = 0 pair of planes x = 0 and y = 0.
Therefore, the origin is binode.
But x = 0 = y is the z-axis and origin lies on oz i.e. z-axis is the nodal line. Consider a point
(0, 0, – 2) on z-axis and shifting the origin at (0, 0, – 2) by substituting
x = X, y = Y, z = Z – 2.
The equation of surface is reduced to
– 2X2 – 2Y2 + 4XY + X3 – 3X2Y – 3XY2 + Y3 + Z (X2 – XY + Y2) = 0
The locus of inflexional tangents is,
– 2X2 – 2Y2 + 4XY = 0
or (X – Y)2 = 0 X – Y = 0 and X – Y = 0.
These are two coincident planes, therefore (0, 0, – 2) is a unode.
By similar treatment, we can prove (0, 0, 2/3) is also a unode.
3.7.4 Self-learning exercise-5.
1. If the locus is proper cone then singular point is called ..... .
2. Write the other name of singular point.
75
3.8 Indicatrix
3.8.1 Definition :
Let the plane z = 0 be taken as the tangent plane and the z-axis as normal at a given point of the
surface.
If z = f (x, y) is the equation of surface, expanding it by Maclaurin’s theorem we get
1 2
z px qy
2
rx 2sxy ty 2 ... .....(3.8.1)
z z 2 z 2 z 2 z
where p , q , r 2, s , t 2 are the values at the origin.
x y x y x y
Since the tangent plane at the origin is z = 0, we have p = 0 and q = 0 and therefore at the
origin
2z = rx2 + 2sxy + ty2 + ... .....(3.8.2)
If we neglect the third and higher powers of x and y, the shape of the surface in the neighbourhood
of the origin is approximately a conicoid given by
2z = rx2 + 2sxy + ty2 .....(3.8.3)
This conicoid is a paraboloid or parabolic cylinder according as rt s2 or rt = s2, respectively.
The section of the surface by the plane z = h is the same as the section of the conicoid therefore
it is a conic, given by
z = h, 2h = rx2 + 2sxy + ty2 .....(3.8.4)
and is called the indicatrix.
Thus the conic in which a surface is cut by a parallel plane at an infinitesimal distance near the
tangent plane at any point is called the indicatrix at the point.
3.8.2 Examples :
Ex.1. Prove that the indicatrix at a point of the surface z = f (x, y) is a rectangular hy-
perbola if
(1 + p2) t + (1 + q2) r – 2pq s = 0.
Sol. The equation of the surface is given by
z = f (x, y) .....(1)
The direction cosines of the inflexional tangents are given by
lp + mq – n = 0 .....(2)
and l2r + 2lms + m2t = 0 .....(3)
z z 2 z 2 z 2 z
where p , q , r 2 , s , t 2.
x y x y x y
76
Equation (3) may be written as
2
l l
r 2 s t 0
m m
l1 l2 t
which gives . .....(4)
m1 m2 r
Also eliminating l between (2) and (3), we get
2
n mq n mq 2
r 2ms m t 0
p p
or m2 (rq2 + tp2 – 2pqs) + 2nm (– qr + ps) + n2r = 0
m1 m2 r
for which 2 2 .....(5)
n1 n2 q r p t 2 pqs
from equations (4) and (5), we have
l1 l2 m1 m2 n1 n1
.....(6)
t r q r p 2t 2 pqs
2
These inflexional tangents will be right angle if
l1 l2 + m1 m2 + n1 n2 = 0 .....(7)
Substituting (6) into (7), we get
t + r + q2r +p2t – 2pqs = 0
or (1 + p2)t – 2pqs + (1 + q2) r = 0
which is the required condition for the indicatrix to be rectangular hyperbola.
Ex.2. Prove that the points of the surface
xyz – a (yz + zx + xy) = 0
at which the Indicatrix is a rectangular hyperbola, lie on the cone
x4 (y + z) + y4 (z + x) + z4 (x + y) = 0.
Sol. The given surface equation can be written as,
axy
z .....(1)
xy ay ax
z xy ay ax ay axy y a z 2
Therefore, p 2 .....(2)
x xy ay ax 2 x
z2 2
2 z 2z z x
Similarly q , r
y2 x 2 x4
2
2 z 2z3 2 z 2 z z y
s and t 2 .....(3)
x y x 2 y 2 y y4
77
Substituting these values into the equation of indicatrix, we have
z4 2z2 z y 2
z 4 2z z x 2z 2 z 2 2z3
1 4 1 4 2 2 2 2 0
x y4 y x4 x y x y
or 2z2 {x4 (y + z) + y4 (z + x) + z4 (x + z)} = 0
Therefore the required locus is
x4 (y + z) + y4 (z + x) + z4 (x + y) = 0 (since z 0) .....(4)
which is a cone.
3.9 Answers to self-learning exercises
Self-learning exercise-1
1. Three
2. Linear relation
3. Constant angle
Self-learning exercise-2
12
1. K1
2
K2 , where c s
K
12
2. 1
K K
, where c s
K 2 K 2
3. r1 r c s tˆ
Self-learning exercise-3
1. r1 r nˆ tan c bˆ.
K 3 cos 3 c
2. K1
K sin c K cos c
K 3 sin c cos 2 c
3. 1
Kt sin c K cos c
1
4. tan c .
K1
Self-learning exercise-4
1. Right conoid : If the given line is at right angles to the given plane, the locus is a right conoid.
2. z = f (y/x).
3. By intersection of fixed line and given plane.
78
Self-learning exercise-5
1. Conic node.
2. First order on the surface.
3.10 Exercises
1. State and prove existence and uniqueness theorems.
2. Prove that the distance between corresponding points of two curves is constant.
3. Show that the involutes of a circular helix are plane curves.
4. Write down the equation of conoid.
5. Find the equation to the conoid generated by lines parallel to the plane XOY, which are drawn to
intersect OZ and the curve
x2 y2 z2
x2 y 2 z 2 b2 , 1.
a2 b2 c 2
2 2 2
[Ans. b 2 z 2 x y 1 z x 2 y 2 ]
a2 b2 c 2
6. Find and classify the singular points of the surfaces
(i) xyz = ax2 + by2 + cz2 [Ans. (0, 0, 0) is a conic node]
(ii) xyz – a2 (x + y + z) + 2a3 = 0. [Ans. (a, a, a) is a conic node]
7. Prove that the indicatrix at every point of the helicoid z = c tan–1 (y/x) is a rectangular
hyperbola.
8. Prove that every point on a cone or cylinder is a parabolic point.
79
UNIT 4 : Envelope, Edge of Regression, Ruled Surfaces,
Developable Surface, Tangent Plane to a Ruled
Surface
Structure of the Unit
4.0 Objectives
4.1 Introduction
4.2 Envelope
4.2.1 Family of surfaces (one parameter)
4.2.2 Characteristic of family of surfaces
4.2.3 Envelope
4.2.4 Edge of regression
4.2.5 Family of surfaces (two parameters)
4.2.6 Self-learning exercises-1
4.3 Ruled surface
4.3.1 Equation to a ruled surface
4.3.2 Criterion for a surface to be developable
4.3.3 Self-learning exercises-2
4.3.4 Equation of tangent plane to a ruled surface
4.4 Summary
4.5 Answers to self-learning exercises
4.6 Exercises
4.0 Objectives
After studying this unit you will be able to understand :
1. characteristic, envelope and edge of regression of family of surfaces,
2. ruled surfaces, their classification and associated properties.
4.1 Introduction
Family of surfaces admit certain geometrical features such as characteristic and edge of regres-
sion which are in fact curves lying on the surface. Their study is of vital importance in the theory of dif-
ferential geometry. Similarly, envelope of family of surfaces has a unique property that it touches each
member of the family of surfaces.
There are many surfaces which are generated due to motion of straight lines. Such surfaces are
called ruled surfaces. This includes their classification as developable, skew surfaces and associated prop-
erties.
80
4.2 Envelope
4.2.1 Family of surfaces (One parameter) :
An equation
F (x, y, z, a) = 0 .....(4.2.1)
where a is a parameter, represents a family of surfaces. By assigning different real values to the param-
eter a we get different surfaces belonging to family given by (4.2.1). For specific value to a, we get a
specific surface of the family and is called member of the family of the surfaces.
4.2.2 Characteristic of a family of surfaces :
Characteristic of a surfaces is the curve of intersection of two consecutive surfaces.
Let F (x, y, z, ) = 0, F (x, y, z, ) = 0 .....(4.2.2)
be two consecutive surfaces of the family given by (4.2.1). Then the curve of intersection of the con-
secutive surfaces (4.2.2) is given by
F x, y, z , F x, y , z ,
F x, y, z , 0, 0 .....(4.2.3)
The limiting position of the curve as 0 is obtained as
F
F x, y , z , 0, 0, .....(4.2.4)
which determines the characteristic curve corresponding to the value .
4.2.3 Envelope :
The concept of envelope of a family of surface is very important. The envelope of a family of
surfaces touches every member of the family, at all points of its characteristic. Geometrically, the enve-
lope of the family of surfaces is the locus of characteristic for all values of the parameter. Hence, the
envelope is obtained from the equation
F
F 0, 0, .....(4.2.5)
by eliminating .
Theorem 1. The envelope of a family of surfaces touches each member of the family at
all points of its characteristic.
Proof : Let F (x, y, z, a) = 0 be the family of surfaces, where a being the parameter.
Let (x, y, z, ) = 0, F (x, y, z, ) = 0 be any two consecutive surfaces of the given family.
Then the envelope is obtained by eliminating from the equations
F
F 0, 0.
Consider (x, y, z, ) = 0 as equation of the envelope, where is not merely a constant but a
F
function of x, y, z satisfying 0.
81
Now, the normal to the envelope
F
F x, y , z , 0 where 0
is parallel to the vector F i.e. parallel to the vector
F F ˆ F F ˆ F F ˆ
i j k
x x y y z z
F ˆ F ˆ F ˆ F
or i
x
j k. 0
y z
F
The vector x iˆ is parallel to normal to the surface F (x, y, z, ) = 0. This reveals that
at all common points, the surface and the envelope admit the same normal, and consequently the
same tangent plane. This concludes that surface and envelope touch each other at all points of the
characteristic.
Note : Characteristic of the envelope is the curve in which two consecutive surfaces intersect.
Thus, each characteristic lies on the envelope.
4.2.4 Edge of regression :
Edge of regression is a curve that lies on the envelope. We have seen that the characteristic is
the curve in which two consecutive surfaces intersect.
Two consecutive characteristics meet in one or more points. The locus of points of intersection
of consecutive characteristics is called the edge of regression of the envelope. Obviously, the edge of
regression (a curve) lies on the envelope simply because every characteristic lies on the envelope. Edge
of regression may have the following formal definition :
“Edge of regression is the locus of the ultimate points of intersection of consecutive characteris-
tics of one parameter family of surfaces”.
Equation of the edge of regression of the envelope :
Let F (x, y, z, a) = 0 ......(4.2.6)
be the family of surfaces, a being the parameter.
Let F (x, y, z, ) = 0 and F (x, y, z, + ) = 0,
be two consecutive surfaces. Then the characteristic to the surface F (x, y, z, ) = 0 is given by
F
F x, y , z , 0, 0 ......(4.2.7)
The characteristic to the surface F (x, y, z, + ) = 0 is given by
F x, y , z ,
F x, y, z , 0, 0 ......(4.2.8)
82
Expanding the equations (4.2.8) be Taylor’s series, we get
F
F x, y , z , ..... 0
F 2 F
.... 0 .....(4.2.9)
2
From equations (4.2.7) and (4.2.9), we obtain
F 2 F
F 0, 0, 0 .....(4.2.10)
2
the edge of regression is obtained by eliminating from the equations (4.2.10).
Theorem 2. Each characteristic touches the edge of regression.
Proof : Let (x, y, z, a) = 0, be family of surfaces. Then for a = and a = + ,
F (x, y, z, ) = 0 .....(4.2.11)
and F (x, y, z, + ) = 0 .....(4.2.12)
are two consecutive surfaces.
Thus the characteristic curve corresponding to the surface F (x, y, z, ) = 0 is given by
F
F 0, 0 .....(4.2.13)
and the edge of regression is given by
F 2 F
F 0, 0, 0 .....(4.2.14)
2
We can consider edge of regression given by
F
F 0, 0
provided is a function of x, y, z given by
2 F
0.
2
Note that the tangent at any point P (x, y, z) to the edge of regression is nothing but the line of
intersection of the tangent planes to the surface. Consequently the tangent is normal to the vectors F
and F, where is function of x, y, z.
Thus this tangent is perpendicular to the vectors
F F ˆ
x i,
x
.....(4.2.15)
F F ˆ
and x
i,
x
.....(4.2.16)
83
F 2F
Using 0, 0, the equations (4.2.15) and (4.2.16) are reduces to
2
F
x iˆ, .....(4.2.17)
2F
and x iˆ. .....(4.2.18)
Vectors (4.2.17) and (4.2.18) are perpendiculars to tangent planes at P (x, y, z) to the charac-
F x, y, z ,
teristic F (x, y, z, ) = 0, 0. This concludes that the tangent to the edge of regression
is parallel to the tangent to the characteristic and consequently the two curves touch at their common
points.
4.2.5 Family of surfaces (two parameters) :
We now proceed for the case of envelope of two-parameter family of surfaces.
Envelope of two parameter family of surfaces.
Let F (x, y, z, a, b) = 0 .....(4.2.19)
where a, b are parameters, denote a family of surfaces.
Then the consecutive surfaces for a = , b = are
F (x, y, z, , ) = 0 .....(4.2.20)
F (x, y, z, + , + ) = 0 .....(4.2.21)
On expanding (4.2.21) by Taylor’s series, we get
F F
F x, y, z , , ..... 0
when 0, 0 we ought to have at a point of intersection in the limiting case :
F F
F 0, 0. .....(4.2.22)
Further since , are mutually independent then the identity (4.2.22) is line if
F F
F 0, 0, 0. .....(4.2.23)
Thus, we conclude that the criterion given by (4.2.23) is mandatory for two consecutive sur-
faces given by (4.2.20) and (4.2.21) to interest. On elimination of , we get the equation of the enve-
lope of two parameter family of surfaces.
Ex.1. Suppose that a tangent plane to the ellipsoid
x2 y 2 z2
1
a2 b2 c 2
84
meets the coordinate axes in points P, Q, R. Prove that the envelope of the sphere OPQR is
(ax)2/3 + (by)2/3 + (cz)2/3 = (x2 + y2 + z2)2/3
where 0 is the origin.
Sol. Let us consider
x y z
1 .....(1)
to be the tangent plane to the ellipsoid
x2 y 2 z2
1 .....(2)
a2 b2 c 2
Then the condition of tangency ensures that
a 2 b2 c 2
1 .....(3)
2 2 2
Given that the tangent plane (1) meets the axes in points P, Q, R, then the equation to the sphere
OPQR is
x2, y2 + z2 – x – y – z = 0 .....(4)
Fig. 4.1
Note that for variable values of , , we would have different tangent planes to the ellipsoid
and consequently different spheres of the form (4), i.e., (4) constitutes a family of surfaces, where
are parameters.
We denote F ( x, y , z , , , ) x 2 y 2 z 2 x y z 0 .....(5)
a 2 b2 c 2
and , , 1 0 .....(6)
2 2 2
The equation of the surfaces given by (5) is obtained on elimination of the parameters , , and
from the equations
F F F .
x y z
This gives
2a 2
/ 3
2b 2
/ 3
2c 2
/ 3
85
x1 3 y1 3 z1 3
or 2 3 2 3 k (say)
a2 3 b c
k a2 3 k b2 3 k c2 3
or , , ......(7)
x1 3 y1 3 z1 3
The values of as obtained in (7) are now put in (3) and (5) to yield
a 2 x 2 3 b2 y 2 3 c 2 z 2 3
43
4 3 4 3 K2
a b c
or (ax)2/3 + (by)2/3 + (cz)2/3 = K2 ......(8)
k a2 3 k b2 3 k c 2 3
and x 2 y 2 z 2 x 1 3 y 1 3 z 1 3 0
x y z
or
x 2 y 2 z 2 k (ax) 2 3 (by )2 3 (cz )2 3
x2 y2 z2
or K . .....(9)
( ax ) 2 3 (by ) 2 3 (cz ) 2 3
From (8) and (9), we get the required result.
Ex.2. Find the envelope of the family of planes
x y z
F ( x, y , z , , ) cos sin sin sin cos 1 0.
a b c
Sol. We have,
x y z
F ( x, y , z , , )
cos sin sin sin cos 1 0 .....(1)
a b c
The required envelope is obtained by the elimination of the parameters and from the equa-
tions
F F
F 0, 0, 0
On differentiating (1) partially with respect to and , respectively, we obtain.
F x y
sin sin cos sin 0
a b
x y
sin cos 0 [ sin 0] ......(2)
a b
F x y z
cos cos sin cos sin 0 ......(3)
a b c
ay
Equation (2) gives, tan ......(4)
bx
x y z
Equation (3) gives, cos sin tan ......(5)
a b c
86
The equation (1) can be rewritten as
x y z
cos sin sin 1 cos ......(6)
a b c
z z
using (5) in (6), we get tan sin 1 cos
c c
z sin z
or sin 1 cos
c cos c
z sin 2 cos 2
or 1
c cos
z
or cos ......(7)
c
Now using (7) in (5), we get
z c2
x y z c2 cos tan 2 1
cos sin 1 .....(8)
a b c z2 c z
On squaring (2) and (8) and then on adding we get
x 2 y 2 z 2 c2
1
a 2 b2 c 2 z 2
x2 y 2 z2
or 1
a2 b2 c 2
as the required envelope of the given family of planes.
Ex.3. Find the equation of the developable surface whose generating line passes through
the curve y2 = 4ax, z = 0; x2 = 4ay, z = c and show that its edge of regression is given by
cx2 – 3ayz = 0 = cy2 – 3ax (c – z).
Sol. Recall that a developable surface is generated by one parameter family of planes.
In order to find the equation of the required developable surface we first find the family of planes
F (m) = 0 (where m is parameter). The developable is obtained eliminating m from F (m) = 0 and
F (m) 0. We proceed as follows.
The equation to the tangent to the curve y2 = 4ax, z = 0 is
a
y mx , z0 .....(1)
m
Then any plane touching the parabola y2 = 4ax, z = 0 is
a
y mx z 0, (where being scalar) .....(2)
m
87
The section of the plane (2) by the plane z = c is
a
y c mx 0 ......(3)
m
y c a
or x
m m m2
If (3) touches the parabola x2 = 4ay, z = c, then the equation
2
y c a
m m 2 4ay,
m
must have equal roots
2
2c2 y a c y a
i.e., 2
2 2 2 4ay 0,
m m m mm m
must have equal roots.
am2 a
This gives .
c mc
Putting this value of in (2), we get the plane touching both the given curves and it is
a am 2 a
F m y mx z 0
m c cm
or
F m am3 a cz my m x a 0
2 .....(4)
z
This gives F (m) 3am 2 y 2mx 0. .....(5)
c
Elimination of m from (4) and (5) will give the required developable surface.
Edge of regression :
Differentiating (5) partially with respect to m, we get
z
F (m) 6am 2 x 0
c
cx
This gives m . ......(6)
3az
Note that the edge of regression is given by F (m) = 0, F (m) 0, F(m) 0. Hence putting the
value of m From (6) in F (m) 0, we find
2
z cx cx
3a y 2 x0
c 3az 3az
On simplification, we get
6x2 – 3ayz = 0 ......(7)
88
Again, on putting the value of m in (4) and performing simplification, we get
27a3z3 + 9azxyc2 – 2c3x3 – 27ca3z2 = 0.
Using (7) in the above equation, we obtain
27a3z2 (z – c) – 2c2x (3ayz) + 9azxyc2 = 0.
c 2 x4
or 27a 3 2 2
z c 3ayz c 2 x 0
9a y
3ac 2 x 4
or
y 2
z c cx 2 c 2 x 0
or cy2 – 3ax (c – z) = 0
Hence the edge of regression is given by
cx2 – 3ayz = 0 = cy2 – 3ax (c – z).
Ex.4. Find the equation of the developable surface which contains the two curves
y2 = 4ax, z = 0 and (y – b)2 = 4cz, x=0
and show that its edge of regression lies on the surface
(ax + by + cz)2 = 3abx (b + y).
Sol. The given curves are
y2 = 4ax, z = 0 .....(1)
(y – b)2 = 4cz, x = 0 ......(2)
The equation to the tangent to the curve (1) is
a
y mx , z 0, (where m is the slope) .....(3)
m
Now, the equation to the plane that touches the parabola (1) is
a
y mx z 0, (where is a scalar) .....(4)
m
a
If the plane (4) touches the curve (2), then it means that the line y z touches the curve
m
(y – b)2 = 4cz.
2
a
That is, z b 4cz must have equal roots.
m
2
2 2 a a
z z 4c 2 b b 0 must have equal roots.
m m
2 2
i.e.
a 2 a
4c 2 m b 4 m b
89
On simplification, we get
mc
......(5)
bm a
On putting the value of from (5), in (4), we find the equation of the plane touching the curve
(2), and it is
a mcz
y mx 0 .....(6)
m bm a
or on simplifying it becomes
bm3x – m2 (ax + by + cz) + am (y + b) – a2 = 0 .....(7)
We denote the surface (7) by F (m).
Developable surface :
We know that a developable surface of the surface F (m) = 0 is obtained by eliminating the
parameter m from the equations F (m) = 0 and F (m) 0 .
Differentiating (6) with respect to m, we get
F ( m ) 3bm 2 x 2m ( ax by cz ) a (b y ) 0 .....(8)
From (7) and (8), we obtain
a (b y ) (ax by cz ) 9a 2bx
m .....(9)
2 (ax by cz )2 6abx ( y b)
Using this value of m in (7), we get the required developable surface.
Edge of regression :
We know that edge of regression for the surface F (m) = 0 is obtained on elimination of m
from, F (m) = 0, F (m) 0 and F(m) 0, we have
F(m) 6 bmx 2 (ax by cz ) 0 [from (7)]
ax by cz
This gives, m . .....(10)
3 bx
Putting this value of m is (7), we get
2
ax by cz ax by cz
3bx
3 bx 2
3 bx ax by cz a (b y ) 0
or (ax + by + cz)2 = 3ab (b + y), .....(11)
which gives the surface on which edge of regression lies.
Ex.5. Find the developable surface which passes through the curves
y2 = 4ax, z = 0 and y2 = 4bz, x = 0.
Sol. The equation to the tangent to the curves
y2 = 4ax, z = 0 .....(1)
90
a
is y mx , z0 .....(2)
m
Therefore, equation to the plane touching (1) is
a
y mx z 0 .....(3)
m
[Remember that here m is the slope of the tangent and is a parameter].
Equation (3) touches the curve
y2 = 4bz, x = 0 .....(4)
a
That means y z [on putting x = 0 in (4)]
m
2
a a
and z 4bz has equal roots i.e., discriminant is zero. [putting y z in the equation
m m
y2 = 4bz]
2a a2
or 2 z 2 4b z 0,
m m2
2
2 a 4 2 a 2
has equal roots. That is 4b
m m2
bm
or .....(5)
a
Putting the value of in the equation (3), we get the equation of the plane touching the given
curve as
a bm
F (m) y mx z 0. .....(6)
m a
Developable surface :
We know that the developable surface is obtained by eliminating m from the equations
F (m) = 0, F (m) 0. Now, differentiating (6) partially with respect to m, we obtain
a bz
F (m) x 2 0 .....(7)
m a
Equation (6) can be written as
a bz
y m x 2 0
m a
a a bz a
or ym 2 2 0 From (7) x a m 2
m m
2a
or y 0
m
91
2a
or m .....(8)
y
On putting the value of m in equation (7), we get the developable surface as
y2 = 4ax + 4bz.
Ex.6. Show that the edge of regression of the developable that passes through the pa-
rabolas x = 0, z2 = 4ay; y2 = 4az, x = a is given by
3x y z
.
y z 3(a x)
Sol. In order to find the required edge of regression we have to first find the plane that touches
both the given curves.
Equation to the tangent to the parabola x = 0, z2 = 4ay is
a
z my , x 0.
m
Then the plane through this tangent (i.e. touching the parabola x = 0, z2 = 4ay) is
a
z my x 0 .....(1)
m
z a x
or y 2 0 .....(2)
m m m
Equation (2) meets the parabola y2 = 4az, x = a therefore its section by x = a is
z a a
y 2 ......(3)
m m m
Now, if equation (3) touches the parabola y2 = 4az, x = a.
2
z a a
Then the equation 2 4az
m m m
must have equal roots, i.e. (mz – a + am)2 = 4azm2
must have equal roots, i.e. m2 z2 + [2am2 – 2ma – 4am4] z + a2 (m – 1)2 = 0 has equal roots.
Thus we must have [2am2 – 2ma – 4am4]2 = 4a2 m2 (m – 1)2.
1 m3
On solving we get .....(4)
m
Putting this value of in (2) we find the plane that touches both the given parabolas as
z a 1 m3
y 2 x .....(5)
m m m
or F (m) = m3 x – m2y + mz + (x – a) = 0 .....(6)
92
Edge of regression :
The edge of regression is given by
F (m) 0, F (m) 0, F(m) 0
From equ. (4), we find F ( m ) 3m 2 x 2my z 0 .....(7)
F(m) 6mx 2 y 0
y
m .....(8)
3x
Equation (7) can be written as
y z
m 2 2m 0
3x 3 x
z y
or m 2 2m ( m ) 0 [putting m ]
3x 3x
z
or m2 .....(9)
3x
Note that to obtain the answer in required form we have to perform some tricky mathematical
manipulation as follows.
Dividing (4) by x, we obtain
y z a
m3 m 2 m 1 0
x x x
a
or m3 3m3 3m3 1 0 [using (6), (7)]
x
ax
or m3 .....(10)
x
Now, we write m3 m2 m
ax z y
[using (6), (7), (8)]
x 3x 3x
3x z
or .....(11)
y 3(a x )
Again, m3 m = (m2)2
2
a x y z
x 3 x 3x
y z . .....(12)
z 3(a x)
From (11) and (12), the required edge of regression is obtained as
3x y z .
y z 3(a x )
93
4.2.6 Self-learning exercises-1
1. Characteristic of the family of surfaces F (x, y, z, ) = 0 is given by
2 F 2 F 3 F
(a) F = 0, 0 (b) 0, 0
2 2 3
F
(c) F = 0 (d) F 0, 0
2. Which of the following is not true ?
(a) characteristic may not lie on the envelope
(b) envelope touches each member of the family
(c) edge of regression is a curve
(d) edge of regression lies on the envelope.
3. Find the envelope of the plane lx + my + nz = 0, where al2 + bm2 + cn2 = 0.
4.3 Ruled surface
You are familiar with the surfaces such as cones, cylinders, hyperboloid of one sheet and hyper-
bolic paraboloid. All these surfaces are generated by single parameter family of straight lines. But things
are not that simple and we need a further analysis. Hence we define ruled surfaces.
Ruled surface : A ruled surface is a surface which is generated by single parameter family of
straight lines. The line is called the generating line or ruling or generator of the ruled surface. All the
surfaces mentioned above are obviously ruled surfaces. Ruled surfaces are classified into two categories
depending upon intersection/non intersection of their consecutive generators.
Ruled surface on which consecutive generators intersect is called developable surface. Cones,
cylinders and conicoid are developable surfaces. A ruled surface on which two consecutive generators
do not intersect is called a skew surface or a scroll. Hyperboloid of one sheet and hyperbolic parabo-
loid are scrolls.
4.3.1 Equation to a ruled surface :
To find the equation of a ruled surface let us first explain directrix or base curve of a ruled sur-
face. A curve C on the ruled surface is called the base curve if it meets each generator exactly once.
Note that a ruled surface has many base curves. Now note that a ruled surface is determined by a base
curve C, say and the direction of the generator at the point of intersection of the generator and the base
curve C. After understanding the above now we find expression for a ruled surface.
Let P be a general point on the ruled surface and Q be a point on the base curve C. Let r ( s )
and R be position vectors of Q and P respectively with respect to the origin. Further we assume that
gˆ ( s ) be the unit vector along the generator at Q then the equation to the ruled surface is
94
R r ( s ) gˆ ( s ) .....(4.3.1)
where is a parameter that determines the directed distance along the generator from C.
g ( s)
P Q
r ( s)
r = r (s)
O
Fig. 4.2
The equation (4.3.1) can also be written in the Cartesian form.
Let (x, y, z) and (X, Y, Z) be the coordinates of the points Q and P, respectively. Then we can
write r ( s ) x iˆ y ˆj z kˆ, Rˆ X iˆ Y ˆj Z kˆ
gˆ ( s) g1 iˆ g 2 ˆj g3 kˆ
where | gˆ (s ) | g12 g 22 g32 1
using above in the equation (4.3.1) we find the equation of the ruled surface as
X x Y y Zz
( ) ......(4.3.2)
g1 g2 g3
Equation (4.3.2) emphasises that a ruled surface is determined by single parameter () family of
straight lines. Note that the equation (4.3.2) can also be written as
X = aZ + , Y = bZ +
where a, b, , are functions of .
Uptil now we have gone through the idea of ruled surface and its further classifications as devel-
opable and skew surface. The following theorem is the criterion to determine whether the ruled is devel-
opable or skew.
4.3.2 Criterion for a surface to be developable :
Theorem 1. A ruled surface is developable or skew if and only if [tˆ, gˆ , gˆ ] 0 or 0
accordingly where tˆ is the unit tangent vector at a point on the base curve, and ĝ is unit vector
along the generator through the point.
Proof : The condition is necessary :
Let C be a base curve given by r r ( s ) on the ruled surface. Let RS be an arc on the curve C
such that R r and S r dr be two consecutive points. Let arc RS = ds.
95
For the neighbouring point S of R, we have the position vector
dr ˆ d r
r dr r ds r t ds . (where tˆ unit tangent vector at R)
ds ds
Let g1 and g2 be generators through R and S, respectively (i.e. g1 and g2 are consecutive gen-
erators since R, S are consecutive points on the base curve C) and ĝ and gˆ dgˆ are unit vectors along
the generators g1 and g2. Let MN be shortest distance between g1 and g2 then MN is perpendicular to
both g1 and g2. Then MN is parallel to gˆ dgˆ gˆ or to
g2 g1
g g + dg
M N
C
ds
R S
r+dr
r
Fig. 4.3
dgˆ
gˆ gˆ ds gˆ [ Note gˆ dgˆ gˆ ds gˆ gˆ ds ]
ds
or MN is parallel to gˆ gˆ ds [ gˆ gˆ 0]
Shortest distance MN = Projection of RS on MN
d r . (unit vector along MN)
gˆ gˆ d r gˆ gˆ
dr ds
| gˆ gˆ | ds | gˆ gˆ |
ds
[tˆ, gˆ , gˆ ] ......(4.3.3)
| gˆ |
[Note : Since ĝ is perpendicular to ĝ and | gˆ | 1 | gˆ gˆ | | gˆ | | gˆ |sin 90 | gˆ | ]
Recall that in developable surface two consecutive generators intersect that is the shortest dis-
tance between the generators is zero. Thus we find that if the surface is developable, shortest distance
MN is zero.
ds
Hence tˆ, gˆ , gˆ 0 | g | 0 .....(4.3.4)
96
This is the necessary condition for the ruled surface to be developable. In ruled surface to be
developable.
In the case when the surface is skew, the consecutive generators don’t interest and therefore the
shortest distance is not zero. Thus, tˆ, gˆ , gˆ 0 is the necessary condition for a ruled surface to be
skew.
The condition is sufficient :
If tˆ, gˆ , gˆ 0 the shortest distance between the consecutive generators is zero, hence the
surface is developable.
Similarly tˆ, gˆ , gˆ 0 consecutive generators don’t intersect, hence the surface is skew..
Theorem 2. A ruled surface generated by x = az + , y = bz + is developable or skew if
b a 0 or 0 respectively..
Proof. The ruled surface is given to be generated by single parameter family of straight lines
x = az + , y = bz + .....(4.3.5)
where a, b, and are functions of single parameter (say). The equation (4.3.5) can be written as
x y z 0
.....(4.3.6)
a b 1
Equation of ruled surface in vector form is
R r gˆ .....(4.3.7)
Then r iˆ ˆj 0kˆ ( , , 0)
dr ˆ
t ( , , 0)
ds
and gˆ aiˆ bjˆ 1 kˆ ( a, b, 1)
dgˆ
g ( a, b, 0)
ds
Thus
tˆ, gˆ , gˆ iˆ ˆj 0 kˆ g gˆ
iˆ ˆj kˆ
iˆ ˆj 0.kˆ a b 0
a b 1
iˆ ˆj 0.kˆ b iˆ ˆja kˆ (ab ab)
b a. .....(4.3.8)
Thus the surface is developable if b a 0 or skew if
b a 0. .....(4.3.9)
97
Aliter : The given generator can be written as
x y z 0
......(4.3.10)
a b 1
x y z
Let
a b 1
x ( ) y ( ) z 0
and
a a b b 1
be consecutive generators of the surface. If the surface is developable then these generators intersect
hence shortest distance between them is zero.
0
a a b b 1 0
i.e.
a b 1
0
a b 0 0
or
a b 1
or ba
b a 2
or (t ) 0
t t t t
b a 0 [ t 0] (4.3.11)
Ex. Find the equation to the edge of regression of the developable
y = xt – t3, z = t3y – t 6.
Sol. In the given surface “t” is the parameter. The find the edge of regression recall that the
point of intersection of the two consecutive generators of a developable surface is a point on the edge of
regression.
The given equation of the developable is nothing but composed of the generators
y = xt – t3, z = t3y – t 6.
The two consecutive generators are
y = xt – t3, z = t3y – t 6 .....(1)
and y = x (t + t) – (t + t)3, z = (t + t)3 y – (t + t)6 .....(2)
Now, solving y = xt – t3 and y = x (t + t) – (t + t)3 and on neglecting higher powers of t,
we get
(x – 3t2) t = 0
x = 3t2 [ t 0]
Again solving z = t3y – t6 and z = (t + t)3 y – (t + t)6, we get on neglecting higher powers of
t y = 2t3.
98
Using this value of y in the given generator we get
z = t3 (2t3) – t6 = t6.
Thus x = 3t2, y = 2t3, z = t6 is the required edge of regression.
4.3.3 Self-learning exercises-2
1. Developable surface is generated by :
(a) cones (b) cyclinders
(c) spheres (d) straight lines
2. Prove that the line x = 3t2z + 2t (1 – 3t4), y = t2 (3 + 4t2) – 2tz generates a skew surface.
3. Prove that xyz = 2 is a developable surface.
4. Explain that a developable surface can be found to pass through two given curves.
5. Name two skew surfaces.
4.3.4 Equation of a tangent plane to a ruled surface :
(A) Equation in the vectorial notation :
Let r be the position vector of any point P on the directrix, ĝ be unit vector along the genera-
tor at P and r and ĝ are function of single parameter . Then the ruled surface is given by
R r gˆ ......(4.3.13)
where R is the position vector of the current point on the ruled surface. Note that R is function of two
independent parameters and .
The equation to the tangent plane to the ruled surface (4.3.13) is given by
R * R R1 R2 0 .....(4.3.14)
where suffixes ‘1’ and ‘2’ denote differentiation of R with respect to and respectively and R* is the
position vector of the current point on the tangent plane. Thus (4.3.14) can be written as
R * R , r1 gˆ1, gˆ 2 0
......(4.3.15)
(B) Equation in cartesian notation :
Let x = az + , y = bz + .....(4.3.16)
be the generator of the ruled surface, where a, b, and are functions of single parameter s.
Let us assume that () be a point on the ruled surface generated by (4.3.13). Then obvi-
ously can be regarded as functions of s and z
where = az + , = bz + , = z. ......(4.3.17)
Thus, the equation of the tangent plane at the point (s, z) is given by
99
x y z
x y z
s s s 0.
x y z .....(4.3.18)
z z z
Now x = az + , y = bz +
x y
az , bz [Note that a, b, , are functions of s only]
s s
where a, b, , denote differentiation of respective quantities with respect to s.
Thus (4.3.18) becomes
az bz z
az bz 0 0.
......(4.3.19)
a b 1
Performing column operations c1 – ac3 and c2 – bc3 in (4.3.19), we get
a b z
az bz 0 0.
.....(4.3.20)
0 0 1
On simplifying the above determinant we get
( a ) (bz ) ( b ) ( az ) 0
az
or a ( b ) .....(4.3.21)
bz
Note that the equation (4.3.21) represents the equation of the plane passing through the line
= az + , = bz + . Recall that the line = az + , = bz + is a generator of the ruled surface at
the point (s, z). The above discussion reveals an important fact that the tangent plane at any point of a
ruled surface contains the generator through that point.
Note 1 : If the ruled surface is developable, then the tangent plane is same at all points of the
generator and involves only one parameter. This is evident from the following discussion :
Let the ruled surface is developable then b a 0
a a az
K (say)
b b bz
where K is function of s. In view of this, the equation (4.3.21) takes the form
– az – = K (– b – ) ......(4.3.23)
and involves only one parameter s and is independent of z. Further note that the parameter s has a fixed
value for a particular generator, therefore the tangent plane will be the same at all point of the generator.
100
Note 2 : If the surface is skew, then at different points of a generator, we have different tangent
planes.
Let the surface is skew, then b a 0 .
That means the equation (4.3.21) contains both s and z. If s is kept fixed (for a particular gen-
erator) then (4.3.21) would give different tangent planes for different values of z. That means, the tan-
gent planes are different at different points of the generator of skew surface.
Theorem. Prove that the generators of a developable surface are tangents to curve.
Proof. Let x = az + , y = bz +
be generator of the developable surface, where a, b, , are functions of single parameter s.
Let x = az + , y = bz + .....(4.3.25)
and x = (a + a) z + ( + ), y = (b + b) z + ( + ), .....(4.3.26)
be two consecutive generators of the developable surface. We know that, two consecutive generators
of a developable surface do intersect, hence point of intersection of (4.3.25) and (4.3.26) is given by
x a , y b , z .....(4.3.27)
a b a b
Here note that (4.3.27) represents a curve since a, b, , are functions of parameter s only.
Thus x, y, z are functions of single parameter, hence (4.3.27) is a curve.
We have to show that the generators to the developable surface are tangents to the curve given
by (4.3.27).
The equation to the tangent to the curve (4.3.27) at point (x, y, z) is given by
x y z
......(4.3.28)
x y z
where dot denotes differentiation with respect to s.
From (4.3.27), we have
s s s s
x a , y b , z [note]
a s a s a s b s
a b
x , y , z .....(4.3.29)
a a a a
Now differentiation of (4.3.29), with respect to s, we get
( a a
) a ( b
b) b
x a
z ,
y b z .....(4.3.30)
a 2 b2
using x az, y bz in equation (4.3.28), we get
x y z
a z b z z
101
or = a – az + x, = b – bz + y
or = a + , = b + .....(4.3.31)
Equation (4.3.30) is nothing but the generator through (x, y, z).
4.4 Summary
In this unit, you learnt that characteristic and edge of regression are the curves that lie on the
envelope. An envelope of family of the surfaces is that surface that touches every member of the family.
In the process of learning you came across with idea of ruled surfaces (developable and skew surfaces)
and associated issues.
4.5 Answers to self-learning exercises
Self-learning exercise-1
x2 y 2 z 2
1. (d) 2. (d) 3. (a) 4. 0
a b c
Self-learning exercise-2
1. (a) 5. Hyperboloid of one sheet, hyperbolic paraboloids.
4.6 Exercises
x y z
1. Find the envelope of the plane 1 where u is the parameter..
a u bu cu
[Ans. (9 – )2 = 4 (2 – 3) (2 – 3), where a + b + c – (x + y + z),
ab + bc + ca – x(b + c) – y (a + c) – z (a + b), abc – (bcx – acy – abz)]
2. Find the envelope of the surface lx + my + nz = p where a2l2 + b2 m2 + 2np = 0.
x2 y2
Ans. 2 2 2 z
a b
( ) (1 ) (1 )
3. Find the envelope of the plane x y z where and are
a b c
the parameters.
x2 y2 z2
Ans. 2
2 2 1
a b c
4. From a point P on the conicoid a2x2 + b2y2 + c2z2 = 1 perpendiculars PL, PM, PN are drawn
to the coordinate planes. Find the envelope of the plane LMN.
[Ans. (ax)2/3 + (by)2/3 + (cz)2/3 = 22/3]
102
UNIT 5 : Necessary and Sufficient Condition that a Surface
= F (, ) should Represent a Developable Surface,
Metric of a Surface
Structure of the Unit
5.0 Objectives
5.1 Introduction
5.2 Condition for surface = f () to be a developable surface
5.3 Metric of surface
5.3.1 Curves on a surface and curvilinear coordinates
5.3.2 Parametric transformation
5.3.3 Regular and singular points
5.3.4 Parametric equation of some surfaces
5.3.5 Metric of a surface
5.3.6 Theorem
5.4 Summary
5.5 Answers to self-learning exercises
5.6 Exercises
5.0 Objectives
After reading this unit you will be able to understand :
1. derivation of necessary and sufficient condition for the surface = F (, ) to be developablem
2. some important concepts such as parametric transformation,
3. oarametric equations of a few surfaces,
4. notion of metric of a surface.
5.1 Introduction
Last unit aimed to present an idea of developable surface. In this unit we would find a criterion
for the surface of the form = F (, ) to be developable. Sometimes this criterion proves to be quite
handy. Many surfaces can be written in parametric form and therefore parametric transformation and
103
parametric representations of many surfaces have been dealt with in this unit. The above
notions are useful in theory of metric of surface. A metric of a surface is the measure of an arc lying on
the surface.
5.2 Necessary and sufficient condition that a surface = F (, ) should represent
a developable surface
Let = F (, ) .....(5.2.1)
be a given surface. In order to seek condition that the given surface is developable, we have to use the
fact ‘‘if the surface is developable, then the tangent plane is same at all points of the generator and
contains only one parameter’’.
We now proceed as follows :
The equation to the tangent plane at the point P (x, y, z) on the given surface (5.2.1) is given by
z z
x y z 0, .....(5.2.2)
x y
z z
Denoting p ,q ,
x y
then (5.2.2) is reduced to
p + q – = px + qy – z
or p + q – = , .....(5.2.3)
where px + qy – z.
We will find the required condition making use of the tangent plane (5.2.3).
Necessary condition :
Let = F (, ) be the developable surface, then the tangent plane (5.2.3) involves only one
parametric t (say). Thus we can write p, q and as functions of t. Thus, let
p = f (t), q = g (t), = h (t) .....(5.2.4)
On elimination of t in the equation (5.2.4), p and can be written in terms of q, as given below
p = f1 (q), = f2 (q) .....(5.2.5)
p f q
Thus 1 f1 s.
x q x
or r = f1 s .....(5.2.6)
p 2 z q 2 z p
where r , s
x x 2 x xy y
p f q
and 1
y q y
104
q 2 z
or s f1 t , where t .....(5.2.7)
y y 2
From (5.2.6) and (5.2.7), on elimination of f1 , we obtain
r t = s2 .....(5.2.8)
Thus, the condition given by equation (5.2.8) is the required necessary condition.
Sufficient condition :
Let r t = s2
Then we will show that the tangent plane (5.2.3) involves only one parameter.
, q
For this we shall show that the Jacobian 0 implying that is function of parameter
x, y
(single parameter) q.
We now consider,
q
, q x x
. .....(5.2.9)
x, y q
y y
Since = px + qy – z,
p q z
Therefore p x y = x r + s y,
x x x x
p q z
and x y q = x s + t y. .....(5.2.10)
y y y y
Using the above expressions in the Jacobian (5.2.9), we find
, q rx sy s r s s s
x y = x (rt – s2) .....(5.2.11)
x, y sx ty t s t t t
, q
0 [ r t – s2 = 0]
x, y
is function of q.
Tangent plane (5.2.3) to the given surface (5.2.1) involves only one parameter.
Surface = F (, ) is developable.
Hence, we conclude that r t – s2 = 0 is the necessary and sufficient condition for the surface
= F (, ) to be developable, where r, t and s have their usual meanings.
Ex.1. Examine whether the surface z = y sin x is developable
Sol. Given that z = y sin x. .....(1)
The surface (1) is developable iff r t – s2 = 0.
105
2 z 2 z 2 z
Now r y sin x , t 0, s cos x
x 2 y 2 xy
Thus r t – s2 = 0 – cos2 x. .....(2)
The surface is developable if
r t – s2 = – cos2 x = 0
2n 1
x , n being integer..
2
2n 1
Hence, the surface is developable when x , n being integer..
2
Ex.2. Show that the surface z c x y is developable.
Sol. Given that z c x y .....(1)
2z 1 3 2 2 z 1
We compute r 2 y x , t 2
x y 3 2
x 4 y 4
2 z 1
s
xy 4 xy
Then, rt s 2 1 1 0 .....(2)
16 xy 16 xy
Therefore, the surface (1) is developable.
5.3 Metric of a surface
In the coming test, you will see that a surface has two family of curves on it. You are familiar
with the idea of arc length. A metric of a surface is the measure of the arc lengths of the curves on the
surface. In order to derive the formula for the metric, we need to go through some concepts and termi-
nology as follows :
We know that a curve is the locus of a point P (x, y, z) whose cartesian coordinates are func-
tions of single parameter t (say). On the same line, we define a surface as the locus of point P (x, y, z)
whose cartesian coordinates are functions of two independent parametric u and v (say). If r is the po-
sition vector of the point P (x, y, z). Then the surface in vectorial notation is represented as
r r u, v . .....(5.3.1)
5.3.1 Curves on a surface and curvilinear coordinates :
Let r r u, v be the surface. Then by keeping one of the parameters among u and v fixed
and varying the other we get a family of curves on the surface. If we keep u = u0 (constant), then v will
vary and the locus of the point p (x, y, z) as v varies would give a parametric curve called v-curve on
106
the surface. For different values of v, we get different v-curves. This constitutes a system of curves
u = constant. Similarly, we get a system of curves v = constant i.e. u-curves on the surface r r u, v .
Notes :
1. If u0, v0 be fixed values of u and v, then (u0, v0) is a point on the surface r r u, v
2. (u0, v0) is called curvilinear coordinates of the surface
3. Through every point of a surface, there passes one and only one curve of each system
4. No two curves of the same system interest and two curves of different system meet only
once.
5.3.2 Parametric transformation :
Let r r u, v be a surface whose parameters u and v be transformed to another set of
parameters u*, v* as given below
u* = u* (u, v), v* = v* (u, v) .....(5.3.2)
where u*, v* are single valued and derivable.
The above transformation is called proper if the Jacobian
u*, v *
0. .....(5.3.3)
u, v
r r
We, now denote r1 , r2 . .....(5.3.4)
u v
These partial derivatives are important since r1 and r2 have tangential directions to u-curves and
v-curves, respectively (in the sense of u and v increasing).
5.3.3 Regular and singular points :
Behaviour of the surface r r u, v in the neighbourhood of the point p (x, y, z) or p r is
closely dependent on r1 and r2 .
The point p r on the surface is called regular point or ordinary point, if
r1 r2 0 .....(5.3.5)
If r1 r2 0, then the point is called singular point or singularity of the surface.
5.3.4 Parametric equations of some surfaces :
This section pertains to equations of some surfaces in parametric form, you are advised to
remember these as it would help you in solving the questions.
(i) Sphere : The equation of a sphere with radius a and centre at the origin is
x = a sin cos , y = a sin sin , z = a cos
where and are parameters
107
Fig 5.1
(ii) Surface of revolution : Note that a surface of revolution is the surface which is generated
by revolving a plane curve about an axis in the plane of the curve.
As an illustration, let us consider a plane curve
z = f (y), x = 0 in YZ-plane.
Let p (0, w, z) be any point on this curve. Then obviously z = f (w).
Fig 5.2
Let us suppose that the curve is revolved about Z-axis, then the point p will traverse a circle in a
plane normal to Z-axis with the centre on the Z-axis (the axis of rotation).
Let c be the centre of the circle and p* be new position of the point p, then the point p* (x, y, z)
is obtained as
x = OL cos = CP* cos = w cos
y = w sin , z = f (w) .....(5.3.6)
where is the angle between XZ-plane and CP*LO.
108
Note that here w and are the parameters.
Note : On the above surface, we have two family of curves. The parametric curves u = con-
stant are called parallels and the curves = constant are called meridians.
Geometrically, the parallels are curves of intersection of surface of revolution and the planes
perpendicular to the axis of revolution. The meridians are the curve of section by the planes through the
axis of revolution.
(iii) Anchor sing : Anchor ring is a surface generated by revolution of a circle in a particular
setup.
Consider a circle of radius a and with center c on the Y-axis in YZ-plane. Let the centre c is
at a distance d from the origin, then the coordinates of any point P on the circle are (0, d + a cos ,
a sin ), where PCY = and OC = d.
Fig 5.3
When this circle is revolved about Z-axis, then it would give rise to the surface
r = ((d + a cos ) cos , (d + a cos ) sin , a sin ) .....(5.3.7)
known as anchor ring. Here and are parameters.
(iv) Helicoid : Helicoid is the surface generated by the screw motion of a curve about a fixed
line. In generation of a helicoid, a curve undergoes with two motions–curve is first translated through a
distance (say) parallel to the fixed line (the axis) and subsequently it is revolved through an angle
(say) about the axis such that is constant.
Let b (constant), then 2b is called the pitch of the helicoid.
As an illustration, let us consider a curve in XZ-plane given by
x = f (v), y = 0, z = g (v).
109
Let Z-axis be the axis of helicoid, then the equation of the helicoid is
r = (f (v) cos , f (v) sin , g (v) + b ) .....(5.3.8)
where v and are the parameters.
(v) Right helicoid : Right helicoid is generated by screw motion of a straight line about the
fixed line (axis) such that the straight line meets the axis at right angle.
Let X-axis be the straight line generating the right helicoid, then its equation is given by
r = (v cos , v sin , a ), .....(5.3.9)
a being the pitch
(vi) Right circular cone : A right circular cone is a locus of a variable straight line passing through
a fixed point (vertex) and making a constant angle (semi-vertical angle) with a line (axis) through the
vertex.
Fig. 5.4
As an illustration,
r = (v cos , v sin , v cot ) .....(5.3.10)
represents a right circular cone, whose vertex is at the origin, Z-axis being the axis of the cone and the
variable straight line lies in the YZ-plane as shown in the figure.
5.3.5 Metric of a surface : A metric of a surface is the measure of the arc lengths of the curves
on the surface.
Let r r u, v be a surface on which a curve u = u (t), v = v (t) lies. Let P and Q be two
neighbouring points on this curve such that the position vectors of P and Q are r and r dr , respec-
tively. Let A be a fixed point on the curve such that arc AP = s and arc PQ = ds where s is the measure
of arc length. Since P, Q are neighbouring points, therefore ds is the infinitesimal distance between these
two points.
110
Q (u + dv, v + dv)
dr dr ds
r+
P (u, v)
r
O
Fig. 5.5
Now r r
dr du dv
u v
r1 du r2 dv, .....(5.3.11)
dr dr
where r1 , r2 .
du dv
Since P and Q are neighbouring points therefore,
PQ = chord PQ i.e. ds dr
ds 2 dr 2 . .....(5.3.12)
Then, ds 2 r1 du r2 dv 2 .
2 2
r1 du 2 r2 dv 2 2r1 r2 du dv
ds 2 E du 2 2 F du dv G dv 2 , .....(5.3.13)
2 2
where E r1 , F r1 r2 , G r2 .
Equation (5.3.13) is quadratic differential form in du and dv, and is called metric or first
fundamental form. The quantities E, F and G are called first fundamental coefficients. Note that
the values of E, F and G vary, in general, for different points on the surface simply because these are
functions of surface parameters u and v. Alternatively we can say that the metric is the relation between
the differentials of the arc of the curve and curvilinear coordinates u, v. A metric is also referred to as
linear element.
Ex.1. Prove that for the curve x = r cos , y = r sin , z = 0, ds2 = dr2 + r2d 2
Sol. r xiˆ yjˆ zkˆ or r r cos , r sin , 0 .
Here r and are two parameters i.e. ‘‘u’’ r, ‘‘v’’ =
dr dr
r1 cos , sin , 0 2
; r r sin , r cos , 0 .
dr d
2
Here E r1 cos2 sin 2 1; F r1 r2 r cos sin r cos sin
111
2 2 2
2 2
G r2 r sin cos r .
Then ds2 = E du2 + 2F du dv + G dv2 = dr2 + r2 d2. Q u r , v
5.3.6 Theorem : The metric of a surface is invariant under parametric transformation.
Proof : Let r r u, v be a surface and the parameters (u, v) undergo parametric transforma-
tion as given below
u* = u* (u, v) ; v* = v* (u, v) .....(5.3.14)
* r r u r v
Now, r1
u * u u * v u *
u v
r1* r1 r2 .....(5.3.15)
u * u *
r r u r v
r2*
v * u v * v v *
r2* r1 u r2 v .....(5.3.16)
v * v *
From (5.3.14), u = u (u*, v*), v = v (u*, v*),
u u
then du du * dv * .....(5.3.17)
u * v *
v v
and dv du * dv *. .....(5.3.18)
u * v *
Let E* du*2 + 2F* du* dv* + G* dv*2 be the metric of the given surface for the parameters
(u*, v*). Then
E* du*2 + 2F* du* dv* + G* dv*2 r1* du *2 2r1* r2*du * dv * r2* dv *2
2 2
2
r1*du * r2*dv *
2
u v u u
r1 r2 du * r1 r2 dv *
u * u * v * v *
2
u u v v
r1 du * dv * r2 du * dv *
u * v * u * v *
2
r1 du r2 dv
r12 du 2 2r1 r2 du dv r22dv 2
= E du2 + 2F du dv + G dv2. .....(5.3.19)
Thus we have shown that if the parameters u and v are transformed to new set of parameters
u* and v*, then the metric does not change i.e., the metric is invariant.
112
Self-learning exercise-1
1. A point P r on a surface is a regular point if
(a) r1 r2 0 (b) r1 r 0 (c) r2 r2 0 (d) r1 r2 0
2. Parametric transformation u* = u* (u, v), v* = v* (u, v) is proper if
u * v * 2u * 2v * u*, v *
(a) (b) 0 (c) 0 (d) it is conviently defined.
u v u 2 v 2 u, v
3. With usual notations, r1 signifies for the surface r r u, v
(a) unit normal vector to the surface (b) normal vector to the skew surface
(c) normal vector to the envelope (d) None of these.
4. In the expression ds2 = E du2 + 2 F du dv + G dv2, F stands for.
12
(a) r1 r1 (b) r2 r2 (c) r1 r2 (d) None of these
5.4 Summary
In this unit you came across with the criterion of surface = F (, ) to be developable. This
criterion is an important tool to examine the surface being developable or screw. Further, you learnt that
a surface has two distinct family of curves on it. Before going to the core topic of metric of a surface,
you have learnt many essential concepts such as-parametric equation of some standard surfaces.
5.5 Answers to self-learning exercises
1. (d) 2. (c) 3. (d) 4. (d)
5.6 Exercises
1. Explain curvilinear equation of a curve lying on a surface
2. Derive a formula for metric of a surface
3. Find metric of a point P (x, y, 0) in XY-plane. [Ans. ds2 = dx2 + dy2]
4. Find metric of a point P (x, y, 0) in XY-plane where x = r cos , y = r sin
[Ans. ds2 = dr2 + r2 d2]
113
UNIT 6 : First, Second and Third Fundamental Forms,
Fundamental Magnitudes of Some Important
Surfaces, Orthogonal Trajectories, Normal
Curvature
Structure of the Unit
6.0 Objectives
6.1 Introduction
6.2 Fundamental forms
6.2.1 First fundamental form
6.2.2 Second fundamental form
6.2.3 Geometrical significance of second fundamental form
6.2.4 Weingarten equations
6.3 Fundamental magnitudes of some important surfaces
6.4 Orthogonal trajectories
6.4.1 Direction coefficients
6.4.2 Direction ratios
6.4.3 Orthogonal curves
6.4.4 Angle between two tangental directions on the surface
6.4.5 Family of curves and associated differential equations
6.4.6 Orthogonal trajectories
6.5 Normal curvature
6.5.1 Curvature of normal section
6.6 Summary
6.7 Self-learning exercises
6.8 Exercises
6.0 Objectives
In this unit you will study about :
1. fundamental forms of a surface,
2. fundamental magnitudes of some standard surfaces,
3. directors and orthogonal trajectories,
4. normal curvature.
114
6.1 Introduction
A surface is associated with there important forms which are infold quadratic differential expres-
sions in du, dv. Each form has its definite geometrical significance and serves as a founding stone in the
development of differential geometry. To distinguish, these are called first, second and third fundamental
forms. This unit aims to discuss the forms and the related properties.
6.2 Fundamental forms
6.2.1 First fundamental form :
In the last unit you have studied about the metric. The metric of a surface determines the first
fundamental form of the surface. Thus the quadratic differential from
E du2 + 2F du dv + G dv2
is called the first fundamental form and the quantities E, G, H are called the first order fundamental mag-
nitudes or first fundamental coefficients. Here it should be noted that since the quantities E, F, G depend
on u and v therefore, in general, they vary from point to point on the surface.
6.2.2 Second fundamental form :
The second fundamental from of a surface r r u, v is a quadratic differential in du and dv
together with the resolved parts of the second order partial derivatives of r (with respect to parameter u
and v) in the direction of the normal at the point p r on the surface.
Let r be position vector of any arbitrary point p on the surface r r u, v . Then we denote
2r 2r 2r
r11 2 , r12 , r22 2 .....(6.2.1)
u uv v
Let L, M, N be the resolved parts of r11 , r12 , r22 , respectively in the direction of normal vector
N̂ at the point p r , then
L r11 Nˆ , M r12 Nˆ , N r22 Nˆ .....(6.2.2)
The quantities L, M, N are called the second order fundamental magnitudes or second funda-
mental coefficients.
The quadratic differential form in du and dv
L du2 + 2M du dv + N dv2 .....(6.2.3)
is called the second fundamental form.
Note : Since the normal at the point p r on the surface is parallel to the vector r1 r2 ,
therefore unit normal vector N̂ at point p r is given by
115
ˆ r1 r2 r1 r2
N , where H r1 r2 . .....(6.2.4)
r1 r2 H
The second fundamental coefficients L, M, N given by (6.2.3) can be expressed alternatively as
follows
We know that the unit normal vector N̂ is parallel to the vector r1 r2 , therefore N̂ is perpen-
dicular to both r1 and r2 .
Thus, Nˆ r1 0, .....(6.2.5)
and Nˆ r2 0. .....(6.2.6)
On differentiating (6.2.5) with respect to u, we find
Nˆ ˆ r1
r1 N 0, or Nˆ1 r1 Nˆ r11 0, .....(6.2.7)
u u
Nˆ
2r r r1
where N1 and r11 2
u u u u u
or r11 Nˆ N1 r1
L N1 r1 , L r11 Nˆ .....(6.2.8)
Similarly on differentiating (6.2.6) with respect to v, are get
Nˆ ˆ r2
r2 N 0
v v
or N 2 r2 Nˆ r22 0
or N̂ r22 N 2 r2
or N N2 r2 N Nˆ r
22 .....(6.2.9)
Further on differentiating (6.2.5) and (6.2.6) with respectively with v and u respectively, we get
M N 2 r1 and M N1 r2 . .....(6.2.10)
The alternative expressions for L, M, N obtained above give rise to another expression for the
second fundamental form as follows :
We have, dr r1du r2 dv .....(6.2.11)
Nˆ Nˆ
dNˆ du dv
u v
dNˆ N1du N 2 dv, .....(6.2.12)
Nˆ Nˆ
where N1 , N2 .
u v
116
We compute dNˆ dr N1du N 2dv r1du r2 dv
N1 r1 du 2 N1 r2 N 2 r1 du dv N 2 r2 dv 2
= – [L du2 + 2M du dv + N dv2]. [using (6.2.8), (6.2.9), (6.2.10)]
Thus the second fundamental form
L du2 + 2m du dv + N dv2 dNˆ dr .....(6.2.13)
6.2.3 Geometrical significance of second fundamental form :
The following theorem entails the geometrical interpretation of the second fundamental form :
Theorem. Let P (u, v) and Q (u + du, v + dv) be two neighbouring points on the surface
r r u, v . Then second fundamental form L du2 + 2M dudv + N dv2 at the point P(u, v) is
twice the length of perpendicular from Q (u + du, v + dv) on the tangent plane at P (u, v), to a
second order approximation in du, dv.
Proof : Let r and r dr be the position vectors of two neighbouring points P (u, v) and
Q (u + du, v + dv) of the surface r r u, v .
Now, r dr r u du , v dv .
Then by Taylor’s series for two variables, we have
r r 1 2 r 2r 2r
r u du, v dv r u , v du dv 2 du 2 du dv 2 dv 2 ...
u v 2 u uv v
1
or
r dr r r1 du r2 dv r11du 2 2r12du dv r22dv 2 + higher order terms
2
1 2
or
dr r1 du r2 dv r11du 2r12du dv r22dv
2
2
.....(6.2.14)
(on neglecting higher order terms)
Let QA be perpendicular from Q on the tangent plane at the point P (u, v) to the surface
r r u , v , then we have
QA = Projection of PQ on normal vector Nˆ at P
1
2
N̂ dr Nˆ r1 du r2 dv r11du 2 2r12du dv r22dv 2
1
Nˆ r1 du Nˆ r2 dv Nˆ r11 du 2 2 Nˆ r12 du dv N̂ r22 dv 2
2
1 Nˆ is perpendicular to both r1 and r2
0 0 L du 2 2 M du dv N dv 2
2 Nˆ r1 0 Nˆ r2 and L r11 Nˆ etc.
2QA = L du2 + 2M du dv + N dv2 .....(6.2.15)
117
N
r = r (u, v)
Q
r
r+d
dr
O P
r
A
Fig. 6.1
6.2.4 Weingarten equations :
Let r r u, v be the surface, N̂ be the unit normal at a point p r on the surface then we
denote
Nˆ ˆ Nˆ
Nˆ1 , N2 .
u v
Here it is emphasized that N̂ is perpendicular to both Nˆ1 and Nˆ 2 , which means that the vectors
Nˆ1, Nˆ 2 are tangential to the surface. Therefore, the vectors Nˆ1 and Nˆ 2 , are spanned by the vectors
r1 and r2 and hence there exist scalars a, b, c, d such that
N̂1 a r1 b r2 .....(6.2.16)
N̂ 2 c r1 d r2 .....(6.2.17)
From (6.2.16), we find
r1 Nˆ1 a r1 r1 b r1 r2
or –L=aE+bF .....(6.2.18)
From (6.2.16), we find
r2 Nˆ 1 a r1 r2 b r2 r2
or – M = aF + bG .....(6.2.19)
Thus solving the system of linear equations given by (6.2.18) and (6.2.19), we get
FM GL FL EM
a 2
,b
EG F EG F 2
FM GL FL EM
or a 2
,b .....(6.2.20)
H H2
118
Using the values of a, b in (1), we get
FM GL FL EM
Nˆ1 2 r1 2 r2
H H
or H 2 Nˆ1 FM GL r1 FL EM r2 .....(6.2.21)
Similarly by taking the scalar product of (6.2.17) with r1 and r2 successively, and solving the
resultant linear system, we get the values of c and d as follows :
FN GM FM EN
c , d . .....(6.2.22)
H2 H2
Thus from (6.2.17) we get
H 2 Nˆ 2 FN GM r1 FM EN r2 . .....(6.2.23)
6.3 Fundamental magnitudes of some important surfaces
(i) Anchor ring :
The parametric equation of an anchor ring is
.....(6.3.1)
r b a cos cos , b a cos sin , a sin
Then we compute,
r
r1 a sin cos , a sin sin , a cos
r
r2 b a cos sin , b a cos cos , 0
2r
r11 2 a cos cos , a cos sin , a sin
2r
r12 a sin sin , a cos cos , 0
2r
r22 2 b a cos cos , b a cos sin , 0 .....(6.3.2)
Thus we have
E r12 a 2 sin 2 cos 2 a 2 sin 2 sin 2 a 2 cos 2
= a2 sin2 (cos2 + sin2 ) + a2 cos2
= a2 (sin2 + cos2 ) = a2
2
F r1 r2 0, G r22 b a cos .....(6.3.3)
2 2
H 2 EG F 2 a b a cos .....(6.3.4)
119
Fig. 6.2
i j k
r1 r2 a sin cos a sin sin a cos
b a cos sin b a cos cos 0
= – i (b + a cos ) cos cos + j (b + a cos ) sin cos
+ k – a (b + a cos ) sin cos2 + a sin (b + a cos ) sin2 .....(6.3.5)
ˆ r1 r2
Thus N cos cos , cos sin , sin
H
Hence L Nˆ r11 a, M Nˆ r12 0 and N Nˆ r22 b a cos cos .....(6.3.6)
(ii) Conoidal surface :
Let the surface of revolution be
r u cos v, u sin v, f v .....(6.3.7)
r
Then r1 cos v, sin v,0
u
r
r2 u cos v, u cos v, f
v
2r
r11 2 0, 0, 0 ,
u
2r
r12 sin v, cos v, 0
uv
2r
r22 2 u cos v, u sin v, f .....(6.3.8)
v
120
Therefore E r12 cos 2 u sin 2 v 1
F r1 r2 u cos v sin v u cos v sin v 0
G r22 u 2 sin 2 v u 2 cos 2 v f 2 u 2 f 2 .....(6.3.9)
H EG F 2 u 2
f 12 0 u 2 f 2 .....(6.3.10)
r r
Nˆ 1 2
f sin v, f cos v, u
.....(6.3.11)
H u 2 f 2
Further L Nˆ r11 Nˆ 0, 0, 0 0
M Nˆ r12
f sin v, f cos v, u
sin v, cos v, 0
u 2 f 2
f
u 2 f 2
f sin v, f cos v, u u cos v, u sin v, f
and N Nˆ r22
u 2 f 2
uf
.....(6.3.12)
u 2 f 2
(iii) Monge’s form surface :
The equation of surface given in the form z = f (x, y) is called Monge’s form.
Let the position vector of a current point on the surface be given by
r x, y , z , .....(6.3.13)
where z = f (x, y).
Since z in a function of x and y, therefore the equation (6.3.13) may be regarded as the para-
metric equation of the surface with parameters x and y.
Hence r1 1, 0, p , r2 0,1, q , r11 0, 0, t ,
r12 0, 0, s and r22 0, 0, t .....(6.3.14)
z z 2z 2 z 2z
where p ,q ,r 2 ,s ,t 2 .
x y x yx y
and suffixes 1 and 2 denote differentiation w.r. to x and y, respectively.
Therefore E r12 1 p 2 , F r1 r2 pq, G r22 1 q 2
H EG f 2 1 p 2 q 2 ,
121
and ˆ
N
r1 r1
p, q,1 . .....(6.3.15)
H 1 p2 q2
r
L Nˆ r11 ,
Also
1 p2 q2
s
M Nˆ r12 ,
2 2
1 p q
t
N Nˆ r22 .
.....(6.3.16)
1 p2 q2
6.4 Orthogonal trajectories
6.4.1 Direction coefficients :
Analogous to concept of direction coefficients in analytical geometry, we have the same notion
in differential geometry.
The discussion undertaken earlier has shown that at a point p r of the surface r r u , v ,
these exist three independent vectors r1 , r2 and Nˆ . Among these vectors, r1 and r2 lie in the tangent plane
to the surface at p r and the vector N̂ is along the normal direction at p r . Consequently, we
conclude that any vector at p r , can be expressed uniquely as a linear combination of r1 , r2 and Nˆ .
Let b̂ be any vector at p r , then we can write
b r1 r2 Nˆ .....(6.4.1)
where , and are scalars. Here and are called tangential components of b and is called nor-
mal component of b .
Here, we are concerned with tangential vectors i.e. the vector in which normal component is
zero.
Let b r1 r2 be a vector along the tangent at p r .
2 2
Then b r1 r2
2 2
2 r1 2 r1 r2 2 r2
E 2 2 F G 2
12
b E 2 2 F G 2 .....(6.4.2)
Eqn. (6.4.2) provides magnitude of the tangential vector b .
122
Direction coefficients, determine the direction, and these are determined by using a unit vector.
Let e lr1 mr2 be a unit vector in a tangential direction at p r of the surface, then the
components l and m are called direction coefficients. Since e is unit vector, therefore
2 2 2 2
e 1 lr1 mr2 l 2 r1 m2 r2 2lm r1 r2
12
e 1 El 2 2lmF Gm2 .....(6.4.3)
We know that a metric on the curve is given by
ds2 = E du2 + 2F du dv + Gdv2
2 2
du du dv dv
or 1 E 2 F G .....(6.4.4)
ds ds ds ds
On comparing the equation (6.4.3) and (6.4.4), we find that
du dv .
l and m
ds ds
du dv
Hence, we may conclude that and are the actual direction coefficients of the tangent at
ds ds
the point p r to the curve (u, v) = c lying on the surface r r u, v .
Note : If (l, m) constitute the direction coefficients of the direction, then the direction coeffi-
cients of the direction opposite to the given direction are (– l, – m).
6.4.2 Direction ratios :
Direction ratios are the quantities which are proportional to the direction coefficients.
Let (, ) be the numbers which are proportional to the direction coefficients (l, m). Then we
may find expressions for and as follows :
From (6.4.2), we have e lr1 mr2 .....(6.4.5)
b r1 r2
e
b 12 .....(6.4.6)
E 2 2 F G 2
On comparing (6.4.5) and (6.4.6), we find that
l ,m
2 12 12
E 2
2 F G E 2
2 F G 2
12
or E 2 2 F G2 l
.....(6.4.7)
12
and E 2
2 F G 2
m
Equation (6.4.7) provides relation between direction coefficients (l, m) and associated direction
ratios (, ).
123
The above expressions are very useful in determining the directions and are readily used.
To illustrate, let us consider a surface r r u, v . Recall that we have two distinct family of curves
(u-curves and v-curves) on the surface. We know that r1 and r2 are the vectors along the tangents to
u-curves and v-curve respectively.
Then we have
r1 1 r1 0 r2 .....(6.4.8)
= 1, = 0.
1
Consequently, l , m 0 (using (6.4.7))
E
r1
Thus the unit vector in the tangential direction to u-curve is .....(6.4.9)
E
r2
Similarly, the unit vector along the tangential direction to the v-curve is .....(6.4.10)
G
6.4.3 Orthogonal curves :
The parametric curves u = const., v = const. are said to be orthogonal if they intersect at right
angle i.e. the angle between their tangents at the point of intersection of the curves is 90°. Before finding
the condition for orthogonality, we find formula for the angle between parametric curves.
Let the curves u = const., v = const. do intersect in a point p r at an angle (0 ).
r1 r2 F
Then cos .....(6.4.11)
r1 r2 EG
r1 r2
and sin H .....(6.4.12)
r1 r2 EG
H
and obviously, tan .....(6.4.13)
F
From the above we see that when = 90°, then cos and a is zero i.e. F = 0. Conversely if
F = 0 then = 90°. Thus we have F = 0 to be the necessary and sufficient condition for the parametric
curves to be orthogonal.
6.4.4 Angle between two tangential directions on the surface :
Let be the angle between two tangential directions (l1, m1) and (l2, m2) on the surface
r r u , v , where li, mi, l = 1, 2 are actual directions coefficient. Let tˆ1 and tˆ2 denote the unit vectors
in the tangential directions, then we have,
tˆ1 l1r1 m1r2 ,
tˆ2 l2 r1 m2 r2 ,
124
ˆ
Then tˆ1 tˆ2 t1 tˆ2 cos .....(6.4.14)
11cos tˆ1 tˆ2 and unit vectors therefore tˆ1 tˆ2 1
cos tˆ1 tˆ2 .....(6.4.15)
or cos l1r1 m1r2 l2 r1 m2 r2
l1 l2 r1 r1 l1m2 m1l2 r1 r2 m1m2 r2 r2
l1 l2 E l1m2 m1l2 F m1m2 G .....(6.4.16)
Similarly, sin tˆ1 tˆ2 A B | A | | B | sin nˆ
l1r1 m1r2 l2r1 m2r2
l1m1 m1l2 r1 r2 r1 r1 0, r2 r2 0
l1m1 m1l2 H H r1 r2 .....(6.4.17)
sin
Thus, tan
cos
H l1m2 m1l2
.....(6.4.18)
l1l2 E l1m2 m1l2 F m1m2G
The angle between tangential directions can also be determined in terms of direction ratios.
Let (1, 1) and (2, 2) be the direction ratios corresponding to direction coefficients (l1, m1)
and (l2, m2), then we have
1 1
l1 12
, m1 12
E 12 2 F 11 G 22 El 1
2
2 F 11 G12
2 2
l2 12
, m2 12
.....(6.4.19)
E 22 2 F 2 2 G 22 E 22 2 F 2 2 G22
Using the above we find
1 2 E F 1 2 1 2 1 2G
cos .....(6.4.20)
12 12
E 12 2 F 11 G12 E 22 2 F 22 G 22
H 1 2 1 2
sin .....(6.4.21)
12 12
E 12 2 F 11 G12 E 22 2 F 22 G 22
and consequently,
125
H 1 2 1 2
tan . .....(6.4.22)
1 2 E F 1 2 1 2 1 2G
From the above discussions, we conclude that the two directions be at right angles (orthogonal)
i.e. = /2 if
12 E + (12 + 12)F + 12 G = 0 .....(6.4.23)
or l1l1 E + (l1m2 + m1l2) F + m1m2G = 0. cos / 2 0 .....(6.4.24)
Ex. Prove that the equation Edu2 – Gdv2 = 0 denote the curves bisecting the angles be-
tween the parametric curves u = constant, v = constant on a surface r r u, v
du dv
Sol. Let 1 be the angle between the direction , , and the u-curve (i.e. v = const.).
ds ds
Let (l2, m2) denote the direction (tangent) to the u-curve, then
1
l2 , m2 0 .
E
Then, ‘‘cos 1 = l1l2 E + (l1m2 + m1l2) F + m1m2 G’’ gives
du 1 dv 1 du dv
cos 1
ds E
E 0
ds E
F 0 l1 ds , m1 ds
1 du dv
E F .
E ds ds
du dv
Let 2 be angle between the direction , , and v-curve (i.e. u = const.) and we sup-
ds ds
pose that (l2, m2) denote the direction of v-curve, then
1
l2 0, m2 .
E
Then, we have cos 2 = l1l2 E + (l1m2 + l2m1) F + m1m2 G
du 1 dv 1
0 0 F G
ds G ds G
1 du dv
F G .
G ds ds
du dv
Given that the direction , bisects the angle between the parametric curves, therefore
ds ds
we must have,
cos 1 = cos 2. .....(3)
Here + and – signs correspond to internal and external bisector respectively.
126
Putting the values of cos 1 and cos 2 from the equation (1) and (2) in equation (3), we get
1 du dv 1 du dv
E F F G
E ds ds G ds ds
du dv du dv
or G E F E F G
ds ds ds ds
2 2
du dv du dv
or GE F EF G
ds ds ds ds
or GE2 du2 + GF2 dv2 + 2EFG dudv = EF2 du2 + EG2 dv2 + 2EFG dudv
or E (EG – F2) du2 – G (EG – F2) dv2 = 0
or (EG – F2) (E du2 – G dv2) = 0
E du2 – G dv2 = 0. [ EG – F2 > 0].
6.4.5 Family of curves and associated differential equations :
For a surface r r u, v .....(6.4.25)
an implicit relation of the form (u, v) = c .....(6.4.26)
give rise to family of curves on the surface (6.4.25), where c is a parameter, (u, v) is single valued
function of u, v and (u, v) possesses continuous derivatives , which don’t vanish simultaneously..
u v
By assigning different values to c, we have different members of the family of curves given by (6.4.26).
Here it should be noted that through every point of the surface, there passes one and only one curve of
the family of curves given by (6.4.26).
The family of curves can be expressed in the form of differential equation as explained below.
Consider equation (6.4.26) as family of curves on the surface (6.4.25). Differentiating (6.4.26),
we get
du dv 0, .....(6.4.27)
u v
1 du + 2 dv = 0,
where 1 and 2 .
u v
Obviously on integrating 1 du + 2 dv = 0 we would get (6.4.26), we assume
1 = P (u, v) and 2 = Q (u, v) .....(6.4.28)
Thus equation (6.4.27) becomes
P du + Q dv = 0 .....(6.4.29)
The equation (6.4.29) constitutes a differential equation of the family of curves (6.4.26).
127
From (6.4.29), we have
du dv
.....(6.4.30)
Q P
equation (6.4.30) emphasises that tangent at (u, v) to the curve has (– Q, P) as its direction ratios.
6.4.6 Orthogonal trajectories :
Trajectory : A trajectory of the given family of curves is a curve which intersects every member
of the family of curves by following some definite law. If the trajectory intersects the members of the
family of curves at a constant angle , then it is called -trajectory. If = 90, then it is called an
orthogonal trajectory. Following proposition underlines the fact that every family of curves on a surface
has orthogonal trajectories.
Differential equation of orthogonal trajectory :
Let (u, v) = c .....(6.4.31)
be a family of curves on the surface r r u, v . .....(6.4.32)
Recall that here has continuous derivatives , which don’t vanish simultaneously. On
u v
differentiating (6.4.31), we have
du dv 0 or 1 du + 2 dv = 0, .....(6.4.33)
u v
where 1 2 .
u v
We prescribe 1 = P (u, v), 2 = Q (u, v)
Then (6.4.33), becomes
P du + Q dv = 0 or du dv . .....(6.4.34)
Q P
Then (– Q, P) are direction ratios of tangent at any point (u, v) of a member of family of curves
given by (6.4.31).
Let (u, v) be direction ratios of the tangent at the point (u, v) of a member of orthogonal
trajectories of (6.4.31).
Recalling that the directions (u1, v1) (u2, v2) are orthogonal if
E u1u2 + F (u1 v2 + v1 u2) + G v1v2 = 0 .....(6.4.35)
We have for the present case,
E (– Q) u + F (– Q v + Pu) + G Pv = 0
On simplification, it reduces to
(FP – EQ) u + (GP – FQ) v = 0 .....(6.4.36)
128
Equation (6.4.36) represents the differential equation of the orthogonal trajectories of the family
of curves.
Note that equation (6.4.36) is integrable simply because (FP – EQ) and (GP – FQ) (the coeffi-
cients of u, v) in (6.4.36)} are continuous.
On integrating (6.4.36), we get the equation of orthogonal trajectory.
Theorem. On a given surface, a family of curves and their orthogonal trajectories can
always be chosen as parametric curves.
Proof : We know that the differential equation
P du + Q dv = 0 .....(1)
where P, Q are functions of u and v, represents a family of curves on the surface r r u, v
Let (u, v) = c1 (c1 being constant) .....(2)
be the solution of (1). Then P , Q , (where 0)
u v
As discussed earlier, we know that the differential equation
(FP – EQ) u + (GP – FQ) v = 0 .....(3)
gives the orthogonal trajectories of the family of curves given by (1).
Let (u, v) = c2 (c2 being constant) .....(4)
is the solution of (3), then we may have
FP – EQ ,
u
GP – FQ (where 0) .....(5)
v
In order to prove the theorem, we will show that
,
0. .....(6)
u, v
That means that the two family of curves (u, v) = c1 and (u, v) = c2 are mutually indepen-
dent.
, u u 1 P FP EQ
Thus we examine,
u, v Q GP FQ
v v
1
P GP FQ Q FP EQ
1 2
EQ 2 FPQ GP 2 0 . .....(7)
129
,
Thus 0 ensures that a family of curves and their orthogonal trajectories can always
u, v
be chosen as parametric curves. Since is independent of , hence proper transformation u* =
(u, v), v* = (u, v) transforms the given family of curves and their orthogonal trajectories into the two
families of parametric curves.
Double family of curves :
We have seen that the equation P du + Q dv = 0 give rise to a family of curves on the surface
r r u, v . Similarly, the quadratic equation
P du2 + 2Q du dv + R dv2 = 0 .....(1)
where P, Q, R are continuous functions of the parameters u and v and do not vanish together and are
such that Q2 – PR > 0, then the equation (1) give rise to two distinct family of curves as illustrated
below.
Equation (1) can be written as
2
du du
P 2Q R 0 .....(2)
dv dv
du
Equation (2) is quadratic in and has two solutions
dv
du Q Q 2 PR
.....(3)
dv P
which infact correspond to directions of the tangents to two distinct family of curves.
l1 l2
Let , be the two directions, then
m1 m2
l1 Q Q 2 PR l2 Q Q 2 PR
, .....(4)
m1 P m2 P
Thus we have l1 l 2Q l1 l2 R
2 , . .....(5)
m1 m2 P m1 m2 P
If is the angle between these two directions then
2 H Q 2 PR
tan .....(6)
ER 2 FQ GP
Obviously these directions are orthogonal if = 90 i.e.,
ER – 2 PQ + GP = 0. .....(7)
Note that (7) is the necessary and sufficient condition for the curves described by (1) to be
orthogonal.
130
Ex.1. Examine whether the parametric curves
x = b sin u cos v, y = b sin u sin v, z = b cos u
on a sphere of radius b constitute an orthogonal system.
Sol. The given sphere is
r = (b sin u cos v, b sin u sin v, b cos u) .....(1)
The parametric curves would constitute an orthogonal system if F = 0 i.e. r1 r2 0 ,
we compute r1 = (b cos u cos v, b cos u sin v, – b sin u)
r2 = (– b sin u sin v, b sin u cos v, 0)
F r1 r2 = – b2 cos u sin u cos v sin v + b2 cos u sin u sin v cos v
= 0. .....(2)
Thus the given parametric curves are orthogonal.
Ex.2. On the paraboloid x2 – y2 = z, find the orthogonal trajectories of the sections by the
planes z = constant.
Sol. Let x = u and y = v, then for the given paraboloid we have u2 – v2 = z.
Thus r = (u, v, u2 – v2) .....(1)
is the parametric equation of the given paraboloid.
The curves of section by the planes z = constant on (1) are given by
u2 – v2 = constant [note] .....(2)
Differentiating (2), we get
u du – v dv = 0. .....(3)
Thus, we have to find out the equation of trajectories orthogonal to family of curves given
by (3).
Recall that if P du + Q dv = 0 .....(4)
is the given family of curves, then its orthogonal trajectories are given by
(PF – QE) u + (PG – QF) v = 0 .....(5)
where (u, v) are direction ratios of the orthogonal trajectories.
Comparing (3), (4) we find
P = u, Q = – v .....(6)
r1 1, 0, 2u , E r12 1 4u 2
. .....(7)
r2 0,1, 2v , F r1 r2 4uv, G r2 2 1 4v 2
Using (6) and (7) in equation (5), we get the orthogonal trajectories as
(– 4u2v + v + 4u2v) u + (u + 4uv2 – 4uv2) v = 0
or vu + 4v = 0
131
or (uv) = 0
or uv = constant
or xy = constant. [ x = u, y = v] .....(8)
Thus the hyperbolic cylinders xy = constant are the required orthogonal trajectories.
Ex.3. Let v2 du2 + u2 dv2 be the metric of a given surface. Then find
(i) The family of curves orthogonal to the curves uv = constant
(ii) The metric corresponding to the new parameters so that these two families are para-
metric curves.
Sol. (i) Let s represent the arc length on the given surface. Then as given, we have
ds2 = v2 du2 + u2 dv2 .....(1)
Thus, E = v2, F = 0, G = u2
We have to find family of curves orthogonal to the curves
uv = constant. .....(2)
Differentiating (2), we get
du u
u dv + v du = 0 . .....(3)
dv v
Hence, the direction ratios of the tangent to the curve (2) are (– u, v).
Let (du, dv) be the direction orthogonal to the direction, then the condition of orthogonality i.e.
1 2 E + (1 2 + 2 1) F + 1, 2 G = 0.
or E (– u) du + G v dv = 0, Q F 0
du dv
– uv2 du + u2v dv = 0 .....(4)
u v
On integration of equation (4), we get
u
log u = log v + log (const) const. .....(5)
v
Hence equation (5) is the equation of the orthogonal trajectory of family of curves (2).
(ii) If the family of curves (2) and their orthogonal trajectories (5) are taken as parametric curves,
then the new parameters u* and v* are given by
u v*
u* and v* = uv u2 = u* v* and v2 . .....(6)
v u*
r r u r v v* u*
Now r1* r1 r2 . .....(7)
u * u u * v u * 2u vu *2
r r u r v u * 1
and r2 * r1 r2 . .....(8)
v * u v * v v * 2u 2vu *
132
The new coefficients E*, F*, G* are given by
2 2
E* r1 *2 u * r12 v * r22 , r1 r2 F 0
4u 2 4v 2u *2
u *2 v *2 1 v *2
E G ,
4u 2 4v 2u *2 2 u *2
F * r1 * r2 * 0,
2
u* 1 1
G* r2*2 r1 r2 .....(9)
2u 2vu * 2
Therefore, the metric referred to new parameters u*, v* is given by
ds*2 = E* du2 + 2F* du* dv* + G* dv*2
1 v *2 1
2
du *2 dv *2 .
2 u* 2
Ex.4. Show that the curves du2 – (u2 + c2) dv2 = 0 from an orthogonal system on the right
helicoid
r = (u cos v, u sin v, cv)
Sol. The given surface is r = (u cos v, u sin v, cv) .....(1)
2 r 2
Then E r1 cos 2 v sin 2 v 1
u
F r1 r2 cos v, sin v, 0 u sin v, u cos v, c
= – u cos v sin v + u cos v sin v + 0 = 0
2 r 2
G r2 u sin v, u cos v, c
v
= u2 (sin2 v + cos2 v) + c2 = u2 + c2 .....(2)
Recall that the two family of curves given by the quadratic differential equation
P du2 + 2Q du dv + R dv2 = 0 .....(3)
constitute an orthogonal system if and only if
ER – 2FQ + GP = 0 .....(4)
The given curves are du2 – (u2 + c2) dv2 = 0 .....(5)
On comparing (3) and (5), we get
P = 1, Q = 0, R = – (u2 + c2). .....(6)
Using these values and the values of E, F, G computed above in the equation (4) we see that
(4) is identically satisfied
1 [– (u2 + c2)] – 0 + (u2 + c2)1 = 0.
133
Ex.5. Show that on a right helicoid, the family of curves orthogonal to the curves
u cos v = constant is the family (u2 + a2) sin2 v = constant
Sol. Let the given right helicoid be
r = (u cos v, u sin v, cv) .....(1)
Then from previous example
E = 1, F = 0, G = u2 + c2 .....(2)
The given family of curves is
u cos v = constant. .....(3)
On differentiating equation (3), we get
cos v du – u sin v dv = 0. .....(4)
Equation (3) implies that the direction ratios of the tangent to given curve at the point (u, v) is (u
sin v, cos v). Let (du, dv) be direction of the required orthogonal curves. Then by the condition of or-
thogonality we have
E (u sin v) du + F (u sin v dv + cos v du) + G cos v dv = 0 .....(5)
Putting the values E = 1, F = 0, G = u2 + c2 in (4), we get
u sin v du + (u2 + c2) cos v dv = 0
u du cos v dv
or 0
u2 c2 sin v
On integration, it gives
1
2
log u 2 c 2 log sin v constant.
or log (u2 + c2) 2 log sin v = constant .....(6)
Equation (6) represents the required orthogonal curves family.
6.5 Normal curvature
Before embarking on the idea of normal curvature, we first have to go through some basic things
as follows :
(i) Plane section of a surface :
A plane drawn through a point on a surface cuts the surface, in general, in a plane curve.
This plane curve is called the plane section of the surface.
(ii) Normal section of the surface :
If the plane section of the surface is such that it contains the normal to the surface at that
point, the section is called the normal section. The section, which is not normal section
is called the oblique section.
134
(iii) Curvature at a point on a given surface :
The curvature at a point on a given surface is closely related to the plane section at the
point.The curvature at a point P (u, v) of the given surface r r u, v in a direction
(du, dv) is the curvature of the plane section (curve) of the surface which passes through
the point P (u, v) and contains the direction (du, dv).
6.5.1 Curvature of normal section :
Let r r u, v be the given surface and P (u, v) be any point on it. Let kn denotes the curva-
ture of the normal section. By convention, we presume that the sense of the unit principal normal to the
curve i.e. n̂ and the unit surface normal i.e., N̂ are the same. Further note that kn is considered positive
when the curve is concave on the side towards which N̂ points out.
We, now, have
dt
r kn nˆ kn Nˆ Here nˆ Nˆ
ds
Therefore,
k n N̂ r . .....(6.5.1)
Again, we know that
dr r du r dv du dv
r r1 r2
ds du ds dv ds ds ds
du dv
r1 u r2 v, where u ds , v ds .....(6.5.2)
Again differentiating with respect to s, we get
d
r r1 u r2 v
ds
r r
r dr1 r dr2
r1 u u r2 v v
ds ds
r r r r
r r r1 du r1 dv r2 du r2 dv
r1 u r2 v
u v
u ds s ds u ds v ds
r r r r r r
r1 u r2v r11u 2 r12u v r21uv r22v2
r1 u r2v r11 u 2 2r12uv r22v2 . .....(6.5.3)
Thus, we have
k n N̂ r
Nˆ r1 u r2 v r11 u 2 2r12uv r22 v2
135
Nˆ r1 u Nˆ r2 v Nˆ r11 u 2 2 Nˆ r12 u v Nˆ r22 v2
0 0 Lu 2 2Muv N v2 .....(6.5.4)
Nˆ r1 Nˆ r2 0 and Nˆ r11 L etc.
2 2
du du dv dv
Thus, k n L 2M N
ds ds ds ds
L du 2 2 M du dv N du 2
or kn
ds 2
L du 2 2 M du dv N du 2
or kn .....(6.5.5)
Edu 2 2 F du dv G dv 2
[ ds2 = E du2 + 2F du dv + G dv2]
Equations (6.5.5) provides the curvature of the normal section, parallel to the direction (du, dv)
in terms of fundamental magnitudes.
Notes :
1. kn depends purely on the direction (du, dv) of the curve drawn on the surface and the quan-
tities E, F, G, L, M, N which are determined at given point P.
This reasoning helps us to conclude that all the curves tangent to the same direction on the
given surface have the same normal curvature, since normal curvature at a point on the
surface is the property of the surface which depends on the direction at the point on the sur-
face.
2. The reciprocal of kn is called the radius of normal curvature, and is denoted by n.
6.6 Summary
In this unit you came across with the notion of the fundamental forms which are quadratic equa-
tion in du and dv. Each form has its definite geometrical significance. Further the directions on the sur-
face where explained and the criterion was extended to the orthogonal trajectories on the surface. Lastly
the notion of normal curvature was given.
6.7 Self-learning exercises
1. Write down first and second fundamental forms
2. Define direction coefficients.
3. Define direction ratios.
4. What are orthogonal trajectories ?
5. Define oblique section and normal section.
136
6.8 Exercises
1. Compute the fundamental magnitudes for the surface
r = (u cos v, u sin v, f (u) + cv).
2. Prove that the curves du2 – (u2 + c2) dv2 = 0 form an orthogonal system on the right helicoid
r = (u cos v, u sin v, cv).
3. Compute E, F, G, H for the surface
(i) x = u, y = v, z = u2 – v2,
(ii) 2z = ax2 + 2hxy + by2.
4. Prove that : HNˆ N1 M r1 L r2 .
137
Unit 7 : Meunier’s Theorem, Principal direction and
Principal curvatures, First Curvature, Mean
Curvature, Gaussian Curvature, Umbilics, Radius
of Curvature of any Normal Section at an Umbilic
on z = f (x, y). Radius of Curvature of a given
Section through any Point on z = f (x, y), Lines of
Curvature
Structure of the Unit
7.1 Objectives
7.2 Introduction
7.3 Definitions
7.4 Curvature of normal section (i.e., Normal curvature)
7.4.1 Radius of normal curvature
7.4.2 Normal curvature
7.5 Meunier’s Theorem
7.5.1 Important result
7.5.2 Self-learning exercise-1
7.5.3 Illustrative example
7.6 Principal directions and principal curvatures
7.6.1 Definitions
7.7 Equations giving the principal directions at a point of surface and to derive the differential
equation of the principal section.
7.8 There are two principal directions at every point on a surface which are mutually orthogonal.
7.9 Umbilics
r r
7.10 The equation giving the principal curvature at a point A (u, v) of the surface r r (u , v ) .
7.11 Some important definitions
7.12 Radius of curvature at an umbilic on the surface z = f (x, y).
7.12.1 Self-learning exercise-2
7.12.2 Illustrative examples
138
7.13 Radius of curvature of a given section through any point of a surface z = f (x, y).
7.14 Lines of curvature
7.14.1 There are two systems of lines of curvature
7.14.2 To find the differential equation of lines of curvature at point (u, v) of the surface
r r
r r (u , v) .
7.14.3 Illustrative examples
7.14.4 Self-learning exercise-3
7.15 Summary
7.16 Answers to self-learning exercises
7.17 Exercises
7.1 Objectives
This unit provides a general overview of the following and after reading this unit you will be able
to learn
1. about Meunier’s theorem,
2. about Principal direction and Principal curvature of the surfaces,
3. about first curvature, Mean curvature and Gaussian curvature,
4. about Umbilics, radius of curvature of any normal section at an umbilics z = f (x, y), radius of
curvature of a given section through any point on z = f (x, y),
5. about lines of curvatures.
7.2 Introduction
In this unit we shall study local non-intrinsic properties of a surface. We shall also study curva-
ture of surfaces, plane section of surfaces and oblique section of surfaces. After that we shall establish a
relationship between curvature of normal section (n) and curvature of oblique section (), which is known
as Meunier’s theorem. In the end of the unit we shall study about lines of curvature.
7.3 Definitions
(i) Intrinsic property
Property of a surface deducible from the metric alone, without using the surface equation
r r
r r (u , v) is called an intrinsic property..
(ii) Plane section of a surface
A plane drawn through a point A of a given surface intersects it in a plane curve, known as the
r r
plane section of the surface. In Fig. 7.1 ABC and ADE are two plane section of the surface r r (u , v ) .
139
(Unit surface normal)
N=n
(Unit principal to normal section)
n
Normal section
D B
E
Oblique C
section
Surface
r = r (u, v)
A
t
(Unit tangent
vector)
Fig. 7.1
(iii) Normal section
The plane is so drawn that it contains the normal N̂ to the surface at the point (say A), then the
curve of intersection is called normal section. In the Fig. 7.1 the curve bounding shaded area ABC is the
normal section. Thus the normal section is parallel to the normal N̂ to the surface.
(iv) Oblique section
The plane is so drawn that it does not contain the normal N̂ to the surface at the point (say A),
then the curve of intersection is called oblique section.
Note : There exist infinite number of planes of normal sections through the principal normal at
point A, but there will be only one such plane of normal section having directions (du, dv).
Principal normal n̂ for normal section is parallel to surface normal N̂ and principal normal n̂
for oblique section is inclined at angle to surface normal N̂ .
We adopt the convention that vector n̂ has the same direction as that of vector N̂ , and with this
convention n̂ Nˆ .
(v) Curvature at a point on a given surface
Let A be a point with position vector r u, v on the surface r r u, v the normal section at A
in the direction (dn, dv) is equal to the curvature at A of the normal section at A parallel to the direction
(dn, dv) of tˆ .
140
(vi) Fundamental Magnitudes
r r
Let r r be the equation of the surface and let 1
u , v r , r2 then E r12 r1 r1 ,
u v
2
F r1 r2 , G r2 r2 r2 are called first order fundamental magnitudes and if N̂ be unit normal vector
at A r , then L r11 Nˆ , M r12 Nˆ and N r22 Nˆ are called second order fundamental magni-
2
r 2r 2r
tudes, where r11 , r12 , r22 2 .
u 2 uv v
7.4 Curvature of section (i.e., normal curvature)
Let r r u, v be the equation of a given surface and A (u,v) is any point on the surface.
Here we assume that Nˆ nˆ .....(7.4.1)
Let n represents the curvature of normal section, which will be positive when the curve is con-
cave on side towards which N̂ points out.
Now r tˆ dr .....(7.4.2)
ds
where s is arc length.
Again differentiating, we have
dtˆ
r n nˆ
ds
[ Nˆ nˆ from equation (1)]
or r n Nˆ .....(7.4.3)
Taking dot product by N̂ , we get
ˆ ˆ
N̂ r n N N
N̂ r n Nˆ Nˆ 1 .....(7.4.4)
Also we know that
dr r u r v
r
ds u s v s
du dv du
r1 r2 r1u r2 v (where u etc.) .....(7.4.5)
ds ds ds
Differentiating this relation again with respect to s, we have
2
r u d r1 u r v d r2 v ( u d u etc.)
r 1 2
ds ds ds 2
r1 du r1 dv r2 du r2 dv
or r r1u r2 v u v
u ds v ds u ds v ds
r1u r2v r11u2 r12uv r21uv r22v2 .....(7.4.6)
141
where r12 r1 , r
r11 1 etc.
v u
Now taking dot product by N̂ , we get
2 2 ˆ
r Nˆ r1u r2 v r11u r12u v r21u v r22v N
r1 Nˆ u r2 Nˆ v r11 Nˆ u2 r12 Nˆ u v
r21 Nˆ u v r22 Nˆ v2 .....(7.4.7)
But we know that
r1 Nˆ 0, r2 Nˆ 0 .....(7.4.8)
( unit normal vector N̂ is to both direction vectors of the tangents r1and r2)
r11 Nˆ L
and r12 Nˆ M .....(7.4.9)
r22 Nˆ N
(The resolved parts of r11, r12, r22 in the direction of normal to the surface are L, M, N called
second order fundamental magnitudes)
Using values from equation (7.4.8) and (7.4.8) in equation (7.4.7), we get
2 2
r Nˆ n Lu 2Muv Nv
2 2
du du dv dv Ldu 2 2Mdu dv Ndv 2
or n L 2 M N
ds ds ds ds ds 2
Ldu 2 2Mdu dv Ndv 2
or n .....(7.4.10)
Edu 2 2 Fdu dv Gdv 2
( ds2 = Edu2 + 2Fdu dv + Gdv2, the first fundamental form)
Equation (7.4.10) gives the curvature of the normal section, usually called normal curvature par-
allel to the direction (du, dv) in terms of the fundamental magnitudes.
Remark : Since n depends only on the direction (du, dv) of the curve drawn on the surface,
as fundamental magnitudes E, F, G, L, M, N are determined by the given point A. So normal curvature
at a point on a surface is a property of the surface which depend on the direction at the point on the
surface. Hence all curves tangent to the same direction on a surface have the same normal curvature.
7.4.1. Radius of normal curvature
Reciprocal of the normal curvature (n) is called the radius of normal curvature and it is denoted
by n i.e.
1
n . .....(7.4.11)
n
142
7.4.2. Normal curvature definition
Let A (u, v) be a point on the surface r r u, v . The normal curvature at A in the direction
(du, dv) is equal to the curvature of the normal section at A, parallel to the direction (du, dv).
7.5 Meunier’s Theorem (or Meusnier’s theorem)
Statement : If and n are the curvatures of oblique and normal sections through the same
tangent line and be the angle between these sections, then n= cos .
Proof : Refer Fig. 7.1, let tˆ be the tangent vector to the normal section of the given surface.
Let ADE be the oblique section of the surface by a plane through tˆ . Here N̂ is the surface normal
which is also principal normal of the normal section. Let n̂ be the principal normal of the oblique sec-
tion, then we have
cos n̂ Nˆ , .....(7.5.1)
because is the angle between the planes of sections as shown in the fig. 7.1
But, if is the curvature of oblique section
then, for any section
r n̂ . .....(7.5.2)
Now taking dot product by N̂ , we have
r Nˆ n̂ Nˆ
= cos [by equation (7.5.1)] .....(7.5.3)
Now r Nˆ = normal curvature at A in the direction
(du, dv) curvature of the normal section at A parallel to be direction
(du, dv) = n
by equation (7.5.3),
n r Nˆ cos
n cos .
Hence Proved.
7.5.1. Important result
If a sphere is described with n as diameter then all centers of curvature lie on this sphere, pro-
vided unit tangent vector tˆ is the same.
Proof : Let C be center of oblique section and Cn that of normal section for all planes contain-
ing tˆ , as shown in the Fig. 7.2. From figure, we have
ACn =n and AC = and Cn AC .
Join Cn and C.
143
n N
Cn
C
90°
n
Shpere
A
Fig. 7.2
Now, we know that n = cos (by Meunier’s Theorem)
1 1 1 1
or cos n ,
n n
= n cos .....(7.5.4)
From this we conclude from ACnC , that Cn AC 90 .
This result is interpreted as if a sphere is described with n as diameter, all centres of curvature
lie on this sphere, provided tˆ is the same.
7.5.2. Self-learning exercise-1
1. Define intrinsic property of a surface.
2. Define plane section of a surface.
3. Define normal section and oblique section of a surface.
4. Define curvature at a point on a given surface.
5. Write formula for curvature of normal section in terms of fundamental magnitudes.
6. Define normal curvature and radius of normal curvature.
7. Write the statement of Meunier’s Theorem.
7.5.3. Illustrative Examples
Ex.1. Find the curvature of a normal section of the right helicoid
x = u cos , y = u sin , z = c .
Sol. The curvature of normal section is given by
Ldu 2 2Mdu dv Ndv 2
n .....(1)
Edu 2 2 Fdudv Gdv 2
To find n, we shall first evaluate fundamental magnitudes E, F, G, L, M, N.
Let r = (u cos , u sin , c ) .....(2)
with u and as parameters, C is constant.
144
Let suffixes 1 and 2 represent partial differentiations of r with respect to u and
r r
r
(i.e. 1 , r2 ).
u
Then on differentiating, (2) with respect to u
r1 = (cos , u sin ) .....(3)
Now differentiating equation (2) with respect to , we have
r2 = (–u sin , u cos c) .....(4)
Now differentiating (3) with respect to u, we get
r11 = (0,0,0) .....(5)
Differentiating (3) with respect to , we get
r12 = (–sin , cos , 0) .....(6)
Now differentiating equation (4) with respect to
r22 = (–u cos , –u sin , 0) .....(7)
E r1 r2 r12 cos2 sin 2 1, .....(8)
F r1 r2 cos ,sin , 0 u sin , u cos , 0
u sin cos u sin cos 0 0 .....(9)
and G r2 , r2 u sin , u cos , c u sin , u cos , c
u 2 sin 2 u 2 cos 2 c 2
or G = u2 + c2 .....(10)
Now H2 = EG – F2 = 1 (u2 + c2) – 02 = u2 + c2 .....(11)
r1 r2 1
and ˆ ˆ ˆ ˆ
ˆ ˆ
N̂ H H cos i sin j 0k u sin i u cos j ck
iˆ ˆj kˆ
1
cos sin 0
H
u sin u cos c
1
u 2
c 2
iˆ c sin 0 ˆj 0 c cos kˆ u cos 2
u sin 2
1
2
u c 2
c sin iˆ c cos ˆj u kˆ
or c sin , c cos , u .....(12)
N̂
H u 2 c2
145
Now L Nˆ r11 Nˆ 0, 0, 0 0 [From equation (5)] .....(13)
c sin , c cos , u
M = Nˆ r12 sin , cos , 0
u 2 c2
c sin 2 c cos 2 0
u 2
c2
c
or M .....(14)
u2 c2
N Nˆ r22
c sin , c cos , u u cos , u sin , 0
u 2 c2
cu sin cos uc sin cos 0
0. .....(15)
u 2
c 2
Now using values from equations (8), (9), (10), (13),(14) and (15) in equation (1), we have
c
0 2 du d 0
2 2
u c
n
du 0 u c 2 d 2
2 2
2c du d .
u c2 2
du u
2 2
c 2
d
2
Ex.2. Show that the curvature at any point P of the curve of intersection of two sur-
faces is given by 2 sin 2 12 22 21 2 cos , where 1 are 2 the normal curvatures of the
surfaces in the direction of the curve at P and is the angle between their normals at that point.
Sol. Let S1 and S2 be the two given surfaces and N̂1 and N̂ 2 be the unit normals to them at
any common point P, respectively.
Curve of intersection of S1 and S2.
Let n̂ be the unit principal normal to the curve of intersection at P,, as shown in the Fig. 7.3.
Since Nˆ1, nˆ and N̂ 2 are to be drawn through the same tangent line, so clearly Nˆ1, nˆ , Nˆ 2 are
coplanar.
Let be angle between n̂ and N̂1 , then (–) is the angle between N̂ 2 and n̂ . Now using
Meunier’s theorem
1 = cos .....(1)
146
and 2 = cos – .....(2)
N2 N1
S2
P S1
Curve of
intersection of
S1 and S 2
Fig. 7.3
From equation (1),
1
cos =
12
and sin 1 cos 2 = 1 2 . .....(3)
Now from equation (2),
2 = cos cos + sin sin
= cos ( cos + sin sin
12
2 = cos sin 1 2 [From (1) and (3)]
k
2 2
or (– cos ) = sin 1
Squaring both sides
22 12 cos 2 2 2 1 cos sin 2 2 12
or 22 12 cos 2 12 sin 2 2 2 1 cos 2 sin 2
or 12 22 2 1 2 cos 2 sin 2
which is the required result.
7.6 Principal direction and principal curvatures
We have seen that curvature of normal section of a surface at a point varies with the direction
(du, dv) on the surface r r u, v . Among all the normal sections there are two directions for which
the curvature is maximum or minimum.
147
7.6.1. Definitions :
(i) Principal section
The normal section of a surface through a given point having maximum or minimum curva-
tures at the point are called principal sections of the surface at that point.
(ii) Principal Direction
Tangents to the principal section, i.e., directions of the principal section are called principal
directions at the given point. We shall see that, in general, there are two principal directions
at a point and these are orthogonal.
(iii) Principal Curvature
The maximum and minimum curvatures of the two principal sections of a surface are called
the principal curvatures.
(iv) Principal radius of curvature (i.e. principal radii)
Radii of curvatures of principal curvatures are called principal radius of curvature.
(v) Surface of centres
The locus of centres of principal curvatures at all points of a given surface called surface of
centres.
7.7 Equation giving the principal directions at a point of surface and to derive the
differential equation of the principal sections
We know that normal curvature n at point A (u, v) in the direction (du, dv) of surface
r r u, v is given by
Ldu 2 2 Mdudv Ndv 2
n .....(7.7.1)
Edu 2 2 Fdudv Gdv 2
If (l, m) be actual direction coefficients of the direction (du, dv),
du dv
where l , m ,
ds ds
Then from (7.7.1), we have
Ldu 2 2 Mdudv Ndv 2 du
2
du dv dv
2
L 2M N
n ds 2 ds ds ds ds
2
Edu 2 2 Fdudv Gdv 2 du du dv dv
2
2 E 2F G
ds ds ds ds ds
2 2
du du
or n Ll 2 Mlm Nm l ,m .....(7.7.2)
El 2 2 Flm Gm 2 ds ds
148
or n Ll 2 2 Mlm Nm 2 .....(7.7.3)
El2 + 2Flm + Gm2 = 1 .....(7.7.4)
Since L, M, N are fixed at A, so value of n at A depends upon the values l, m at A. Hence n is
a function of two variables l, m, which are connected by equation (7.7.3) Taking l as a function of m,
we find for stationary values by
d n dl dl
0 2 Ll 2M l m 2 Nm 0
dm dm dm
[on differentiating equation (3)]
dl dl
or Ll M l m Nm 0 .....(7.7.5)
dm dm
and by differentiating (7.7.4), we have
dl dl
2 El 2F l m 2Gm 0
dm dm
dl dl
or El F l m Gm 0 .....(7.7.6)
dm dm
By equation (7.7.5), rearranging the terms
dl
Ll Mm Ml Nm 0 .....(7.7.7)
dm
and by equation (7.7.6)
dl
El Fm Fl Gm 0 .....(7.7.8)
dm
dl
Eliminating , 1 between equation (7.7.7) and (7.7.8), we get
dm
Ll Mm Ml Nm
0
El Fm Fl Gm
or Ll Mm Fl Gm El Fm Ml Nm 0 .....(7.7.9)
Simplifying, we get
LF EM l 2 GL EN lm MG FN m2 0
or EM LF l 2 EN GL lm FN MG m2 0 .....(7.7.10)
This equation determines the principal directions of the principal section.
To obtain the differential equation of the principal section replace l, m by direction ratios du, dv
in equation (7.7.10), we get
(EM – LF) du2 + (EN – GL) du dv + (FN – GM) du2 = 0,
149
which can be expressed in the form of determinant as follows :
dv 2 du dv du 2
E F G
= 0. .....(7.7.11)
L M N
This equation gives differential of the principal section.
7.8 There are two principal directions at every point on a surface which are mu-
tually orthogonal
We know that the equation determining the principal directions at a point A (u, v) of the surface
r r u, v is given by
EM LF l 2 EN GL lm FN MG m2 0 [equation (7.7.10)]
which can be expressed as
2
l l
EM LF EN GL FN MG 0 .....(7.8.1)
m m
(on dividing above equation by m2)
l
This being a quadratic equation in , which provides two directions, i.e. there are two roots
m
l1 l2
(say) m and m of equation (7.8.1)
1 2
l1 l EN GL GL EN
Then sum of roots 2 ......(7.8.2)
m1 m2 EM LF EM LF
l1 l2 FN GM
and product of roots .....(7.8.3)
m1 m2 EM LF
l1l2 m1m2
or . .....(7.8.4)
FN GM EM LF
Now by equation (7.8.2)
l1m2 l2 m1 GL EN l m l m m1m2
1 2 2 1 .....(7.8.5)
m1m2 EM LF GL EN EM LF
From equation (7.8.4) and (7.8.5), we get
l1m2 l2 m1 l1l2 m1m2
c1 (say)
GL EN FN GM EM LF
l1m2 l2 m1 c1 GL EN , l1l2 c1 FN GM ,
m1m2 c1 EM LF .....(7.8.6)
150
Now if is the angle between these directions, than
H l1m2 m1l2
tan
El1l2 F l1m2 m1l2 Gm1m2
which can be expressed as
2
H l1m2 m1l2 4l1l2 m1m2
tan
El1l2 F l1m2 m1l2 Gm1m2
2
H c12 GL EN 4c1 FN GM c1 EM LF
or tan [by equation (7.8.6)]
Ec1 FN GM Fc1 GL EN Gc1 EM LF
or
H GL EN 4 EM FL FN GM
tan
E FN GM F GL EN G EM LF
2
H GL EN 4 EM FL FN GM
EFN EGM FGL FEN GEM GLF
or
H GL EN 4 EM FL FN GM
tan
0
. Hence the two principal direction are mutually orthogonal.
2
7.9 Umbilics
To derive the condition that a point be umbilic on the surface r r u, v .
The equation determining the principal directions at a point of surface r r u, v is given by
EM LF l 2 EN GL lm FN MG m2 0 .....(7.9.1)
in this, if
E F
EM LF 0
L M
E G E F G
EN GL 0
L N L M N .....(7.9.2)
F G
and FN MG 0
M N
l1 l l1 l2
then sum of roots 2 and product of roots of the equation
m1 m2 m1 m2
2
l l
EM LM EN GL FN MG 0 .....(7.9.3)
m m
151
0 l
becomes , i.e. values of becomes inderminate, which means that in this situation the normal curva-
0 m
ture becomes independent of directions (du, dv) and so has the same value for all directions through the
given point A (u,v) of the surface. Such a point is called an umbilic or a navel point on the surface
r r u, v .
Definition : A point A (u,v) on the surface r r u, v is called an umbilic, if at the point
E F G EG F 2 H
.....(7.9.4)
L M N LN M 2 T
An umbilic can also be taken as a circular section of zero radius.
Since at each point of a sphere, the normal curvature is same, so every point of a sphere is an
umbilical point.
7.10 The equation giving the principal curvatures at a point A (u, v) of the surface
r r u, v
By the equation (7.7.9) which determines the principal directions, we have
Ll Mm Fl Gm El Fm Ml Nm 0 .....(7.10.1)
Ll Mm Ml Nm
El Fm Fl Gm
Ll Mm Ml Nm l Ll Mm m Ml Nm
El Fm Fl Gm l El Fm m Fl Gm
Ll 2 2 Mml Nm 2
2 n
El 2 Flm Gm 2
Ll Mm Ml Nm .....(7.10.2)
n
El Fm Fl Gm
Ll Mm
Hence n Ll + Mm = n (El + Fm) .....(7.10.3)
El Fm
Ml Nm
and n Ml Nm n Fl Gm .....(7.10.4)
Fl Gm
Equation (7.10.3) and (7.10.4) can be rewritten as
L E n l M F n m 0 .....(7.10.5)
and M F n l N G n m 0 .....(7.10.6)
On eliminating l, m, we have
L E n M F n
0
M F n N G n
152
L E n N G n M F n M F n 0
or EG F EN LG 2FM LN M 0
2 2
n n
2
or H 2 2n EN LG 2 FM n T 2 0 , .....(7.10.7)
where H2 = EG – F2 and T2 = LN – M2.
This equation being quadratic in n gives two roots say n = a and n = b, which are called
two principal curvatures. Thus from equation (7.10.7), we have
EN LG 2 FM
sum of roots a b .....(7.10.8)
H 2 EG F 2
T 2 LN M 2
and product of roots = a b . .....(7.10.9)
H 2 EG F 2
7.11 Same important definitions
(i) Mean curvature or mean normal curvatures
The arithmetic mean of the principal curvatures at a point is called the mean curvature. It is de-
noted by symbol
1
i.e. a b EN LG 22FM .....(7.11.1)
2 2 EG F
(ii) Amplitude of normal curvatures
Amplitude of normal curvature is denoted by A is defined as
1
A a b .....(7.11.2)
2
(iii) First curvature
The sum of principal curvatures at a point is called the first curvature at the point, denoted by J
and given by
EN LG 2 FM
J a b .....(7.11.3)
EG F 2
1
then clearly J.
2
(iv) Gaussian curvature (or second curvatures)
The product of the principal curvatures at a point is called the Gaussian curvature at the point,
denoted by symbol
LN M 2
and given by a b . .....(7.11.4)
EG F 2
It is also called, specific curvature or total curvature.
153
(v) Minimal surface
The surface for which the first curvature is zero (or the mean curvature is zero) at all points, is
called a minimal surface.
Hence the surface will be minimal, if and only if
EN GL 2FM 0 0
a b 0
1 2 0
1 2 .....(7.11.5)
where and are radius of normal curvatures.
(vi) Developable surface
The surfaces for which Gaussian curvature is zero, are called developable surfaces.
Thus for developable surfaces
= 0 LN – M2 = 0 .....(7.11.6)
Remarks :
(a) The necessary and sufficient condition for a surface to be developable is that its Gaussian
curvature should be zero.
(b) If there is a surface of minimum area passing through a closed space curve, it is necessarily a
minimal surface, i.e., a surface of zero mean curvature.
7.12 Radius of curvature at an umbilic on the surface z f x , y
The Principal radii of curvature are given by the equation (7.10.7), when n is replaced by 1/
(principal radius of curvature) i.e.,
H2 1
2
EN LG 2 FM T 2 0 .....(7.12.1)
For the surface z f x, y , we know that
E 1 p2 , F pq , G 1 q2 , H 1 p2 q2
r s t rt s 2 .....(7.12.2)
L , M , N , T2 ,
H H H H2
z z 2 z 2 z 2 z 2 z
where p , q , r 2 , t 2 , s .
x y x y xy yx
Then substituting values from equation (7.12.2) in (7.12.1) we get
H2 1 1 rt s 2
2 H
t 1 p r 1 q 2 pqs
2 2
H2
0. .....(7.12.3)
154
But when equation of surface is given in the Monge’s form z f x, y , the condition for a
point to be umbilic is
1 p 2 pq 1 q 2 1 p 2 q2 H
. .....(7.12.4)
r s t rt s 2 rt s 2
From this,
1 p2 H 1 p2 r
, .....(7.12.5)
r rt s 2 H rt s 2
1 q2 H r 1 q 2 rt
, .....(7.12.6)
t rt s 2 H rt s 2
pq H pq s
and . .....(7.12.7)
s rt s 2 H rt s 2
Using values from equation (7.12.5) to (7.12.7) in (7.12.3), we get
H2 1 r t 2 s s rt s 2
t r
2 rt s 2 rt s 2 rt s 2 H2
H 2 1 2 rt s rt s 2
2
or 0,
2 rt s 2 H2
2
H rt s 2
or 0
H
H rt s 2 H2 1 p2 q 2
or . .....(7.12.8)
H rt s 2 rt s 2
Thus from (7.12.4) and (7.12.8), it follows that for an umbilic
1 p 2 pq 1 q 2
. .....(7.12.9)
r s t H
7.12.1. Self-learning exercise-2
1. Define the following :
(i) Principal sections of the surface.
(ii) Principal direction.
(iii) Principal curvature.
(iv) Principal radius of curvature.
(v) Surface of centres.
2. Write the differential equation of the principal section of the surface r r u, v .
155
3. Write the equation giving the principal directions at a point of a surface r r u, v .
4. Are the two principal directions at every point on a surface r r u, v , mutually ortho-
gonal ?
5. Write the Condition that a point be umbilic on the surface r r u, v .
6. Define the following :
(i) Mean curvature at a point of a surface.
(ii) First curvature.
(iii) Amplitude of normal curvature.
(iv) Gaussian curvature.
(v) Minimal surface.
(vi) Developable surface.
7. Write the formula for radius of curvature at are umbilic on the surface z = f (x, y).
7.12.2. Illustrative Examples
Ex.1. Find the principal sections and principal curvatures of the surface
x = a(u+ v), y = b(u – v), z = uv.
Sol. The position vector r of any point on the surface is given by
r ( xiˆ yjˆ zkˆ )
or r a (u v) iˆ b (u v) ˆj uvˆ [vector equation of surface]
or r = [a(u + v), b(u – v), uv] ......(1)
Here a and b are constants and u, v are parameters.
Differentiating (1) partially with respect to u, we get
r
r1 a, b, v . .....(2)
u
Again differentiating with respect to u, we get
2r
r11 0,0, 0 .....(3)
u 2
Now differentiating equation (2) partially with respect to v, we get
2r
r12 0, 0,1 .....(4)
vu
Now differentiating equation (1) partially with respect to v, we get
r
r2 a, b, u .....(5)
v
156
Differentiating this with respect to v again, we get
2r
r22 0, 0, 0 .....(6)
v 2
Now r1 r2 a, b, v a, b, u
aiˆ bjˆ vkˆ aiˆ bjˆ ukˆ
iˆ ˆj kˆ
a b v
a b u
iˆ bu bv ˆj av au kˆ ab ab
or r1 r2 b u v , a v u , 2ab .....(7)
Now E r1 r1 r12 a 2 b 2 v 2 .....(8)
F r1 r2 a, b, v , a, b, u a 2 b 2 uv .....(9)
and G r2 r2 r22 a 2 b 2 u 2 .....(10)
2
Now H 2 EG F 2 a 2 b 2 v 2 a 2 b 2 u 2 a 2 b 2 uv
or H 2 a 4 a 2 b 2 a 2u 2 a 2 b 2 b 4 b 2 u 2 a 2 v 2 b 2 v 2 u 2 v 2
a 4 b 4 u 2 v 2 2a 2b 2 2a 2uv 2b 2uv
a 2 u 2 b 2u 2 a 2 v 2 b 2 v 2 2a 2uv 2b 2 uv
a 2 u 2 v 2 2uv b 2 u 2 v 2 2uv 4a 2b 2 .
2 2
or H 2 a 2 u v b 2 u v 4a 2b2 .
1/ 2
or 2 2
H a 2 u v b 2 u v 4a 2 b 2 .....(11)
r r 1
Now, Nˆ 1 2
H H
aiˆ bjˆ vkˆ aiˆ bjˆ ukˆ
iˆ ˆj kˆ
1
a b v
H
a b u
1 ˆ
H
i bu bv ˆj av au kˆ ab ab
1
or N̂
H
b u v , a v u , 2ab .....(12)
157
Now L Nˆ r11 Nˆ 0, 0, 0 0 .....(13)
M Nˆ r12
1
H
b u v , a v u , 2ab 0, 0,1 2Hab .....(14)
and N Nˆ r22 Nˆ 0, 0, 0 0 .....(15)
2 2 2
2ab
2 4a b
T2 LN M 0 .....(16)
H H2
(i) Principal sections
The differential equation determining the principal sections is
EM LF du 2 EN LG dudv FN GM dv2 0 .....(17)
Using values of E, F, G, L, M, N from above equations, we get
2 2 2 ab 2 2 2 ab
a b v H 0 du 0 0 dudv 0 a b u H dv 0
2 2 2 2
Simplifying
du dv
2 2 2
a b u a b2 v 2
2
On integrating, we get
du dv
or 2
c1
2
a b 2
u 2
a b2 2
2
v 2
u v
or sinh 1 sinh 1 c1 .....(18)
2 2
a b a b2
2
where c1 is constant of integration.
Equation (18) is the equation of principal section.
(ii) Principal curvatures
The differential equation determining the principal curvatures is
H 2 2 EN LG 2 FM T 2 0 .....(19)
Using values of E, F, G and L, N, M and T2, we get
2ab 4 a 2 b 2
H 2 2 0 0 2 a 2 b 2 uv 2 0
H H
or H 4 2 4 abH a 2 b 2 uv 4a 2 b 2 0 .....(20)
where H 2 EG F 2 a 2 b 2 v 2 a 2 b 2 u 2 a 2 b 2 uv 2
2 2
or H2 a 2 u v b 2 u v 4a 2 b 2 .....(21)
158
On using equation (21) in (20), we get a quadratic equation in We can find principal curva-
tures.
Ex.2. For the hyperboloid 2z = 7x2 + 6 xy – y2, prove that the principal radii at the ori-
1 1
gin are and , and that the principal sections are x = 3y,, 3x = – y.
8 2
Sol. The given surface is
1
z
2
7 x 2 6 xy y 2 , .....(1)
which is the Monge’s form of equation of surface.
Here first we shall calculate p, q, r, s, t at the origin and then we shall calculate fundamental
magnitudes of this surface. Here x, y will be treated as parameters.
Now differentiating (1) partially with respect to x and y, we get
z z
p 7x 3y , q 3x y
x y
2 z 2 z 2 z
and r 2 7, s 3, t 2 1
x xy y
We shall find these values at the origin (0, 0, 0), so
z z
p 0, q 0, .....(2)
x (0,0,0) y (0,0,0)
2 z 2 z 2 z
r 2 7, s 3, t 2 1 .....(3)
x 0,0,0 x y 0,0,0 y 0,0,0
Here consider r x, y , f x, y z
z z 2
r1 1, 0, p , r2 1, 0, q , r11 z 0, 0, r
x y x 2
2 z 2 z
r12 0, 0, s , r22 2 0, 0, t .
xy y
E r1 r1 1 p 2 , F r1 r2 pq, G r2 r2 1 q 2
r s t 2 2 .....(4)
L , M , N and H 1 p q
H H H
Using values of p, q, r, s, t from equations (2) and (3) in equation (4), we get
E = 1, F = 0, G = 1, H = 1, L = 1, M = 3, N = –1 .....(5)
Now equation giving the principal curvatures is
EG F EN 2FM LG LN M 0
2 2
n n
2
.....(6)
159
Putting the values of E, F, G, L, M, N and H from equation (5), we get
2n 6 n 16 0 n 8 n 2 0
n = 8, – 2 .....(7)
1 1 1 1 1
Hence principal radii are 1 and 2
n 8 n 2 2
Again the equation of principal section is
EM FL dx2 EN GL dxdy FN GM dy 2 0 .....(8)
or 3dx 2 8dxdy 3dy 2 0 [on using values of E, F, G, L, M and N]
or 3dx dy dx 3dy 0
3dx + dy = 0 or dx – 3 dy = 0
On integrating 3x + y = c1 , x – 3y = c2 .....(9)
But at the origin (0,0,0), using x = 0, y = 0, we get c1 = 0, c2 = 0 [form (9)]
Principal sections of the origin are
3x + y = 0, x – 3y = 0 [From (9)]
or 3x = – y, x = 3y .....(10)
Ex.3. Show that the points of intersection of the surface xm + ym + zm = am and the line
x = y = z are umbilics and that the radius of curvature at an umbilic is given by
a
3( m 2) / 2 m .
m 1
Sol. The equation of given surface is
xm + ym + zm = am,, .....(1)
where a is a constant.
This surface may be regarded as Monge’s from [z = f(x, y)] by taking z as a function of x and y.
Differentiating equation (1) partially with respect to x and y respectively, we get
m 1
m 1 m 1 z z x
mx 0 mz 0 p .....(2)
x x z
m 1
z
0 my m 1 mz m1 0 z q y
.....(3)
y y z
Now, from equation (2), on taking log
m 1
x
log p log log p m 1 log x log z .....(4)
z
Differentiating this partially with respect to x, we get
1 p 1 1 z
m 1
p x x z x
160
1 p 1 1 z
or m 1 p p
p x x z x
r 1 p p 2 z
or m 1 r x x 2
p x z
Now differentiating equation (4) partially with respect to y
1 p 1 z
m 1 0
p y z y
z
m 1
1 p y
or
p y z
s m 1 q p 2 z
z s
or
p z y yx ......(6)
From equation (3), on taking log
m 1
y
log q log log q m 1 log y log z
z
Differentiating this partially with respect to y, we get
1 q 1 1 z
m 1
q y y z y
1 q 1 1 z
or m 1
q y y z y
t 1 1
or m 1 q .....(7)
q y z
1 p 2 pq
Now for an umbilic .....(8)
r s
pq z
But from (6), .....(9)
s m 1
1 p 2 pq z
From equation (8),
r s m 1
1 p2 z
or
r m 1
z
or 1 p
2
r
m 1
z 1 p
m 1 p [From equation (5)]
m 1 x z
161
pz
or 1 p
2
x
p2
pz
1
x
or p (x / z) .....(10)
Using this value of p in equation (2), we get
m 1 m 2
x x x
1 z m 2 x m 2 .....(11)
z z z
Similarly we can find, y m 2 z m 2 .....(12)
Therefore from equation (11) and (12), we get
x m 2 y m 2 z m 2 x y z
Therefore for an umbilic, x=y=z .....(13)
Then from equation (1),
z m z m z m a m 3z m a m x z, y z
a
z 1/ m .....(14)
3
Now from equation (2),
m 1
z
p p 1 x z
z
Similarly from equation (3),
m 1
z
q q 1 y z
z
From equation (6),
s m 1 q m 1
s p 1 q
p z z
m 1 m 1 31/ m
or s [From equation (14)]
a a
1/ m
3
2 2
and H 1 p 2 q 2 1 1 1 3
Then radius of curvature is given by
pqH
1 1 3 a (1/ 2)(1/ m )
3
s
m 1 1/ m m 1
3
a
162
a ( m 2) / 2 m
or 3
m 1
a
or 3( m 2) / 2m . (numerically)
m 1
Ex.4. Show that the surface ez cos x= cos y is minimal surface.
Sol. The given surface can be expressed as
cos y
ez or z log cos y log cos x [on taking log] .....(1)
cos x
i.e., the equation of the surface is in the form
z = f (x, y) [Monge’s form]
The position vector r of any current point (x, y, z) on this surface is given by
r x, y , z x, y, log cos y log cos x .....(2)
On differentiating it partially with respect to x and y respectively
r
r1 1, 0, tan x .....(3)
x
r
and r2 0, 1, tan y .....(4)
y
Again differentiating (3) partially with respect to x and y, we get
r11 0, 0,sec2 x , r12 0, 0, 0 .....(5)
Differentiating (4) partially with respect to y
r22 0, 0, sec 2 y .....(6)
Now, E r1 2 1 tan 2 x sec2 x,
F r1 r2 tan x tan y, .....(7)
G r22 1 tan 2 y sec 2 y
r1 r2 1, 0, tan x 0,1, tan y tan x, tan y,1
ˆ r1 r2 tan x, tan y,1
Now N
H H
tan x, tan y ,1 0, 0, sec 2 x sec 2 x
Hence L Nˆ r11 .....(8)
H H
M Nˆ r12
tan x , tan y ,1 0, 0, 0 0 and
H
N Nˆ r22
tan x , tan y ,1 0, 0, sec y
2 sec 2
y
H H
163
The condition for the surface to be minimal is
EN – 2FM + GL = 0 .....(9)
Putting values of E, F, G and L, M, N form equation (7) and (8) in (9), we get
sec 2 x sec2 y sec2 y sec2 x
EN 2 FM GL 0 0
H H
Hence the given surface is minimal.
Ex.5. Find the values of (i) First curvature. (ii) Gaussian curvature, at any point of right
helicoid x = u cos , y = u sin , z = c.
Hence show that a right helicoid is a minimal surface.
Sol. The position vector r of any current point (x, y, z) on this surface is given by
r xiˆ yjˆ zkˆ x, y , z u cos , u sin , c .....(1)
Then on differentiating r partially with respect to u and , we get
r r
r1 cos ,sin , 0 , r2 u sin , u cos , c .....(2)
u
Similarly
2r 2r
r11 2 0, 0, 0 , r12 sin , cos , 0
u u
2r
r22 2 u cos , u sin ,0 . .....(3)
Then the fundamental magnitudes are
E r1 r1 r12 1, F r1 r2 0, G r2 r2 u 2 c 2
and H 2 EG F 2 1 u 2 c 2 0 u 2 c 2 H u 2 c 2
ˆ r1 r2 c sin , c cos , u
N , L Nˆ r11 0
H H
c a2
M Nˆ r12 , N Nˆ r22 0, T 2 2 . .....(4)
u2 c2 H
(i) First curvature
The first curvature J of the given surface at any point (u,) is obtained by
EN GL 2 FM 0 0 2 0
J 0, .....(5)
1 u 2 c 2 0
2
EG F
which shows that first curvature for right helicoid is zero, hence it is a minimal surface.
164
(ii) Gaussian curvature
The Gaussian curvature K at any point (u,) is obtained by
T 2 LN M 2 c2
K .
H2 h2 2 2 .....(6)
u 2
c
7.13 Radius of curvature of a given section through any point of a surface
z = f (x, y).
Suppose the surface is z = f (x, y).
Tangent to
T N2
2, n2
m
l2 ,
Curve by a plane 90°
intersection called
plane section A
Curve
AN1 principal normal
to surface AN 2 is
normal to AT
Surface
z = f (x, y)
Fig 7.4
A plane cut in it is a curve . Suppose A be a point of , AT be tangent to at A having direc-
tion cosines (say) l1, m1, n1. Let AN2 be one of the normal to AT lying in the plane of section. Let d.c.’s
of AN2 be l2, m2, n2. Also let AN1 be principal normal to surface at A. If equation of surface is taken in
the form F (x, y, z) = 0, then direction ratios of the normal to surface are
f F F z z
, , i.e., – p, – q, 1, where p , q .
x y z x y
Therefore d.c.’s of principal normal are
p q 1
, ,
.....(7.13.1)
p2 q2 1 p2 q2 1 1 p2 q2
Let angle between the plane of the section and normal section through AT be , then
165
pl2 qm2 n2
cos .....(7.13.2)
1 p2 q2
Now, AT and AN1 are perpendicular, therefore
pl1 + qm1 – n1 = 0 .....(7.13.3)
Differentiating this equation with respect to s, we get
dp dl1 dq dm1 dn1
l1 p m1 q 0
ds ds ds ds ds
dl1 dm dn dp dq
or p q 1 1 l1 m1 0 .....(7.13.4)
ds ds ds ds ds
dp dq
Now we shall find values of and
ds ds
dp p dx p dy
rl1 sm1 .....(7.13.5)
ds x ds y ds
p z 2 z dx dy
Q r 2 etc. l1 , m1
x x x x ds ds
dq q dx q dy
Similarly sl1 tm1 .....(7.13.6)
ds x ds y ds
multiplying equation (7.13.5) by l1 and equation (7.13.6) by m1 and adding, we get
dp dq
l1 m1 = l1 (rl1 + sm1) + m1 (sl1 + tm1)
ds ds
rl12 sl1m1 sl1m1 tm12
rl
2
1 2sl1m1 tm12 .....(7.13.7)
Now using Serret-Frenet formulae, we find
dl1 dm dn l m n 1 .....(7.13.8)
p q 1 1 p 2 q 2 2 pl1 qm2 n2
ds ds ds
Using values from equation (7.13.7) and (7.13.8) in (7.13.4), we get
1
pl2 qm2 n2 rl12 2sl1m1 tm12 0
or pl2 qm2 n2 rl12 2sl1m1 tm12 .....(7.13.9)
Using this value in equation (7.13.2), we have
cos
rl12 2sl1m1 tm12 cos rl 2
1 2sl1m1 tm12
.....(7.13.10)
1 p2 q2 1 p2 q2
166
This relation gives radius curvature for surface z = f (x, y) at a point A of it by a plane section,
where l1, m1, n1 are direction cosines of its tangent at the point (x, y, z). If = 0, then from (7.3.10)
1 rl12 2sl1m1 tm12
.....(7.13.11)
1 p2 q2
then this expression gives the principal radius of curvature corresponding to a given principal section.
Ex. Find the principal radii at the origin of the surface
2z = 5x2 + 4xy + 2y2.
Find also the radius of curvature of the section x = y.
Sol. Equation of surface can be expressed as
1
z
2
5 x 2 4 xy 2 y 2 .....(1)
Deafferentating it partially with respect to x and y respectively.
z z
p 5 x 2 y, q 2x 2 y
x y
2z 2 z 2 z
and r 5, s 2, t 2 .....(2)
x 2 x y y 2
p = 0, q = 0, r = 5, s = 2, t = 2 and H2 = 1 + p2 + q2 = 1 .....(3)
Now the principal curvatures for a surface in Monge’s form is given by
H42 – H [(1 + p2) t + (1 + q2) r – 2spq] + (rt – s2) = 0
Using values form equation (2), we get
2 – 7 + 6 = 0 or ( – 1) ( – 6) = 0 = 1, 6 .....(4)
then principal radii of curvature (say), 1 and 2 are
1 1 1 1
1 1, and 2 . .....(5)
1 6
Second part : The given point is origin (0, 0, 0) and the plane of the section is x – y = 0. The
equation of tangent plane of the surface z = f (x, y) at (0, 0, 0) is z = 0
Hence the direction cosines of the line
x – y = 0, z = 0
x y z 0 x y z
or
or 1 1 0 1 1 0 .....(6)
2 2
1 1
are , , 0 l1, m1, n1 .....(7)
2 2
where l1, m1, n1, are the d.c.’s of tangent through origin for curve of section.
167
Now, equation of normal plane to the surface at origin (0, 0, 0) through the tangent line is,
x–y=0 .....(8)
which is same as given plane.
So, = 0, then radius of curvature is given by
5 1 1 1
22 2
1 rl12 2sl1m1 tm12 2 2 2 2
2
1 p q 2 1 0 0
or 1 (5 / 2) 3 11 2
radius of curvature . .....(9)
1 2 11
7.14 Lines of curvature
Definition : A curve on a surface is called a line of curvature if the tangent at any point of it is
along the principal direction at that point.
7.14.1 There are two systems of lines of curvature
At each point of the surface there are two principal directions which are at right angles. Hence,
we have two orthogonal systems of lines of curvature on the surface and through each point on the sur-
face there pass two lines of curvature, one corresponding to each system.
7.14.2 To find the differential equation of lines of curvature at point (u, v) of the sur-
r r
face r r (u, v ).
By definition of line of curvature, the direction of line of curvature at any point is along the prin-
cipal direction at that point, so the differential equation of the two systems of line of curvature is the
same as the differential equation of the principal section and is given by
(EM – FL) du2 + (EN – GL) du dv + (FN – GM) dv2 = 0 .....(7.14.1)
this equation can be expressed in the following form
Pdu2 + 2Qdu dv + Rdv2 = 0 .....(7.14.2)
where P = EM – LF, 2Q = EN – GL, and R = FN – GM.
Here ER – 2FQ + GP = E (FN – GM) – F (EN – GL) + G (EM – LF) = 0, ...(7.14.3)
r r
which shows that lines of curvature cut orthogonally at a point on the surface r r (u, v).
Remark :
(a) Equation (1) can be put in the following determinant form
dv 2 du dv du 2
E F G 0 (Here u and v are parameters) .....(7.14.3)
L M N
168
(b) If we take the parameters x, y in place of (u, v), the above equation (7.14.3) of line of
curvature through a point on z = f (x, y) is given by
dy 2 dx dy dx 2
1 p2 pq 1 q2 0 .....(7.14.4)
r s t
Q E = 1 + p2, F = pq, G = 1 + p2, H2 = 1 + p2 + q2
r s t
and L , M , N .
H H H
7.14.3 Illustrative examples
Ex.1. Prove that the cone xy = z {(x2 + z2)1/2 + (y2 + z2)1/2} passes through a line of
curvature of the paraboloid xy = az.
Sol. From the given equation of paraboloid, we have
1
z xy .....(1)
a
Differentiating partially with respect to x and y respectively
z y
p . .....(2)
x a
z x
q .....(3)
y a
2 z 2 z 1 2 z
and again differentiating r 0, s , r0 .....(4)
x 2 x y a y 2
Here parameters are x and y, so differential equation of lines of curvature is given by
dy 2 dx dy dx 2
1 p2 pq 1 q2 0
.....(5)
r s t
Using value of p, q, r, s, t, we get
dy 2 dx dy dx 2
y2 xy x2
1 1 0
a2 a2 a2 .....(6)
1
0 0
a
expanding the determinant, we get
169
1 x2 1 y
dy 2 0 1 2 dx xy 0 dx 2 1 2 0 0
a a a
a
or
a 2 x 2 dy 2 a 2 y 2 dx 2 0
dx dy
0
or 2 2 .....(7)
a x a y2
2
1 x y
On integrating, we get sinh sinh 1 c1 constant .....(8)
a a
where c1 is arbitrary constant if integration.
This relation in equation (8) provides us the surface on which the lines of curvature lie.
Equation (8) can be expressed as
x y2 y x2
sinh 1 1 2 1 2 c1
a a a a
x y2 y x2
or 1 2 1 2 sinh c1 constant .....(9)
a a a a
Now from equation az = xy of the paraboloid, we have
x z y z
, , [using these in equation (9)]
a y a x
z z2 z z2
we get 1 2 1 2 constant K say
y x x y
12 12
or
z x2 z2
z y2 z2 K constant
xy
12 1 2
or z x 2 z 2
y2 z2 Kxy .....(10)
which is a can passing through a line of curvature of xy = az.
Ex.2. Prove that in general three lines of curvature pass through an umbilic.
Sol. Let the umbilical point be taken as origin (0, 0, 0); and xy-plane as the tangent plane at
origin, and z-axis normal at the origin.
Now tangent plane is xy-plane, so we expect that first degree terms to be zero in the equation
of surface z = f (x, y). Further, the section at umbilic being circle, so the equation of surface z = f (x, y)
will have, coefficient of x2 = coefficient of y2 and no term of xy will be present. Under these restrictions
we express the equation of surface z = f (x, y) as follows :
170
x2 y 2 1
2z
3
ax 3 3bx 2 y 2cxy 2 dy 3 ... . .....(1)
Now, the condition that the normal (0, 0, 0) and other point (x, y, z) should intersect, i.e., the
two lines whose equation are
0 0 0
.....(2)
0 0 1
x y z
and .....(3)
p q 1
x y z
should be coplanar is p q 1 0 . .....(4)
0 0 1
On expanding, we get qx = py .....(5)
Now, differentiating equation (1) partially with respect to x and y respectively we get
2z 2x
2p ax 2 2bxy cy 2 ... .....(6)
x
2z 2y
and 2q bx 2 2cxy dy 2 ... .....(7)
x
Using values of p and q from (5) and (6) in (4), we get
2 xy 2 xy
ax 2 y 2bxy 2 cy 3 ... bx3 2cx 2 y dxy 2 ...
or bx3 + x2y (2c – a) + xy2 (d – 2b) – cy3 = 0
3 2
y y y
3
divide by – x , we get c 2b d a 2c b 0 .....(8)
x x x
Let be the angle which tangent to a line of curvature makes with z-axis, then tan lim ( y / x )
therefore from equation (8), we have
c tan3 + (2b – d) tan2 – (2c – a) tan – b = 0. .....(9)
This is a cubic equation in tan , so it given three lines of curvature through the umbilic, corre-
sponding to three values of tan .
7.14.4 Self-learning exercise-3
1. Write formula for radius of curvature of a given section through any point of a surface
z = f (x, y).
2. Define lines of curvature.
3. What is the differential equation of lines of curvature at point (u, v) of the surface
r r u, v ?
4. In general how many lines of curvature pass through an umbilic ?
171
7.15 Summary
1. In this unit you have studied Meunier’s theorem, Principal direction and principal curvature at
r r
point (u, v) on surface r r u , v , about first curvature, mean curvature, and Gaussian Curva-
r r
tures at point (u, v) on surface r r u , v , about umbilic point, about formula of radius of
curvature of any normal section at an umbilic on z = f (x, y), about radius of curvature of a given
section through any point on surface of the form z = f (x, y), about lines of curvature and its
differential equation.
2. Sufficient number of examples have been solved in the unit.
3. Partial derivative p, q, r, s, t and fundamental magnitudes E, F, G and L, M, N, H, T will help
the students to easily understand the text of the unit.
4. Examples in the text have been inserted frequently to help students to understand the text of the
unit.
7.16 Answers to self-learning exercises
Self-learning exercise-1
Ldu 2 2 Mdu dv Ndv 2
1. See definitions of §7.3 5. n
Edu 2 2 Fdu du Gdv 2
6. See §7.4 (i) & (ii) 7. See Meurier’s Theorem.
Self-learning exercise-2
1. See §7.6.1 (i) to (v)
2. (EM – LF) du2 + (EN – GL) du dv + (FN – GM) dv2 = 0
3. (EM – LF) l2 + (EN – GL) lm + (FN – GM) m2 = 0
E F G
4. Yes 5.
L M N
1 p2 Hpq H 1 q2
6. See §7.11 (i) to (vi) 7. H .
r s t
Self-learning exercise-3
cos rl12 2 sl1m1 tm12
1. 2. See 7.14.2
1 p 2 q2
3. Pdu2 + 2Qdu dv + Rdv2 = 0, where P = Em – LF, Q = EN – GL, R = FN – GM.
4. Three.
172
7.17 Exercises
1. Find the curvature of the normal section of the helicoid x = u cos , y = u sin , z = f (x) + c
uf du 2 2cdu dv u 2 f fv2
[Ans. n ]
H 1 f 2 du 2 2cf du dv u 2 c 2 dv 2
x u v y u v uv
2. For the surface , , z ; prove that the principal radii are given by
a 2 b 2 2
a2b22 + ab (a2 – b2 + uv) – 4 = 0
where 42 = 4a2b2 + a2 (u – v)2 + b2 (u + v)2 and that the line of curvature are given by
du dv .
2 2 2 2 2 2
a b u a b v
3. Prove that for helicoid x = u cos , y = u sin , z = c ,
u 2 c2
1 2 , where u 2 x 2 y 2
c
and that the lines of curvature are given by
du
d
u c2
2
2 2 2
4. Find the umbilics of the ellipsoid x y z 1. If P is an umbilic of this ellipsoid, then prove
a2 b2 c2
that the curvature at P of any normal section through P is ac .
3
b
[Ans.
xa
y b z z ]
2
a b 2 0 2
b c 2
5. Find the Gaussian curvature at point (u, v) of the anchor ring,
x = (b + a cos u) cos v, y = (b + a cos u) sin v, z = a sin u,
where the domain of u, v is 0 u 2, 0 v 2. Verify that the total curvature of the whole
cos u
surface is zero. [Ans. Gaussian curvature ]
a b a cos u
173
UNIT 8 : Principal Radii, Relation between Fundamental
Forms, Asymptotic Lines, Differential Equation
of an Asymptotic line, Curvature and Torsion of
an Asymptotic Line
Structure of the Unit
8.0 Objectives
8.1 Introduction
8.2 Principal radii through a point of the surface z = f (x, y).
8.3 Fundamental forms
8.4 Relation between three fundamental forms
8.5 Asymptotic lines
8.6 To show that to a given direction there is one and one only conjugate direction. Also derive the
condition for the directions (du, dv) and (Du, Dv) to be conjugate.
8.7 To show that the direction given by P du2 + 2Q dudv + R du2 = 0, are conjugate if
LR – 2MQ + NP = 0.
8.8 Family conjugate to the family of curves P du + Q dv = 0.
8.9 Conjugate directions and parametric curves.
8.10 Principal directions (lines of curvature) at a point are always orthogonal and conjugate.
8.11 A characteristic property of conjugate directions
8.11.1 Self-learning exercise-1.
8.12 Definitions
8.13 Differential equation of the asymptotic lines at any point (u, v) on the surface r r (u , v)
8.14 Conditions for two asymptotic directions at a point to be real and distinct, coincident or
imaginary.
8.14.1 Asymptotic lines are orthogonal if the surface is minimal.
8.15 An asymptotic line is a curve on a surface such that the normal curvature of the surface in its
direction is zero.
8.15.1 Illustrative examples.
8.16 The necessary and sufficient condition for the parametric curves to be asymptotic lines.
8.17 If the parametric curves are asymptotic lines, then find differential equation of line of curvature
2FM M2
and show that first and second curvatures are and .
H2 H2
174
8.18 Osculating plane at any point of a curved asymptotic line is the tangent plane to the surfaces.
8.19 Torsion of an asymptotic line r r ( s ) on the surface r r (u , v) .
8.20 Curvature of an asymptotic line r r ( s ) on the surface r r (u , v) .
8.21 Beltrami-Enneper theorem
8.21.1 Illustrative examples
8.21.2 Self-learning exercise-2
8.22 Summary
8.23 Answers to self-learning exercise
8.24 Exercises
8.0 Objectives
This unit provides a general overview of Principal Radii, relation between three fundamental forms
of surface, Asymptotic lines and their differential equations, also curvature and torsion of an asymptotic
line. After reading this unit you will be able to learn :
1. about Principal Radii.
2. about relation between fundamental forms.
3. about Asymptotic lines and their differential equations.
4. about curvature and torsion of asymptotic lines.
8.1 Introduction
In this unit we shall study principal radii through a point of the surface z = f (x, y), relation be-
tween three fundamental forms, asymptotic lines and differential equation of asymptotic lines. In the end
of unit we shall study curvature and torsion of an asymptotic lines.
In the earlier Unit-7 we have already obtained the principal curvatures and principal sections for
the Monge’s surface z = f (x, y) from which expression for principal radii can be directly written. We
shall discuss here an alternative method.
8.2 Principal radii through a point of the surface z = f (x, y)
Equation of surface is z = f (x, y) .....(8.2.1)
If l1, m1, n1 are the direction cosines of the tangent to a normal section of the surface through
the point (x, y, z), then the radius of curvature of the section is given by
1 rl12 2 sl1m1 tm12
, .....(8.2.2)
2
1 p q 2
175
z z 2 z 2 z 2z
where p ,q ,r 2, s , t 2.
x y x xy y
Equation (8.2.2) can be expressed as
rl12 2sl1m1 m12t
1 , where H 1 p 2 q 2
H H H
r 2 s 2 t
or l12 l1m1 m1 1 .....(8.2.3)
H H H
Now, – p, – q, 1 are the direction ratios of the surface normal at (x, y, z), so by the condition of
perpendicularity
pl1 + qm1 – n1 = 0
or n1 = pl1 + qm1. .....(8.2.4)
On squaring, we get
n12 = (pl1 + qm1)2
or 1 – (l12 + m12) = p2l12 + q2m12 + 2pql1m1 [ l12 + m12 + n12 = 1]
or (1 + p2) l12 + 2pql1m1 + m12 (1 + q2) = 1 .....(8.2.5)
Subtracting equation (8.2.3) from equation (8.2.5), we get
r s t
l12 1 p 2 2l1m1 pq m12 1 q 2 0. .....(8.2.6)
H H H
l1 l
The equation (8.2.6) is quadratic equation in , therefore it gives two values of 1 , corre-
m1 m1
sponding to a given radius of curvature. If is a principal radius, these sections coincide (i.e., values
coincide). Therefore for the principal radii (by condition of roots both equal B2 – 2AC = 0).
2
s r t
4 pq 4 1 p 2 1 q 2 0. .....(8.2.7)
H H H
On simplifying, we get
2 (rt – s2) – H {(1 + p2) t + (1 + q2) r – 2pqs} + H4 = 0. .....(8.2.8)
This equation gives the principal radii.
8.3 Fundamental forms
In the earlier units, we have studied first and second fundamental forms. Here we shall define
third fundamental form and then we will derive relation between them.
Three fundamental forms are given by first fundamental form
I dr dr E ( du ) 2 2 Fdudv G (dv ) 2 (ds ) 2 , .....(8.3.1)
176
where equation of surface is regarded as r r (u, v ), u, v are its parameters, and E, F, G are first
order fundamental magnitudes.
Second fundamental form
II d r dNˆ L du 2 2 Mdudv Ndv 2 ( n ds 2 ), .....(8.3.2)
where L, M, N are second order fundamental magnitudes.
Third fundamental form.
The quadric A du2 + 2B dudv + C dv2, .....(8.3.3)
is called the third fundamental form for surface r r (u , v) and is denoted by III, where
Nˆ ˆ Nˆ
A Nˆ12 , B Nˆ1 Nˆ 2 and C Nˆ 2 2 and Nˆ1 , N2 .
u v
III dNˆ dNˆ A du 2 2 B dudv C dv 2 . .....(8.3.4)
8.4 Relation between three fundamental forms
If K is the total curvature and J is the first curvature then to show that
KI – J.II + III = 0 .....(8.4.1)
Proof. Suppose that the lines of curvature be the parametric curves, then
F=M=0 .....(8.4.2)
Then, the two fundamental forms are reduced to
I = E du2 + G dv2 .....(8.4.3)
and II = L du2 + N dv2. .....(8.4.4)
If a, b be the principal curvatures at the point (u, v) of surface r r (u , v) we have for two
parametric curves by Rodrigue’s formula
dr dNˆ
a 0 a r1 Nˆ 1 0 Nˆ1 a r1 .....(8.4.5)
du du
dr dNˆ
and b 0 b r2 Nˆ 2 0 Nˆ 2 b r2 .....(8.4.6)
dv dv
So, the third fundamental form is given by
2
III (dNˆ ) 2 Nˆ1du Nˆ 2 dv
Nˆ12 du 2 2 Nˆ1 Nˆ 2 dudv Nˆ 22dv 2
2 2
a r1 du 2 2 a r1 b r2 dudv b r2 dv 2 .
177
Using equations (8.4.5) & (8.4.6)
2 2 2 2 2 2
III a r1 du 2a b (r1 r2 ) dudv b r2 dv
2a Edu 2 2a b F dudv b2G dv 2
2a Edu 2 b G dv 2 F 0
Now 2 2
2
K.I. – J.II a b Edu Gdv a b L du N dv
2
.....(8.4.8)
Using K = ab, and J = a + b
L N
a L E a , b N b G. .....(8.4.9)
E G
Using these values of L and N in equation (8.4.8), we get
2 2
2
K.I. – J.II a b Edu a bGdv a b E a du bGdv
2
a Edu 2 b Gdv 2 III . .....(8.4.10)
Hence K.I – J.II + III = 0. .....(8.4.11)
This equation gives a relation between three fundamental forms of the surface r r (u, v ). This
can also be expressed in the form
J
III 2 II K I 0. .....(8.4.12)
2
8.5 Asymptotic lines
Before defining the asymptotic lines we shall define conjugate directions at a point of a surface
and we shall derive expressions for conjugate directions.
Definition. Conjugate directions
Conjugate directions at a given point (u, v) on a surface r r (u , v ) are defined as follows :
Let P be any point (u, v) on a surface (say) S and Q be a neighbouring point on it [i.e., Q is
(u + du, v + dv)].
Let tangent planes at P and Q to the surface S intersect in a line (say) LM. Then the limiting
directions of the line PQ and LM as Q P are called conjugate directions at P.
8.6 To show that to a given direction there is one and only one conjugate direc-
tion. Also derive the condition for the two directions (du, dv) and (Du, Dv) to
be conjugate
Proof. Let r r (u , v ) be equation of surface. Let r be position vector of the point P (u, v)
on this surface and r dr be position vector of point Q (u + du, v + dv), a point adjacent to P in the
178
direction (du, dv). Let N̂ and Nˆ dNˆ be unit normals at the points P ( r ) and Q ( r dr ) on the
surface, respectively. Then clearly d r is the limiting position of the vector PQ as Q P.. Let Dr be
the vector along the limiting position of the line LM as Q P.
(where D is used to denote the other direction in place of d)
r r
then PQ dr du dv r1du r2 dv
u v
or dr r1du r2dv .....(8.6.1)
Nˆ Nˆ
and dNˆ du dv Nˆ1du Nˆ 2 dv .....(8.6.2)
u v
r r
Also Dr Du Dv r1Du r2 Dv .....(8.6.3)
u v
Now the vector Dr lies in the tangent plane of P ( r ) as well as that of Q ( r dr ) so it will be
to both normals N̂ and Nˆ dNˆ .
Dr Nˆ 0 .....(8.6.4)
and
Dr Nˆ dNˆ 0 .....(8.6.5)
or Dr Nˆ Dr dNˆ 0 or Dr dNˆ 0 .....(8.6.6)
Using values from (8.6.2) and (8.6.3), we get
r1 Du r2 Dv Nˆ1du Nˆ 2 dv 0
r Nˆ Dudu r Nˆ Dudv r Nˆ Dvdu r Nˆ Dvdv 0
1 1 1 2 2 1 2 2
L Dudu + M Dudv + M Dvdu + N Dvdv = 0
or L Dudu + M (Dudv + Dvdu) + N Dvdv = 0, .....(8.6.7)
where Nˆ1 r1 L, Nˆ 1 r2 M Nˆ 2 r1 , r2 Nˆ 2 N .
The equation (8.6.7) is the required condition for the two directions (du, dv) and (Dv, Du) to be
conjugate.
du Du
Also the equation (8.6.7) is linear in each of the ratio and which shows that to a given
dv Dv
direction there is one and only one conjugate direction.
Remarks.
du
(i) The symmetry of equation (8.6.7) shows that if the direction is conjugate to directions
dv
Du Du du
, then is conjugate to direction i.e., the property of conjugate direction is
Dv Dv dv
reciprocal.
179
(ii) It follows from equation (8.6.7) that the directions (l1, m1) and (l2, m2) at point P are con-
jugate iff Ll1l2 + M (l1m2 + m1l2) + Nm1m2 = 0.
8.7 To show that the directions given by Pdu2 + 2Qdudv + Rdv2 = 0, are conjugate
if LR – 2MQ + NP = 0.
Proof. Given LR – 2MQ + NP = 0. .....(8.7.1)
The equation Pdu2 + 2Qdudv + Rdv2 = 0 can be expressed as
2
du du
P 2Q R 0. .....(8.7.2)
dv dv
du du Du
This equation being quadratic in , gives two roots, say , then
dv dv Dv
du Du 2Q
sum of roots , .....(8.7.3)
dv Dv P
du Du R
and product of roots . .....(8.7.4)
dv Dv P
Now from equation (8.6.7), we have
L Dudv + M (Dudv + Dvdu) + NDvdv = 0.
which can be expressed as
Du du Du du
L M N 0. .....(8.7.5)
Dv dv Dv dv
Using values from equation (8.7.3) and (8.7.4), we get
R 2Q
L M N 0
P P
LR – 2MQ + NP = 0, .....(8.7.6)
which is the required condition.
8.8 Family conjugate to the family of curves P du + Q dv = 0
The equation of family of curves given by
Pdu + Qdv = 0, .....(8.8.1)
can be expressed as Pdu = – Qdv
du dv
k (say)
Q P
du = kQ, dv = – kP. .....(8.8.2)
180
Now condition for two direction to be conjugate is
L Du du + M [Du (– dv) + Dvdu] + N Dvdv = 0 .....(8.8.3)
Using equation (8.8.2) into the equation (8.8.3), we get
L Du (kQ) + M [Du (– kP) + Dv (kQ)] +N Dv (– kP) = 0
k [LQ Du – MPDu + MQ Dv –NPDv] = 0
or (LQ – MP) Du + (MQ – NP) Dv = 0 .....(8.8.4)
which is the required result.
8.9 Conjugate directions and parametric curves.
Necessary and sufficient condition that the parametric curves through a point
to have conjugate directions is that M = 0.
Proof : The condition is necessary : The parametric curves, u = c1 (constant) and v = c2
(constant), in the differential form will be
du dv = 0, .....(8.9.1)
which can be expressed as 0 du2 + 1du dv + 0 dv2 = 0. .....(8.9.2)
Comparing it with Pdu2 + 2Q du dv + Rdv2 = 0,
we find that P = 0, 2Q = 1, R = 0. .....(8.9.3)
Now the direction of parametric curves are conjugate if
LR – 2MQ + NP = 0 .....(8.9.4)
Using equation (8.9.3), we get
0 – M + 0 M = 0. .....(8.9.5)
Thus the parametric curves have conjugate direction when M = 0.
The condition is sufficient : If M=0 .....(8.9.6)
and for parametric curves P = 0, R = 0. .....(8.9.7)
Then the condition LR – 2MQ + NP = 0 is satisfied by using (8.9.6) and (8.9.7).
Hence, if M = 0, the directions are conjugate for parametric curves.
8.10 Principal directions (line of curvature) at a point are always orthogonal and
conjugate.
Proof : We know that parametric curves u = const. and v = const., whose combined equation
is given by
du dv = 0.
181
The necessary and sufficient condition that these parametric curves be line of curvature are
F = 0, M=0 .....(8.10.1)
and condition for parametric curves to be conjugate is
M = 0. .....(8.10.2)
Also the necessary and sufficient condition for parametric curves to be orthogonal is that
F = 0. ...(8.10.3)
From these we conclude that the direction of line of curvature are always orthogonal as well as
conjugate.
8.11 A characteristic property of conjugate direction.
To show that conjugate direction at a point P on a surface are parallel to con-
jugate diameters of the indicatrix at P.
Proof : Take the lines of curvature as parametric curves, so that
F = 0, and M = 0. .....(8.11.1)
Then by equation (EG – F2) kn2 – (EN + LG – 2FM) kn + (LN –M2) = 0
becomes EG kn2 – (EN + LG) kn + LN = 0
or E (Gkn – N) kn – L (Gkn – N) = 0
(Ekn – L) (Gkn – N) = 0
L N L N
kn , (say) ka and kb .....(8.11.2)
E G E G
are the principal curvatures at the point P.
Now by setting point P as origin, x-axis along the principal direction v = const. at P, y-axis
along the principal direction u = const. at P and z-axis along the normal to the surface at P, then equa-
tion of indicatrix (Dupin’s indicatrix) at P is given by
x2 y2
1, z h. .....(8.11.3)
2h ka 2h kb
But we know from coordinate geometry that the lines y = m1x and y = m2x are conjugate
diameters of the conic
x2 y2 b2
1, if m1m2 , .....(8.11.4)
a 2 b2 a2
so the directions of conjugate diameters of curve given by equation (8.11.3) are given by
ka
m1m2 .....(8.11.5)
kb
182
m1m2
L E LG .
N / G EN
Let 1 and 2 be the angles which the conjugate directions (du, dv) and (Du, Dv) make with the
parametric curve v = const., then
G dv
and m1 tan 1 .....(8.11.6)
E du
G Dv
m2 tan 2 . .....(8.11.7)
E Du
Using these values of m1, m2 in to equation (8.11.5), we get
G dv G Dv LG
E du E Du EN
G dv Dv LG
E du Du EN
N dv D = – L du Du
or L du Du + N dv Dv = 0, .....(8.11.8)
which is the equation determining conjugate direction, when M = 0. Hence the result.
Alternate definition of conjugate directions :
Two directions at a point of a surface are said to be conjugate if they are parallel to the conju-
gate diameters of the indicatrix of the surface at that point.
8.11.1 Self-learning exercise-1.
1. Write a formula to find principal radii through a point of the surface z = f (x, y).
2. Write the third fundamental form of the surface.
r r
3. Write the relation between three fundamental forms of a surface r r u , v .
r r
4. Define conjugate directions at a given point (u, v) on a surface r r u , v .
5. Write the condition for two directions (du, dv) and (Du, Dv) to be conjugate.
6. What is the necessary condition for parametric curves through a point to have conjugate
direction ?
8.12 Definitions
r r
(i) Asymptotic directions : A self conjugate direction on a surface r r u , v is called
asymptotic direction.
r r
(ii) Asymptotic lines : A curve on a surface r r u , v whose direction at each point is self
conjugate, is called an asymptotic line.
183
r r
Alternate definition of asymptotic line : A curve drawn on a surface r r u , v so as to
touch at each point one of the inflexional tangents through the point, is called an asymptotic line on the
surface.
8.13 Differential equation of the asymptotic lines at a point (u, v) on the surface
r r
r r u, v .
r r
Let (du, dv) be directions of an asymptotic line at any point (u, v) on the surface r r u , v .
Hence (du, dv) is a self-conjugate direction.
du Du
So for asymptotic line , .....(8.13.1)
dv Dv
where (du, dv) and (Du, Dv) are conjugate directions.
The condition of conjugacy of these two directions are
L du Du + M (du Dv + dv Du) + N dv Dv = 0
du Du du Du
or L M N 0. .....(8.13.2)
dv Dv dv Dv
On using (8.13.1), we have
2
du du
L 2M N 0
dv dv
L du2 + 2M du dv + N dv2 = 0, .....(8.13.3)
r r
this is differential equation of the asymptotic lines at a point (u, v) on the surface r r u , v in curvilin-
ear coordinates. Equation (8.13.3) can also be expressed in the following form
r
dNˆ dr 0. .....(8.13.4)
8.14 Conditions for two asymptotic directions at a point to be real and district,
coincident or imaginary
Differential equation of asymptotic line is
2
du du
L 2M N 0 .....(8.14.1)
dv dv
du
which is quadratic in . It follows that in general asymptotic direction through every point of surface
dv
are self conjugate.
(i) These directions will be real and distinct if the discriminant (B2 – 4 Ac) = 4 (M2 – LN) or
(M2 – LN) is positive or (LN – M2) < 0.
184
But the Gaussian curvature is
LN M 2 ve
K K ve . .....(8.14.2)
H2 H2
Hence asymptotic lines are real and distinct if
K<0 .....(8.14.3)
and when K < 0, the point is called hyperbolic point and the asymptotic lines at the point are parallel to
the asymptotes of the indicatrix at the point.
(ii) The directions given by equation (8.14.1) will be real and coincident, if the discriminant of
equation (8.14.1) (say)
(B2 – 4 Ac) = 4 (M2 – LN) = 0 or M2 – LN = 0. .....(8.14.4)
Then Gaussian curvature
LN M 2
K 0,
H2
then the point is called parabolic point.
(iii) The directions given by equation (8.14.1) will be imaginary if the discriminant
‘‘B2 – 4 Ac < 0’’ i.e., 4 (M2 – LN) < 0
(M2 – LN) < 0
(LN – M2) > 0. .....(8.14.5)
Then the Gaussian curvature
LN M 2 ve
K 0
H2 H2
i.e. K > 0, then the point is called an elliptic point.
In this case the asymptotic lines are imaginary.
8.14.1 Asymptotic lines are orthogonal if the surface is minimal
Proof : When both roots of the differential equation
Ldu2 + 2M du dv + N dv2 = 0 .....(8.14.6)
of the asymptotic lines are real and district, then
(M2 – LN) > 0
(LN – M2) < 0. .....(8.14.7)
Then Gaussian curvature
LN M 2
K 0, .....(8.14.8)
H2
then the point is hyperbolic and the indicatrix at the point is rectangular hyperbola and asymptotic lines
are orthogonal. In this condition, the first curvature
185
J = 0 EN – 2MF + GL = 0,
which is condition of minimal surface.
Hence asymptotic lines are orthogonal if surface is minimal.
8.15 To show that an asymptotic line is a curve on a surface such that the normal
curvature of the surface in its direction is zero.
Proof. Differential equation of asymptotic line is
L du2 + 2M du dv + N dv2 = 0 .....(8.15.1)
and normal curvature at any point (u, v) on surface r r (u , v) is given by
Ldu 2 2 Mdudv Ndv 2
Kn . .....(8.15.2)
Edu 2 2 Fdudv Gdv 2
Using equation (8.15.1), we get
K n = 0. .....(8.15.3)
Hence in case of asymptotic lines Kn = 0, an asymptotic line is a curve on a surface such that
the normal curvature of the surface in its direction is zero.
Remarks.
1. A curve drawn on a surface so that its osculating plane at any point contains the binormal to
the curve at the point is an asymptotic line on the surface.
2. Two asymptotic lines through any point have the same osculating plane.
3. When the principal curvatures ka and kb equal and opposite i.e., ka = – kb, the indicatrix is
a rectangular hyperbola and so asymptotic lines are at right angles.
4. The normals to a surface at points of an asymptotic line generate a skew surface whose line
of striction is the asymptotic line.
8.15.1 Illustrative Examples
Ex.1. Prove that on the surface z = f (x, y) (Monge’s form) the equation of asymptotic
lines are
r dx2 + 2s dxdy + t dy2 = 0.
Sol. The given surface is z = f (x, y), then position vector r of the current point is r = (x, y, z),
where z = f (x, y), x and y are taken parameters.
On differentiating partially with respect to x and y respectively
r r
r r r r
r1 1, 0, p , r2 0, 1, q .....(1)
x y
z z
where p , q etc.
x y
186
r 2r r 2r
and r12 0, 0, s , r11 2 0, 0, r ,
yx x
2
r r .....(2)
r22 2 0, 0, t ,
y
r r r r r r
Therefore E r1 r1 1 p 2 , F r1 r2 pq, G r2 r2 1 q 2 .....(3)
r r
ˆ
N
r1 r2
p, q,1 , H 2 EG F 2 1 p 2 q 2 .
.....(4)
H H
r r r r s s
L Nˆ r11 , M Nˆ r12
H 1 p2 q2 H 1 p2 q2
ˆ r t .....(5)
N N r22
2
1 p q 2
Now differential equation of asymptotic line is
Ldu2 + 2M du dv + N dv2 = 0 .....(6)
Here parameters u = x, and v = y, then differential equation (6) of the asymptotic lines at point
(x, y) on the surface z = f (x, y) reduces to the form
Ldx2 + 2M dx dy + Ndy2 = 0. .....(7)
Using values from equation (5), we get
rdx 2 2 s dx dy tdy 2
0
1 p2 q2 1 p2 q 2 1 p2 q2
or rdx2 + 2s dx dy + tdy2 = 0, .....(8)
which is the required equation of asymptotic line.
Ex.2. Find the asymptotic lines on the surface
z = y sin x.
Sol. Equation of the surface is of the form z = f (x, y) [monge’s form], so differential equation of
asymptotic line will be
r dx2 + 2s dx dy + t dy2 = 0, .....(1)
2 z 2 z 2 z
where r , s , t .
x 2 xy y 2
Surface is z = y sin x. .....(2)
Differentiating partially with respect to x and y, we get
z
p y cos x .....(3)
x
187
z
and q sin x .....(4)
y
2 z 2 z 2 z
and r 2 y sin x, s cos x, t 2 0 .....(5)
x xy y
Using values of r, s and t form equation (5) in equation (1), we get
(– y sin x) dx2 + 2 cos x dx dy + 0 = 0
2
tan x dx dy dx 0
y
2
Then either dx = 0 or – tan x dx dy 0
y
when dx = 0 x = constant = c1 (say) .....(6)
2
and when – tan x dx dy 0
y
log (cos x) + 2 log y = (constant) = log c2 (say) [on integrating]
or log (cos x) + log (y2) = log (c2)
or log (cos x y2) = log (c2)
or cos x y2 = c2 .....(7)
Hence form equation (6) and (7) are the equations of asymptotic lines as given x = c1 and
y2 cos x = c2.
Ex.3. Prove that for the surface x = 3u (1 + v2) – u3, y = 3v (1 + u2) – v3, z = 3u2 – 3v2,
the asymptotic line are u v = constant.
r
Sol. Let r be the position vector of any current point on the surface, then
r
r = (3u (1 + v2) – u3, 3v (1 + u2) – v3, 3u2 – 3v2) .....(1)
On differentiating with respect to u, we get
r
r r
r1
u
3 1 v 2 u 2 , 2uv, 2u .....(2)
On differentiating (1) partially with respect to v
r
r r
r2
v
3 2uv,1 u 2 v 2 , 2u .....(3)
From equation (2), on differentiating with respect to x again
r r
r11 6 u , v,1 , similarly r12 6 u , v, 0
r .....(4)
r22 6 u, v, 1
188
iˆ ˆj kˆ
r r
r r 1
Now Nˆ 1 2
H H
3 1 v2 u 2 6uv 6u
6uv
3 1 u 2 v2 6v
9
or Nˆ
H
2u 1 u 2 v 2 , 2v 1 u 2 v 2 ,1 u 4 v 4 2u 2 v 2
2
r
L Nˆ r11
54 u 2 v 2 1 r
, M Nˆ r12 0
H
2
r
and N Nˆ r22
54 u 2
v 2
1
H
Now differential equation of asymptotic line is
L du2 + 2M du dv + N dv2 = 0. .....(6)
Using values of L, M and N from equation (5) into equation (6), we get
54 2 2 54 2 2 2
H
u v 1 du 2 0
H
u v 1 dv 2 0
or du2 – dv2 = 0 or du = dv du dv = 0
On integrating (u v) = constant which is the required equation. Hence proved.
8.16 Necessary and sufficient condition for the parametric curves to be asymptotic
lines
Proof : The parametric curves are u = constant, and v = constant, then the combined differen-
tial equation of the parametric curves is given by
du dv = 0. .....(8.16.1)
And differential equation of asymptotic lines is
Ldu2 + 2M du dv + N dv2 = 0 .....(8.16.2)
(i) The condition is necessary : When the parametric curves are asymptotic lines, the above
two equations (1) and (2) must be same, so on comparing these we get
L = 0, N = 0, M 0. .....(8.16.3)
(ii) The condition is sufficient : When L = 0, N = 0 and M 0, then equation (2) reduces
to
0 + 2M du dv + 0 = 0 du dv = 0. .....(8.16.4)
which is equation (1), this shows that the asymptotic lines are parametric curves.
189
8.17 If the parametric curves are asymptotic lines, then find differential equa-
tion of line of curvature and show that first and second curvatures are
2FM M 2
and .
H2 H2
Proof : If the parametric curves are asymptotic lines, then
L = 0, N = 0, M 0 .....(8.17.1)
and the differential equation (EM – LF) du2 + (EN – GL) du dv + (FN – GM) dv2 = 0
of lines of curvature reduces to the following form
EMdu2 + 0 + (0 – GM) dv2 = 0
EMdu2 – GM dv2 = 0 ( Q M 0) .....(8.17.2)
Now the equation of the principal curvatures
H2 2 – (EN + GL – 2FM) + (LN – M2) = 0
reduces to the following form H2 2 + 2FM – M2 = 0 .....(8.17.3)
which is a quadric in , so it gives two roots. i.e., a, b, say.
2FM
Hence, sum of roots = (a + b) = J = first curvature
H2
2
and product of roots = (a b) = K M = second curvature.
H2
Hence the result.
8.18 Osculating plane at any point of a curved asymptotic line is the tangent plane
to the surface.
r r r r
Proof : Let equation of surface be r r u , v , and let r r s be curved asymptotic line
r
ˆ dr
lying on this surface. Let t be the unit tangent to the asymptotic line at any point P (r) and N̂ be
ds
the unit normal to surface at point P, then
Nˆ tˆ 0. .....(8.18.1)
Differentiating with respect to s, we get
r
dNˆ ˆ ˆ dtˆ dNˆ d r ˆ
t N 0 or N nˆ 0 .....(8.18.2)
ds ds ds ds
r
dtˆ d r
Q nˆ and tˆ , where curvature, nˆ unit principal normal vector
ds ds
190
But by Serret-Frenet formulae. The differential equation of asymptotic line is
r
r dNˆ d r
dNˆ d r 0 or 0 .....(8.18.3)
ds ds
then equation (8.18.2) reduces to
Nˆ nˆ 0 Nˆ nˆ 0 Q 0 .....(8.18.4)
By equation (8.18.1) and (8.18.4), we have
Nˆ tˆ 0 and Nˆ nˆ 0,
this means N̂ is perpendicular to both tˆ and nˆ so Nˆ is parallel to
tˆ nˆ bˆ or N̂ bˆ .....(8.18.5)
where b̂ is unit vector along binormal.
From this, we conclude that at any point of a curved asymptotic line, the binormal is also normal
to the surface.
Let Nˆ bˆ. .....(8.18.6)
r r
Now the equation of osculating plane to curve r r s at a point P (s) is given by
r
R rr bˆ 0 .....(8.18.7)
and equation of tangent plane to the surface at this point is given by
r r
R r Nˆ 0 . .....(8.18.8)
The relation (8.18.6) makes the above planes coincident at the point on the curve on the
surface.
Hence the result.
r r r r
8.19 Torsion of an asymptotic line r r s on the surface r r u, v .
We know that the unit vector along binormal b̂ to an asymptotic line is the unit surface
normal N̂
i.e. N̂ bˆ .....(8.19.1)
Differentiating both sides with respect to arc length s, we get
dNˆ dbˆ dNˆ
nˆ, [by Serret-Frenet formulae]
ds ds ds
Nˆ nˆ , .....(8.19.2)
where Nˆ dN .
ds
191
Now taking dot product of both sides of equation (8.19.2) by nˆ, we have
Nˆ nˆ nˆ nˆ
or Nˆ nˆ Q nˆ nˆ 1 .....(8.19.3)
now n̂ bˆ tˆ
or n̂ Nˆ tˆ .....(8.19.4)
Now, using value of n̂ in equation (8.19.3), we get
Nˆ Nˆ tˆ
or Nˆ Nˆ tˆ
or Nˆ Nˆ tˆ
r
or Nˆ Nˆ r , .....(8.19.6)
which is torsion of an asymptotic line.
r r r r
8.20 Curvature of an asymptotic line r r s on the surface r r u, v .
r
ˆt dr is the unit vector along the tangent then we know that
ds
dtˆ ˆ
t nˆ . .....(8.20.1)
ds
Now, taking dot product of both sides by n̂
tˆ nˆ nˆ nˆ
or tˆ nˆ Q nˆ nˆ 1
tˆ Nˆ tˆ Q nˆ bˆ tˆ Nˆ tˆ
or tˆ nˆ tˆ Nˆ tˆ tˆ
r r
or Nˆ r r , .....(8.20.2)
r r
where tˆ r and tˆ r , is curvature of an asymptotic line.
8.21 Beltrami-Enneper theorem
Statement : At a point on a surface, where the Gaussian curvature is negative and
equal to K, the torsion of the asymptotic lines is K .
r r
Proof : The torsion of an asymptotic line r r s is
r
Nˆ Nˆ r .....(8.21.1)
192
r
or
Nˆ Nˆ r .....(8.21.2)
dNˆ Nˆ du Nˆ dv ˆ
Now Nˆ N1u Nˆ 2 v .....(8.21.3)
ds u ds v ds
r r r
r dr r du r dv r r
and r r1 u r2 v .....(8.21.4)
ds v ds v ds
r r r
N̂ r Nˆ1u Nˆ 2 v r1 u r2v
r r r r
Nˆ1 r1 u2 Nˆ1 r2 Nˆ 2 r1 uv Nˆ 2 r2 v2 . .....(8.21.5)
r
Using value of N̂ r from equation (8.21.5) in equation (8.21.2), we have
r r r r
Nˆ Nˆ r u Nˆ Nˆ r Nˆ Nˆ r u v Nˆ Nˆ r v .
1 1
2
1 2 2 1 2 2
2
r r r r
Nˆ Nˆ1 r1 u 2 Nˆ Nˆ 1 r2 Nˆ Nˆ 2 r1 uv Nˆ Nˆ 2 r2 v2
EM FL 2 EN GL FN GM 2
or u uv v .....(8.21.6)
H H H
r EM FL
by using Weingarton equations, where Nˆ Nˆ 1 r1 etc.
H
Now let the asymptotic lines be taken as parametric curves, then
L = 0, N = 0, M 0. .....(8.21.7)
Using L = 0, N = 0 into equation (8.21.6), we get
EM 0 GM 2
0 u 2 0 v
H H
M
or
H
Eu 2 Gv2 .....(8.21.8)
First for asymptotic line u = constant, u = 0 .....(8.21.9)
MG 2
then from equation (8.21.8), v .....(8.21.10)
H
Now from the first fundamental form, we have
Eu2 + 2Fu v + Gv2 = 1 .....(8.21.11)
From this for the curve u = constant u = 0, we have from equation (8.21.11),
0 + 0 + Gv2 = 1 Gv2 = 1,
then from equation (8.21.12), we get
M
.....(8.21.12)
H
193
Again for the asymptotic line v = constant, we have v = 0 and from equation (8.21.11),
Eu2 = 1. .....(8.21.13)
And from (8.21.8),
ME 2 M
u , . .....(8.21.14)
H H
The Gaussian curvature
LN M 2 M 2 [ Q Here L = 0, N = 0]
H2 H2
M2 M
2
H H
or [by equation (8.21.12) and (8.21.14)]
Hence the theorem is proved.
8.21.1 Illustrative examples
Ex.4. Prove that on the surface z = f (x, y) torsion of the asymptotic lines are
s 2
rt .
1 p 2
q2
Sol. By Example 1, for this surface
z = f (x, y).
r
Let r = (x, y, z = f (x, y)) be position vector of a point then
r r
r r r r
r1 1, 0, p , r2 0,1, q
x y
r r r
r12 0, 0, s , r11 0, 0, r , r22 0, 0, t
r r r r r r
therefore E r1 r1 1 p 2 , F r1 r2 pq, G r2 r2 1 q 2
r r
ˆ r1 r2 p, q, 1 2
N , H EG F 2 1 p 2 q 2
H H
r r r s r t
and L Nˆ r11 , M Nˆ r12 , N Nˆ r22 .
2 2 2 2 2 2
1 p q 1 p q 1 p q
Then on using values of L, M and N in the differential equation
Ldu2 + 2M du dv + Ndv2 = 0,
of the asymptotic line, we get rdx2 + 2s dx dy + tdy2 = 0, (Here parameters are x, y)
Also K = Gaussian curvature
LN M 2
rt s 2
H2 H4
194
s 2 st s 2 st
Torsion K
H4 H2
s 2 st
or ,
(1 p 2 q 2 )
which is the required result.
Ex.5. Show that the curvature of an asymptotic line may be expressed as
r r r r r r r r
r1 r r2 r r2 r r1 r .
H
Sol. We know that for an asymptotic line curvature is given by
r r
N̂ r r (box product)
r r
r r r1 r2 r r
r r
N̂ r r
H
r r r r
1 r r r r 1 r1 r r1 r
r1 r2 r r r r r r [By Lagrange’s identity]
H H r2 r r2 r
1 r r r r r r r r
H
r1 r r2 r r2 r r1 r
Hence the result.
Ex.6. Prove that on the surface of revolution x = u cos v, y = u sin v, z = f (u), the
asymptotic lines are f11 du2 + u f1 dv2 = 0. Also show that the values of their torsions are
u f f .
3
1 11
u 1 f
2
1
2
r
Sol. The position vector r of any point on the given surface is
r
r = (u cos v, u sin v, f (u))
Differentiating partially with respect to u and v
r r
r r r r
r1 cos v, sin v, f1 u , r2 u sin v, u cos v, 0
u v
r
2r r r
Similarly r11 0, 0, f11 u , r12 sin v, cos v, 0
u 2
r
and r22 = (– u cos u, – u sin v, 0)
r r
r1 r2 u cos v f1 , u sin v f1, u
r r r
E r12 1 f12 , F r1 r2 u cos v sin v u cos v sin v 0
195
r r
r2 2 ˆ r1 r2 u cos v f1 , u sin v f1 , u
G r2 u , N
H H
where
H2 = EG – F2 = u2 1 f1
2
r uf r r u 2 f1
Also L Nˆ r11 11 , M Nˆ r12 0, N Nˆ r22
H H
Therefore the equation of asymptotic lines is
Ldu2 + 2M du dv + N dv2 = 0
which on using values of L, M, N reduces to the following form
uf11 2 u 2 f1 2
du dv 0 or f11du 2 uf1dv 2 0
H H
which is the required equation.
Again
LN M 2
torsion K or
u f f .
3
1 11
u 1 f
2 2 2
H 1
8.21.2 Self-learning exercise-2.
1. Define :
r r
(i) Asymptotic direction on a surface r r u , v .
(ii) asymptotic lines.
2. Write the differential equation of the asymptotic lines at a point (u, v) on the surface
r r
r r u, v .
3. Write the condition for asymptotic lines to be orthogonal.
4. For Monge’s form of surface z = f (x, y) write the equation of asymptotic line.
r r
5. Write formula for torsion and curvature of an asymptotic line r r s on surface
r r
r r u, v .
6. State Beltrami-Enneper theorem.
8.22 Summary
1. In this unit you have studied about principal radii through a point of surface z = f (x, y), relation
between three fundamental forms, asymptotic lines and differential equation of asymptotic lines
r r
at a point (u, v) on surface r r u , v in curvilinear coordinates, curvature () and torsion ()
r r r r
of an asymptotic line r r s on the surface r r u , v .
2. Sufficient number of examples have been solved in the unit.
196
3. Fundamental magnitudes E, F, G and L, M, N, H and T along with differential equation which
gives the principal radii, expressions of curvature and torsion will help the students to easily
understand the text of the unit.
8.23 Answers to self-learning exercises
Self-learning exercise-1
1. 2 (rt – s2) – H {1 + p2) t + (1 + q2) r – 2 pqs} + H4 = 0.
2. III = Adu2 + 2B du dv + c du2, where A Nˆ 12 , B Nˆ1 Nˆ 2 , C Nˆ12
3. KI – JII + III = 0
4. See §8.5.
5. LDu du + M (Du dv + Dv du) + N Dv dv = 0.
6. M = 0
Self-learning exercise-2
1. See §8.12 (i) and (ii)
2. Ldu2 + 2M du dv + N dv2 = 0
3. First curvature J = 0; i.e. surface is minimal.
4. rdx2 + 2s dx dy + tdy2 = 0
r r r
5. Torsion Nˆ Nˆ r (box product) and curvature N̂ r r (box product).
6. See §8.21.
8.24 Exercises
1. Prove that the asymptotic lines of the surface x = v – 2u – e–u, y = ev–u, z = eu–v lie on the
cylinders yz + ay – ea = 0, xy + by + e –b = 0, where a, b are arbitrary constant.
2. Show that the asymptotic lines of helicoid x = u cos , y = v sin , z = c consist of the genera-
tors and the curves of intersection with coaxial right cylinders.
r r
3. Derive the formula for torsion and curvature of an asymptotic line r r s on surface
r r
r r u, v .
r r
4. Derive the differential equation of the asymptotic lines at a point (u, v) on the surface r r u , v in
curvilinear coordinates.
197
Unit 9 : Geodesics, Differential Equation of a Geodesic,
Single Differential Equation of a Geodesic,
Geodesic on a Surface of Revolution, Geodesic
Curvature and Torsion, Gauss-Bonnet Theorem
Structure of the Unit
9.0 Objectives
9.1 Introduction
9.2 Normal property of a geodesic
9.3 Definition
r r
9.4 General differential equation of geodesics on a surface r r (u, v) .
r r
9.5 Cannonical equations of a geodesic on the surface r r (u, v) .
9.6 Differential equation of a geodesic in Gauss coefficients.
9.7 Single differential equation of a geodesic on surface.
9.8 On the general surface, a necessary and sufficient condition that the curve v = c (const.) be a
geodesic is EF2 + FE1 – 2EF1 = 0.
9.9 The curve u = c (const.) is a geodesic if and only if GG1 + FG2 – 2GF2 = 0.
9.9.1 Self-learning exercises-1
9.9.2 Illustrative examples
9.10 Differential equations of a geodesic on a surface f (x, y, z) = 0.
9.11 Differential equation of geodesic on the surface z = f (x, y), the Monge’s form.
9.12 Geodesic on a surface of revolution.
9.13 Clairut’s theorem
9.14 Geodesic on surface of revolution cuts the meridian at a constant angle, then the surface is a
right circular cylinder.
9.15 A curve on sphere is a geodesic if only if it is a great circle.
9.16 Geodesic Curvature and torsion of geodesic
r
9.17 An expression for g and that it is intrinsic.
9.18 The geodesic curvature vector of any curve is orthogonal to the curve.
9.19 Formulae for geodesic curvature
198
9.20 Geodesic curvature in terms of Gauss coefficients
9.21 Geodesic curvature for parametric curves
9.22 Normal angle
9.23 Geodesic curvature in terms of normal angle
9.24 Expression of the torsion of a geodesic on any surface and that the torsion of an asymptotic line
is equal to the torsion of its geodesic tangent.
9.25 Expressions for the torsion of a geodesic in terms of fundamental magnitudes and also in terms
of principal curvatures.
9.26 Some important definitions
9.27 Gauss-Bonnet theorem
9.27.1 Self learning exercise-2
9.27.2 Illustrative examples
9.28 Summary
9.29 Answers to self-learning exercises
9.30 Exercises
9.0 Objectives
This unit provides a general overview of geodesics, differential equation of a geodesic, single
differential equation of a geodesic, geodesic on a surface of revolution, geodesic curvature and torsion,
Gauss-Bonnet theorem. After reading this unit you will be able to learn :
1. about geodesics,
2. about the general differential equation of a geodesic on a surface,
3. about single differential equation of a geodesic when curve on surface is given by a single
relation between the parameters u and v (either v = v/u or u = u/v),
4. about geodesic on a surface of revolution,
5. about geodesic curvature and torsion on a surface,
6. about Gauss-Bonnet theorem, which gives us the relation between torsion of a curve c
and torsion of its geodesic tangent.
9.1 Introduction
We know that in Euclidean space curves of shortest distance between any two points are
r r
straight lines. But curves on a surface r r (u , v ) having shortest length are called geodesics. So roughly
199
speaking “a geodesic on a surface may be defined as a curve of shortest distance between two points
on that surface”
But to find the arc of shortest distance between two points on a given surface is a very compli-
cated affair.
9.2 Normal property of a geodesic
The normal to the surface coincides with the principal normal to the curve (geodesic).
Consider a tightly stretched string on the smooth convex side of the surface to lie along the curve
r r
(geodesic) joining two points on the surface r r (u , v ) , very close to each other. The forces which
keep this small string in equilibrium are the tensions at its extremities and the reaction normal to the sur-
face. Because the string is very small so these tensions are in the osculating plane of the string and there-
fore for equilibrium the force of reaction must also lie in the same plane, which implies that the normal to
the surface coincides with the principal normal to the curve (geodesic). This property is termed as the
normal property of a geodesic.
9.3 Definition:
Geodesic (or geodesic curve) : Geodesic on a surface is defined as the curve of stationary
length (rather than strictly shortest distance) on a surface between any two points in its plane.
or
A geodesic on a surface is a curve whose osculating plane at each point contains the normal to
the surface at that point.
Hence the normal to the surface coincides with the principal normal to the geodesic.
r r
9.4 General differential equations of geodesics on a surface r r ( u, v )
Let r be the position vector of any point P (u, v) on the geodesic drawn on the surface
r r
r r (u , v) . Let n̂ and N̂ be the principal normals to the geodesic curve at P and the normal to the
given surface at the same point P, respectively. Then by definition of geodesic
n̂ Nˆ .....(9.4.1)
We know that d r ˆ,
r t
ds
so again differentiating with respect to s,
r tˆ nˆ r Nˆ , .....(9.4.2)
where is the curvature of geodesic at point P.
200
r r
Now r r (u , v) .
On differentiating with respect to s
dr r du r dv
r ru
1
r2v . .....(9.4.3)
ds u ds v ds
Again differentiating, we get
ru d r1 u r v d r2 v
r 1 2
ds ds
r1 du r1 dv r2 du r2 dv
r1u u r2v v
u ds v ds u ds v ds
ru
1
r11u r12 v u r2v r21u r22v v
r1u r11u 2 r12 u v r2 u r12u v r22 v2
r ru
1
r11u 2 2r12u v r2 v r22 v2 . .....(9.4.4)
Using value of r from equation (9.4.2) in (9.4.4), we get
N̂ ru
1
r11u 2 2r12u v r2 v r22 v2 . .....(9.4.5)
Now taking scalar product of equation (9.4.5) with r1 , we get
r1 Nˆ r1 r1 u r1 r11 u 2 2 r1 r12 uv r1 r2 v r1 r22 v2
1 2 1 2
or 0 Eu E1u E2u v Fv F2 G1 v .....(9.4.6)
2 2
ˆ E 2 r1
r1 N 0, E1 u u r1 2r1 u 2 r1 r11
E 2 r1
E2 r1 2r1 2 r1 r12 etc.
v v v
Now taking scalar product of equation (9.4.5) with r2 , we get
r2 Nˆ r2 r1 u r2 r11 u 2 2 r2 r12 uv r2 r2 v r2 r22 v2
1 2 1 2
0 Fu F1 E2 u G1u v Gv G2v .....(9.4.7)
2 2
ˆ 1 G
r2 N 0, r2 r11 F1 E2 , G1 2r2 r12 etc.
2 u
Hence equation (9.4.6) and (9.4.7) are the general differential equation of geodesics on a sur-
r r
face r r (u , v ) .
201
r r
9.5 Canonical equations of a geodesic on the surface r r ( u, v )
r r
The general differential equations of geodesics on surface r r (u , v ) are
1 1
Eu Fv E1u2 E2u v F2 G1 v2 0 .....(9.5.1)
2 2
1 1
and Fu Gv F1 E2 u 2 G1uv G2v2 0 .....(9.5.2)
2 2
E E G G
where E1 , E2 , G1 , G2 etc.
u v u v
u 2u v 2v
and u , u 2 , v , v 2 etc.
s s s s
These equation (9.5.1) and (9.5.2) may be written in a more compact form if we denote
1
T
2
Eu 2 2 Fu v Gv2 .....(9.5.3)
u v
where u and v .
s s
Differentiating partially equation (9.5.3) with respect to u and v respectively, we get
T
Eu Fv .....(9.5.4)
u
T
and Fu Gv .....(9.5.5)
v
Now differentiating (9.5.3) partially with respect to u and v, we get
T 1
E1u 2 2 F1 u v G1v2 .....(9.5.6)
u 2
T 1
and E2u2 2 F2 uv G2 v2 .....(9.5.7)
v 2
Differentiating equation (9.5.4) with respect to s, we get
d T d E du E dv F du F dv
Eu Fv Eu u Fv v
ds u ds u ds v ds u ds v ds
Eu E1u E2v u Fv Fu
1
F2 v v
d T
Eu Fv E1u E2u v F1u v F2u
2 2
or .....(9.5.8)
ds u
Similarly differentiating equation (9.5.5) with respect to s, we get
d T
Fu Gv F1u 2 F2 u v G1u v G2 v 2 .....(9.5.9)
ds v
202
Now subtracting equation (9.5.6) from equation (9.5.8), we get
d T T 1 1
Eu Fv E1u 2 E2 uv F2 G1 v2 0
ds u u 2 2
d T T
0. .....(9.5.10)
ds u u
Similarly subtracting equation (9.5.7) from equation (9.5.9), we get
d T T 1 1
Fu Gv F1 E2 u 2 G1uv G2v2 0
ds v v 2 2
d T T .....(9.5.11)
0
ds v v
Hence, the equations (9.5.10) and (9.5.11) are called the canonical equations of a geodesic.
9.6 Differential equations of a geodesic in Gauss coefficients
It will be another simple form of the general differential equations of a geodesic given below
1 1
Eu Fv E1 u 2 E2 u v F2 G1 v2 0 .....(9.6.1)
2 2
1 1
and Fu Gv F1 E2 u2 G1 uv G2 v2 0 .....(9.6.2)
2 2
Multiplying equation (9.6.1) by G and equation (9.6.2) by F and subtracting, we get
EG F u 12 GE 2FF FE u
2
1 1 2
2
1
GE2 FG1 uv 2GF2 GG1 FG2 v2 0 .....(9.6.3)
2
Dividing whole equation by (E G – F2) (= H2), we get
1 GE1 2 FF1 FF2 2 GE2 FG1 1 2GF2 GG1 FG2 2
u 2
u 2
u v v = 0
2 H H 2 H2
.....(9.6.4)
Using the Gauss coefficients into the equation (9.6.4), we get
and u lu2 2m uv n v2 0, .....(9.6.5)
1
where l GE1 2 FF1 FF2 ,
2H 2
1
m GE2 FG1 ,
2H 2
1
n 2GF2 GG1 FG2 .
2H 2
203
Similarly multiplying equation (9.6.1) by F and (9.6.2) by E and subtracting, we get
u u 2 2 u v v2 0 , .....(9.6.6)
where , are Gauss coefficients given by
1 1
2
2 EF1 EF2 FE1 , EG1 FE2 ,
2H 2H 2
1
EG2 2 FE2 FG1 .
2H 2
Hence equation (9.6.5) and (9.6.6) are differential equation of a geodesic in Gauss coefficients.
9.7 Single differential equation of geodesics on surface
r r
When a curve on the surface r r ( u, v ) may be determined by a single relation
between the parameters u and v either by v = v (u) or by u = u (v).
r r
Proof : Let the equation of surface be r r (u , v ) , u, v are parameters. A curve on a surface
may be determined by a single relation between the parameters u and v, which paves us our way to
develope a single relation between the parameters from pair of following differential equation of a geo-
desic in Gauss coefficients given by
u l u 2 2m uv n u 2 0 .....(9.7.1)
and v u 2 2 uv v2 0 .....(9.7.2)
First taking a single relation v = v (u). .....(9.7.3)
then we shall reduce pair of equation (9.7.1) and (9.7.2) in to single differentiating equation.
On a geodesics, on differentiating equation (9.7.3) with respect to s
dv dv du dv
v u . .....(9.7.4)
ds du ds du
Again differentiating, we get
dv d 2v
v u u 2 2 . .....(9.7.5)
du du
Putting values from equation (9.7.4) and (9.7.5) in to equation (9.7.2), we get
2 2
dv 2 d v dv dv
u u 2
u2 2 u u u 0
du du du du
2
d 2v dv dv dv
or u 2 2
u u2 2 u 2 u 2 0 . .....(9.7.6)
du du du du
Now from equation (9.7.1),
dv
u lu2 2muv nv2 , v u
du
204
Using it in equation (9.7.6), we get
2 3 2
d 2v dv dv dv dv dv
u 2 2
lu 2 2mu 2 nu2 u 2 2u2 u 2 0
du du du du du du
.....(9.7.7)
Cancelling u 2 throughout, we get
3 2
d 2v dv dv dv
2
n 2m l 2 (as u 0 ) .....(9.7.8)
du du du du
which is single differential equation of geodesic.
Equation (9.7.8) is a second order ordinary non-linear differential equation, so it has a unique
dv
solution of v [v = v (u)] which takes a given value v0 when u= u0, also takes a value
du
dv dv
at u = u .
du u u0 du 0 0
Therefore through each point of a surface there passes a unique geodesic in a given direction.
If we start by taking the relation between parameters u and v, of the from u = u (v), then we get
the single differential of geodesic as
3 2
d 2u du du du
2
2 l 2m n . .....(9.7.9)
dv dv dv dv
On comparing equation (9.7.9) with equation (9.7.8), we see that equation (9.7.9) can be ob-
tained from (9.7.8) on interchanging u and v, and changing l, m, n by and vice-versa.
Remark :
1. Note that unlike lines of curvature and asymptotic lines, geodesics are not determined uniquely
by the nature of surface.
2. Through any point there passes an infinite number of geodesic so each geodesic being de-
cided by its direction at the point.
3. A geodesic is uniquely determined by an initial point and tangent at that point of the
surface.
4. The differential equation of geodesic [equation (9.4.6) and (9.4.7)] are in terms of E, F, G
and E1, E2, F1, F2, G1, G2, (derivatives of E, F, G). Therefore if a surface is deformed
without stretching such that the length ds of each arc element does not change, then the geo-
desics remain geodesics on the deformed surface.
205
9.8 On the general surface, a necessary and sufficient condition that the curve
v = c (const.) be a geodesic is EF2 + FE1 – 2EF1 = 0, when v = c, for all values
of u.
On the curve v = c (constant), we may take u as a parameter, so that v = c, u = t (say)
v 0, u 1 . .....(9.8.1)
On differentiating with respect to t
du
v 0, u 1 u .....(9.8.2)
dt
1
Hence T
2
Eu 2 2Fu v Gv2 . .....(9.8.3)
Now differentiating (9.8.3) partially with respect to u and u , we get
T 1 E F G 2
u 2 2 u v v
u 2 u u u
1 E
E1 1 2 F1 1 0 G1 0 u 1, v 0, E1 etc.
2 u
T 1
or E1 .....(9.8.4)
u 2
T 1
and 2 Eu 2 F 1 v E u 1, v 0 .....(9.8.5)
u 2
Now differentiating equation (9.8.3) partially with respect to v and v , we get
T 1 E 2 F F 2 1
u 2 u v v E2 .....(9.8.6)
v 2 v v v 2
E
[by equation (9.8.2) and E2 ]
v
T 1
2 Fu 2Gv F .....(9.8.7)
v 2
d T T d E 1 dE 1 E E E1
Now U (say) dt 2 E1 du 2 1 1
2
dt u u
E1
U .....(9.8.8)
2
d T T dF 1 1
and V (say) E2 F1 E2 .....(9.8.9)
dt v v dt 2 2
dF dF
F1
dt du
206
Now the necessary and sufficient condition for the curve u = u (t) , v = v (t) (t-parameter) to
be geodesic is
T T for all values of t.
V U 0, .....(9.8.10)
u v
On using values from equation (9.8.5), (9.8.7), (9.8.8) and (9.8.9), we get
T T 1 1
V U F1 E2 E E1 F 0
u v 2 2
EF2 FE1 2 EF1 0 . .....(9.8.11)
Hence the result.
Note : If the parametric curves are orthogonal then F = 0 and obviously F1 = 0, then from
above equation (9.8.11) we get, EE2 + 0 – 0 = 0
E2 = 0, when v = c, for all values of u,
E is independent of v, so E is function of u only i.e., E = E (u).
9.9 The curve u = c (constant) is a geodesic if and only if GG1 + FG2 – 2GF2 = 0.
On the curve u = c, we may take v as a parameter, so
u = c, u = t t-parameter, then
u 0, v 1 .
Now proceeding exactly on the same lines as in 9.8, we get
GG1 + FG2 – 2 GF2 = 0. .....(9.9.1)
Note : If the parametric curves are orthogonal then F = 0 and F2 = 0, then from above equa-
tion (9.9.1), we get
G dG
GG1 = 0 G1 = 0 0 0 G is independent of u.
u dc
Hence G is function of v only i.e., G = G (v).
9.9.1 Self-learning exercise-1
1. What is normal property of geodesic ?
2. Define geodesic.
r r
3. Write the general differential equations of geodesics on a surface r r (u , v ) .
r r
4. Write the canonical equations of a geodesic on the surface r r (u , v ) .
5. Like lines of curvature and asymptotic lines, can geodesic be determined uniquely by the na-
ture of surface ?
r r
6. On the general surface r r (u , v ) what is the necessary and sufficient condition that the
curve v = c (constant) be a geodesic ?
207
9.9.2 Illustrative Examples
Ex.1. Prove that the curves u + v = constant are geodesics on a surface with metric
(1 + u2) du2 – 2uv and v + (1 + v2) dv2.
Sol. The parametric equation of the given curve u + v = constant, can be taken as u = t,
v = c – t. .....(1)
So on differentiating these with respect to t
u 1, v 1 .....(2)
Here E = 1 + u2, F = – uv, G = (1 + v2) .....(3)
Now we know that
1
T
2
Gv 2
Eu 2 2 Fuv .....(4)
Using values of E, F and G from equation (3), we get
1
T 1 u 2 u 2 2uvuv
11 v 2 v 2 .....(5)
2
On differentiating (5) with respect to u and v, we get
T
u u 2 vu v t 1 c t 1 c ......(6)
u
T
u u v v v 2 t c t c ......(7)
v
Now differentiating (5) with respect to u and v , we get
T
1 u 2 u uv v 1 t 2 t c t 1 1 ct ......(8)
u
T
v
u v u 1 v 2 v t c t 1 c t
2
1 ct 1 c
2
......(9)
d T T d
Now U 1 ct c c c 0 ......(10)
dt u u dt
d T T d
and V ct 1 c 2 c 0 ......(11)
dt v v dt
T T
then V U 0 1 ct 0 ct 1 c 2 0 [by equation (8), (9), (10) and (11)]
u v
T T
Hence the relation V U 0,
u v
for all values of t.
Therefore the given curve u + v = constant is a geodesic.
208
v3
Ex.2. Prove that the curves of the family constant, are geodesics on a surface with
u2
metric v2 du2 – 2 uv dv du + 2 u2 dv2 ; (u > 0, v > 0).
Sol. The parametric equation of the given curve can be conveniently chosen to be
u = ct3, v = ct2, c = (constant) .....(1)
The differentiating equation (1) with respect to t
u 3ct 2 , v 2ct .....(2)
then E = v2, F = – uv, G = 2u2 .....(3)
1
Therefore, T
2
Gv 2
Eu 2 2 Fuv .....(4)
Using values of E, F and G from (3) in (4), we get
1 2 2
T
2
2u 2 v 2
v u 2uv uv .....(5)
On differentiating equation (5) partially with respect to u
T
2uv 2 ct 2 3ct 2 2ct 2 ct 3 2ct 2
vuv
u
6c 3t 5 8c 3t 5 2c 3t 5 [by equation (1) and (2)]
Now partially differentiating equation (5) with respect to v, we get
T
vu 2 uuv
3c 3t 6 [by (1), (2)]
v
Now differentiating equation (5) partially with respect to u and v , we get
T
v 2 u uvv c3t 6
u
T
and uvu 2u 2 v c 3t 7
v
d T T d
Now U c 3t 6 2c3t 5 4c 3t 5
dt u u dt
d T T d
and V c 3t 7 3c 3t 6 4c 3t 6
dt v v dt
T T
V U 4c t c t 4c t c t 4c t 4c t 0
3 6 3 6 3 5 3 7
then 6 12 6 12
u v
T T
Hence V U = 0 , for all values of t.
u v
v3
This shows that the given family 2 constant (c) is a geodesic for all values of c.
u
209
9.10 D i f f er ent i al equat i on of a geodesi cs on t he sur f ace F (x, y, z) = 0.
We have discussed about the differential equation of a geodesic on a surface whose equation
was given in parametric form. Now we shall find equation of geodesic when equation of surface is given
in the implicit form F (x, y, z) = 0.
Equation of surface is F (x, y, z) = 0. .....(9.10.1)
We know that if r be the position vector of a point
dr
then r tˆ
ds
Again differentiating with respect to s
r tˆ n
or r nˆ Nˆ , ( for geodesic n̂ Nˆ ) .....(9.10.2)
where is the curvature.
But
r xiˆ yjˆ zkˆ r xiˆ yˆj z kˆ
.....(9.10.3)
d 2x
where x etc.
ds 2
and N̂ be normal at a point on the surface F (x, y, z) = 0, so
F ˆ F ˆ F ˆ F ˆ F ˆ F
iˆ j k i x j y k z
x y z
Nˆ
.....(9.10.4)
2 2 2 2 2 2
F / x F / y F / z Fx Fy Fz
where Fx F / x etc.
F ˆ F ˆ F
iˆ j k
d x 2
d y d z ˆ
2
x 2
y z
Therefore ˆ
r N 2 ˆ
i ˆ
j k .....(9.10.5)
ds ds 2 ds 2 Fx2 Fy2 Fz2
Comparing coefficients of iˆ, ˆj and k̂ , we get
d2x
F / x d 2
x / ds 2
.....(9.10.6)
ds 2 Fx2 Fy2 Fz2 F / y Fx2 Fy2 Fz2
d 2
y / ds 2
Similarly F / y .....(9.10.7)
Fx2 Fy2 Fz2
d 2
z ds 2
. .....(9.10.8)
F y F Fy2 Fz2
x
2
210
Hence from equation (9.10.6) to (9.10.8), we get
d 2
x ds 2
d 2
y ds 2
d 2
z ds 2
. .....(9.10.9)
Fx Fy Fz
If the integral of one of these equations is found, it will contain two arbitrary constants and with
the equation to the surface F (x, y, z) = 0, will represent the geodesics.
9.11 Differential equation of geodesics on the surface z = f (x, y), the Monge’s form.
Taking x, y as parameters, let r be the position vector of any point on the surface z = f (x, y),
then
r = (x, y, f (x, y) = z). .....(9.11.1)
On differentiating with respect to x and y partially. we get
r r z z
r1 1, 0, p , r2 0,1, q , p ,q
x y x y
2r 2r
2r
r11 2 0, 0, r , r12 0, 0, s , r22 2 0, 0, t . .....(9.11.2)
x xy y
Hence E = r1 r1 = 1 + p2, F = r1 r2 = pq, G = r2 r2 = 1 + q2
H2 = EG – F2 = ( 1 + p2 ) ( 1 + q2 ) –p2 q2 = ( 1 + p2 + q2 ) .....(9.11.3)
Further, the Gauss coefficients are obtained as
pr ps pt
l 2
, m 2 , n 2
H H H
qr qs qt
and 2
, 2, 2 , .....(9.11.4)
H H H
1 2
where l H GE1 2 FE1 FE2 etc.
2
Hence the single differential equation
3
d 2v dv dv dv
2
n 3m l 2
du du du du
with u = x, and v = y becomes
3 2
d2y dy dy dy
H2 2
pt 2 ps qt pr 2qs qr
dx dx dx dx
2
d 2 y dy dy dy
1 p 2 q 2 dx 2
p
dx
q 2s r
t
dx dx .....(9.11.5)
which is differential equation of geodesic for surface z = f (x, y).
211
9.12 Geodesic on a surface of revolution
Let the surface of revolution be
x = u cos , y = u sin , z = f (u) .....(9.12.1)
Let r be position vector of a point on this surface
then r = (u cos , u sin , f (u)). .....(9.12.2)
Differentiating with respect u and partially, we get
r r
r1 cos ,sin , f , r2 u sin , u cos , 0
u r
2r 2r
and r11 2 0, 0, f , r12 sin , cos , 0 ,
u r
2r
r22 2 u cos , u sin , 0 .....(9.12.3)
then E r1 r1 1 f 2 , F r2 r2 0, G r2 r2 u 2
r1 r2 u f cos , f sin ,1 , H2 = EG – F2 = u2 (1 + f2) ......(9.12.4)
1 EE2
2
Also 2 EF1 EE2 FE1 ( F = 0)
2H 2H 2
but E2 E / 0
1 1 1
2
0 and EG1 FE2 , EG2 2 FF2 FG1 = 0. ......(9.12.5)
2H u 2H 2
For the present form of geodesic, we use the equation of geodesics given below
v u 2 2u v v2 0 ,
d 2 2 du d d 2v
2
0 0 0, where v etc. ......(9.12.6)
ds u ds ds ds 2
d 2 du d
Multiplying by u2 , we get u2 2
2u 0,
ds ds ds
which can be expressed as d 2 d . ......(9.12.7)
u 0
ds ds
d
On integrating, we get u2 h1 . (say) ......(9.12.8)
ds
which is called the first integral of the equation of geodesic, where h1 is constant of integration.
It is independent of form of f (u).
212
Now we shall find the complete integral of equation (9.12.8), for this we proceed as follows :
The metric of surface r r u, v is
ds2 = Edu2 + 2Fdu dv + Gdv2 . ......(9.12.9)
In the present case, u = u, v = , then above equation reduces to,
ds2 = Edu2 + 2Fdu d + G d2. ......(9.12.10)
Using values of E, F and G from equation (9.12.4), we get
ds2 = (1 + f2) du2 + u2 d2 ......(9.12.11)
Now equation (9.12.8) may be expressed as
u2 d = h1ds
or u4 d = h12 ds 2
or
u4 d h1 1 f du u d
2 2 2 2 2
u4 d – h12u2 d h1 1 f du
2 2 2
u2 (u2 – h12) d h1 1 f du
2 2 2
1/ 2 1/ 2
or u u 2 h12 d h1 1 f 2 du
1/ 2
h1 1 f 2
d 2 du ......(9.12.12)
u u h12
On integrating, we get
1/ 2
h1 1 1 f 2
du c1
1 u u 2 h12
......(9.12.13)
where c1 is again another arbitrary constant of integration.
As the differential equation of geodesic is of second order and its solution (9.12.13) involves
two arbitrary constants h1 and c1.
Hence, it is the complete integral of the differential equation of geodesic on the surface of revo-
lution.
Note : If the arbitrary constant h1 = 0 in equation (9.12.13), then = 0 + c1 = c1
(constant), which in this case is geodesics and are the meridians.
Hence every meridian is a geodesic on the surface of revolution.
9.13 Clairut’s theorem
If the geodesic on the surface of revolution intersects the meridian ( = constant) at any point P
at an angle , then u sin is constant, where u is the distance of point P from the axis.
213
merid the
ian
nt to
ic
O d es
o
Tange
ge
t o
i a n e nt
id ng
Mer
Ta
cti lar
se ircu
on
C
90 Q
P
geodesic
Fig . 9.1
Geometrical derivation of the result :
The projection of arc ds of the geodesic on the circular section (See Fig. 9.1) through point P is
ds cos (90 – ) = PQ .....(9.13.1)
This arc PQ subtend an angle dat the centre O of the circular section of radius uˆ, but by
formula,
arc
angle =
radius
arc PQ
d arc PQ u d .....(9.13.2)
u
Using (9.13.2) into (9.13.1), we get
ds sin = u d
d
sin u .....(9.13.3)
ds
d
But u h1 (constant)
ds
d
uu h1
ds
u sin h1 constant,
this is called Clairut’s theorem.
It may be stated explicitly as follows :
At every point of a geodesic on a surface of revolution, the radius (u) of the circle of latitude
multiplied by the sine of the angle between the geodesic and the meridian is constant.
214
Remark : From above equation (9.13.4), we draw one important conclusion that h1 is the mini-
mum distance from the axis of a point on the geodesic, and is attained at the point where the geodesic
cuts a meridian at right angles.
9.14 A geodesic on a surface of revolution cuts the meridian at a constant angle, then
the surface is a right circular cylinder.
The equation of the surface of revolution is given by
x = u cos , y = u sin , z = f (u). .....(9.14.1)
Then by Clairut’s theorem, if a geodesic on a surface of revolution cuts the medium through any
point P on it at angle , then we have
u sin = h1 (constant), .....(9.14.2)
2 2
where u x y , is the distance of the point P from the axis.
But, it is given that a geodesic cuts all the meridians at constant angle, so = constant and
h1 constant
u = (constant) = a (say)
sin constant
u2 = a2
x2 + y2 = a2. .....(9.12.4)
which is equation of a right circular cylinder, whose axis is z-axis and radius is a.
9.15 A curve on sphere is a geodesic if and only if it is a great circle.
(i) The condition is necessary : Let C be a geodesic curve on a sphere. Let n̂ be the normal
to curve C at a point P and let N̂ be the normal to the surface of the sphere at P. Then by the normal
property of the geodesic
n̂ Nˆ . .....(9.15.1)
At point of the sphere, the normals pass through the centre of sphere, also the principal normal
at every point of C will pass through the centre of sphere, which is a fixed point for the sphere.
Let r be the position vector of point P on the geodesic and let â be the position vector of the
centre of the sphere.
Then
r aˆ nˆ , .....(9.15.2)
where is a scalar parameter which is function of arc lengths.
Now differentiating equation (9.15.2) with respect to s, we have
dr dnˆ d
nˆ
ds ds ds
215
d
ˆ
tˆ b tˆ nˆ ds [by Serret-Frenet’s formulae]
ˆ ˆ d
or tˆ b t nˆ ds .....(9.15.3)
Equating the coefficients of b̂ on both sides, we get
0 = 0 as 0 torsion = 0. .....(9.15.4)
Hence the curve C is a plane curve. So the curve C (geodesic) on the sphere is a plane curve
whose normals at each point of it pass through the centre of the sphere. Hence C is a section of the
sphere by plane passing through its center. That is, C is a great circle.
(ii) The condition is sufficient : If C is a great circle on the sphere, then at each point of C
the principal normal to C coincides with the surface normal to the sphere. Therefore by normal property
curve C is a geodesic.
9.16 Geodesic curvature and torsion of a geodesic
Geodesic (tangential) curvature : Let S be any surface r r u, v and C be a curve on this
surface. Let r be position vector of a point P on the curve C.
N
surface
normal D
n Principal
normal to the
curve C at P.
n
B
P geodesic
g
r2
Q n e
t C t pla
n
n ge face
Ta sur
r1 to
r
Fig 9.2
216
Then the curvature vector of a curve C on surface at a point P, with the tangent direction tˆ is
dtˆ
r nˆ .....(9.16.1)
ds
and it lies in the plane (see -plane in Fig. 9.2), through P perpendicular to tˆ , this plane also contains
the surface normal N̂ .
Then according to the Meusnier’s theorem, the projection of the curvature vector on this
surface normal N̂ is, the curvature vector of the normal section in direction tˆ . It is represented by n .
Our main aim is to study the projection of on the tangent plane, which is called the vector of tangential
curvature. It has been denoted by symbol g . So we have the relation
n g . .....(9.16.2)
The equation (9.16.2) implies that the curvature vector is the sum of the normal curvature and
tangential curvature vectors.
The tangential curvature vector g is generally called as geodesic curvature vector..
9.17 An expression for g and that it is intrinsic
We know that the curvature vector r at any point P on a curve C can be expressed as a linear
combination of vectors
r r
Nˆ , r1 , and r2
u v
as given below r n Nˆ r1 r2 , .....(9.17.1)
where and are scalars.
Taking dot product by N̂ , we get
r Nˆ n Nˆ Nˆ r1 Nˆ r2 Nˆ
r Nˆ n 0 0 Nˆ Nˆ 1, r1 Nˆ 0 r2 Nˆ
n r Nˆ .....(9.17.2)
So it is deduced that g r1 r2 .....(9.17.3)
Now taking dot product of equation (9.17.1) by r1 , we get
r r1 n Nˆ r1 r1 r1 r2 r1
0 r12 r2 r1
or r r1 E F E r12 , F r1 r2 .....(9.17.4)
but d T T
r r U ,
ds u u
217
then from equation (9.17.4),
U E F . .....(9.17.5)
Now taking dot product of equation (9.17.1) by r2 , we get
r r2 n Nˆ r2 r1 r2 r2 r2
or r r 0 F G G r2 r2
d T T
and r r ds v v V
U F G. .....(9.17.6)
Solving equations (9.17.5) and (9.17.6), we get
1 1
2
GU FV , 2 EV FU . .....(9.17.7)
H H
Equation (9.17.7) shows that values of and are intrinsic.
Hence the geodesic curvature g is intrinsic.
Now in case of geodesic r is parallel to N̂ , therefore geodesic curvature vector (, is zero
for geodesic.
9.18 The geodesic curvature vector of any curve is orthogonal to the curve
dr ˆ
Proof : r t , again differentiating
ds
r tˆ nˆ .....(9.18.1)
and r can be expressed as a linear combination of the vectors N̂ , r1 and r2 , so we can write
r n Nˆ r1 r2 . .....(9.18.2)
Now, form (9.18.1) and (9.18.2), we have
n nˆ n Nˆ r1 r2 .....(9.18.3)
Taking to product by tˆ n nˆ tˆ n Nˆ tˆ r1 r2 tˆ
0 0 r1 r2 tˆ nˆ tˆ 0, Nˆ tˆ 0
0 g tˆ
g is orthogonal to tˆ . g
r1 r2
Hence geodesic curvature vector g is orthogonal to the curve.
Hence the result.
218
9.19 Formulae for geodesic curvature
If parameter s (arc length), then show that geodesic curvature g N̂ r r , and if
we replace parameter s by t, then show that
1 T T
g V t U t .
3
Hs u v
Proof : In the article 9.18, we have proved that the geodesic curvature vector g of a curve is
orthogonal to the curve. Also we know that vector g lies in the tangent plane (See Fig. 9.2), so it is
perpendicular to the surface normal vector N̂ also. Thus g is orthogonal to both the unit vectors r tˆ
and N̂ . Hence it is parallel to the unit vector N̂ r . Hence
g g Nˆ r , where g g . .....(9.19.1)
Now r n Nˆ g
r n Nˆ g Nˆ r .....(9.19.2)
Now taking scalar product by N̂ r , we get
N̂ r r Nˆ r Nˆ Nˆ r Nˆ r
n g
or N̂ r r = 0 + g 1
Nˆ r Nˆ 0 and Nˆ r Nˆ r 1
g Nˆ r r .....(9.19.3)
which is the required result.
Now, if we replace the parameter s by t,
dr dr / dt r
then
r .....(9.19.4)
ds ds / dt s
Again differentiating with respect to ‘s’, we get
d r d r dt s
r r
s s
r 0 r
r Q & 0
s& .....(9.19.5)
ds s dt s ds s3 s3 s2
Using equation (9.19.4) and (9.19.5) in (9.19.3), we get
ˆ r
r 1 ˆ
g N 2
3 N r r .....(9.19.6)
s
s s
219
r r
But we know that Nˆ 1 2
H
1 r1 r2 1
g 3 r r r r2 r r
Hs
3 1
s H
1 1 r1 r r1 r
or g 3
Hs
r1 r2 r r 3
Hs r2 r
r2 r
or g
1
Hs3
r r r r r r r r
1 2 1 2 (by Lagrange’s Identity) .....(9.19.7)
1 du
Now, if we take T
2
Eu 2 2 Fuv
Gv 2 (where u
dt
etc.) .....(9.19.8)
dr r du r dv
then r r1 u r2 v .....(9.19.9)
dt u dt v dt
2
Squaring, we get r 2 r1 u r2 v ,
then r 2 r1 u r2 v r1 u r2 v
r1 r1 u 2 r1 r2 u v r2 r1 u v r2 r2 v 2
Eu 2 Fu v Fu v Gv 2
Eu 2
2 Fu v Gv 2
r 2 = 2T [by equation (9.19.8)]
1 2
or T r .....(9.19.10)
2
Differentiating with respect to u
T 1 r
2r r r1 u r2 v [by equation (9.19.9)]
u 2 u u
r r1 0 r r1
T
or r r1
.....(9.19.11)
u
T
Similarly, r r2
.....(9.19.12)
v
Now differentiating equation (9.19.10) with respect to u, we get
T 1 r
2r r r1 u r2 v [by equation (9.19.9)]
u 2 u u
d
r r11 u r21 v r r1
[just as equation (9.19.9)]
dt
220
T d
or r r1 .....(9.19.13)
u dt
T d
Similarly, r r2 .....(9.19.14)
v dt
d T T
Now we know that U (t)
dt u u
d d
dt
r r1 r r1
dt
d r1 d r1
or U (t) r r1 r r r r1 .....(9.19.15)
dt dt
Similarly, V (t)
r r2 .....(9.19.16)
Now, using equation (9.19.11), (9.19.12) and (9.19.15), (9.19.16) in equation (9.19.7), we get
1 T T
g V t U t .....(9.19.17)
Hs u v
If parameter t = s, so that s 1 then we get
1 T T
g V s U s .....(9.19.18)
H u v
Hence, equation (9.19.17) and (9.19.18) are formulae for g.
Another form : We know that u U (s) + v V (s) = 0
v
U (s) V s .....(9.19.19)
u
Putting this value in (9.19.18), we get
1 T v T
g V s V s
H u u v
V s T T V s T T
u
H u u
v
v Hu u u v v 2T 1
V s
or g .....(9.19.20)
Hu
which is value of g in terms of V (s).
Similarly in terms of U (s), g will be
U s
g . .....(9.19.21)
Hv
221
9.20 Geodesic curvature in terms of Gauss coefficients.
r r r
We know that r r1 u r2 v .......(9.20.1)
Again differentiating with respect to ‘s’, we get
r r r r r r
r r11u 2 2r12 u v r22v2 r1 u r2 v ......(9.20.2)
r r r r r 2 r r r r
2
then r r1 r1u r2 v r11 u 2r12 uv r22 v r1 u r2v
r r r r r r r r r r
r1 r11 u3 2r1 2r12 r2 r11 u2v r1 r22 2r2 r12
r r r r
u v2 r2 r22 v3 r1 r2 uv uv .....(9.20.3)
r r
Now, g N̂ r r
r r r r r r
Nˆ r r11 u 3 Nˆ 2r1 r12 Nˆ r2 r11 u 2v
r r r r r r r r
Nˆ r1 r22 Nˆ 2r2 r12 uv2 Nˆ r2 r22 v3 N̂ r1 r2 u v u v
Hu 3 2H l u2 v vH 2 mH u v2 nH v3 H uv vu
or 2 2
2
k g Hu v u 2u v v Hv u lu 2muv nv
2
......(9.20.4)
which is the formula for geodesic curvature in terms of Gauss coefficients , , , l, m and n.
r r
Since N̂ r1 r2 Nˆ HNˆ H
r r r1 r2 r r 1 r r r r r r r
N̂ r1 r11 r1 r11 r11 r2 r12 r1 r11 r2 r1
H H
1 1 1 1
H F1 2 E2 E 2 E1F 2 H 2 EF1 EE2 FE1 H
Similarly
r r r r
Nˆ r1 r12 H , Nˆ r1 r22 H
r r r r
Nˆ r2 r11 lH , Nˆ r2 r12 mH and
r r
Nˆ r2 r22 nH . ......(9.20.5)
9.21 Geodesic curvature for parametric curves.
(i) For the parametric curve v = c (constant) i.e., u = curve, we have
v= 0, v= 0 ......(9.21.1)
But
k g Hu v u2 2uv v2 Hv u lu2 2mu v nv2
kg u k g v c Hu 0 u2 0 0 H 0 u lu 2 0 0
222
kg u Hu3 ........(9.21.2)
2
1 du 1 1
Now ds 2 Edu 2 u 2 u ........(9.21.3)
E ds E E
Then equation (9.21.2) reduces to
kg u H E 3 / 2 . ......(9.21.4)
(ii) For the parametric curve u = (constant) c (say) i.e., v = curve, then u = 0, u = 0 and
proceeding as above we get
kg v kg u c nHv3 .
kg v nHG 3 / 2 Q v G 1/ 2
.......(9.21.5)
Remark : If the parametric curves are orthogonal, we have
E2 G1
F = 0, this gives H2 = EG, and n
2G 2E
Hence kg u 2EE2 G 1
EG v
E ........(9.21.6)
and kg v 2G G1E 1
EG u
G. .....(9.21.7)
Ex.3. Find the geodesic curvature of the curve u = constant, on the surface
1 2
x = u cos , au .
y = u sin , z
2
Sol. Let rr be position vector of any point on the surface then
r 1
r u cos , u sin , au 2
2
Differentiating with respect to u and
r
r r r
r1 cos ,sin , au , r2 u sin , u cos , 0
u
r r r r
then E r1 r1 1 a 2u 2 , F r1 r2 0 .....(1)
r r G
G r2 r2 u 2 , then G1 2u .....(2)
u
Since kg uconsant GHn ,
3/ 2 but n
G1
2E
G1 2u
u c
kg
2G E 2u 2 1 a 2u 2
u
, which is the required result.
2 2
1 a u
223
9.22 Normal angle
Definition : The angle between the principal normal n̂ and the surface normal N̂ is known as
normal angle, it is denoted by symbol w.
(i) Here angle w is positive if the rotation from n̂ to N̂ is in the sense from n̂ to binormal bˆ.
b
al
rm
no N
ce
rfa
Su
w
O
n
t Principal normal
Fig. 9.3
(ii) Angle w is negative if the rotation from n̂ to N̂ is in the sense from b̂ to nˆ.
All three vectors Nˆ , nˆ and b̂ lie in the same plane. So angle between N̂ and b̂ is w .
2
Nˆ bˆ cos 90 w sin w ......(9.22.1)
and Nˆ nˆ cos w. ......(9.22.2)
9.23 Geodesic curvature in terms of normal angle.
r r
We know that g Nˆ r r
r r r
or g Nˆ tˆ tˆ Q r tˆ, r t
or g Nˆ tˆ nˆ tˆ nˆ
Nˆ tˆ nˆ
Nˆ tˆ nˆ Nˆ bˆ tˆ nˆ bˆ
or g sin w .....(9.23.1)
1
or g sin w, .....(9.23.2)
1
where is the radius of curvature of the curve.
224
Remark : We also know that
r Nˆ n g
Taking dot product with Nˆ , we gets
r Nˆ n Nˆ Nˆ g Nˆ
cos w n 1 0 g Nˆ 0 r Nˆ nˆ Nˆ cos w
or n cos w . .....(9.23.3)
Now, on dividing equation (9.23.1) by (9.23.3), we get
g
tan w g n tan w .....(9.23.4)
n
On squaring and adding equations (9.23.1) and (9.23.3), we get
2g n2 2 . .....(9.23.5)
9.24 Expression of the torsion of a geodesic on any surface, and that the torsion of
an asymptotic line is equal to the torsion of its geodesic tangent.
Let c be a curve on a surface S, let r be position vector of any point P of curve c, then
bˆ tˆ nˆ. .....(9.24.1)
Differentiating with respect to s (arc length), we get
dbˆ dtˆ dnˆ
nˆ tˆ .....(9.24.2)
ds ds ds
By Serret-Frenet formulae, equation (9.24.2) is reduced to
dnˆ
nˆ nˆ nˆ tˆ
ds
dnˆ
or nˆ 0 tˆ .....(9.24.3)
ds
Now, if the curve C is geodesic on surface S, then nˆ Nˆ , we also denote by g as the torsion
of the geodesic, then by equation (9.24.3), we get
dNˆ .
g Nˆ tˆ
ds
Taking dot product by Nˆ , we get
g Nˆ Nˆ Nˆ tˆ Nˆ Nˆ Nˆ tˆ
or
g Nˆ Nˆ tˆ Nˆ Nˆ tˆ , .....(9.24.4)
which is the basic expression for the torsion of a geodesic on a surface S.
225
Torsion found in equation (9.24.4) is same as the torsion of an asymptotic line. The geodesic
which touches a curve at any point is often called its geodesic tangent at the point. Hence the torsion of
an a asymptotic line is equal to the torsion of its geodesic tangent.
9.25 Expressions for the torsion of a geodesic in terms of fundamental magnitudes
and also in terms of principal curvatures.
We know that Torsion of a geodesic is
g Nˆ Nˆ r .....(9.25.1)
dNˆ Nˆ du Nˆ dv ˆ
Now Nˆ N1 u Nˆ 2 v .....(9.25.2)
ds u ds v ds
dr r du r dv
and r r1u r2 v ......(9.25.3)
ds u ds v ds
N̂ r Nˆ1 u Nˆ 2 v r1 u r2 v
Nˆ1 r1 u2 Nˆ 1 r2 uv Nˆ 2 r1 u v Nˆ 2 r2 v2
Taking dot product of both sides with vector Nˆ , we get
Nˆ Nˆ r Nˆ Nˆ 1 r1 u 2 Nˆ Nˆ1 r2 u v Nˆ
Nˆ
r1 u v Nˆ Nˆ 2 r2 v2
2
2
or Nˆ Nˆ r g Nˆ Nˆ1 r1 u Nˆ Nˆ 1 r2 uv Nˆ Nˆ 2 r1 u v Nˆ Nˆ 2 r2 v2
.....(9.25.4)
But Nˆ Nˆ 1 r1 EM FL , Nˆ Nˆ r FM GL
1 2
H H
Nˆ Nˆ 2 r1 EN FM , Nˆ Nˆ r FN GM . .....(9.25.5)
2 2
H H
Using equation (9.25.5) in to (9.25.4), we get
1
g EM FL u 2 FM GL EN FM u v FN GM v 2
H
1
or g
H
EM GL u 2
EN GL u v FN GM v2 , .....(9.25.6)
which is an expression for g in terms of fundamental magnitudes.
To find expression in terms of principal curvatures chose the lines of curvature as parametric
curves so that F = 0, M = 0 and H2 = EG H EG , then above equation (9.25.5) reduces to the
following form
g
1
EN GL uv EG GN EL uv. .....(9.25.7)
EG
226
Now let be the angle which the geodesic makes with the parametric curve v = c (constant),
then
cos u E and sin v G .....(9.25.8)
Then using these in equation (9.25.7), we get
N L
g sin cos .....(9.25.9)
G E
But the principal curvatures a and b are given by
L N
a and b , .....(9.25.10)
E G
So that (9.25.9) can be written as
1
g a b sin 2, .....(9.25.11)
2
which is expression of g in terms of principal curvatures.
From equation (9.25.11), it follows that g is maximum when
2 = 90° / 4,
hence the geodesic bisecting the angle between the line of curvature has maximum torsion.
Remark : If g and g be the torsions of two orthogonal geodesics then from equation (9.25.11)
above, we have
1
g b a sin 2 .....(A)
2
1
and g 1 b a sin 2 b a sin 2
2 2 2
1
g b a sin 2 .....(B)
2
In adding equation (A) and (B), we get
g g 0 g g ,
which shows that two orthogonal geodesics have their torsion equal but opposite in sign.
9.26 Some important definition
(i) Simply connected region R : The region R, in which every closed curve lying in the region
R on a surface can be contracted continuously into a point without leaving R, is called simply
connected region [See Fig 9.4].
(ii) Excess of a closed curve C [ex (c)] : Suppose a simply connected region R (See Fig 9.3)
be enclosed by a closed curve (say) C, consisting of n arcs A0 A1, A1 A2, ...., An–1 An, where
227
A0 = An, making at the vertices exterior angles 1, 2, ..., n; then the excess of closed curve C is
denoted by ex (c) is defined as
3
A3
C
2
A2
R
A 1
n A1
A0 = A1
Fig.9.4
ex (c) = 2 – (1 + 2 + ... + n) g ds
c
n
or ex (c) = 2 r g ds .....(9.26.1)
r 1 c
where g being the geodesic curvature of the arcs.
(iii) Total curvature of R : The total curvature or Gaussian curvature of an arc on a surface is
denoted by K and is given by
LN M 2 T 2
K .....(9.26.2)
EG F 2 H 2
1 FE2 G1 1 2 F1 E2 FE1
or K .....(9.26.3)
2 H u EH H 2 H v H H EH
Therefore, the total curvature of a simply connected region R is given by
ds. .....(9.26.4)
R
9.27 Gauss-Bonnet theorem
Statement : Any curve which encloses a simply connected region R, the excess of the closed
curve C is equal to the total curvature of R, i.e., ex c ds.
R
Proof : Let us consider a surface r r u, v of class 3 with u, v as parameters, let c be a closed
curve, which is boundary of a simply connected region R on the surface. (see Fig 9.4). Let c consists of
n (finite) smooth arcs
A0 A1, A1 A2, ..., An–1 An ; (where A0 = An)
228
such that each arc is of class 2 and these are positively described in anti-clockwise direction. Let
r(r = 1, 2, 3,..., n) be the exterior angle between the tangents to the arcs Ar–1 Ar and Ar Ar +1 at the
vertex Ar, measured with usual convention. So that – < r <
The geodesic curvature g exists at every point of c except possibly at the vertices A r
(r = 1, 2, ..., n).
Then by Liuville’s formula, we have
d H du dv
g .....(9.27.1)
ds E ds ds
or ds d P du Q dv (on integrating over c) .....(9.27.2)
c c
where is the angle between the curve c and the parametric curve v = constant (u-curve) and P, Q are
functions of parameter, u, v, which are given by
H 1
P (u, v) 2 EF1 EE2 FE1 . .....(9.27.3)
E 2 HE
H 1
and Q (u, v) EG1 FE2 . .....(9.27.4)
E 2 HE
Now, the parametric curves v = c (constant) form a family in the region R enclosed by curve C,
the tangent to C turns through 2 relative to these curves, so that
n
d r 2
r 1
c
n
or 2 r d .....(9.27.5)
r 1 c
But by definition of excess of a closed curve C, we have
n
ex (c) 2 r g ds
r 1 c
or ex (c) d g ds [by equation (9.27.5)]
c c
P du Q dv [by equation (9.27.2)]
c
Q P
ex (c) du dv [using Green’s theorem] .....(9.27.6)
R
u v
But in curvilinear coordinates, area of surface element ds (say) is given by
r r r r
ds r1 du r2 dv r1 r2 dudv H dudv .....(9.27.7)
229
ds
or du dv .
H
Using it in (9.27.6), we get
1 Q P
ex (c) ds . .....(9.27.8)
R
H u v
But we know that the intrinsic formula for Gaussian curvature or total curvature at any point
(u, v) on the surface is obtained by
1 1 1 1
K FE2 EG1 2 EF1 EF2 FE1 .....(9.27.9)
H u 2 HE H v 2 HE
1 Q P
or K .
H u v
Using this value in (9.27.8), we get
ex (c) Kds , .....(9.27.10)
R
which shows that the excess of the closed curve c is equal to the total curvature of R.
Hence the theorem.
9.27.1 Self-learning exercise-2
1. What is differential equation of a geodesic on a surface F (x, y, z) = 0 ?
2. Write the differential equation of geodesic on the surface z = f (x, y).
3. State Clairut’s theorem.
4. If a geodesic on a surface of revolution cuts the meridian at a constant angle, is surface a
right cylinder ?
5. Is a curve on sphere a geodesic if it is a great circle ?
6. Define geodesic curvature.
7. Define normal angle.
8. Write geodesic curvature in terms of normal angle w.
9. Are two orthogonal geodesic have their torsion equal in magnitude and sign ?
10. Define excess of a closed curve C.
11. State Gauss Bonnet theorem.
9.27.2 Illustrative examples
Ex.4. Geodesic are drawn on a catenoid so as to cross the meridians at an angle whose
sine is c/u, where u is the distance of the point of crossing from the axis. Prove that the polar
u c
equation to their projections on the xy-plane is e 2( ) , where is an arbitrary constant.
uc
230
Sol. A catenoid is a surface of revolution, obtained by revolving the catenary about its directrix.
Let its equation be
u
x = u cos , y = u sin , z = c cos h–1 .....(1)
c
Then for a geodesic curve on its surface
d
u2 h1 .....(2)
ds
It is given that geodesic cuts the meridians at an angle, say ,
c c
whose sine is , so sin . .....(3)
u u
We know that sin u d .....(4)
ds
c d d
Using (3) in (4), we get u u2 c .....(5)
u ds ds
Then form equation (2) and (5), we have h1 = c.
Now we find that the equation of the geodesic becomes
du u c
c 2 2
log
u c 2 uc
uc u c
or 2 log e 2( )
uc uc
where is constant of integration.
x2 y 2 z 2
Ex.5. A geodesic on the ellipsoid of revolution 2 1, crosses a meridian at an
a2 c
angle at a distance u from the axis. Prove that at the point of crossing it makes an angle
1 cu cos
cos with the axis.
a 4
u 2 a 2 c2
Sol. The given equation of the ellipsoid of revolution may be expressed as
2
x = u cos v, y = u sin v, z c 1 u .....(1)
a2
r
where u, v are parameters. Let r be position vector of a point, then
r u2
r u cos v, u sin v, c 1 2 .
a
231
r uc r
r1 cos v, sin v,
a 2 1 u 2 / a 2 u
r r
and r2 u sin v, u cos v, 0
v
r r r c 2u 2 r r r r
E r1 r1 r12 1 , F r1 r2 0, G r2 r2 u 2
and .....(2)
a 2
a 2
u 2
Then surface element ds is given by
ds2 = Edu2 + 2 Fdu dv + G dv2
2 2
du du dv dv
1 E 2F G
ds ds ds ds
2 2
c 2u 2 du 2 dv
or 1 2 2 u 1 Q F 0 .....(3)
a a u
2
ds ds
Now we know that the first integral of a geodesic on a surface of revolution [of the form
x = u sin v, y = u sin v, z = f (u)] is
dv
u2 G1 .....(4)
ds
Also, it is given that the geodesic crosses a meridian at an angle , therefore
dv
sin u . .....(5)
ds
On using equation (5) in (3), we get
2
c 2u 2 du 2
1 sin 1
2
a a u
2 2
ds
2
c 2u 2 du 2
1 cos .....(6)
2
a a u
2 2
ds
u2
Now z2 c 2 1 2 , on differentiating with respect to s, we get
a
dz 2u du dz c 2u du
2z c2 0 2 .....(7)
ds a ds ds a 2 z ds
232
dz c 2u du
or cos 2 .....(8)
ds a z ds
where angle (say) is the angle which geodesic makes with z-axis.
du a 2 z cos
From equation (8), using in equation (6), we get
ds c 2u
4 2
1 c 2u 2 a z 2 2
4 2 cos cos
2 2
1 a a z
2
c u
a 4 a 2u 2 c 2u 2 a4 z 2
cos 2 cos 2
or
a2 a 2 u 2 4 2
c u
2
cos
c 2u 2 a 2 u 2 2
cos
c 2 u 2 cos 2
2
Q z
c2 a 2 u 2
a2
a 2 z 2 a 4 a 2u 2 c 2u 2 a4 u 2 a2 c2
cu cos cu cos
or cos cos 1
a 4 u 2 a 2 c2
a 4 u 2 a 2 c2
which is the required result.
Ex.6. Prove that the projection on the xy-plane of the geodesics on the catenoid
z
u = c cos h are given by
c
adu
d ,
u 2
c 2
u 2
a 2
where a is an arbitrary constant.
Sol. A catenoid is obtained by revolving a catenary about its directrix, hence its equation is
u
x = u cos , y = u sin , z c cosh 1 f u (say) .....(1)
c
The equation of geodesics on the surface of revolution is given by
a 1 f 2
d 2 du [by equation (9), §9.12] .....(2)
u u a 2
where f df .
du
z du z 1 z
Now by u c cos h , c sin h sin h
c dz c c c
233
dz 1 df 1
or f .....(3)
du sin h z / c du sin h z / c
then by equation (2), we have
1
1 2
a sin h ( z c )
d du [on using values of u and f ]
c cos h z / c u 2 a2
a du 2 z 2 z
c sin h z / c u 2 a2 Q 1 sin h c cos h c
z z u2
or d
a du Q sin h cos b2 1 2 1
c c c
u 2
c2 u 2 a2
which is the required result.
Ex.7. Show that for a geodesic
1 1 1 1 1 .
2 a b or
2 a b
Sol. The torsion and curvature of a geodesic are given by
1
b a sin 2 .....(1)
2
and = a cos2 + b sin2 (by Euler’s theorem) .....(2)
where is the angle between the line of curvature and the geodesic tangent.
Now (b – ) = b – (a cos2 + b sin2 )
= b (1 – sin2 ) – a cos2 = b cos2 – b cos2
or (b – ) = (b – a) cos2 .....(3)
Similarly ( – a) = (a cos2 + b sin2 ) – a
or ( – a) = (b – a) sin2 .....(4)
multiplying equation (3) and (4), we get
(b – ) ( – a) = (b – a)2 sin2 cos2 = 2
or (b – ) ( – a) = 2, which is the required result.
1 1 1
But , a , b , then above equation reduces to
a b
1 1 1 1 1 1
2 where .....(5)
b a
which is another form of the result.
234
Ex.8. Prove that at the origin the geodesic curvature of the section of the surface
2z = ax2 + by2, by the plane lx + my + nz = 0, is
n (bl2 + am2)/(l2 + m2)3/2.
Sol. The given point is (0, 0, 0) and the plane of the given section is
lx + my + nz = 0. .....(1)
Then the equation of the tangent plane of the surface
2z = ax2 + by2 at the origin is, z = 0
i.e., 0x + 0y + z = 0. ......(2)
Therefore the direction cosines of the line of intersection of planes (1) and (2), which will be the
direction cosines of a tangent through origin to the given section of surface, are obtained as
m 1
l1 , m1 , n1 0 . .....(3)
2 2 2 2
l m l m
Let the equation of the normal plane to the given surface at the origin through the tangent line be
x + y + z = 0 .....(4)
then l1+ m1 + n1 = 0
or m l 0 .....(5)
l m
Also = 0. .....(6)
Now plane given in equation (4) passes through z-axis.
Hence
l m 0
2 2 2 1
or
l m 0 l 2 m 2 02 l 2 m2
l m
, , 0 . .....(7)
l 2 m2 l 2 m2
Now the direction cosines of the normal to the given section (1) are
l m n
, , .....(8)
2 2 2 2 2 2
l m n l m n l m2 n2
2
If be the angle between (7) and (8) then
l 2 m2 . .....(9)
cos 2
l m2 n2
The radius of curvature of a given section through any point of a surface is obtained by the ex-
pression
235
cos rl 2 2 sl1m1 tm12
1 .....(10)
1 p2 q 2
For the given surface 2z = ax2 + by2, we have at the origin p = 0, q = 0, r = a, s = 0 and t = b.
Using value of cos form (9) and these values of p, q, r, s, t in equation (10), we get
1 l 2 m2 am 2 bl 2
2
l 2 m2 n2 l m2
or = (l2 + m2)3/2 (am2 + bl2)–1 (l2 + m2 + n2)–1/2
1 3 2 12
but
, so l 2 m 2 am 2
bl 2 l 2 m2 n 2 .....(11)
Also, if is the angle between the given section and the normal section, then
cos = (l2 + m2)1/2 (l2 + m2 + n2)–1/2
Hence by Meunier’s theorem
n = cos = (l2 + m2)–1 (am2 + bl2) ......(12)
therefore, the geodesic curvature g of the required section is given by 2 2n 2g
2 2 2
2g 2 2n
am 2
bl 2 l 2
m 2 n2 am bl
2
3 2
l 2
m2 l m
2 2
am 2
n
bl 2 2
3
l 2
m 2
or g
n am 2 bl 2 ,
32
l 2
m2
which is the required result.
Ex.10. Find the Gaussian curvature at the point (u, v) of the anchor ring
r
r g (u ) cos v, g (u ) sin v, f (u ) .
where g (u) = (b + a cos u), f (u) = a sin u and the domain of u, v is
0 < u < 2, 0 < v < 2,
verify that the total curvature of the whole surface is zero.
r
Sol. We have position vector of a point (say) r ,
r
r b a cos u cos v, b a cos u sin v, a sin u
r r r r r
then by finding r1 , r2 , r11 , r12 , r22 we can get
236
E = a2, F = 0, G = (b + a cos u)2, H2 = a2 (b + a cos u)2 .....(1)
L = a, M = 0, N = cos u (b + a cos u) .....(2)
Then the Gaussian curvature is given by
LN M 2 cos u
.....(3)
H2 a b a cos u
Then the total curvature of the whole surface is
Kds
s
2 2
H du dv
u 0 v 0
2 2
cos u
a b a cos u a b a cos u du dv [on using values form (3) and (1)]
0 0
2 2 2
2
cos u du dv cos u v 0 du
0 0 0
2
2 cos u du 2 0 0
0
total curvature = 0. Hence verified.
9.28 Summary
1. In this unit you have studied about geodesic, differential equation of a geodesic, single
differential equation of a geodesic, when the relation between parameters u and v be of the form
u = u (v) or v = v (u), geodesic on a surface of revolution. About geodesic curvature and tor-
sion and their expressions in different forms, about Gauss-Bonnet theorem.
2. Sufficient number of examples have solved in the unit.
3. Differential equation of geodesic in different forms and formulae for geodesic curvature and tor-
sion will help the students to easily understand the text of the unit.
4. Examples in the text have been inserted frequently to help students to understand the text of the
unit.
9.29 Answers to self-learning exercises
Self-learning exercise-1
1. See §9.2.
237
2. See §9.3.
1 1
3. Eu E1u 2 E2 uv Ev F2 G1 v2 0 and
2 2
1 1
Fu F1 E2 u2 G1 u v Gv G2v2 0
2 2
d d
4. 0 and 0
ds u u ds v v
2 2 2 2
5. u l u 2m u v nu 0 and v u 2 u v v v 0
3 2
d 2u du du du
6. 2
2 l 2m n
dv dv dv dv
7. No.
8. EF2 + FE1 – 2EF1 = 0
Self-learning exercise-2
d 2x d 2 y d 2z
2 2 2
ds ds ds
1. Integral of one of the equation , with the equation F (x, y, z) = 0
Fx Fy Fz
2
d 2 y dy dy dy
2. 1 p 2 q 2
dx 2 dx
p q t
dx
2 s
dx
r
3. See §9.13. 4. Yes. (see §9.14) 5. Yes. (see §9.15)
1
6. See §9.16. 7. See §9.22. 8. g sin w
l
9. No (see §9.25) 10. See §9.26(ii). 11. See §9.27.
9.30 Exercises
r r
1. Derive the general differential equations of geodesic on a surface r r u , v . [Ans. See §9.4]
r r
2. Derive the canonical equation of a geodesic on the surface r r u , v . [Ans. See §9.5]
3. Derive the differential equations of a geodesic in Gauss coefficient. [Ans. See §9.6]
4. Find the single differential equation of geodesics, on surface r = r (u, v), when a curve on the
surface may be determined by a single relation between the parameters, u and v either by
v = v (u) or by u = u (v). [Ans. See §9.7]
5. Show that for surface, a necessary and sufficient condition that the curve v = c (constant) be a
geodesic is EF2 + FE1 – 2EF1 = 0, when v = c, for all values of u. [Ans. See §9.8]
238
6. Find the differential equation of geodesics on the surface z = f (x, y), the Monge’s form.
[Ans. See §9.11]
7. State and prove Clairant’s theorem. [Ans. See §9.13]
8. Find an expression of g (geodesic curvature) and show that it is intrinsic. [Ans. See §9.17]
r r
9. Derive the formula for geodesic curvature of the form g N̂ r r [Ans. See §9.19]
1
10. Find the geodesic curvature in terms of normal angle. [Ans. g sin w ]
11. Derive the basic expression for the torsion of a geodesic on a surface. [Ans. g Nˆ Nˆ tˆ ]
12. State and prove Gauss-Bonnet theorem. [Ans. See §9.27]
13. Prove that on a surface with metric ds2 = a2 du2 + b2 dv2 the geodesic curvature of the curve
b
u = c is (ab)–1 .
u
239
Unit 10 : Gauss Formulae, Gauss’s Characteristic Equation
Weingarten Equations, Mainardi-Codazzi Equa-
tions. Fundamental Existence Theorem for Surfaces,
Parallel Surfaces, Gaussian and Mean Curvature for
a Parallel Surface, Bonnet’s Theorem on Parallel
Surfaces.
Structure of the Unit
10.0 Objective
10.2 Introduction
10.3 Gauss’s formulae
10.4 Weingarten equations
10.5 Mainardi-Codazzi equations.
10.6 Illustrative examples
10.7 Fundamental existence theorem for surfaces.
10.8 Parallel surfaces
10.9 Gaussian and mean curvature for the parallel surface.
10.10 Bonnet’s is theorem for parallel surfaces
10.11 Self-learning exercises
10.12 Summary
10.13 Answers to self-learning exercises
10.14 Exercises
10.0 Objectives
Six fundamental magnitudes E, F, G and L, M, N and their partial derivatives play an important
role in the surface theory. Gauss’s formulae and Gauss’s characteristic equations are some of the rela-
tions between them.
10.1 Introduction
This unit is devoted to the study of some relations between E, F, G and L, M, N and their par-
tial derivatives. We also study the fundamental existence theorem for surfaces and Bonnets theorem on
parallel surfaces.
240
10.2 Gauss’s formulae
The Gauss’s formulae or Gauss’s equations are given below
r11 L Nˆ l r1 r2
r12 M Nˆ m r1 r2
.....(10.2.1)
r22 N Nˆ n r1 r2
where l, m, n; are called Christoffel symbols and are suitable functions of E, F, G and their
partial derivatives with respect to u and v.
Proof : Second order partial derivative of r w. r. to u and v can be expressed linearly in terms
of r1 , r2 and N̂ as given below..
r11 A Nˆ l r1 r2 , .....(10.2.2)
r12 B Nˆ m r1 r2 , .....(10.2.3)
r22 C Nˆ n r1 r2 , .....(10.2.4)
where A, B, C ; l, m, n ; are the coefficients to be determined.
Taking scalar multiplication of (10.2.2) by N̂ , we get
r11 Nˆ A Nˆ Nˆ A as r1 Nˆ 0 r2 Nˆ
A = L as r11 Nˆ L .
Similarly B = M, C = N. .....(10.2.5)
Hence the relation (10.2.2) to (10.2.3) assume the form (10.2.1).
Now to determine l, m, n ; we proceed as follow :
E 2 E
Clearly E1
u u
r1 2r1 r11 ; E2
v
2r1 r12 ;
F F
F1 r11 r2 r1 r12 ; F2 r12 r2 r1 r22 .....(10.2.6)
u v
G
G1 2r2 r12 ; G2 2 r2 r22
u
1 1
Hence r1 r11 E1 ; r11 r2 F1 E1
2 2
1 1
r1 r12 E2 ; r2 r12 G1 .....(10.2.7)
2 2
1 1
r1 r22 F2 G1 ; r2 r22 G2
2 2
241
Multiplying (10.2.2) by r1 and r2 successively scalarly, we get
r1 r11 l E F
.....(10.2.8)
r2 r11 l F G
Using (10.2.7) in (10.2.8) and then on solving we get
1 1
E1 l E F and F1 E2 l F G , .....(10.2.9)
2 2
where values of l and are given by
1 1
2
l GE1 2 FF1 FE2 , 2 EF1 EE2 FE1 .
2H 2H 2
Proceeding exactly with (10.2.3) and (10.2.4), we get
1 1
2
m GF1 FG1 ; EG1 FE2
2H 2H 2
1 1
2
n 2GF2 GG1 FG2 , EG2 2 FF2 FG1 . .....(10.2.10)
2H 2H 2
Corollary : In the case, the parametric curves are orthogonal, then F = 0 and H2 = EG.
GE1 E E G
Hence l 2
1 , m 2 , n 1 ;
2H 2E 2E 2E
E2 G G
, 1 , 2 . .....(10.2.11)
2G 2G 2G
Thus Gauss’s formulae become
1 E E
r11 L Nˆ r1 2 r2 ,
2 E1 2G
E G
r12 M Nˆ 2 r1 1 r2 ,
2E 2G
G G
and r22 N Nˆ 1 r1 2 r2 .....(10.2.12)
2E 2G
10.3 Gauss’s characteristic equations
It states that
1
T2 2 F12 E22 G11 m 2 E 2m F 2G ln E l n F G
2
where T 2 L N M 2 .
Proof : Taking scalar product of first and third Gauss’s formulae and subtracting the square of
the second, we get
242
2
r11 r22 r122 L Nˆ l r1 r2 N Nˆ n r1 r2 M Nˆ m r1 r2
L N M 2 ln m 2 r22 l n 2m r1 r2 2 r22
T 2 ln m 2 E l 2m n F 2 G .....(10.3.1)
1 1 2
But E r12 i.e., E2 r1 r12 and E22 r12 r1 r122
2 2
F r1 r2 i.e., F2 r12 r2 r1 r22 and F12 r12 r21 r121 r2 r11 r22 r1 r221
1 1
G r22 G1 r2 r12 and G11 r21 r12 r2 r121
2 2
1
Hence E22 G11 2 F12 r122 r11 r22 . .....(10.3.2)
2
From (10.3.1) and (10.3.2), we get desired characteristic equation as
1
T2 2 F12 E22 G11 m 2 E 2m F 2G ln E l n F G .
2
Corollary 1. Gauss’s characteristic equation can be put in the following form using the values of
l, m, n ; form 10.2
1 F 1 1 2 F E FE
LN M2 H E2 G1 H 1 2 1 .....(10.3.3)
2 u EH H 2 v H H EH
Corollary 2. Suppose r r u, v represents a surface. We know that at any point (u, v) on
the surface, the Gaussian curvature K is given by
LN M2 1 FE2 G1 1 2 F1 E2 FE1
2
H 2 H u EH H 2 H v H H EH
[Using (10.3.3) equation of corollary 1]
1 FE2 EG1 1 2 EF1 FE1 EE2
.....(10.3.4)
H u 2 HE H v 2 HE
where H2 = EG – F2.
The formula (10.3.4) gives Gaussian curvature K in terms of first fundamental magnitudes E, F,
G and their partial derivatives with respect to u and v.
Thus equation (10.3.4) is intrinsic formula for Gaussian curvature. In the case, the parametric
curves are orthogonal, F = 0 and then F1 = 0.
1 G1 1 E2
LN M2 2 H F 0
H u H 2 v H
But H2 = EG – F2 = EG.
243
LN M2 1 G1 E2
2 H u H v H
2
H
1 1 G1 1 E2
2 u
H EG 2 v EG
1 1 G 1 E
.
EG u E u v G v
10.4 Weingarten equations
The formulae H2 N1 = (FM – GL) r1 + (FL – EM) r2
and H2 N2 = (FN – GM) r1 + (FM – EN) r2
are known as Weingarten formulae or Weingarten equations.
Proof : Since Nˆ 1 Nˆ Nˆ1 0, Nˆ Nˆ 2 0 .
Thus N1 and N 2 are perpendicular to N̂ . Hence, these lie in the plane of r1 and r2 .
Hence N1 Ar1 Br2 .....(10.4.1)
Taking scalar multiplication of (10.4.1) with r1 and r2 successively, we get
– L = AE + BF, – M = AF + BG.
Solving for A and B, we get
A = (FM – GL)/(EG – F2), B = (FL – EM)/(EG – F2)
But EG – F2 = H2, therefore by (10.4.1)
H2 N1 = (FM – GL) r1 + (FL – EM) r2
Similarly we can get second equation.
10.5 Mainardi–Codazzi equations
The three fundamental magnitudes L, M, N are not functionally independent. They are related
through the equations.
L2 – M1 = mL – (l –) M –N
M2 – N1 = nL – (m – )M – N.
which are called Mainardi-Codazzi equations.
Proof : Consider the identity
r11 r12 .....(10.5.1)
v u
244
Putting the values of r11 and r12 from Gauss’s formula in (10.5.1), we get
v
L Nˆ l r1 r2 u
M Nˆ m r1 r2
L2 Nˆ l2 r1 2 r2 L N 2 l r12 r22 M 1 Nˆ m1 r1 1 r2 MNˆ 1 m r11 r12 .
....(10.5.2)
Now substituting the values of r11 , r22 , r12 form Gauss’s formulae and the values of N̂1 and N̂ 2
form Weingarten equations, namely
1 1
2
N̂1 FM GL r1 2 FL EM r2
H H
1 1
2
and N̂ 2 FN GM r1 2 FM EN r2 .....(10.5.3)
H H
The identity (10.5.2) is expressed in terms of vectors N̂ , r1, r2 (non-coplaner vectors).
On equating the coefficients of N̂ on both sides, we get
L2 l M N M 1 m L M
or L2 M 1 m L l M N .....(10.5.4)
which is the first Mainardi-Codazzi equation.
Now consider the identity
r12 r22 . .....(10.5.5)
v u
From Gauss’s formula, we have
r12 MNˆ m r1 r2 ; r22 NNˆ n r1 r2 .....(10.5.6)
Putting these values in (10.5.5), we get
M 2 Nˆ m1 r1 M N 2 m r12 r22 N1 Nˆ n1 r1 r2 NN1 n r11 r12 .....(10.5.7)
Substituting in (10.5.7), the values of r11 , r22 , r12 from Gauss’s formulae and for Nˆ 1 , Nˆ 2 from
Weingarten equation, we get a vector identity and then equating coefficients of N̂ on both sides of the
identity, we get
M2 – mN + N = N1 + nL + M
or M2 – N1 = nL – (m –) M + N, .....(10.5.8)
which is second Mainardi-Codazzi equation.
10.6 Illustrative examples
Ex.1. Show that for the surface z = f (x, y) with x, y are parameters.
pr ps pt qr qs qt
l 2
, m 2 , n 2 ; 2 , 2 , 2 .
H H H H H H
245
Sol. We have for the surface z = f (x, y).
E = 1 + p2, F = pq, G = 1 + q2,
E1 = 2pr, F1 = rq + pq, G1 = 2qs,
and E2 = 2ps, F2 = sq + pt, G2 = 2qt. .....(1)
2
Again l GE1 2 FF1 FE2
H2
1
2H 2
1 q 2 pr 2 pq rq ps pq 2 ps
2
1 pq
2
2 pq 2 prq 2 2 prq 2 2 p 2 qs 2 p 2 qs 2 .....(2)
2H H
Similarly, on putting these values in experience for m, n, we get the required results.
Ex.2. For any surface prove that
log H l , log H m .
u v
where u and v are parameters and symbols have their usual meaning.
Sol. We have H2 = EG – F2
1 1 1
log H u 2 log H 2 2 H 2 u H 2
u
1 E G G1 E 2 FF1 .
2
2 H u
EG F 2 1
2H 2
.....(1)
Using values of Christoffel symbols from §10.2, we get
1
l GE1 FE2 2 FE1 EG1 FE2
2H 2
1
E1G EG1 2 FF1 . .....(2)
2H 2
From (1) and (2), log H l .
u
Similarly, we can prove the second result.
Ex.3. Show that for the right helicoid r = (u cos v, u sin v, cv).
u
l = 0, m = 0, n = – u ; = 0, , v = 0.
n c2
2
Sol. We have r = (u cos v, u sin v, cv)
r1 cos v,sin v, 0 , r2 u sin v, u cos v, c
E r1 2 1, F r1 r2 0, G r22 u 2 c 2
H 2 EG F 2 u 2 c 2 .....(1)
246
Therefore E1 = 0, E2 = 0, F1 = 0 = F2 ; G1 = 2u, G2 = 0.
1
Again l GE1 2 FF1 FF2 0
2H 2
1
2 EF1 EE2 FE1 0
2H 2
1 1 u
2
m GE2 FG1 0 , 2
EF1 FE2 2 2
2H 2H u c
1 1
n 2
2GF2 GG1 FG2 [– (u2 + c2)2u] = – u,
2H 2 u c2
2
1
EG2 2 FF2 EG1 0
2H 2
Ex.4. From the Gauss’s characteristic equation deduce that, when the parametric curves
are orthogonal
1 1 G 1 E
.
EG u E u v G v
Sol. In case parametric curves are orthogonal, we have
F = 0, F1 = 0 .....(1)
Hence equation (3) of §10.3 corollary 1, we have
1 G1 1 E2
LN M 2 H H .....(2)
2 u H 2 v H
But H2 = EG – F2 = EG.
2
LN 2M
H
1 G1 E2
u H v H
2H
1 1 G1 1 E2
2 u
H EG 2 v EG
1 1 1
u G E . .....(3)
EG E u v G v
Ex.5. Show that the surface whose metric is given by
ds2 = du2 + D2 dv2,
D1 D
l = 0, m = 0, n = – DD1, = 0, , 2.
D D
247
Sol. Comparing ds2 = du2 + D2 dv2 with
ds2 = E du2 + 2F dudv + G dv2 .....(1)
Here, we have E = 1, F = 0, G = D2, H2 = EG – F2 = D2
E1 = 0, G1 = 2DD1; E2 = 0, G2 = 2DD1
E2 G D G D
0, 1 1 , 2 2 .....(2)
2G 2G 2 D 2G D
Ex.6. For a surface given by ds2 = (du2 + dv2) prove that
1
l , m 2 , n 1 , 2 , 1 , 2
2 2 2 2 2 2
and further show that Mainardi-Codazzi relations become
1 2 1
L2 M 1 L N , N1 M 2 1 L N
2 2
Also show that the Gauss characteristics equation then
1 1
LN M 2 12 2 2 11 22 .
2 2
Sol. Comparing ds2 = (du2 + dv2) with
ds2 = E du2 + 2F dudv + G dv2
E = , F = 0, G = .
As = 0, the Gauss’s coefficient reduces to
1 1 1 2 1 1 2 1 1 1 2
l , m , n 1, , , .
2 2 2 2 2 2
Hence, the Mainardi-Codazzi relations become
1 2 1 1 1 2 1 2
L2 M1 L 1 1 M N L N
2 2 2 2 2
1 1 1 2 1 2 1 1 1 1
and M 2 N1 L M N L N .
2 2 2 2 2
Ex.7. Show that when the lines of curvature are chosen as parametric curves, the Codazzi
relations expressed in terms of E, G, L, N and their derivatives are
1 L N 1 L N
L2 E2 , N1 G1 .
2 E G 2 E G
Show also that the equation of Gauss may be written as
LN 1 G 1 E
0.
EG u G u v G v
248
Sol. The Codazzi equations are
L2 – M1 = mL – (l – ) M – N .....(1)
M2 – N1 = nL – (m – ) M – N .....(2)
Since lines of curvature are parametric curves, we have F = 0, M = 0.
Also then the Gaussian coefficients are
E2 E G E G G
l , m 2 , n 2 , 2 , 1 , 2 .
2E 2E 2E 2G 2G 2G
Thus, the equation (1) reduces to
E2 E 1 L N
L2 L 2 N E2
2E 2G 2 E G
and equation (2) reduces to
G1 G 1 L N
N1 L 1 N or N1 G1 .
2E 2G 2 E G
For the second part, from Ex.4 above, we have
1 1 G 1 E
. .....(3)
LG u E u v G v
LN M 2 LN
In this case
EG F 2 EG
. Q F 0 M
Putting this value of k in (3), we get the required result.
10.7 Fundamental existence theorem for surfaces
Statement : When the coefficients of the two quadratic differential forms
E du2 + 2 F dvdu + G dv2 and L du2 + 2M dudv + N dv2
are such that the first form is positive definite and the six coefficients satisfy the Gauss’ characteristic
equation and the Mainardi-Codazzi equations, then there exists a surface, uniquely determined to within
a Euclidean displacement, for which these forms are respectively the first and second fundamental forms.
Proof. The proofs of this theorem depends on the existence and uniqueness theorems of the first
order differential equations which can be obtained from the two given fundamental quadratic differential
forms.
r r
It is easy, if we choose the principal directions and the normal N̂ to the surface r r (u, v ) in
the curvilinear coordinates, as the coordinate axes. In this case F = 0 = M and quadratic differential
forms reduces to
E du2 + G dv2 and L du2 + N dv2 .....(10.7.1)
Let ˆ (u , v ), ˆ (u , v) denote the unit tangent vectors along the parametric curves v = constant
(u-curve) and u = constant (v-curve), respectively.
249
1 1
Then ˆ r1 , ˆ r2 . .....(10.7.2)
E G
In this statement of the theorem, we have given that the fundamental coefficients E, F, G and L,
M, N satisfy Gauss’s characteristics equation and the Mainardi-Codazzi equations and the Weingarten
equations which are Gauss’s equations
r 1 Er E E r
r11 r1 r2 LNˆ
E u G v
r 1 E r 1 Gr
r12 r1 r2
E v G u
r G Gr 1 Gr
r22 r1 r2 NNˆ .....(10.7.3)
E u G v
Weingarten equations
Lr N r
N1 r1 , N 2 r2 . .....(10.7.4)
E G
Differentiating (10.7.2) partially with respect to u and v. Using (10.7.3) and (10.7.4), we find
ˆ 1 E ˆ L ˆ ˆ 1 G ˆ
N, ;
u G v E v E u
ˆ 1 E ˆ 1 G N ˆ
ˆ , ˆ N;
u G v v E u G
Nˆ L Nˆ N ˆ
ˆ and . .....(10.7.5)
u E v G
Thus six first order partial differential equations for the triad ˆ u , v , ˆ u , v , Nˆ u , v are
r r r
necessary conditions to be satisfied so that their exists a surface r r u , v exists with r12 E ,
r r r
r1 r2 0, r2 2 G and the equations (10.7.5) admits at least one solution (ˆ , ˆ , Nˆ ) which assumes the
prescribed values u0, v0 for u and v in a given interval such that u(0) = u0, v(0) = v0.
Now, if there are two surfaces S, S* with the same prescribed fundamental forms then an Eu-
clidean displacement we can arrange the triad ˆ , ˆ , Nˆ and (ˆ *, ˆ *, Nˆ *) to coincide when u = u0,
v = v0 and therefore they all coincide for all values of u and v.
Hence, the surfaces S and S* differ by atmost a Euclidean motion. This proves the uniqueness
of the solutions limited within the Euclidean displacement.
250
10.8 Parallel surfaces
A surface S is said to be parallel to another surface S* if the points of S* are at a constant
distance along the normal to S.
r
Clearly if P(r ) is any point on S, then corresponding point Q on S* is given by
r r
r * r cNˆ .....(10.8.1)
where c is a scalar constant whose magnitude represents the distance along normal i.e.,
uuur
PQ c.
Fundamental magnitudes of S* :
Clearly on differentiating (10.8.1) with respect to u and v successively we can get fundamental
r r r r
magnitudes provided we know N1 , N 2 . N1 , N 2 are given by Weingarten formula.
As a particular case if lines of curvature are parametric curves i.e., F = 0 = M, the Weingarten
formulae reduces to
r r
EGN1 GLr1 , Q H 2 EG when F 0 M .....(10.8.2)
L N
Writing a , b , we shall have fundamental magnitudes E*, F*, G*, L*, M*, N* as
E G
2 2
E* E 1 c a , F * 0, G* G 1 c b
2
L* 1 c a L, M * 0, N * 1 c b
Also H * H 1 2c c 2 . .....(10.8.3)
It at once follows that curves on S* corresponding to lines of curvature of S are also lines of
curvature on S*.
10.9 Gaussian and mean curvature for the parallel surfaces
L* N * M *2
We have *
H *2
LN 1 c a 1 cb 0
2 2
H 2 1 c a 1 cb
1 ca 1 cb
LN 0 .....(10.9.1)
. Q H 2
1 2c c 2
1 1
Remark : In the above proof, we have not taken into account the case when c or .
a b
251
This happens in case parallel surfaces are spheres of radii 2B concentric with S. Then parallel
surfaces degenerate into points.
Thus first part of the Bonnet’s theorem can be modified into the following :
“For every surface with constant positive Gaussian curvature k there exists at least one (non-
singular) parallel surface with constant mean curvature.
10.10 Bonnet’s theorem on parallel surfaces
For every surface with constant positive Gaussian curvature B–2, in general, there are associated
two surfaces of constant mean curvatures ( 2B)–1, which are parallel to the surface S and distant B
from it, and for every surface S with constant mean curvature (2B)–1 there is a parallel surface of con-
stant Gaussian curvature B–2 distant B from it.
Proof. Here = B–2 and c = B.
1 * *b 1 L* N *
a * *
2 2 2 E G
1 c
, from above
2 1 2c c 2
1 B
1
1
2B
2 1 B
Conversely, using = (2B)–1, c = – B, we have
L* N * M *2 LN
* , Q H 2
H *2
1 2c c 2
1 2(2 B) 1
B B 2
1
2
B 2 (constant) . .....(10.10.2)
B
Hence proved.
Remark : In the above proof also, we have not taken into account the case when
1 1
c or .
a b
10.11 Self-learning exercises
1. Write Gauss’s characteristic equation.
2. What are the Weingarten formulae ?
252
3. Define parallel surfaces.
4. Write Mainardi-Codazzi equations.
5. State Bonnets theorem on parallel surfaces.
10.12 Summary
In this unit we have derived. Gauss’s formulae in the form of partial differential equations and
Gauss’s characteristic equation. We have also studied the Weingarten formula and Mainardi-Codazzi
equations. Fundamental existence theorem for parallel surfaces has also been studied in this unit. For
parallel surfaces Bonnet’s theorem has also been discussed. Some solved question on above theorems
have been given in the exercises.
10.13 Answers to self-learning exercises
1. § 10.3
2. § 10.4
3. § 10.8
4. § 10.5
5. § 10.10
10.14 Exercises
1. Obtain the fundamental equation of surface theory.
2. Obtain the equation of Weingarten and use them to establish Mainardi-Codazzi equations.
3. Prove that the Gaussian curvature at a point is expressible in terms of the fundamental magni-
tudes of the first order and their derivatives of the first two orders.
4. Prove the Gauss characteristic equation and deduces that, when parametric curves are ortho-
gonal
1
1
EG
u
E u
G v 1
G v
E .
253
Unit 11 : Tensor Analysis, Kronecker Delta, Contravariant and
Covariant Tensors, Symmetric Tensors, Quotient law
of Tensors, Relative tensor
Structure of the Unit
11.0 Objective
11.1 Introduction
11.2 Space of N-dimensions
11.3 Coordinate transformation
11.4 Summation convention
11.5 Kronecker delta
11.6 Contravariant vectors
11.7 Covariant vector
11.8 Invariant
11.9 Second order tensors
11.10 Higher order tensors
11.11 Zero tensor
11.12 Symmetric tensor
11.13 Skew symmetric tensor
11.14 Algebraic operations with tensor
11.15 Illustrative examples
11.16 Quotient law of tensor
11.17 Illustrative examples
11.18 Relative tensors
11.19 Conjugate (or Reciprocal) symmetric tensor
11.20 Illustrative examples
11.21 Self-learning exercise
11.22 Summary
11.23 Answer to self-leaning exercises
11.24 Exercise
254
11.0 Objective
Tensor calculus is the generalisation of the differential geometry of Gauss and Riemann. Einstien
used it as a most suitable tool for the study of his general theory of relativity. The reason behind it is that
a physicist wants to formulate the laws of physics which remain same (i.e. invariant) when we go from
one frame of reference to another. The objective of this unit is to define the tensorial quantities and their
properties. We also study the algebra of tensorial quantities in this unit.
11.1 Introduction
The tensor formulation became popular when Einstien (1879-1955) used it as an excellent tool
for the presentation of his general theory of relativity. It has now become an excellent tool in the study of
many branches of theoretical physics, such as mechanics, Fluid Mechanics, Elasticity, Plasticity, Electro-
magnetic theory etc.
Tensor analysis is the generalization of vector calculus. It handles the answers to the questions
such as :
(i) are all basic physical laws expressible in terms of scalars and vectors ?
(ii) which transformation is suitable for the invariant character of physical laws ?
(iii) how a certain physical law be written if wider class of transformation is introduced ?
It is a basic principal of tensor analysis that we should not tie ourselves down to any our system
of coordinates, we seek statements which are true, not for one system of coordinates but for all. The
transformation laws for the components of an entity from one coordinate system to another are the basic
criteria to determine the tensor character of that entity. In other words :
“A tensor is an entity whose components, when are being transformed from one coordi-
nate system to another, obey certain basic transformation laws.” The study of these laws is the
prime aim of this unit.
11.2 Space of N-dimensions
We know that in the three dimensional rectangular space, the coordinates of a point are given by
triplets in the form (x, y, z) where x, y, z are three numbers. But this representation is not suitable if we
want to generalizes the concept of space from three dimension to N-dimensions. That is why it is
advisable to use a triplet (x1, x2, x3) in place of (x, y, z) where 1, 2, 3 are the superscripts not power
indices. In general, the coordinate of a point in N-dimensional space are given by the N-tuples of the
form (x1, x2, x3, ... xN) where 1, 2, ... N are not powers of x but are the superscripts of x and N 2.
This type of N-dimension space is denoted by VN.
255
11.3 Coordinate transformation
Consider two different frames of references of N-dimensions. Let the coordinates of a point with
1 2 N
respect to these frames be respectively (x1, x2, ... xN) and x , x ,...x . Suppose these coordinates
of the two systems have the following independent relations :
x i x i x1 , x 2 ,... x N (i = 1, 2,..., N) .....(11.3.1)
where x i are single valued, continuous functions and have continuous derivatives for certain ranges of
x1, x2, ... xN. Under these conditions equations (11.3.1) can be solved for xi as functions of x i given by
xi xi x 1 , x 2 ,... x N i 1, 2,3,... N .....(11.3.2)
The relations given by (11.3.1) and (11.3.2) define a transformation of coordinates from one
frame of reference to another.
Differentiating (11.3.1), we get
x i 1 x i 2 x i N
N
x i
i
dx 1 dx 2 dx ... N dx x j dx j . .....(11.3.3)
x x x j 1
11.4 Summation convention
We know that the expression
a1x1 + a2x2 + a3x3 + ... + aNxN .....(11.4.1)
N
is represented by ai xi .
i 1
According to summation convention we drop sigma sign and merely write the above sum
as aixi.
(a) Thus by summation convention we mean that if a small latin index (superscripts or subscripts)
is repeated in a term then it is understood that we are to sum over this index from 1 to N unless other-
wise stated. This summation convention was first used by Einstien.
(b) Indicial (or Range) convention : When a small latin index is used either as superscript or
subscripts occurs unrepeated in a term, it takes all values from 1 to N unless otherwise stated, N being
the number of dimensions of the space.
The unrepeated latin index used in a term is called free or real index and takes all values from 1
to N. For example ‘i’ is the free index in the following expressions :
xi xi x 1 , x 2 , x 3 ,... x N , .....(11.4.2)
and xi x x , x , x ,... x ,
i 1 2 3 N
.....(11.4.3)
(c) Dummy index : Any index, which is repeated in a given term so that the summation con-
vention applies, is called a dummy index or dummy suffix. This is also called umbral or dextral index.
256
11.5 Kronecker delta
The Kronecker delta which is denoted by ij , is defined as :
1 , if i j
ij . .....(11.5.1)
0 , if i j
Thus, we have 11 22 33 ... NN 1 (no summation over N)
12 32 ... 0 ......(11.5.2)
and ii 11 22 ... NN 1 1 ... 1 N . ......(11.5.3)
An important properly of Kronecker delta is that
ij A j Ai , .....(11.5.4)
since in the L.H.S. summation is carried over j.
xi
If may also be noted that j
ij , since the coordinates x1, x2,... xN are independent.
x
x i
Similarly j
ij .
x
Ex.1. Use Einstien is summation convention to write the following :
(i) A1k B1 A2k B 2 ... ANk B N
2 2
(ii) ds 2 g11 dx1
g 22 dx 2 dx 2 ... g NN dx N
g12 dx1dx 2 g 21dx 2 dx1 ... g1N dx1 g N 1dx N dx1
Sol. (i) A1k B1 A2k B 2 ... ANk B N Aik Bi
N N
(ii) ds 2 gij dx1dx 2 .
i 1 j 1
Ex.2. Show that
xk x i
(i) ij kj ik (ii) kj
i j
x x
Sol. (i) ij kj 1i 1k i2 k2 i3 3k ... ik kk ... iN kN
0 0 ... ik (1) 0 ... 0 [no summation over k]
ik .
xk x i xk x 1 x k x 2 xk x N
(ii) ...
xi x j x1 x j x 2 x j xN xj
x k
[By chain rule]
x j
kj .
257
11.6 Contravariant vectors
If a set of N quantities Ai in a coordinate system xi are transformed to the set of another N
quantities A j in the coordinate system x j by the equations
x p q
Ap A
x q
then Ai are said to be components of a contravariant vector or contravariant tensor of the first order or
first rank.
Note : It is a convention that contravariant tensors are denoted by superscripts, with the
exception of the coordinates xi, which may behave as contravariant vector in special conditions (see
Theorem 2)
Theorem 1. The law of transformation of a contravariant vector is transitive.
Proof. Let Ai be the components of a contravariant vector in the coordinate system xi and they
are related to the components A j of same vector in the coordinate system x j , then we have by the
law of transformation
x j i
j
A i A .....(11.6.1)
x
Now, a further change of coordinates from ...... to x*k, the new components A*k by contravariant
law is given by
*k x*k j
A A .....(11.6.2)
xj
Combining (11.6.1) and (11.6.2), we get
x*k x j i
A *k A
x j xi
x*k i
A .....(11.6.3)
xi
This shows that the law of transformation of contravariant vector is transitive.
Theorem 2. The coordinates xi behave like a contravariant vector with respect to linear
transformation of the type x j a ij xi , where aij are a set of N2 constants.
Proof. We have x j aij xi .....(11.6.4)
x j
Differentiating, we get i
aij . .....(11.6.5)
x
Combining (11.6.4) and (11.6.5), we get
xj i
xj x , .....(11.6.6)
xi
which shows that xi behaves like a contravariant vector.
258
Illustrative examples
dx dy
Ex.3. If a vector has components x , y x , y in rectangular cartesian coordi-
dt dt
nates then r, are its components in polar coordinates.
Sol. Here, the space is two dimensional.
Let for rectangular cartesian coordinates x1 = x, x2 = y for polar coordinates
x1 r, x 2
y
where x 2 y 2 r 2 , tan 1 .
x
xx yy rr, .....(1)
xx yy x 2 y 2 rr r 2 . .....(2)
1 xy yx
2 2
r 2 xy yx.
1 2
y x .....(3)
x
Using (1) and (3)
r 2 r 2 r 4 2 x 2 y 2
x 2
y 2
r 2 r 2 2 x 2 y 2 . .....(4)
Using contravariant law
x j i x j 1 x j 2
j
A i A 1 A 2 A . .....(5)
x x x
x 1 1 x 1 2 r r
A1 A A
x y
x1 x2 x y
x y rr
x y r. .....(6)
r r r
x 2 1 x 2 2
A2 A 2 A
x1 x
y x
x y 2 x 2 y
x y r r
xy yx
2
. .....(7)
r
Ex.4. A vector has components
x,
y in rectangular cartesian coordinates then its respec-
tive components in polar coordinates are
2
r r 2 ,
r .
r
Sol. Assuming A1
x, A2
y .....(1)
259
We find r r
A1
x
y
x y
xx yy rr r 2 2
rr r 2 . .....(2)
r r
A2
x
y
x y
xy yx r 2
2r r
r2 r2
2
r . .....(3)
r
Note : It is noted form above examples that the velocity and acceleration components are
contravariant vectors.
11.7 Covariant vectors
If a set of N quantities Ai in a coordinate system xi are transformed to a set of another N quanti-
ties A j in the coordinate system x j by the equations
x q
Ap Aq , .....(11.7.1)
x p
then Ai are said to be the components of a covariant vector or covariant tensor of first order or first
rank.
Note : The components of covariant vectors are denoted by subscript as a convention.
Theorem 3. The law of transformation for a covariant vector is transitive.
Proof. Let the components of a covariant vector in the coordinate system xi be Ai and compo-
nents of same vector in coordinate system x j be A j , then by covariant law of transformation.
x i
A j j Ai . .....(11.7.2)
x
Now for the further change of coordinates from system x j to x*k the new components Ak*
by covariant law are given by
x j
Ak* *k A j . .....(11.7.3)
x
x j x i
Combining (11.7.2) and (11.7.3) Ak* Ai
x*k x j
xi
Ai , .....(11.7.4)
x*k
which shows that the law of transformation of covariant vectors is transitive.
260
Theorem 4. These exists no distinction between contravariant and covariant vectors when
we restrict ourselves to coordinate transformations of the type
x i ami x m bi ,
where bi are N constants which do not necessarily form the components of a contravariant vec-
tor and ami are N2 constants which do not necessarily form the components of a tensor such that
ari ami rm .
Proof. We have x i ami x m bi . ....(11.7.5)
Multiplying by ari and summing over index i, we get
ari x i ari ami x m ari bi .....(11.7.6)
Using given relation ari ami rm ,
we have ari x i rm x m ari bi x r ari bi . .....(11.7.7)
Now, replacing the free index r by m on both sides, we obtain
x m ami x i ami bi . .....(11.7.8)
From (11.7.5) and (11.7.6), it follows that
x i i x m
am ,
x m x i
which sows that transformation laws for contravariant and covariant vectors respectively, define the same
type of entity in the present ease.
11.8 Invariant
A function I of N coordinates xi [I = I (xi)] is called an in variant or a scalar or tensor of zero
order with respect to coordinate transformations if I I , where I I I x j is the value of I in
new coordinate system x j .
Ex.5. A covariant tensor of first order has components xy, 2y – z2, xz in rectangular coor-
dinates. Determine its covariant components in spherical polar coordinates.
Sol. Here we have three dimensional space
x1 = x, x2 = y, x3 = z,
x 1 r , x 2 , x 3 ,
where x = r sin cos , y = r sin sin , z = r cos . .....(1)
Taking A1 = xy, A2 = 2y – z2, A3 = xz. .....(2)
Using covariant transformation law
x j
Ai Aj
x i
x1 x 2 x3
A1 A2 A3 . .....(3)
x i x i x i
261
x1 x 2 x 3
We find that A1 A1 A2 A3
x1 x1 x1
x y z
xy (2 y z 2 ) ( xz )
r r r
= (sin cos ) r2 sin2 sin cos + sin sin (2r sin sin – r2 cos2 )
+ cos r2 sin cos cos . ....(4)
Similarly from (3)
x1 x 2 x 3
A2 A1 A2 A3
x 2 x 2 x 2
x y z
xy (2 y z 2 ) ( xz )
= (r cos cos ) r2 sin2 sin cos + (r cos sin ) (2r sin sin – r2 cos2 )
– (r sin (r2 sin cos cos ) .....(5)
x1 x 2 x 3
and A3 A1 A2 A3
x 3 x 3 x 3
= (– r sin cos ) r2 sin2 sin cos + (r sin cos ) (2r sin sin – r2 cos2 )
.....(6)
11.9 Second order tensors
(a) Contravariant tensor of rank two : If a set of N2 quantities Aij in a coordinate system xi
are trans formed to another set of N2 quantities A kl in coordinate system x j by the equations
klx k x l ij
A i A , .....(11.9.1)
x x j
then Aij are called components of a contravariant tensor of rank two or second order.
(b) Covariant tensor of second order : If a set of N2 quantities Aij in a coordinate system xi
are transformed to another set of N2 quantities Akl in a coordinate system x j by the relations
xi x j
Akl Aij , .....(11.9.2)
x k x l
then Aij are said to be the components of a covariant tensor of rank two or second order.
(c) Mixed tensor of second order : If a set of N2 quantities Aij in a coordinate system xi are
transformed to another set of N2 quantities Alk in the coordinate system x j by the relations
x k x j i
Alk Aj , .....(11.9.3)
xi x l
then Aij are said to be the components of a mixed tensor (contravariant rank one and covariant rank
one) of second order or second rank.
262
Theorem 5. The Kronecker delta is a mixed tensor of second order whose components in
any other coordinate system again constitute the Kronecker delta.
Proof. The Kronecker delta is
1 , if i j ,
ij .....(11.9.4)
0 , if i j.
Let ij be the components in coordinate system xi and corresponding components in x j be lk
xk xj i xk xj xk
we have i
l j j l
l
lk .....(11.9.5)
x x x x x
ij behaves like a mixed tensor (contravariant rank one and covariant of rank one) of second
order.
Theorem 6. If Aij be a covariant tensor of second order and B i, Ci are contravariant
vectors, prove that Aij Bi Ci is an invariant.
Proof. We have Aij a covariant tensor of rank two
xi x j
Apq Aij , .....(11.9.6)
x p x q
and Bi, Ci are contravariant vectors
x p k
p
B k B , .....(11.9.7)
x
x q l
Cq B, .....(11.9.8)
xl
Multiplying equations (11.9.6), (11.9.7) and (11.9.8), we get
x i x j x p x q
Apq B p C q p q k
A Bk C l
l ij
x x x x
ik lj Aij B k C l
= Aij Bk Cl .....(11.9.9)
which shows invariant character of Aij Bi Ci.
11.10 Higher order tensors
i i .....i
If a set of Nm+n quantities A j11 2j2 ... jmn in a coordinate system xi are transformed to another set of
p p ... p
Nm+n quantities Aq11q2 2.....qnm in, the coordinate system x j by the relations
x p1 x p2 x pm x j1 x j2 x jn i1i2 ...im
Aqp1qp2.....
... pm
q ... i ... A j j ... j , .....(11.10.1)
1 2 n
xi1 xi2 x m x q1 x q2 x qn 1 2 n
i i .....i
then A j11 2j2 ... jmn are said to be the components of a mixed tensor of (m + n)th order contravariant of mth
order and covariant of nth order.
263
Theorem 7. The transformation of the tensors form a group i.e. the law of transformation
of tensors possesses transitive property.
Sol. Without loss of generality, we can consider a mixed tensor Aij in a coordinate system xi
and consider the transformation of coordinates from xi to x j and then x*l. Let the corresponding
components of the tensor be Alk and Aq* p , then
x k x j i
Alk i Aj , .....(11.10.2)
x x l
x* p x l r
and Aq* p Al . .....(11.10.3)
x k x*q
Combining (11.10.2) and (11.10.3), we get
*p k l j *p j
Aq* p x x x x Aij x x Aij , .....(11.10.4)
x k x i x*q x l x i x*q
which shows that the transformation of tensors possesses transitive property i.e. transformation of ten-
sors form a group.
11.11 Zero tensor
A tensor whose components relatively to every coordinate system are all zero is known as zero
tensor.
11.12 Symmetric tensor
A tensor is called symmetric with respect to two contravariant or two covariant in dices
if its components remain unaltered upon interchange of the indices.
e.g. the tensor Astpqr is said to be symmetric in p and q if
Astpqr Astqpr , .....( 11.12.1)
and it is said to be symmetric in s and t if
Astpqr Atspqr . .....(11.12.2)
N N 1
Theorem 8. A symmetric tensor of the second order has atmost . different
2
components in VN.
Sol. Let Aij be a symmetric tensor of order two. The total number of its components in an array,
in a VN
A11 A12 ........... A1N
A21 A22 ........... A2N
...............................
AN1 AN2 ......... ANN, .....(11.12.3)
264
are N2, out of which all the N diagonal terms will be different and the rest (N2 – N) will be equal in
pairs due to symmetric property. The number of such pairs will be
N 2
N . Hence the total number
2
of independent components
N2 N
N
2
1
N N 1 . .....(11.12.4)
2
Theorem 9. If a tensor is symmetric with respect to two contravariant indices (or covari-
ant indices) in any coordinate system it remains symmetric with respect to these two indices in
any other coordinate system.
Proof. Due to involvement of only two indices in symmetric property, there in no loss of gener-
ality if we take contravariant tensor viz. Aij = Aji, it is symmetric in i, j.
We have pq x p x q ij
A A .
xi x j
x p x q ji
A [due to symmetry]
x i x j
x q x p ji
j A
x x i
Aqp .
Hence the proposition.
11.13 Skew symmetric tensor
A tensor is called skew symmetric with respect to two contravariant or two covariant
ijk jik
indices if its components change sign upon interchange of the indices. e.g. Apq A pq is skew
symmetric in i and j and if
Aijk ijk
pq Aqp
is said to be skew symmetric in p and q.
If a tensor is skew symmetric with respect to any two contravariant indices and also any two
covariant indices, then it is called skew-symmetric tensor.
Notes :
(i) The property of skew symmetry (like that of symmetry) in also independent of the choice
of the coordinate system.
(ii) Skew-symmetry, like symmetry cannot be defined with respect to the indices of which one
denotes contravariance and the other covariance.
265
N N 1
(iii) A skew-symmetric tensor Aij of second order has at most different arithmeti-
2
cal components, as all the N diagonal terms Aii (no summation) are zero in this case.
11.14 Algebraic operations with tensors
(i) Addition : The sum of two or more tensors of the same rank and same type is a tensor of
same rank and same type.
x i x q p
Let Aij Aq , .....(11.14.1)
x p x j
x i x q p
and B ij p Bq . .....(11.14.2)
x x j
Adding (11.14.1) and (11.14.2) we get
x i x q p
Aij B ij p
x x
j
Aq Bqp . .....(11.14.3)
This shows that Aqp Bqp Cqp (say) is a tensor of same rank and type.
Remark : It can easily be verified that the addition of tensors is commutative and associative.
(ii) Subtraction : The difference of two tensors of the same rank and same type is also a tensor
of the same rank and same type.
It immediately follows from above equations (11.14.1) and (11.14.2) that Aqp Bqp Dqp is also a
tensor.
(iii) Outer multiplication : The product of two tensors is a tensor whose rank is the sum of the
ranks of given tensors.
This process involving ordinary multiplication of the components of the tensor is called open prod-
ij
uct or outer product of the two tensors, for example : the outer product of a tensor Almn by a tensor
ijk
B kpq is a tensor Clmnpq is a mixed tensor of rank 8, contravariant of rank 3 and covariant of rank 5.
Notes :
(i) The converse of above product rule is not always true i.e. not every tensor can be written as
a product of two tensors of lower ranks, for this, the reason is that the division of tensors is
not always possible.
(ii) The division, in usual sense, of one tensor by another is not defined.
Theorem 10. Outer multiplication of tensors is commutative and associative.
Proof. Commutative law : Let Alij and Bqp be two tensors,
h k l
then Amhk x x x Alij , .....(11.14.4)
xi x j x m
x r xq p
Bsr Bq . .....(11.14.5)
xp xs
266
Multiplying (11.14.4) and (11.14.5), we get
x h x k xl ij p x r x q
Amhk Bsr Al .Bq p s
x i x j x m x x
x h x k xl x r x q ij p
i j m
p s Al .Bq . .....(11.14.6)
x x x x x
Now multiplying (11.14.5) and (11.14.4)
x r x q p x h x k x l ij
Bsr Amhk Bq i j m Al .
x p x s x x x
x h x k xl x r x q p ij
i j m p s Bq Al . .....(11.14.7)
x x x x x
Equations (11.14.6) and (11.14.7) show that the expression within brackets in the R.H.S. are
same therefore we can say
Alij Bqp Bqp Alij . .....(11.14.8)
Associative law : Here we are to prove that
A ij
l
Bqp Cmk Alij Bqp Cmk .
Proceed as usual.
(iv) Contraction : If one contravariant and one covariant index of tensor (mixed tensor) are set
equal, the result indicates that a summation over the equal indices (dummy indices) is to be taken ac-
cording to the summation convention. This resulting sum is a tensor of rank two less than that of the
original tensor. The process is called contraction.
Consider a tensor Aijpqr of rank five. If we put j = r, we get Airpqr a tensor of rank 3 obtained
by contracting Aijpqr .
kl x k x l x p x q x r ij
Astu i j s t u Apqr . .....(11.14.9)
x x x x x
kl x k x l x p x q x r ir
Putting j = r Astu Apqr
xi x r x s x t x u
x k x p x q l ir
u Apqr .....(11.14.10)
xi x s x t
ku x k x p x q ir
Astu Apqr
x i x s x t
x k x p x q i
Astk Apq .....(11.14.11)
xi x s x t
This is a law of transformation of a tensor of rank 3.
267
Thus after contraction we get a tensor of rank 3. Contravariant rank (2 – 1) and covariant rank
(3 – 1).
Notes :
(i) We never contracts the indices of same type as the resulting sum is not necessarily a tensor.
(ii) The process of contraction reduces the order by two and may be repeatedly used, if so
desired, to contract new tensors, whose order will always be non-negative.
(v) Inner multiplication : The process of outer multiplication, followed by a contraction, we
obtain a new tensor called inner product of the given tensor. The process is called inner multiplication.
For example : The outer product of Alij and Bqp is Alij Bqp . Putting p = l, we get the inner product
Alij Bql Cqij .
Again if we put p = l, j = q, we have another inner product Aliq Bql D i .
Note : It can easily be verified that inner multiplication of tensors is commutative and associa-
tive.
11.15 Illustrative examples
Ex.5. If = aij Ai Aj, then we can always write = bij Ai Aj where bij is symmetric.
Sol. = aij Ai Aj. .....(1)
On interchanging the dummy indices
= aji Aj Ai. .....(2)
Adding (1) and (2), we get 2 = (aij + aji)Ai Aj
or = bij Ai Aj, .....(3)
1
where bij
2
aij a ji ,
which is symmetric, i.e. bij = bji.
Ex.6. If Ars is skew-symmetric and Brs is symmetric, prove that Ars Brs = 0
Sol Given that Ars = – Asr and Brs = Bsr on changing the dummy indices in Ars Brs, we get
Ars Brs = Asr Bsr = – Ars Brs
or 2Ars Brs = 0 Ars Brs = 0.
Ex.7. If aij is a symmetric covariant tensor and bi a covariant vector which satisfy the
relation aij bk + ajk bi + aki bj = 0, prove that either
aij = 0 or bi = 0.
Sol. Let aij bk = Aijk , .....(1)
then Aijk is a third order covariant tensor which is symmetric with respect to the pair of indices i and j
due to symmetric property of aij. Also replacing the indices i, j and k by i , k and i respectively on both
sides, we find
aji bi = Ajki, .....(2)
268
is symmetric with respect to j and k and similarly
aki bj = Akij, .....(3)
is symmetric with respect to k and i.
Hence aijk is a symmetric tensor.
Adding (1), (2) and (3), we get
Aijk + Ajki + Akij = 0
3Aijk = 0
aij bk = 0
aij = 0
or bk = 0 i.e. bi = 0. .....(4)
Ex.8. If uij 0 are the components of a tensor of the type (0, 2) and if the equation
fuij + guji = 0.
holds, then prove that either f = g and uij is skew symmetric or f = – g and uij is symmetric.
Sol. Given that fuij + guji = 0. .....(1)
Changing the free indices, we may write it as
fuji + guij = 0. .....(2)
Adding (1) and (2), we get (f + g) (uij + uji) = 0. .....(3)
(i) either uij + uji = 0 i.e. uij is skew symmetric, and then from (1) it follows that f = g.
(ii) or f = – g and then from (1) it follows that uij is symmetric.
11.16 Quotient law of tensors
In the study of tensor analysis some times it becomes necessary that whether a given entity is a
tensor or not. Theoretically we may say that if components of an entity obey tensor transformation laws,
then it is a tensor otherwise not. However in practice it is troublesome and a simple test is provided by a
law known as Quotient law, which is as follows :
Theorem 11. An entity whose inner product with an arbitrary tensor is a tensor, is itself a
tensor.
Proof. Let A (i, j, k) be given entity in a coordinate system xi, and Bmij be an arbitrary tensor
whose inner product with A (i, j, k) is a tensor Cmk i.e.
A i, j, k Bmij Cmk . .....(11.16.1)
We have to show that A (i, j, k) is a tensor.
In the coordinate system x i , we have
A p, q, r Bnpq Cnr . .....(11.16.2)
x p x q x m ij
But we have Bnpq Bm , .....(11.16.3)
xi x j x n
269
x m x k
Cnr Cmk . .....(11.16.4)
x n x r
Using (11.16.3) and (11.16.4) in (11.16.2), we get
x p x q x m ij x m x k
A p, q, r Bm n r Cmk
xi x j x n x x
x m x k
n r
A i, j, k Bmij . [using (11.16.1)]
x x
x m x p x q x k
n A p , q , r i j
r A i, j , k Bmij 0
x x x x
x n
On inner multiplication by s
x
x p x q x k
m
s A p , q , r i j
r A i, j , k Bmij 0
x x x
x p x q x k
A p, q, r i j
r A i, j, k Bsij 0 .....(11.16.5)
x x x
Form above equation we cannot jump to the conclusion that the expression within bracket van-
ishes. Since here i and j are dummy indices which imply summation and it is the sum which is zero.
However since Bsij is an arbitrary tensor we can arrange that only one of its components is non-zero.
Now each component of Bsij may be chosen in turn as that one which does not vanish. Therefore the
expression within brackets is identically zero.
x p x q x k
Hence A p, q, r A i, j, k 0. .....(11.16.6)
x i x j x r
xi x j
Taking inner multiplication with , we get
x m x n
x k xi x j
A p, q, r mp qn r m n A i, j , k
x x x
x k x i x j
A m, n, r , A i, j , k , .....(11.16.7)
x r x m x n
which shows that A (i, j, k) is a tensor of third order, and is covariant in i, j, and k and therefore may be
written as Aijk.
270
11.17 Illustrative examples
Ex.9. Use Quotient law to prove that Knonecker delta is a mixed tensor of order two.
Sol. Let Aj be an arbitrary contravariant vector, then by property of Kronecker delta, we have
ij A j Ai , .....(1)
which is again a tensor of order one (contravariant).
Hence by Quotient law ij is a mixed tensor of order two.
Ex.10. If Ai and Bi are arbitrary contravariant vectors and Cij Ai Bj is an invariant, show
that Cij is a covariant tensor of second order.
Sol. Given that Cij Ai Bj is invariant, we have
Cij Ai B j C pq A p B q . .....(1)
Further, Ai and Bi are contravariant vectors, therefore
x p i
Ap A, .....(2)
xi
qx q j
and B j B . .....(3)
x
Substituting (2) and (3) in (1), we get
x p x q i j
C
ij C pq i AB 0
j .....(4)
x x
Ai, Bj are arbitrary vectors, therefore
x p x q
Cij C pq , .....(5)
x i x j
which shows that Cij is a covariant tensor of rank two.
Ex.11. If Ai is an arbitrary contravariant vector and Cij Ai Aj is an invariant, show that
Cij + Cji is a covariant tensor of second order.
Sol. Proceeding as in Example 10, equation (4) in the present case may be written as
x p x q i j
Cij C pq i A A 0. .....(1)
x x j
This quadratic form, vanishes for arbitrary Ai, but we can not jump to the conclusion that the
expression within bracket is zero. We remember that in the form bij Ai Aj, the coefficient of the product
A1 A2 is mixed up with the coefficient of A2 A1, it is in fact b12 + b21. Thus interchanging the dummy
indices i and j, and adding these two results, we can deduce only that
x p x q x p x q
Cij C ji C pq C pq . .....(2)
x i x j x j xi
271
On changing the dummy indices p and q in last term, equation (2) becomes
p
x q
Cij C ji C pq Cqp xxi
x j
, .....(3)
which establishes the tensor character of (Cij + Cji) as covariant tensor of the order two.
11.18 Relative tensor
p p ... p
If the component of a tensor Aq11q22...qs r transform according to the equation
w
x x u1 x ur x q1 x qs
Avu1vu2......vur Aqp1qp2......qpr ... ...
1 2 s
x 1 2 s
x p1 x pr x v1 x vs
x
then Aqp1qp2.....
... pr
is called a tensor of weight w, where is the Jacobian of transformation. If
1 2 qs x
w = 1, the relative tensor is called a tensor density. If w = 0, the tensor is said to be absolut or simply
tensor.
Note : If the rank of relative tensor is one then it is called relative vector. Hence if
w
px xp
A Au ,
x xu
then Ap is a relative vector of weight w. If w = 1, the relative vector is called a vector density. If w = 0,
the relative vector is called absolute vector or simply vector.
11.19 Conjugate (or Reciprocal) symmetric tensor
Consider a covariant symmetric tensor Aij of rank two. Let d denotes the determinant | Aij | with
elements Aij i.e. d = | Aij | and d 0. We define Aij by
cofactor of Aij in the determinant | Aij |
Aij , .....(11.19.1)
d
Aij is a contravariant symmetric tensor of rank two and is said to be conjugate (or reciprocal) ten-
sor of Aij.
Theorem 12. If (i, j) is the cofactor of Aij in the determinant d = | Aij | 0 and Aij is
defined by
i, j
Aij ,
d
then show that Aij Arj ir .
Proof. From the properties of determinants we have following two results :
i, j
(i) Aij (i, j) = d Aij 1,
d
Aij Aij = 1. .....(11.19.2)
272
(ii) Aij (r, j) = 0
r, j
Aij 0 d 0
d
Aij Arj = 0 if i r
1 if i r
(i) and (ii) Aij Arj
0 if i r
i.e. Aij Arj ir . .....(11.19.3)
Theorem 13. Prove that Aij (defined as above in theorem 12) is a symmetric contravariant
tensor of rank two.
i, j
Proof. Given Aij , .....(11.19.4)
d
where (i, j) is a cofactor of Aij in d = | Aij |.
i, j
Since Aij is covariant symmetric tensor, so (i, j) is symmetric and hence Aij is sym-
d
metric.
Now it remains to prove that Aij is a tensor.
we know Aij Arj ir . .....(11.19.5)
We cannot apply the quotient law directly to this equation to establish the tensor character of Arj
because Aij is not arbitrary.
Now consider the arbitrary contravariant vector k. Then Bp = Akp k is an arbitrary covariant
tensor.
Multiplying this equation by Aip, we have
Aip Bi = Aip Akp k
ik k = i
Aip Bp = i. .....(11.19.6)
Since Bp in arbitrary vector, hence by quotient law Aip is a contravariant tensor of rank two.
Hence Aij is a contravariant tensor of rank two.
11.20 Illustrative example
Ex.12. If Aij is a symmetric covariant tensor of rank two and Bij is formed by dividing the
cofactor of Aij in the determinant | Aij | = a (say) by | Aij | itself, show that
1
(i) B ij and (ii) Aij Bij = N
a
Sol. By theory of determinants
Aij B ik kj . .....(1)
273
(i) Aij B ik kj
a B ik 1
1
B ik . .....(2)
a
(ii) Again, form (1) identifying j and k, we get
Aij B ij jj N .
1
Ex.13. If Aij = 0 for i j, show that the conjugate tensor Bij = 0 for i j and Bii (no
Aii
summation).
Sol. We have Aij B ik kj . .....(1)
(i) Let k j, then 0 = Aij Bik
= A1j B1k + A2j B2k + ... + Ajj Bjk + ... + ANj BNk
= 0 + 0 + ... + Ajj Bjk + ... + 0 = Ajj Bjk (No summation over j).
But Ajj 0 (No summation over j)
Hence Bjk = 0, j k.
i.e. Bij = 0, i j
(ii) Let k = j, then from (1)
1 = Aij Bij
= Aj1 Bi1 + Ai2 Bi2 + ... + Aii Bii + ... + AiN BiN.
= 0 + 0 + ... Aii Bii + ... + 0 = Aii Bii (No summation over i).
But Aii 0 (No summation over i)
1
Hence Bii (No summation).
Aii
11.21 Self-learning exercises
1. What do you mean by Eienstien summation convention ?
2. What are dummy and free indices ?
3. Define Kronecker delta.
4. Define contravariant and covariant vectors.
11.22 Summary
The unit starts with the introduction of tensors in the space of N dimensions. By giving the
concepts of indicial and summation convention we have defined the covariant and contravariant tensors
274
of one or more ranks. Here we study the different properties of tensor entities. In algebra of tensors we
define addition, subtraction, outer multiplications, contraction and inner-multiplication. Some theorems
and examples on above concepts are given. The symmetric and skew-symmetric tensors have also been
studied in this unit. To test whether a given quantity is a tensor or not, the quotient law of tensors is
given. In the end conjugate tensors have been defined.
11.23 Answers to self-learning exercises
1. § 11.4(a)
2. § 11.4(b), (c)
3. § 11.5
4. § 11.6, § 11.7.
11.24 Exercise
1. Prove that the transformation of tensors form a group.
2. Show that a second rank covariant (or contravariant) tensor is expressible as a sum of two tensors
one of which is symmetric and other is antisymmetric.
3. Prove that the contracted tensor Aij is a scalar..
4. Show that the tensor equation aij i j where is an invariant and j an arbitrary tensor,,
demands that
aij ij .
5. Show that the contraction of the outer product of the tensors Ap and Bq is an invariant.
275
Unit 12 : Riemannian Space, Metric Tensor, Indicator,
Permutation Symbol and Permutation Tensors,
Christoffel Symbols and their Properties
Structure of the Unit
12.0 Objective
12.1 Introduction
12.2 Metric tensors and Riemannian space
12.3 Conjugate metric tensor
12.4 Indicator
12.5 Illustrative examples
12.6 Permutation symbols and tensors
12.7 Christoffel’s symbols
12.8 Properties of Christoffel symbols
12.9 Illustrative examples
12.10 Laws of transformation of Christoffel symbols
12.11 Self-learning exercises
12.12 Summary
12.13 Answers to self-learning exercises
12.14 Exercises
12.0 Objective
In this unit our objective is to generalize the concept of distance between any two neighboring
points from three dimensional space to N-dimensional Riemannian space. We introduce a particular type
of tensor, called metric tensor which has a great importance in the theory of tensor analysis. We also
consider two types of expressions due to Christoffel involving the derivatives of the components of
metric tensor of fundamental tensor gij and gij. These expressions will be called Christoffel symbols of
first and second kind.
12.1 Introduction
We know that in Euclidean space of three dimensional rectangular cartesian coordinates the
distance ds between two neighbouring points (x1, x2, x3) and (x1 + dx1, x2 + dx2, x3 + dx3) is given by
ds2 = (dx1)2 + (dx2)2 + (dx3)2 = dxi dxi, i = 1, 2, 3 .....(12.1.1)
276
The distance ds is also called the line element. If we take the coordinates of points in any of the
curvilinear coordinates (e.g. cylindrical or spherical polar coordinates) such as (x*1, x*2, x*3) then xi
are functions of x*i and dxi are linear homogeneous functions of dx*i given by
x i
dx *m dx*m (i, m = 1, 2, 3)
i
.....(12.1.2)
x
when we substitute dxi from (12.1.2) in (12.1.1), we get a homogeneous quadratic function in dx*i viz.,
x i xi
ds 2 *m *n dx*m dx*n (summation over i) .....(12.1.3)
x x
which can be written as ds2 = g mn
*
dx*m dx*n (m, n = 1, 2, 3) .....(12.1.4)
* xi xi
where g mn (summing over i) .....(12.1.5)
x*m x*n
The differential expression of R.H.S. of (12.1.3) which represents ds2 is called the metric form
or fundamental form of the space under consideration.
Motivated by the above fact, the idea of distance was extended by Riemann, originator of ten-
sor calculus, to a space of N-dimensions.
12.2 Metric tensor and Riemannian space
The quadratic differential form
ds2 = gij dxi dxj, .....(12.2.1)
which expresses the distance between two neighbouring points, whose coordinates in a VN are xi and
xi + dxi, is called a Riemannian metric or line element, gij in called metric tensor or fundamental tensor.
The N-dimensional space characterised by a Riemannian metric is called a Riemannian space
and is denoted by ‘Riemannian – VN’.
Here we postulate that the line element ds is independent of coordinate system i.e. ds2 is an
invariant. We will show that gij is a symmetric covariant tensor of order two, it is called the fundamen-
tal covariant tensor or metric tensor of Riemannian space.
Theorem 1. The fundamental tensor gij is a covariant symmetric tensor of the order two.
Proof : The line element or metric is given by
ds2 = gij dxi dxj .....(12.2.2)
Consider a coordinate transformation from the system xi to x i i 1, 2,3, ..., N as
xi xi x1 , x2 , x3 , ..., xN ,
so that the metric gij dxi dxj transforms to gij dx i dx j . But we have ds2 is invariant.
277
ds2 = gij dxi dxj gij dx i dx j .....(12.2.3)
g pq dx p dx q
x p i x q j
g pq dx dx
xi x j
x p x q i j
or gij g pq i dx dx 0 .....(12.2.4)
x x j
As explained in Example 11 of Unit 11, we deduce from (12.2.4)
p q
that (gij + gji) g pq g qp x x
.....(12.2.5)
xi x j
which shows that (gij + gji) is covariant tensor of the second order.
Now we can write gij 1 gij g ji 1 gij g ji
.....(12.2.6)
2 2
1 1
Then gij dxi dxj
2
gij g ji dx i dx j gij g ji dx i dx j
2
.....(12.2.7)
On interchanging the dummy indices in R.H.S., we get
1 1
gij dxi dxj
2
g ji gij dx j dx i g ji g ij dx j dxi
2
.....(12.2.8)
Adding (12.2.7) and (12.2.8), we get
2gij dxi dxj = (gij + gji) dxi dxj .....(12.2.9)
which show that gij is symmetric. Thus combining the two conclusions that (gij + gji) is a covariant ten-
sor of the second order and gij is symmetric, we conclude that 2gij or gij is a symmetric covariant ten-
sor of the second order.
Note : We call a N-dimensional space as Euclidean space of N-dimensions if its metric is
ds2 = (dx1)2 + (dx2)2 + (dx3)2 + ... + (dxN)2
i.e. gij = 0, i j and gii = 1 (no summation).
12.3 Conjugate metric tensor
We know that gij is a symmetric covariant tensor of the second order and g = | gij | 0, we can
define
G (i, j )
g ij .....(12.3.1)
g
where G (i, j) is the cofactor of gij in the determinant g.
It follows from Theorem 13 of Unit 11 that gij is a symmetric contravariant tensor of the second
order and is said to be conjugate of gij i.e. conjugate metric tensor. It is also called the fundamental
contravariant tensor.
278
Hence the fundamental covariant tensor gij and fundamental contravariant tensor gij, being con-
jugate, are related to each other by the equation
gij g ik kj . .....(12.3.2)
12.4 Indicator
It in implied that the metric in Euclidean space is positive definite i.e. ds2 > 0.
But in the theory of relativity, the metric of the four dimensional space (space-time) is given by
ds2 = – (dx1)2 – (dx2)2 – (dx3)2 + c2 (dx4)2, .....(12.4.1)
where c is the velocity of light and x4 is the time coordinate. This metric is not positive definite. We see
that ds2 > 0 when x1, x2, x3 are constants along the curves, it is zero when, say, x2 and x3 are constants
and x1 = cx4 and negative when x4 in constant.
Thus, in general, for some displacement dxi, the form ds2 may be positive and for others it may
be zero or negative. If ds2 = 0, for dxi not all zero, i.e. the two points are not coincident the displace-
ment is called a null displacement. A curve along which the displacement gij dxi dxj is null despite of the
fact that the two points are not coincident, is called a null curve. For any displacement dxi which is not
null, we introduce an indicator e, which is + 1 or –1, so as to make ds2 always positive, i.e.
ds2 = egij dxi dxj, .....(12.4.2)
where e is called an indicator.
12.5 Illustrative examples
Ex.1. If a metric of a V3 is given by
ds2 = 5 (dx1)2 + 3 (dx2)2 + 4 (dx3)2 – 6 (dx1) (dx2) + 4 (dx2) (dx3)
find (i) g and (ii) gij.
Sol. When we compare the given metric with the metric
ds2 = gij dxi dxj (i, j = 1, 2, 3)
we find that
g11 = 5, g22 = 3, g33 = 4, g12 = g21 = –3, g23 = g32 = 2, g13 = g31 = 0. .....(1)
5 3 0
gij 3 3 2
0 2 4
g = | gij | = 4. .....(2)
To get conjugate of gij, we find
G (1, 1) = 8, G (1, 2) = G (2, 1) = 12, G (2, 3) = G (3, 2) = – 10,
G (2, 2) = 20, G (3, 1) = G (1, 3) = – 6, G (3, 3) = 6. .....(3)
279
G ( i, j )
Since g ij , we obtain
g
3 23 32 5 31 13 3
g11 2, g 22 5, g 33 , g12 g 21 3 , g g , g g . .....(4)
2 2 2
3
2 3
2
5
g 3
ij
5 . .....(5)
2
3 5 3
2 2 2
Ex.2. Show that the metric of a Euclidean space, referred to cylindrical coordinates is
given by
ds2 = (dr)2 + (r d)2 + (dz)2.
Determine its metric tensor and conjugate metric tensor.
Sol. We have
ds 2 gij dx i dx j g pq dx p dx q , .....(1)
where in rectangular coordinates
x1 = x, x2 = y, x3 = z
g12 = g21 = g13 = g31 = g23 = g32 = 0, g11 = g22 = g33 = 1, .....(2)
and in cylindrical coordinates
x 1 r , x 2 , x 3 z ; x cos , y r sin , z z and g pq ? .....(3)
By covariant transformation law
x i x j
g pq gij . .....(4)
x p x q
xi x j
Therefore g11 gij
x 1 x 1
2 2 2
x1 x 2 x 3
1 g11 1 g 22 1 g33
x x x
2 2 2
x y z
r r r
= cos2 + sin2 + 0 = 1. .....(5)
2 2 2
g22 x y z
= r2 sin2 + r2 cos2 + 0 = r2 .....(6)
280
2 2 2
x y z
g33
z z z
=0+0+1=1 .....(7)
xi x j x1 x1 x 2 x 2 x3 x3
g12 g ij x 1 x 2 11 x 1 x 2 22 x 1 x 2 g33
g g
x 1 x 2
x x y y z z
r r r
= – r cos sin + r sin cos + 0. .....(8)
Similarly g13 0 g 23 due to symmetric property g 21 0, g31 0, g32 0.
1 2 2 2 3 2
Hence ds2 g11 (dx ) g 22 (dx ) g33 (dx )
= (dr)2 + (rd)2 + (dz)2. .....(9)
The metric tensor g pq in cylindrical coordinates is given by
1 0 0
g pq 0 r 2 0 .
0 0 1
Clearly g g pq r 2 , .....(10)
and conjugate metric tensor g pq , which is the inverse of the matrix (10), is
1 0 0
1
g pq 0 0 .
r2
0 0 1
Ex. 3. Show that the metric of a Euclidean space, referred to spherical coordinates is
given by
ds2 = (dr)2 + (rd)2 + (r sin d )2.
Determine its metric tensor and conjugate metric tensor.
Sol. We have ds2 = gij dxi dxj g pq dx p dx q , .....(1)
where in rectangular coordinates
x 1 = x, x2 = y, x3 = z
g11 = g22 = g33 = 1, gij = 0, i j .....(2)
and in spherical polar coordinates
2 3
x 1 r , x , x
x = r sin cos , y = r sin sin , z = r cos .....(3)
we have to find g pq .
281
x i x j
By covariant law g pq gij. .....(4)
x p xq
2 2 2
x i x j x1 x 2 x 3
Therefore g11 1 1 gij 1 g11 1 g 22 1 g33
x x x x x
2 2 2
x y z
r r r
= sin2 cos2 + sin2 sin2 + cos2 = 1 .....(5)
2 2 2
g22 x y z
= r2 cos2 cos2 + r2 cos2 sin2 + r2 sin2 = r2 .....(6)
2 2 2
x y z
g33
= r2 sin2 sin2 + r2 sin2 cos2 + 0 .....(7)
= r2 sin2
xi x j
g12 gij
x 1 x 2
x1 x1 x 2 x 2 x3 x 3
1 1 g11 1 2 g 22 1 2 g33
x x x x x x
x x y y z z
r r r
= sin cos(rcos cos) + (sin sin) (rcos sin) + cos (– r sin )
= r sin cos (cos2 + sin2) – r sin cos .....(8)
= 0.
Similarly g13 0, g 23 0. .....(9)
By the symmetric property g 21 0 g31 g32 0. .....(10)
1 2 2 2 3 2
Hence ds 2 g11 (dx ) g 22 (dx ) g33 (dx )
= (dr)2 + (rd)2 + (r sin d )2. .....(11)
The metric tensor g pq in spherical polar coordinates is given by
1 0 0
g pq 0 r 2 0 .
.....(12)
0 0 r 2 sin
Clearly g | g pq | r 4 sin 2 . .....(13)
282
The conjugate metric tensor g pq , which is the inverse of matrix g pq , in given by
1 0 0
1
g pq 0 0 .
r2
0 1
0
r 2 sin
Ex. 4. Show that
(i) (ghj gik – ghk gij) ghj = (N – 1) gik
k k k
(ii) j
( g hk gil g hl g ik ) g hj k
gil g ik
x x x l
Sol. We have gij gik = kj . .....(1)
(i) (ghj gik – ghk gij)ghj = ghj ghj gik – ghj ghk gij
Ngik kj gij
= Ngik – gik
= (N – 1) gik.
k k k
(ii) j
( g hk gil g hl gik ) g hj j
g hk g hj g il j
g hl g hj gik
x x x
k k
j
kj gil j
lj gik
x x
k k
k
gil gik .
x x l
Ex. 5. Show that
(i) gij gkl dgik = – dgjl
(ii) gij gkl dgik = – dgjl
Sol. We have
(i) gij gik kj .....(1)
On differentiation gij dgik + gik dgij = 0
or gij dgik = – gik dgij. .....(2)
Taking inner product on both sides of (2) by gkl, we get
gij gkl dgik = – gkl gik dgij
li dg ij
= – dg lj
= – dg jl. [by symmetric property]
283
(ii) Relation (i) may be written as
gik gij kj .
On differentiation gij dgik + gik dgij = 0
or gij dgik = – gik dgij
gij gkl dgik = – gik gkl dgij
il dgij
= – dgij = – dgjl.
Hence proved.
12.6 Permutation symbols and tensors
The permutation symbol is written as eijk and in the Euclidean three dimensional space V3 is
defined as
0 , if any two of i, j,k are equal
eijk 1 , if i, j,k is a cyclic permutation
.....(12.6.1)
1 , if i, j,k is anticyclic permutation
Thus e112 = e113 = e221 = e223 = e331 = e332 = e111 = e222 = e333 = 0
e123 = e231 = e312 = + 1
e132 = e321 = e213 = – 1 .....(12.6.2)
We now introduce an entity defined by
1
ijk geijk ;ijk eijk , .....(12.6.3)
g
where g is the determinant of metric tensor gij of the space referred, which may not necessarily be rect-
angular. We shall now prove that although eijk is not a tensor, in general, both ijk and ijk are tensors
covariant and contravariant respectively. These are called permutation tensors in three dimensional
space, It is clear from the definitions of eijk, ijk and ijk that they are skew-symmetric in all three
indices.
Theorem 2. The entities defined by (permutation tensor)
1
ijk g eijk ; ijk eijk ,
g
are respectively covariant and contravariant tensors, where eijk is a permutation symbol and g is
the determinant of the metric tensor gij.
Proof : We have
xi x j x k x j x i x k
eijk e jik (interchanging the dummy indices i and j)
x l x m x n x l x m x n
x i x j x k
eijk (using skew-symmetric property of eijk)
x m x l x n
284
xi x j x k
This shows that eijk is skew-symmetric in l and m.
x l x m x n
Similarly, it can be shown that it can be shown that it is skew-symmetric in all l, m and n. But
x r
this expression, apart from the sign, is the Jacobian determinant . From the theory of determi-
x s
nants, it therefore follows that
x i x j x k x i
eijk elmn .....(12.6.4)
x l x m x n x j
Now by covariant law we know that
x i x j
g pq p q gij . .....(12.6.5)
x x
g pq x i x j
Therefore gij
x p x q
2
x r
or g g. .....(12.6.6)
x s
Suppose in the coordinate system x i , the entity ijk be denoted by lmn , where
lmn g elmn . .....(12.6.7)
Now, using (12.6.4) and (12.6.6) in (12.6.7), we find
xi x j x k
lmn g eijk
x l x m x n
xi x j x k
ijk , .....(12.6.8)
x l x m x n
which shows that ijk is a third order covariant tensor.
Also writing elmn for elmn and e ijk for eijk we have
r lmn
lmn 1 lmn x e
e [using (12.6.6)]
g x s g
1 ijk x l x m x n
e [using (12.6.4)]
g x i x j x k
xl x m x n
ijk , .....(12.6.9)
x i x j x k
which shows that lmn in a contravariant tensor of third order.
285
12.7 Christoffel symbols
These are the two expressions due to Christoffel involving contravariant fundamental tensor gij
and the partial derivations of the components of the fundament tensor gij.
The Christoffel symbols of the first and second kind are denoted by [i j, n] and ijk , respec-
tively and are defined as
1 g jk g gij
[ij, k] i ikj k .....(12.7.1)
2 x x x
g
k
ij
kh
ij , h .....(12.7.2)
Notes :
(i) The symbols [ij, k] and may also be represented by Tk,ij and T
l
ij
l
ij respectively..
However we shall use only the brackets type representation.
(ii) These symbols, in general, are not tensors.
(iii) All, but one of the indices of the Christoffel symbols are regarded as subscripts. The ex-
ception in the index l which is treated as the superscript in the symbol of second kind.
(iv) Both the symbols are symmetric with respect to the indices i and j.
(v) In Euclidean space of N-dimensions g11 = g22 = ... = gNN = 1 and gii = 0, i j in this
case all the Christoffel symbols are zero.
N ( N 1)
(vi) Since gij is a symmetric tensor and has independent components in the space
2
gij N ( N 1)
VN , then will have N independent components. Therefore the number of
x k 2
N 2 ( N 1)
independent components of Christoffel symbols of a kind are .
2
12.8 Properties of Christoffel symbols
l
Property I : [ij, m] glm ij
Proof : By definition we have
lij = glk [ij, k] .....(12.8.1)
Taking inner product by glm, we get
glm ijl = glm glk [ij, k]
km ij , k = [ij, m].
Hence proved.
286
gik
Property II : = [ij, k] + [kj, i]
x j
Proof : We know by definition
1 g g jk gij
[ij, k] ikj i k .....(12.8.2)
2 x x x
1 g gij g jk
and [kj, i] ikj k i .....(12.8.3)
2 x x x
keeping in the mind the symmetric property of gij, adding above two expressions, we get
gik
[ij , k ] [kj , i] . Hence proved
x j
g mk
Property III :
x l
g mi g
k
il
ki m
il
Proof : We know that gij gik kj . .....(12.8.4)
Differentiating with respect to xl, we get
g ik gij ik
gij l l g 0
x x
g ik ik g ij
or gij g .....(12.8.5)
x l x l
Taking inner product with gim, we obtain
g ik ik jm g ij
im g g .....(12.8.6)
x l xl
Now using property 2, we finally get
g mk
l
g ik g jm [ jl , i ] [il , j ]
x
= –gik gjm [jl, i] – gik gjk [il, j]
g jm g
k
ij
ik m
il
g mj g ,
k
ij
ki m
il .....(12.8.7)
where the dummy index in the first term has been replaced by i.
1 g
Property IV : 2 g x
i
ij j
x j
log g ,if g is positive
x j
log g ,if g is negative
287
Proof : We have the matrix of gij as
g11 g12 ... g1N
g g 2 N
21 g 22 ...
gij .
.....(12.8.8)
g N1 g N 2 g NN
Since g denotes the determinant of gij, we have
g11 g12 ... g1N
g g 2 N
21 g 22 ...
g .
.....(12.8.9)
g N1 g N 2 g NN
Differentiating it with respect to xj, we get
g11 g12 g1N g11 g12 ... g1N
x j ... j
x j
x g
g g 21 g 22 ... g 2 N
21 g 22 ... g 2 N .....
x j .....(12.8.10)
g N 1 g N 2 ... g NN
g N 1 g N 2 g NN x j x j x j
g11
Clearly, cofactor of = cofactor of g11 in g = gg11 etc. .....(12.8.11)
x j
g g g g
Thus j
11j gg11 12j gg12 ... 1Nj gg 1N ...
x x x x
g g g
Nj1 gg N 1 Nj2 gg N 2 ... NN j
gg NN
x x x
gik
gg ik . .....(12.8.12)
x j
Now using property II, we get
g
j = ggik ([ij, k] + [kj, i])
x
i k
g g
ij kj
i i
g g (as k is dummy index so it is replaced by i)
ij ij
i
2g . .....(12.8.13)
ij
288
i 1 g
Hence
ij 2 g x
j
x
j log g , if g is positive
x j
log
g , if g is negative
Hence proved.
12.9 Illustrative examples
Ex.5. Calculate the Christoffel symbols corresponding to the metric
ds2 = (dx1)2 + G(x1, x2) (dx2)2
where G is a function of x1 and x2.
Sol. Here N = 2 i.e. it is a two dimensional space, where
g11 = 1, g22 = G (x1, x2), g12 = 0, g21 = 0.
The number of Christoffel symbols of a kind will be
N 2 N 1 22 (2 1)
6.
2 2
I. First kind :
1 gii
Case I : i = j = k, then [ii, i] (No summations)
2 x i
1 G
Hence [11, 1] = 0, [22, 2] . .....(1)
2 x 2
1 gij
Case II : i = j k, then [ii, k] . .....(2)
2 x k
1 G
Therefore [11, 2] = 0, [22, 1] . .....(3)
2 x1
1 gij
Case III : i = k j, then [ij, i] . .....(4)
2 x i
1 G
Therefore [12, 1] = 0, [21, 2] . .....(5)
2 x1
Case IV : i j k
It is redundant in two dimensional space
1 ij
II. Second kind : g ii , g 0, i j
gii
l ll
g [ij , l ] .....(6)
ij
289
1 2 22 1 G
Hence 0, g [12, 2] , .....(7)
12 12 2G x1
1 1 11 1 G
0, g [22,1] , .....(8)
11 22 2 x1
2 2 22 1 G
0, g [22, 2] , .....(9)
11 22 2G x 2
Ex.6. Surface of sphere is a two dimensional Riemannian space. Compute the Christoffel
symbols
Sol. For a sphere, r is constant, the metric of the surface of a sphere is given by
ds2 = r2 (d)2 + r2 sin2 (d)2.
Here g11 = r2, g22 = r2 sin2, g12 = g21 = 0
1 1
and g11 2
, g 22 2 2
, g12 g 21 0 .....(1)
r r sin
(i) First kind :
1 gii
(a) i = j = k, then [ii, i] . [No summation]
2 xi
Therefore [11, 1] = 0, [22, 2] = 0. .....(2)
1 gii
(b) i = j k, then [ii, k]
2 x k
Therefore [11, 2] = 0, [22, 1] = – r2 sin cos .....(3)
1 gii
(c) i = k j, then [ij, i]
2 x j
Therefore [12, 1] = 0, [21, 2] = r2 sin cos. .....(4)
(d) i j k,
Redundant in two dimensional space
l
(ii) Second kind : = gll [ij, l] (No summation)
ij
The non-zero components are
1
= g11 [22, 1] = – sin cos
22
2
and = g22 [21, 2] = cot .....(5)
21
The remaining four will be zero.
290
Ex.7. Calculate the Christoffel symbols corresponding to the metric
ds2 = (dx1)2 + (x1)2 (dx2)2 + (x1)2 (sin x2)2 (dx3)2
Sol. For the given metric
g11 = 1, g22 = (x1)2, g33 = (x1)2 (sin x2)2 .....(1)
and gij = 0, i j.
(I) Christoffel symbols of the first kind :
N 2 ( N 1)
Since N = 3, the total number of independent components to be determined are 18.
2
1 g g g
We know that [ij, k] ikj jki ijk . .....(2)
2 x x x
Here, there are four cases.
Case I : i = j = k, then (1) becomes
1 gii
[ii, i] (No summations)
2 x i
We find [11, 1] = 0, [22, 2] = 0, [33, 3] = 0 .....(3)
Case II : i = j k, then (1) becomes
1 g g g 1 gii
[ii, k] ikj iki ijk
2 x x x 2 x k
Therefore we find
[11, 2] = 0, [11, 3] = 0, [22, 1] = – x1
[22, 3] = 0, [33, 1] = –x1 (sin x2)2,
[33, 2] = – (x1)2 sin x2 cos x2 .....(4)
Case III : i = k j, then (1) gives
1 g g g
[ij, i] iij iji iji 1 gii
2 x x x 2 x j
Therefore we find
[12, 1] = 0, [13, 1] = 0, [21, 2] = x1, [23, 2] = 0
[31, 3] = x1 (sin x2)2, [32, 3] = (x1)2 sin x2 cos x2
Case IV : i j k, then (2) by virtue of (1), becomes [ij, k] = 0.
Hence, [12, 3] = 0, [23, 1] = 0, [31, 2] = 0 .....(6)
(II) Christoffel symbols of second kind :
1 1 1
We have g11 1, g 22 1 2 , g 33 1 2
g11 (x ) ( x ) (sin x 2 ) 2
and gij = 0, i j .....(7)
291
l
We know that = glk [ij, k] .....(8)
ij
Here too, we have to find out 18 independent components
Since gij = 0 when i j.
l
We have from (2) = gll [ij, l] (No summation over i) .....(9)
ij
The non-zero components are
1 11 1 1 11 1 2 2
g [22,1] x , g [33,1] x (sin x )
22
33
2
g 22 [33, 2] sin x 2 cos x 2
33
2 22 1
g [12, 2] 1
12 x
3 1
g 33[13,3] 1 ,
13 x
3
g 33[32,3] cot x 2 .....(10)
32
The remaining twelve components will be zero.
Ex.8. If the metric of a VN is such that gij = 0 for i j, show that
i i 1 gii
0,
jk jj 2 gii x i
i i
j log gii ; i log gii
ij x ii x
where i, j and k are not equal, and the summation convention does not apply.
1
Sol. Here g ii and g ij 0, i j
gii
i il ii
(i) g [ jk , l ] g [ jk , i], ( g il 0, i l )
jk
=0 ( fundamental tensors are zero when i j k) .....(1)
i il 1 il gil 1 g
(ii) g [ jj , l ] g l
g ii iii
jj 2 x 2 x
1 gii
. .....(2)
2 gii xi
292
i
(iii) = gil [ij, l] = gii [ij, l]
ij
1 ii gii 1 gii
g
2 x j 2 gii x j
x j
log
gii . .....(3)
i
(iv) = gil [ii, l]
ii
1 ii g ii
g ii [ii, i ] g
2 xi
1 gii
ii
2 g x i
i log gii .
x
.....(4)
12.10 Laws of transformation of Christoffel symbols
Theorem 3. The Christoffel symbols are not tensor quantities.
Proof : Let us consider the transformation of the Christoffel symbols from the coordinate sys-
tem xi to x k .
(i) By definition we know that
1 gin g mn glm .....(12.10.1)
[lm, n] n
2 x m x l x
1 gik g jk gij
and [ij, k] i k .....(12.10.2)
2 x j x x
We know that by covariant law
i j
glm x x gij .....(12.10.3)
xl x m
n
Differentiating with respect to x , we get
glm xi xi gij x k 2 x i x j x i 2 x j
g ij gij .....(12.10.4)
x n x l x m x k x n x nx l x m x l x m x n
Similarly (by cyclic order)
g mn x j x k x i g jk 2 x j x k x j 2 x k
g jk g jk .....(12.10.5)
x l x m x n xl xi x l x m x n x m x nx l
g nl x k x i x j g ki 2 x k x i x k 2 x i
g ki g ki .....(12.10.6)
x m x n x l x m x j x m x n x l x n x l x m
293
1 1
Multiplying (12.10.5) and (12.10.6) by and (12.10.4) by then adding keeping in
2 2
view (12.10.1) and (12.10.2) we get on changing the dummy indices appropriately, we get
i j k k 2 i
lm, n x x x ij , k x x g
ik
x l x m x n x n x l x m
xi x j x k x i 2 x j
ij , k .....(12.10.7)
x l x m x n x n x l x m
which shows that Christoffel symbols of first kind do not behave like tensors.
(ii) By contravariant law, we have
g np
x n x p rs .....(12.10.8)
g .
x r x s
np
Taking inner multiplication of (12.10.7) by g and its corresponding equivalent from (12.10.8),
we get
xi x j x p x p 2 x j
g np lm, n ij , k l m rk s g rs gij g rs ri s
x x x x x l x m
p ks x i x j x p is x
p
2 x j
or g ij , k g ij g
lm x l x m x s x s x l x m
xi x j x p x p 2 x j
s
ij
x l x m x s x j x l x m
.....(12.10.9)
which shows that Christoffel symbols of second kind also do not behave like tensors.
Remark : We have proved that Christoffel symbols are not tensor quantities. But in some very
special case of linear transformation of coordinates, viz.
x j amj x m b j ,
where amj and bj are constants, we have
2 x j
0.
x l x m
and the equations (12.10.8) and (12.10.9) become
i
lm, n ij , k x x j x k
, .....(12.10.10)
x l x m x n
p s x p xi x j
s , .....(12.10.11)
lm ij x x l x m
which shows that in the case of linear transformation Christoffel symbols transform like a tensor.
294
x r
Note : Taking inner multiplication of (12.10.9) by , we get
x p
i
p x r x j r 2 x j r
p
lm x
xx
s
ij l
x m
s
x l x m
j
2 xr p x r r x i x j
p l m .....(12.10.12)
x l x m lm x ij x x
It is an important result and should be remembered. It expresses the second order partial
i
derivative of xr with respect to x in terms of the first derivatives and Christoffel symbols of the second
kind.
12.11 Self-learning exercises
1. Define metric tensor.
2. Define permutation tensors.
3. What do you mean by indicator.
4. Show that g dxi dxj in invariant.
5. How many independent components of Christoffel symbols in V3.
12.12 Summary
In this unit we have generalised the concept of distance to the Riemannian space by metric and
defined metric tensor, a covariant symmetric tensor of rank two, some examples are given to calculate
components of metric tensor in different Riemannian spaces. We have defined Christoffel symbols of
first and second kinds, which are the expressions of partial derivatives of fundamental tensor gij. Some
properties of these symbols are given and some examples are given to calculate these symbols. In the
end we have shown that Christoffel symbols are not tensor quantities.
12.13 Answers to self-learning exercises
1. § 12.2
2. § 12.6
3. § 12.4
4. § 12.2
5. 18
12.14 Exercises
1. Show that gij is a covariant tensor of order two.
295
2. Prove that gij is a symmetric contravariant tensor of rank two.
3. What are the fundamental tensors and show that :
Aij dgij = – Aij dgij.
4. Prove that :
(i) gij gkl dgik = – dgil,
(ii) gij gkl dgik = – dgjl.
5. Prove that the permutation tensors are tensor of third order and also show that
ijk = gil gjm gkn lmn,
where the symbols have their usual meanings.
6. Show that the transformation of Christoffel symbols form a group.
7. Evaluate Christoffel symbols in spherical coordinates.
8. Prove that the Christoffel symbols are not tensor.
296
Unit 13 : Covariant Differentiation of Tensors, Ricci Theorem,
Intrinsic Derivative
Structure of the Unit
13.0 Objective
13.1 Introduction
13.2 Covariant differentiation of vectors
13.3 Covariant differentiation of second order tensors
13.4 Ricci’s theorem
13.5 Illustrative examples
13.6 Divergence of a vector
13.7 Gradient of a scalar
13.8 Laplacian of a scalar
13.9 Curl of a covariant vector
13.10 Illustrative examples
13.11 Intrinsic derivative (absolute derivative)
13.12 Self-learning exercise
13.13 Summary
13.14 Answers to self-learning exercises
13.15 Exercise
13.0 Objective
The objective of this unit is to study the behavior of partial derivatives of vectors and tensors
and consequently covariant differentiation. The properties of covariant differentiation and its uses are also
the points of study.
13.1 Introduction
The transformation laws of partial derivatives of covariant, contravariant vectors and tensors are
not like tensor quantities. So we investigate a particular form of partial derivative which behaves like
tensors and it will be called covariant derivative.
297
13.2 Covariant differentiation of vectors
Here we study the transformation laws of the partial derivatives of contravariant and covariant
vectors. We also investigate that these partial derivatives behave like tensors or not.
(i) Covariant derivative of contravariant vector :
We have from contravariant law of vector
x k i
Ak A, ......(13.2.1)
x i
Differentiating partially with respect to xj, we get
Ak x k x n Ai 2 x k x n i
A. .....(13.2.2)
x j x i x j x n x n x i x j
The presence of second form on the R.H.S. of above equation shows that the partial derivative
Ak
. does not behave like a tensor..
x j
Putting the value of
2 xk p x k k x r x s
p i n
x n x i in x rs x x
in the above equation, we have
x k p x k k x r x s x n i x k x n Ai
p i n
A i j n
x j in x rs x x x
j
x x x
Ak k x r s i p x k x n i x k x n Ai
or i jA A i ...(13.2.3)
x j rs x
p
im x x
j
x x j x n
Using (13.2.1) and making suitable changes of dummy indices, we may write
Ak k r A p p i x k x n
A n A p . ......(13.2.4)
x j rj x in x x j
Introducing following comma notation, viz.
Ak k r
A,kj j A .....(13.2.5)
x rj
the above equation may be written as
x k x n
A,kj A,np , ......(13.2.6)
x p x j
which shows that A,kj behaves like a mixed tensor of second order. It is called covariant derivative of a
contravariant vector Ak with respect to xj.
298
(ii) Covariant derivative of covariant vector :
We have from covariant transformation law of vector
x i
Ak Ai . .....(13.2.7)
x k
Differentiating partially with respect to x j, we get
Ak x i x n Ai 2 x i
Ai, ......(13.2.8)
x j x k x j x n x j x k
2 x i p x i i x l x m
Substituting
x j x k jk x p lm x j x k
and making suitable changes in dummy indices.
Ak x i x n Ai p x i i x l x m
Ai
x j x k x j x n jk x p lm x j x k
Ak p Ai l x i x n
or p n Al k j .
A ......(13.2.9)
x j jk x ni x x
Introducing the comma notation, viz.
Ak p
Ak,j Ap , ......(13.2.10)
x j jk
x i x n
we get above relation as Ak,j Ai,n , .....(13.2.11)
x k x j
which shows that Ak,j behave like a covariant tensor of second order. If is called covariant derivative of
covariant vector Ak with respect to x j.
Note : Expression (13.2.6) and (13.2.11) are very important and should be remembered. These
may be taken as definition of covariant derivatives of contravariant and covariant derivatives respec-
tively.
13.3 Covariant differentiation of second order tensors
In order to extend the process of covariant differentiation to tensors of order more than one, we
choose, without loss of generality, a mixed tensor Aij .
We have from transformation law
xi x l m
Aij Al . ......(13.3.1)
x m x j
x m
Taking inner product with , we get
x j
x m i x l m
A j Al . ......(13.3.2)
xi x j
299
Differentiating with respect to x k, we find
i
2 x m i x m A j 2 xl m x l x r Am
Aj i k j Al j k r ......(13.3.3)
x k x i x x k x x x x x
Using value of second order derivative, we get
p x m m x s x t i x m Aij
p k i Aj i
ki x st x x x x k
p x l l x s x t m x l x r Alm
p k A j k
j l
. .......(13.3.4)
kj x st x x x x x r
Now using (13.2.2) and changing appropriate dummy indices
x m Ai i n p i x l x r Alm m t
i k
A j Ap j k r Alt Akm . .....(13.3.5)
x x kn kj x x x rt rl
Introducing the comma notation
Aij i p
Aij ,k k
Anj Aip , .....(13.3.6)
x kn kj
the above relation may be written as
x m i x l x r m
A j ,k Al ,r . .....(13.3.7)
xi x j x k
x n
Taking inner multiplication by , we get
x m
x n x l x r m
in Aij,k Al ,r
x m x j x k
x n x l x r m
or Anj,k m j k Al ,r ......(13.3.8)
x x x
This shows that Anj,k is a mixed tensor of third order, contravariant of rank one and covariant of
rank two. It is called the covariant derivative of Alm with respect to x r.
Note : The covariant derivative Aij ,k defined by (13.3.6) contains three terms :
(i) The partial derivative of Aji with respect of xk.
(ii) A positive sign term similar to that which occurs in the covariant derivative of a contravariant
vector.
(iii) A negative sign term similar to that which occurs in the covariant derivative of a covariant
vector.
13.4 Ricci’s theorem
The covariant derivatives of the tensors gij, gij and ij all vanish identically.
300
Proof : (i) Covariant derivative of gij with respect to xk
gij r r
gij,k g rj gir
k
x ik jk
gij
ik , j jk ,i [Using property (1) of Christoffel symbols]
x k
gij gij
k
[Using property (2) of Christoffel symbols]
x x k
= 0. .....(13.4.1)
(ii) Covariant derivative of gij with respect to xk
g ij i j
g ,kij k
g rj g ir
x rk rk
g ij g ij
x k x k
= 0, [Using property (3) of Christoffel symbols] ......(13.4.2)
(iii) Covariant derivative of ij with respect to xk
ij l i i l
ij,k j i
x k lk jk
iji l
k [Using property of Kronecker delta]
x jk jk
ij
x k
=0 [ ij is a constant either 1 or 0] ......(13.4.3)
Hence gij,k 0, g ,kij 0 , ij ,k 0
which shows that the tensors gij, gij and ij may be treated as constants in covariant differentiation.
13.5 Illustrative examples
Ex.1. Prove that
1 k
Ai,j j
g x j
Ai j g Akj .
ij
Show that if associate tensor Aij is symmetric, then
1 1 g jk
Ai,j j
g x j
Ai j g A jk
2
x i
.
301
Sol. We have from process of covariant differentiation
Ai j j r
Ai,kj Air Arj
k
......(1)
x rk ik
Putting k = j, we get
Ai j r j j r
Ai,j j Ai Ar
x j rj ij
Ai j j r
x j
Air
x r
logg Ar
ij
Ai j 1 k
i Aij
x g x j
g A k
j
ij
[on making suitable changes in dummy indices]
1 j k
g x j
Ai
j
g Ak
ij
......(2)
This proves the first result.
k
Now Akj Akj g lk ij,l
ij
= Ajl [ij, l]
1 jl gil g jl gij
A j i l
2 x x x
1 jl gil 1 g jl 1 jl gij
A A
2 x j 2 xi 2 xl
1 lj gij 1 jl g jl 1 jl gij
A A A , ......(3)
2 xl 2 x i 2 x l
where in the first term the dummy indices suitably have been changed. Since Aij is symmetric the first
and the last term will cancel out and therefore
k 1 g jk
Akj A jk . ......(4)
ij 2 x i
Substituting this result in (1), we get the required result as
1 1 g
Ai,j j
g x j
Aij g A jk iki .
2 x
Ex.2. Prove that
1 i
A,ijj
g Aij A jk
g x j jk
Show that the last term vanishes if Aij is skew-symmetric.
302
Sol. We have from covariant differentiation
Aij j i
A,lij l
Air Arj . ......(1)
x rl rl
Putting l = j, we get
Aij j i
A,ijj j
Air Arj
x rj rj
Aij 1 g i
j
Air r
Arj
x 2 g x rj
[using property (4) of Christoffel symbols]
Aij 1 g i
j
Air r
Akj
x g x kj
1 Aij ij g
kj i
g A A
g x j x j kj
[making suitable changes in dummy indices]
1 i
g x j
g Aij A jk .
jk
.......(2)
This proves the first result.
If Aij is skew-symmetric, then
i i
Akj A jk
kj kj
i
Akj [interchanging the dummy indices]
jk
i
Akj [using symmetric prop. of Christoffel symbols]
kj
i
2 Akj = 0
kj
i
Akj = 0.
kj
Hence, if Aij is skew-symmetric, then
1
A,ijj
g x j
g Aij .
Hence Proved.
303
Ex.3. If Aijk is a skew-symmetric tensor, show that
1
g x k
g Aijk ,
is a tensor.
Sol. Since Aijk is a skew-symmetric tensor, we have
Aijk = – Ajik, Aijk = – Aikj, Aijk = – Akji. ......(1)
We know that
Aijk rjk i irk j ijr k
A,lijk
A A A ......(2)
x l rl rl rl
Contracting over i and l (l = i), we get
Aijk rjk i irk j ijr k
A,iijk A A A .....(3)
x i ri ri ri
j j j
But Airk Arik Airk ,
ri ri ir
k
similarly, 2 Aijr 0. ......(4)
ri
Aijk i
Hence, A,iijk i
Arjk
x ri
Aijk rjk 1 1
x i
A
g x r
g
g x j
Aijk g . .....(5)
Since left hand side, which is a covariant derivative of a tensor is a tensor, the right hand side
will also be a tensor.
13.6 Divergence of a vector
(i) Divergence of a contravariant vector Ai is defined as the contraction of its covariant deriva-
tive. It is denoted by div Ai and is an invariant. Thus
div Ai = Ai,i. .....(13.6.1)
(ii) The divergence of a covariant vector Ai is denoted by div Ai and is defined as
div Ai = gjk Aj,k. .....(13.6.2)
It is also an invariant.
Note : The concept of divergence may be extended to the contravariant tensors of higher order
or to mixed tensors. The divergence of a tensor may be obtained first by taking a covariant derivative of
it and contracting over a superscript and the subscript of covariant derivative.
Thus,
div Aij Aij ,i . .....(13.6.3)
The order is reduced by one in taking its divergence.
304
Theorem 1. If Ai is a contravariant vector, then
1
div Ai
g x i
Ai g .
Proof. The covariant derivative of Ai is given by
Ai i
A,ij
j
Ar .
......(13.6.4)
x rj
Contracting over i and j (j = i), we get
i
div Ai A,i
Ai i
i
Ar
x ri
Ai 1
x i
Ar
g x r
g [changing dummy index r to i]
Ai 1
i Ai
x g xi
g
1
g x i
Ai g . ......(13.6.5)
Theorem 2. To prove that
1
div Ai
g x r
g g rk Ak div Ai
where Ai and Ai are the contravariant and covariant components of the same vector A.
Proof. We have by definition
div Ai = gjk Aj,k
A j r
g jk k Ar
x jk
A j rk j g jr
g jk Ar g j [using property (3) of Christoffel symbols]
x k kj x
A j j g jr
g jk g rk
Ar Ar
x k jk x j
jr
Ar
g rk
x k
g rk Ar
1
g x k
g A gx r j
Ak 1 g rk
g kr
x r
g kr Ak
g x r
g Ak
x r
[on changing the dummy indices]
1
g x r
g g kr Ak . ......(13.6.6)
305
But grk Ak = Ar (associate vector) ......(13.6.7)
Therefore above relation may be written as
1
div Ai
g x r
g Ar
= div Ar
= div Ai.
Hence proved.
Remark : According to Ricci’s theorem gij be have like a constant in covariant differentiation,
we may write
jk jk
div Ai g A j,k g A j ,k
= (Ak),k = div Ak
= div Ai.
13.7 Gradient of a scalar
If I is a scalar function of coordinates xi, the gradient of I is defined by
I
grad I I ,i , ......(13.7.1)
x i
where I,i is a covariant vector.
Theorem 3. The covariant differentiation of invariants is commutative, that is
I ,i , j I , j ,i .
I
Proof. We have I ,i . .....(13.7.2)
xi
I r
Therefore, I ,i , j
x j xi
I ,r ij
2I I r
j i
r
x x x ij
2I I r
i j
r
x x x ji
= (I,j),i. ......(13.7.3)
13.8 Laplacian of a scalar
If I is a scalar functions of coordinate xi, then the divergence of grad I is defined as the Laplacian
of I and it is denoted by 2I.
Thus 2I = div grad I = div I,i. ......(13.8.1)
306
Theorem 4. Prove that
1
2
(i) div grad I I
g x r
g g rk I ,k ,
2
2 jk I I r
(ii) div grad I I g j k r .
x x x jk
Proof. (i) From theorem 2 we have
1
div Ai
g x r
g g rk Ak , .....(13.8.2)
where Ai is a covariant vector.
Since gradient of a scalar I is a covariant vector, setting Ai = I,i in the above equation, we get
1
div A,i
g x r
g g rk A,k
1
2I
g x r
g g rk I ,k . .....(13.8.3)
(ii) 2I = div I,i = gjk (I,j),k
I r
g jk ,ik I ,r
x jk
jk
2 I ,i I r
g j k r . ......(13.8.4)
x x x jk
13.9 Curl of a covariant vector
From a covariant vector Ai in V3, we can form the contravariant vector
Bk = jik Ai,j ......(13.9.1)
and call Bk the curl of vector Ai and written as curl Ai.
Thus curl Ai = B k = jik Ai,j .....(13.9.2)
We may also write on interchanging the dummy indices i and j as
Bk =ijk Aj,i = – jik Ai,j .....(13.9.3)
Hence 2Bk =jik (Ai,j – Aj,i)
1
2
Bk jik Ai, j A j,i .....(13.9.4)
Thus curl Ai may also be defined as
1 jik
curl Ai B k
2
e
Ai, j A j ,i . .....(13.9.5)
307
13.10 Illustrative examples
Ex.4. If Aij is the curl of a covariant vector, prove that
Aij,k A jk ,i Aki, j 0.
Show further that this expression is equivalent to
Aij Aik Aki
k
k
0.
x x x j
If Aij = Bi,j – Bj,i,
prove that Aij,k + Ajk,i + Aki,j = 0.
Sol. Let Bi is a covariant vector and let its curl be Aij.
Thus curl Bi = Aij
Bi,j – Bj,i = Aij. .....(1)
Bi p
we have Bi,j B p . ......(2)
x j ij
Inter changing i and j, we get
B j p B j p
Bj,i B p i Bp
i
.....(3)
x ji x ij
Bi B j
Aij Bi, j B j ,i .....(4)
x j x i
Now, we have
Aij p p
Aij,k Apj Aip
k .....(5)
x ik jk
2 Bi 2 B j B p B j p Bi B p p
j k i k j p p i ......(6)
x x x x x x ik x x jk
Aik p p
Similarly Ajk,i i
Apk Aip .....(7)
x ji ki
2 B j 2 Bk B p Bk p B j B p p
or Ajk,i .....(8)
x k xi x i x j x k x p ji x p x j ki
Aij p p
Similarly, Aki,j Api Aip
j
.....(9)
x kj ij
2 Bk 2 Bi B p B p B B p p
Aki,j i j j k i pi kp k ......(10)
.
x x x x x x kj x x ij
Adding (5), (7) and (9), we get
Ajk,i + Aki,j + Aij,k = 0.
Proved I part.
308
Again, we are given Aij = Bi,j – Bj,i
Aji = Bj,i – Bi,j
= – (Bi,j – Bj,i)
= – Aij
Aij + Aji = 0. .....(11)
Equation (11) shows that Aij is antisymmetric.
Adding (5), (7) and (9) and using (11), we get
Aij A jk Aki
Aij,k + Ajk,i + Aki,j . .....(12)
k i
x x x j
Hence from (12) if follows that
Aij,k + Ajk,i + Aki,j = 0
Aij A jk Aki
is equivalent to k
i
0.
x x x j
Proved II part.
Ex.5. Evaluate div Aj in (i) cylindrical polar coordinates, and (ii) spherical polar coordi-
nates.
Sol. (i) For cylindrical polar coordinates
x 1 = r, x2 = , x3 = z
g11 = 1, g22 = r2, g33 = 1, gij = 0, i j
g = | gij | = r2 ......(1)
The physical components in cylindrical polar coordinates of Ar, A and Az.
Therefore (A)2 = (Ar)2 + (A)2 + (Az)2 .....(2)
Also (A)2 = g11 (A1)2 + g22 (A2)2 + g33 (A3)2 .....(3)
Hence Ar g11 A1 A1 , A g 22 A2 rA2 , A z g33 A3 A3 .....(4)
1
Now, by definition div A j
g x i
g Ai
1
r x i
rAi
1
r x x
i rA1 2 rA2 3 rA3
x
1
r r
rAr
A
z
rA z
Ar 1 A A z Ar
. .....(5)
r r z r
(ii) For spherical polar coordinates
x = r, x2 = , x3 =
g11 = 1, g22 = r2, g33 = r2 sin2, gij = 0, i j
309
g = | gij | = r4 sin2. .....(6)
The physical components are denoted by Ar, A, Aand are given by
Ar g11 A1 A1 , A rA2 , A r sin A3 .....(7)
1
By definition div A j
g x i
g Ai
1
r 2
sin x
r i
2
sin Ai
1 2
2
r sin x i
x x
r sin A1 2 r 2 sin A2 3 r 2 sin A3
1 2
r 2 sin x
r sin A r
r sin A
rA
Ar 1 A 1 A 2 Ar cot
A . .....(8)
r r r sin r r
13.11 Intrinsic derivative (Absolute derivative)
The intrinsic derivative or absolute derivative of a covariant vector Ai along a curve x j = x j (t)
dx j dx j
is defined as the inner product of the covariant derivative of Ai and i.e. Ai, j and is denoted
dt dt
Ai
by .
t
Ai dx j dAi dAi r dx j
Thus Ai , j Ar . .....(13.11.1)
t dt dt dt ij dt
Similarly the intrinsic derivative of contravariant vector Ai.
j i j
Ai Ai dx dA i Ar dx .
,j .....(13.11.2)
t dt dt rj dt
The vectors Ai or Ai are said to move parallelly along a curve if their intrinsic derivatives along
that curve are zero, respectively.
i ,i .....i
Similarly we can define the intrinsic derivative of higher order tensor A j11 , 2j2 .... mjn along a curve
xk = xk (t) defined by
Aij1 ,,i2j .....im
dx k
1 2 .... jn
Aij1 ,,i2j .....im
.... j , .....(13.11.3)
t 1 2 n ,k
dt
where summation is taken over the index k.
Thus the intrinsic derivative is a tensor of the same order and type as the original tensor.
310
The intrinsic derivative of an invariant I is defined as
I dxi I dx i dI
I ,i , .....(13.11.4)
t dt xi dt dt
which is same as its total derivative.
The intrinsic derivatives of higher order can easily be defined as
i
2 A j dx k
t 2
i
Aj
t t
Aij ,k
t dt
dx k dx p
Aij ,k . ......(13.11.5)
dt , p dt
Theorem 8. The intrinsic derivatives of gij, gij and ij are zero.
dx k
Proof.
t
,k
gij gij
dt
0 [using Ricci theorem]
ij dx k
t
g g ij
,k dt
0
i dx k
t
j ij
,k dt
0.
13.12 Illustrative examples
Ex.6. Show that
dx i d 2 xi i dx j dx k
2 .
t dt dt jk dt dt
dxi
Sol. Let Ai (contravariant vector)
dt
dx i i dx j
Now
t dt
A A,i j
t dt
Ai i dx j
j Ar
x rj dt
dAi i r dx j
A
dt rj dt
d dx i i dx r dx j
dt dt rj dt dt
d 2 xi i dx j dx k
.
dt 2 jk dt dt
311
Ex.7. If two unit vectors Ai and Bi are defined along a curve C such that their intrinsic
derivatives along C are zero, show that the angle between them is constant.
dx j dx j
Sol. It is given, that A,i j 0, B,i j 0, .....(1)
ds ds
at every point of C.
dx j dx j
Therefore Ai , j gik A,kj 0, .....(2)
ds ds
dx i dx j
Also Bi, j gik B,kj 0, .....(3)
ds ds
at every point of C.
Let be the angle between unit vectors Ai and Bi, then
cos = Ai Bi
d dx j
ds
cos Ai Bi ,j ds
d dx j
sin
ds
Ai B,i j Ai, j B i
ds
[using (1) & (2)]
d
sin = 0. .....(4)
ds
From (4) it follows that either = 0 or = constant.
But 0 being included in constant, we conclude
= constant.
Hence Proved.
Ex.8. If the intrinsic derivative of a vector Ai along a curve C vanishes at every point of
the curve, then show that the magnitude of the vector A i is constant along the curve.
Sol. Let the equation of the curve C be
x i = xi (s). .....(1)
dx j
It is given that A,i j 0, at every point of C. .....(2)
ds
We know that Ai = gik Ak and (gik), j = 0. .....(3)
dx j dx j
Therefore, Ai , j gik A,kj 0 at every point of C. .....(4)
ds ds
Since, A 2 = Ai Ai
dA2 2 dx j
ds
s
A Ai Ai ,j ds
dx j
Ai , j Ai Ai A,i j ds
312
dx j i i dx j
Ai , j A Ai A, j
ds ds
= 0. [using (1) and (2)] .....(5)
Hence A2 = constant.
i.e. magnitude of vector Ai is constant.
Hence proved.
13.12 Self-learning exercises
1. What do you understand by the covariant derivative of a covariant vector ?
2. Define covariant derivative of a contravariant vector.
3. Show that the covariant derivative of a covariant tensor of second order is a covariant tensor of
third order.
4. Define intrinsic derivatives of a tensor.
Ar
5. Show that is not a tensor even though Ap is a covariant tensor of rank one.
x s
13.13 Summary
In this unit we have studied the partial differentiation of tensors. We have defined a particular
process of it, called covariant differentiation. The properties of covariant differentiation like Ricci’s theo-
rem have also been studied. The use of covariant differentiation to define gradient, divergence and curl
have also been discussed.
13.14 Answers to self-learning exercises
1. § 13.2(ii) 2. § 13.2(i)
3. § 13.3 4. § 13.11
5. § 13.2(i)
13.15 Exercise
1. State and prove Ricci’s theorem on fundamental tensors.
2. Prove that the covariant derivative of the tensor aik with respect to xk, that is a,ikk has the expres-
sion
1 i
a,ikk
g x k
g a ik a im .
km
313
Unit 14 : Geodesics, Differential Equation of Geodesic,
Geodesic Coordinates, Field of Parallel Vectors
Structure of the Unit
14.0 Objective
14.1 Introduction
14.2 Geodesic
14.3 Euler’s condition
14.4 Differential equation of geodesic in a VN
14.5 Curvature of a curve
14.6 Null geodesic
14.7 Illustrative examples
14.8 Geodesic coordinates
14.9 Illustrative examples
14.10 Field of parallel vectors (Parallelism of vectors)
14.11 Parallelism of a vector of variable magnitude along a curve
14.12 Illustrative examples
14.13 Self-learning exercises
14.14 Summary
14.15 Answers to self-learning exercises
14.16 Exercises
14.0 Objectives
The geodesic, a curve of stationary length on the surface, Riemannian coordinates, geodesic co-
ordinates are the points of study of this unit. Parallelism of vectors and fundamental theorem on Rieman-
nian geometry have also been given.
14.1 Introduction
In the calculus, we study the process of finding stationary values of a function. While in calculus
of variation we find a path on which an integral has stationary value. This gives a process to find shortest
path joining any two points on a surface, which we call the geodesic curve.
14.2 Geodesic
“A geodesic, in a Riemannian space VN, is a curve whose length has stationary value with re-
spect to arbitrary small variations of the curve, the end points being held fixed.”
314
Geodesic on a surface in Euclidean three dimensional space may be defined as the curve along
which the shortest distance measured on the surface between any two points in its plane.
The differential equations of a geodesic can be obtained with the help of Euler’s equations, which
are derived by the technique of calculus of variations.
14.3 Euler’s condition
t1
Theorem. The integral f xi , x i dt has stationary value on the curve whose differential
t0
equations are
f d f
i
i 0,
x dt x
dxi
where x i .
dt
Proof. Let xi = xi (t) be a parametric equation of a curve C in VN joining two fixed points
t1
A (t = t0) and B (t = t1) on it. Let the integral I = f xi , x i dt has stationary value on the curve C
t0
and C be a neighbouring curve whose equation is given by
xi = xi (t) + i (t),
where is small and i (t) are arbitrary continuous differentiable functions of t, satisfying i (t0) = 0,
i (t1) = 0 to ensure that the curve passes through A and B. The value of I taken along the curve C is
thus a function of of the form
t1
I f xi i , x i i dt .....(14.3.1)
t0
Since the integral I is stationary on C for which = 0, we have the condition I (0) = 0.
Differentiating (14.3.1) with respect to
t1
f xi i , x i i dt
0t
t1
f xi i , x i i dt
t0
t1 i
i
f xi i , x i i i f x i i , x i i dt
t0
x
The stationary requirement I (0) = 0, now gives
t1 i
i
f x i , x i i i f x i , x i dt 0 .....(14.3.2)
t0
x x
The second term becomes as
1 t
t1 f i t1 d f i
i dt i t
dt
t0 xi x t0 0 dt x i
315
t1 d f
i dt. .....(14.3.3)
t0 dt x i
Since i (t0) = 0 = i (t1), thus equation (14.3.2) becomes
t1 f d f i
i i dt 0. .....(14.3.4)
t0
x dt x
Since i is arbitrary, subject to its being differentiable and vanishing at A and B, equation (14.3.4)
implies that
f d f
i
i 0. .....(14.3.5)
x dt x
Hence (14.3.5) are necessary conditions for the integral I to be stationary. These are called Euler’s
conditions or Euler’s equations.
14.4 Differential equation of geodesics in a VN
Using the property that geodesic curve is a path of stationary length joining two points A and B
in it, we shall now find the differential equations of it in the space VN.
In the Riemannian space VN, we have
ds2 = egij dxi dxj
2
ds dxi dx j
eg ij
dt dt dt
1/ 2
s egij x i x j f x i , x i , (say)
t1 ds t1 1/ 2
Now, s
t0 dt
dt
t0
eg ij x i x j dt
t1
or f x i , x i dt. .....(14.4.2)
t0
In order that s is stationary, the function f must satisfy the Euler’s equations viz.,
f d f
i
i 0. .....(14.4.3)
x dt x
1/ 2
Now we have f egij x i x j s .....(14.4.4)
and gij being a function of xi, therefore
f 1 g ij i j
l e x x . .....(14.4.5)
x 2s x l
f 1
Also, l
egij x i l j x j li
x 2s
e
2 s
x i gil x j g lj
316
e
2 s
x i g il x i gil
e i
x gil . .....(14.4.6)
s
j
d f e i gil dx e
Now l s x j dt gil
x xi 2
s
s x i gil .....(14.4.7)
dt x
Substituting (14.4.5) and (14.4.6) in (14.4.3), we get
1 i j gij i j gil
s
x x x x g il
x i x i g il 0
s
l j
2 x x
s 1 g g g
x i x i gil x i x j ilj x i x j ijl x i x j ijl 0
gil
s 2 x x x
s
x i x i gil x i x j ij , l 0
gil
s
Taking inner product by glm, we get
s m
im
x i im x i x i x j 0
s ij
d 2 x m
s dx m m dxi dx j
0.
dt 2 s dt ij dt dt
These are the differential equations of a geodesic in parameter t. These may further be simpli-
fied, if we choose the arc distance s alone C as a parameter, i.e., s = t. Then,
s 1,
s 0. .....(14.4.10)
d 2 x m m dx i dx j
Hence (14.4.9) reduces to 0. .....(14.4.11)
ds 2 ij ds ds
These are the required differential equations of geodesic. These constitute N-differential equa-
tions of the second order, and in terms of the intrinsic derivative may be written as
dx m
0. .....(14.4.12)
s ds
Theorem 1. In general, one and only one geodesic passes through two specified points
lying in a small neighbourhood of a point O of a VN.
Proof. The differential equations of a geodesic curve in a VN are
d 2 x i i dx j dx k
0. ......(14.4.13)
ds 2 jk ds ds
These are N differential equations of second order, therefore, their general solution will involve
2N constants. The theory of differential equations states that these constants will be uniquely determined
317
dx i
if the initial values of xi and are given at a point. It means geometrically, that at any point of the
ds
space there is a unique geodesic with given direction. Since the geodesic is defined in terms of the curve
passing through two points, it will be unique when the points are sufficiently close to one another.
Notes :
1. The geodesic may be unique unless the points are sufficiently close to one another. On the
surface of the sphere there is a unique geodesic passing through any two points, except when
the two points are at the ends of a diameter. In the latter case, all great circles passing through
the two points are geodesics.
2. In Euclidean space VN, using orthogonal coordinates, all Christoffel symbols vanish. There-
fore, the differential equation of geodesic become
d 2 xi
0,
ds 2
whose solution is xi = Ais + Bi,
when Ai and Bi are constant vectors. These represent straight lines. Hence in Euclidean space
VN, the geodesic are straight lines.
14.5 Curvature of a curve
Let xi = xi (s) be the equation of a curve C in the space VN. The unit tangent vector to C is
defined as (dxi/ds) and it is denoted by tˆ with ti as its coutravariant component, thus
dx i
ti . .....(14.5.1)
ds
First Curvature :
The intrinsic derivative of ti along the curve C is called the first curvature vector or principal
normal of curve C relative to VN and is denoted by pi,
i
Thus, p i t,i j dx .....(14.5.2)
ds
2 i
i j k
or p i d x2 dx dx . .....(14.5.3)
ds jk ds ds
The magnitude of a first curvature vector pi is called first curvature of C relative to VN and is
denoted by . Therefore
2 = pi pi = gij pi pj. .....(14.5.4)
If i denotes the components of the unit principal normal, then
pi
i , pi = i. .....(14.5.5)
318
We have the differential equation of geodesic as
d 2 x i i dxi dx k
0
ds 2 jk ds ds
pi = 0
i = 0 = 0. .....(14.5.6)
Thus we conclude that a geodesic in Riemannian space VN is the curve whose first curvature
relative to VN is zero at all points. It gives the alternative definition of geodesic as :
“A geodesic in the Riemannian space VN is a curve whose first curvature relative VN vanishes at
all points.”
14.6 Null-geodesics
Along any portion of a curve which is not null, we have
j
dxi dx 1
gij e. e .....(14.6.1)
ds ds e
Differentiating with respect to s, we get
d dxi dx j dx i dx j
gij g ij
ds ds ds s ds ds
dxi dx j dx i
2 gij 2 gij 0 0.
ds s ds ds
de
= 0, .....(14.6.2)
ds
dx i
which shows that the indicator e cannot change along a geodesic. Therefore, the unit tangent vector
ds
which is not null at any point, cannot be null at any other point on the geodesic.
Contrary to it, if the initial direction is null, then the curve is null and we cannot introduce the arc
distance s (which is zero) as parameter.
Thus null geodesic is a null curve xi = xi (t) which is the solution of the equation
d 2 xi i dx i dx k
0. .....(14.6.3)
dt 2 jk dt dt
14.7 Illustrative examples
Ex.1. Assume that we live in a space for which the line element is
ds2 = (dx1)2 + [(x1)2 + c2] (dx2)2,
which is the surface of a right helocid immersed in a Euclidean three dimensional space. Deter-
mine the differential equation of geodesic.
319
Sol. Here we have g11 = 1, g12 = 0, g21 = 0, g22 = (x1)2 + c2.
Therefore the Christoffel symbols of second kind are
1 2 1 1
0, 0, 0, 0,
11 11 21 12
2 2 x1
1 log
12 21 x
g 22
1 2
,
x c 2
1 1 g 22 2
1
x1 , 0.
22 2 g11 x 22
Thus the differential equations of the geodesics on the surface are
d 2 x1 1 dx 2
x 0,
ds 2 ds
d 2 x2 2 x1 dx1 dx 2
0.
and ds 2 2
x1 c 2 ds ds
Ex.2. Obtain the differential equations of geodesics for the metric
1
ds 2 f x dx 2 dy 2 dz 2 dt 2 .
f x
Sol. Here x1= x, x2 = y, x3 = z, x4 = t.
1
g11 f ( x1 ), g 22 1, g33 1, g 44
f ( x1 )
gij = 0, i j. .....(1)
1
Therefore g11 1
, g 22 1, g 33 1, g 44 f ( x1 )
f (x )
gij = 0, i j. .....(2)
Thus, the non-zero symbols of second kind are
1 1 g 44 1 1 1 f
1
1
44 2 g11 x 2 f x f 2 f 3 x1
4 4 1 d
1 log g 44
14 41 x 2 dx1
log f
1 1 d
1 log
11 x
g11 2 dx1
log f . .....(3)
The differential equations of geodesics are
d 2 x i i dx j dx k
0. .....(4)
ds 2 jk ds ds
320
Hence, taking
d 2 x1 1 dx1 dx1 1 dx 4 dx 4
(i) i = 1, 0
ds 2 11 dx ds 44 ds ds
2 2
d 2x 1 d dx 1 d dt
2
log f 2 log f 0 .....(5)
ds 2 dx ds 2 f dx ds
d 2 x2 d2y
(ii) i = 2, 0, i.e., 0 .....(6)
ds 2 ds 2
d 2 x3 d 2z
(iii) i = 3, 0, i.e., 0 .....(7)
ds 2 ds 2
d 2 x4 4 dx 4 dx1
(iv) i = 4, 2 0
ds 2 41 ds ds
d 2t d dt dx
or 2
log f 0. .....(8)
ds dx ds ds
These are the required differential equation of the geodesics.
Ex3. Show that the curve given by
x1 C r cos cos dr
x2 C r cos sin dr
x3 C r sin dr
x4 C r dr
where r, are functions of t, is a real null curve in the V4 space whose metric is
ds2 = – (dx1)2 – (dx2)2 – (dx3)2 + C2 (dx4)2,
but not a null geodesic, in general.
Sol. For the given curve, we have
2 1 2 2 32 24 2
ds dx dx dx C 2 dx
dt dt dt dt dt
C 2 r 2 cos 2 cos 2 C 2 r 2 cos 2 sin 2 C 2 r 2 sin 2 C 2 r 2
= 0. .....(1)
t ds
Therefore, S 2 dt 0, .....(2)
t1 dt
along the given curve. Hence it is a null curve.
In a V4 space, whose metric is (1) the Christoffel symbols vanish and therefore the equations of
a geodesic are
321
d 2 xi
0. .....(3)
dt 2
These equation are not satisfied in general, by the given curve. Hence it is not a null geodesic in
general. However if we take and r as constants equations (3) will be satisfied and the null curve will
become a null geodesic.
Ex.4. Show that on the surface of a sphere, all great circles are geodesics while no other
circle is a geodesic.
Sol. The metric on the surface of a sphere of radius a is given by
(ds)2 = a2 (d)2 + a2 sin2 (d)2. .....(1)
Here g11 = a2, g22 = a2 sin2 , g12 = g21 = 0. .....(2)
The non-zero Christoffel symbols of second kind are
2 2 1
cot , sin cos . .....(3)
12 21 22
The geodesic equation is
d 2 xi i dx j dx k
0. .....(4)
ds 2 jk ds ds
Therefore the geodesic equations reduce to (x1 = , x2 = )
2
d 2 d
2
sin cos 0 i 1, j k 2 .....(5)
ds ds
d 2 d d
and 2
2cot 0 i 2, j 1, k 2 or i 2, j 2, k 1 .....(6)
ds ds ds
(i) We consider a great circle on the surface of the sphere and choose the normal to the plane of
the circle as the z-axis ( = 0), so that this great circle is the equator. Its parametric equation is
, C1s C2 , C1 0 .....(7)
2
where C1, C2 are independent of s, and .
Clearly equation (7) satisfies equations (5) and (6).
Therefore the great circle is geodesic, since the choice of the polar axis = 0 is arbitrary, it
follows that any great circle is a geodesic.
(ii) Consider, a circle on the sphere, whose plane does not pass through the centre of the sphere.
Taking the normal to the plane of the circle as = 0, the parametric equation of the circle is
0 , 0 , 0 0
2
= k1s + k2, k1 0, .....(8)
where k1, k2 are independent of s, and .
It may be noted that on substitution of equations (8) in (5) and (6), the equation (6) is satisfied
but equation (5) reduces to
322
k12 sin 0 cos 0 = 0,
which is not true under the given conditions on and . Hence any circle, not being a great circle, on the
surface of a sphere is not a geodesic.
14.8 Geodesic coordinates
In Euclidean space VN the components of the fundamental tensor gij are constants and therefore
all the Christoffel symbols are zero at every point of the Euclidean space. However it is not possible to
have such a coordinate system for an arbitrary VN. Although it is always possible to choose a coordi-
nates system, so that all Chritoffel symbols are zero at a particular point P0, i.e., in which gij are locally
constants, such a coordinates system is known as geodesic coordinate system with the pole at P0.
Thus we can define it as :
“A coordinates system is said to be a geodesic coordinate system with the pole at a point P0 if
relative to this coordinate system the components of the fundamental tensor gij are locally constants in
the neighbourhood of the point P0, i.e.,
gij
0
x k
at P0 for all values of i, j and k.”
It is to be noted that in this case the first covariant derivative at P0 reduces to the corresponding
partial derivative. Hence
Ai r i Ai
A,i j A at P0 .
at P0 x j rj x j
at P0
Theorem 2. It is always possible to choose a coordinate system so that all the Christoffel
symbols vanish at a particular point P0 (Geodesic coordinate system).
i
Proof. Let xi be any coordinate system and at a particular point P0 the value of xi is x(0) . Wee
introduce a new coordinate system x i defined by the equation
1 i
x i xi x(0)
i
2 mn (0)
x m x(0)
m
x n n
x(0) .....(14.8.1)
Here the index (0) is used for the values at P0. Differentiating (14.8.1) with respect to xj
x i x i 1 i m x
n
1 i x m n
x j
x j
2 mn (0)
m
x x(0)
x j
2 mn (0) x j
n
x x(0)
1 i 1 i
ij
2 mn (0)
x m x(0)
m
nj mj x n x(0)
2 mn (0)
n
i
ij
jn (0)
x n x(0)
n
. .....(14.8.2)
323
x i i .
Hence j j .....(14.8.3)
x (0)
x i
This shows that the Jacobian determinant j is not zero and therefore the transforma-
x (0)
tion (14.8.1) is permissible in the neighbourhood of P0.
x j
Taking inner multiplication of (14.8.2) with k , we get
x
x i i x j
ik
x k jn (0)
x n
x n
(0)
x k
, .....(14.8.4)
which implies that at P0
x i
ik k . .....(14.8.5)
x (0)
Differentiating (14.8.4) with respect to x h , we find
2 xi i x n x j i 2 xi
0
x h x k jn x h x k jn (0)
n n
x x(0)
xh xk
. .....(14.8.6)
Hence at P0
2 xi i x n x j
0 h k h k . .....(14.8.7)
x x (0) jn (0) x (0) x (0)
Using (14.8.5), we get
2 xi i n j i
0 h k h k , .....(14.8.8)
x x (0) jn (0) hk (0)
x i i x i i
therefore at P0 j j , k k .
x (0) x (0)
2 xi i
and h k . .....(14.8.9)
x x (0) hk (0)
We know that
p s x p x i x j x p 2 x j
s l m
,
lm ij x x x x j x l x m
p s x p xi x j x p 2 x j
therefore s l m j l m
lm (0) ij (0) x (0) x (0) x (0) x (0) x x (0)
324
s j
sp ij mj pj
ij (0) lm (0)
p p
0. .....(14.8.10)
lm (0) lm (0)
This prove that theorem.
Theorem 3. The necessary and sufficient conditions that a system of coordinates be geo-
desic with the pole P0 are that their second covariant derivatives, with respect to the metric of
the space, all vanish at P0.
Proof. We know that
2 xr p x r r x i x j
. .....(14.8.11)
x l x m lm x p ij x l x m
Interchanging the coordinate system xi and x i the equation (14.8.11) can be written as
2 xr p x r r x i x j
x l x m lm x p ij xl x m
r x i x j xr p xr
l m
m l p
ij x x x x lm x
p
m x,rl x,rp .
lm
x
r x i x j
Hence l
ij x x
r
m x,l ,m
x,rlm . .....(14.8.12)
Necessary condition : If the coordinate system x i be a geodesic coordinate system with pole
at P0, then
r
0 at P0 ,
ij
and therefore from (14.8.12) it follows that x,rlm 0 at P0 .
Sufficient condition : Courversely suppose that
x,rlm 0 at P0 .
r x i x j
then equation (14.8.12) implies l m
0 at P0 .
ij x x
r x i x j
Thus, 0 at P0 , as 0, 0.
ij x l x m
Hence the proposition.
325
14.9 Illustrative examples
Ex.5. Show that the coordinate system x i defined by
1 i
x i xi x m x n ,
2 mn (0)
is a geodesic coordinate system with the pole at the origin.
Sol. Since the pole is at the origin (xm)(0) = 0, then x i becomes a particular case of theorem 2.
We can prove it independently as follows :
x i i 1 i m n 1 i m n
j
j j x x j
x 2 mn (0) 2 mn (0)
i
ij x n . .....(1)
jn (0)
Also, we have
2 x i i i
j k
nk . .....(2)
x x jn (0) jk (0)
x i i 2 x i i
Hence j j and j k . .....(3)
x (0) x x (0) jk (0)
x i x i i
Now x,ijk j r . .....(4)
x k x x jk
2 x i xi r
at the pole
x,ijk
(0)
j k
x x
r
(0) x
(0) jk (0)
i r
ir 0. .....(5)
jk (0) jk (0)
Hence proved the result.
Ex.6. Show that at the pole P0 of a geodesic coordinate system
2 Ai l
Ai, jk j k
Al .
x x x k ij
Ai r
Sol. Since we have Ai, j j
Ar . .....(1)
x ij
Taking covariant derivative of (1), we get
r r r
Ai, j ,k x k xAij Ar ij Ai,r jk Ar, j ik
326
2 Ai r Ar r r r
Ai, jk Ar Ai ,r Ar , j . .....(2)
x j x k x k ij x k ij jk ik
Now, we know that in geodesic coordinate system at the pole P0 the Christoffel symbols vanish
and therefore (2) reduces to
2 Ai r
Ai , jk j k
Ar k . .....(3)
x x x ij
14.9 Riemannian coordinates
The Riemannian coordinate system is a particular case of geodesic coordinate system. Hence
every Riemannian coordinate system is necessarily a geodesic coordinate system but converse is not
always true. To define Riemannian coordinate system we consider an arbitrary fixed point P0 in VN . We
define the quantity i such that
i dx i
, ......(14.9.1)
ds (0)
where suffix (0) indicates the value related to P0. Let c be the geodesic through P0 in VN. Since one and
one geodesic c will pass through P0 in the direction of i, such that
yi = si, .....(14.9.2)
defines the Riemannian coordinate system. Here P(yi) is a point on the geodesic c and s is the arc length
along the curve form P0 to P.
Theorem 4. The Riemannian coordinates are geodesic coordinates with the pole at P0.
Proof. The differential equation of geodesic c in terms of Riemannian coordinates yi relative to
VN are given by
d 2 y i i dy j dy k
0, .....(14.9.3)
ds 2 jk ds ds
i
where is the Christoffel symbol relative to the coordinates yi. Since the P(yi), defined by (14.9.2)
jk
is on the geodesic c given by (14.9.1). This must satisfy it, therefore
i
0 j k 0, .....(14.9.4)
jk
Q y i si , then (14.9.4) becomes
i i k
y y 0. .....(14.9.5)
jk
Equation (14.9.5) holds throughout the space VN. It also implies that
i
0 at P0 . .....(14.9.6)
jk
327
Since at P0, yi 0, yk 0.
Hence the Riemannian coordinates are geodesic coordinates with the pole at P0.
14.10 Field of parallel vectors (Parallelism of vectors)
The vectors Ai constitute a field of parallel vectors along the curve xi = xi (t) in a VN , if Ai is a
solution of the differential equation
Ai dxi dAi l dxi
Ai , j Al 0. .....(14.10.1)
t dt dt ij dt
The concept of parallelism is given by Levi-Civita. These are N-differential equations of first
order, therefore the general solution will involve N-constants. According the theory of differential equa-
tion, if the initial values of Ai are given at a point of the curve these constants will uniquely be deter-
mined, i.e., Ai will be determined uniquely at all other points if it is given at one point of the curve. Thus
we can say that a field of parallel vectors is obtained from a given vector by parallel propagation (dis-
placement) along the curve.
The condition (14.10.1) is in covariant form, we can write it in the contravariant form as
Ai ij A j
t
t
g A j g ij
t
0
Ai dx j dAi i k dx j
A,i j A 0.
t dt dt jk dt
Theorem 5. The magnitude of all vectors of a field of parallel vectors is constant.
Proof. If A be the magnitude of vector Ai, then
(A)2 = e(A) gij Ai Aj.
Differentiating with respect to parameter t, we find
dA d
2A
dt
dt
e( A) gij Ai A j
t
e( A) gij Ai A j
Ai
2e( A) gij A j ,
t
as the total derivative becomes the intrinsic derivative in the case of scalars (e(A) gij Ai Aj is a scalar).
Using (14.10.2), we find that
dA
2A =0
dt
dA 2
dt
A =0
A2 = constant
A = constant.
Hence the theorem.
328
Theorem 6. Prove that the geodesic is an auto parallel curve.
Or
Prove that the unit tangent vectors form a field of parallel vectors along a geodesic.
Proof. The differential equation of a geodesic are
d 2 xi i dx j dx k
0
ds 2 jk ds ds
d dx i i dx j dx k
0
ds ds jk ds ds
dxi dx j i dx j dx k
0
x j ds ds jk ds ds
dxi i dx k dx j
j 0
x ds jk ds ds
dxi dx j
0
ds , j ds
dx j
t,i j 0,
ds
dx i
which shows that the unit tangent vectors t i form a field of parallel vectors along a geodesic.
ds
Hence proved.
14.11 Parallelism of vector of variable magnitude along a curve
We know that Ai or Aj constitute a field of parallel vectors along the curve xi = xi (t) if their
intrinsic derivative with respect to t is zero. Further from Theorem 5, we find that the magnitude of all
vectors of a field of parallel vectors is constant.
We shall now define the parallelism of two vectors whose magnitude need not to be constant.
‘Two vectors at a point are said to be parallel, if their corresponding components are propor-
tional.’
Clearly, a vector Bi is called parallel to a vector Ai at each point of a curve c if
Bi = Ai,
where is an arbitrary scalar function of arc length s.
Theorem 7. The necessary and sufficient condition for a vector B i of variable magnitude
to suffer a parallel displacement along a curve c is that
dx j
B,i j Bi f s .
ds
329
Proof. Necessary condition : First we suppose that Ai constitute a field of parallel vectors
along the curve c, then the magnitude of Ai is constant and
dx j
A,i j 0. .....(14.11.1)
ds
Now, we know that if Bi is parallel to Ai at each point of c, then
Bi = Ai. .....(14.11.2)
dx j dx j
Therefore, B,i j
ds
Ai ,j ds
dx j i dx j
A,i j A
ds x j ds
d Bi
0 [using (14.11.1) & (14.11.2)]
ds
d
Bi log
ds
dx j
B,i j Bi f s , .....(14.11.3)
ds
d
where log s = f (x). .....(14.11.4)
ds
This shows that equation (14.11.3) is necessary condition for the vector Bi of variable magni-
tude to suffer a parallel displacement along c.
Sufficiently condition : Conversely suppose that Bi is a vector of variable magnitude, such that
dx j
B,i j Bi f s . .....(14.11.5)
ds
Taking A i = Bi F (s), .....(14.11.6)
dx j dx j
we have A,i j
ds
FB i ,j ds
dx j F i dx i
FB,ij B
ds x j ds
dF
FB i f s B i
ds
dF
B i Ff s . .....(14.11.7)
ds
Choosing F such that
dF
Ff s 0, .....(14.11.8)
ds
330
because of arbitrary nature of F (s), we find
dxi
A,i j 0,
ds
which shows that Ai form a field of parallel vectors along c and is of constant magnitude, then (14.11.6)
implies that Bi is parallel along c.
14.12 Illustrative examples
Ex.7. Show that the vector Bi of variable magnitude suffers a parallel displacement along
a curve c if and only if
dx j
B l B,i j B i B,l j ds
0.
Sol. We know from Theorem 7, that Bi suffers a parallel displacement along c if and only if
dxi
B,i j Bi f s .
ds
Taking outer multiplication by Bi, we get
dx j
B l B,i j Bl Bi f s . .....(1)
ds
Interchanging the suffixes i and l, we find
dx j
B i B,l j Bi B l f s . .....(2)
ds
Subtracting (2) from (1), the required result is obtained as
dx j
B l B,i j B i B,l j ds
0. .....(3)
Theorem 8. (Fundamental theorem of Riemannian geometry).
With a given fundamental tensor of a Riemannian manifold (Riemannian space VN), there
exists exactly one symmetric connection with respect to which the parallel displacement preserves
scalar product.
Proof. Let Ai and Bi be two unit vectors defined along a curve c in space VN and these vectors
suffer parallel displacement along c, then
dx j
A,i j 0, .....(1)
ds
dx j
B,i j 0, .....(2)
ds
The scalar product of Ai and Bi is gij Ai Bj, where gij is the given fundamental tensor.
The parallel displacement preserves scalar product if the intrinsic derivative of gij Ai Bj is zero.
331
dx k dx k i j i dx k j i j dx
k
We have g i
ij A B j
,k
gij ,k
ds
A B g ij A,k
ds
B g ij A B,k
ds
ds
Using (1), (2) and Ricci theorem i.e., gij,k = 0.
So the R.H.S. is zero and this proves the theorem.
14.13 Self-learning exercises
1. Define Geodesic.
2. Write Euler’s condition of calculus of variation.
3. Define first curvature ?
4. What is null geodesic ?
5. What is field of parallel vectors ?
6. Write fundamental theorem of Riemannian geometry.
14.14 Summary
Geodesic in a surface is the curve of stationary length on a surface between any two points in its
plane. It is the main point of the study in this chapter. We obtained a differential equation whose solution
will give the geodesic curve. The geodesic coordinates and Riemannian coordinates have also been studied.
14.15 Answers to self-learning exercises
1. § 14.2 2. § 14.3 3. § 14.5
4. § 14.6 5. § 14.10 6. Theorem 8.
14.16 Exercises
1. Show that it is always possible to choose a geodesic coordinate system for any VN with an arbi-
trary point P0.
2. Obtain the equations of geodesic for the metric
ds 2 e2 kt dx 2 dy 2 dz 2 dt .
2
3. Show that the great circles on sphere are geodesic.
4. Obtain the differential equation of geodesic for the metric
1
ds 2 f x dx 2 dy 2 dz 2 dt 2 .
f x
5. Give an example of a geodesic coordinate system.
332
Unit 15 : Riemannian-Christoffel Tensor and its Properties,
Covariant Curvature Tensor, Einstein Space,
Bianchi’s Identity, Einstein Tensor, Flat Space,
Isotropic Point, Schur’s Theorem
Structure of the Unit
15.0 Objective
15.1 Introduction
15.2 Properties of Riemann-Christoffel tensor
15.3 Covariant curvature tensor
15.4 Properties of covariant curvature tensor
15.5 Illustrative examples
15.6 Contraction of Riemann-Christoffel tensor-Ricci tensor
15.7 Curvature invariant-Einstein space
15.8 Einstein tensor
15.9 Riemannian curvature of a VN at a point
15.10 Illustrative examples
15.11 Flat space
15.12 Isotropic point
15.13 Illustrative examples
15.14 Self-learning exercises
15.15 Summary
15.16 Answers to self-learning exercises
15.17 Exercises
15.0 Objectives
The objective of this unit is to study the commutativity of the process of covariant differentiation
of vectors and hence we define Riemann’s symbols of first and second kinds. The contraction of
Riemann-Christoffel tensor and Ricci tensor are the points of study. In the end Einstein space, Einstein
tensor and flat space are the points of study.
333
15.1 Introduction
We have studies that the covariant differentiation of invariant is commutative. Now we shall in-
vestigate this property of commutative nature for covariant differentiation of vectors. For the study let us
take the covariant derivative of an arbitrary covariant vector Ai
Ai
Ai, j j
A . .....(15.1.1)
x ij
Differentiating again covariantly with respect to xk, we get
Ai, j ,k x k xAij A ij A, j ik Ai, jk .
2 Ai A
Ai, jk k j
k A k
x x x ij x ij
A A
j A i A
x j ik x i jk
Now interchanging the suffices j and k, we find that
2 Ai A
Ai, jk j k j A j
x x x ik x ik
A A
k A i A
x k ij x i kj
Subtracting (15.1.3) from (15.1.2) and interchanging we get
Ai, jk Ai,kj x j ik x k ij j ik k j A .....(15.1.4)
Since left hand side of (15.1.4) is a covariant tensor of third order and A be the arbitrary
covariant vector in the right hand side, therefore from quotient law it follows that the coefficient of A in
R.H.S. must be a mixed tensor of fourth order, contavariant of rank one and covariant of rank three. Let
us denote this quantity by R.ijk i.e.,
R.ijk . .....(15.1.5)
x j ik x k ij j ik k ij
The tensor R.ijk defined by (15.1.5) is known as Riemann-Christoffel tensor or mixed cur--
vature tensor and thus, we have
Ai, jk Ai ,kj R.ijk A . .....(15.1.6)
The symbol R.ijk is also called Riemann’s symbol of second kind.*
* Riemann’s symbol of the first kind is introduced later on.
334
Now it is clear that the necessary and sufficient condition for covariant differentiation of a vector
Ai to be commutative is that Riemann-Christoffel tensor be identically zero or R.ijk = 0.
Note : R.ijk is formed exclusively from the fundamental tensor gij and its derivatives upto sec-
ond order. It does not depend on the choice of Ai.
15.2 Properties of Riemann-Christoffel tensor
Property–I : The Riemann-Christoffel tensor is skew-symmetric in the last two subscripts, i.e.,
R.ijk R.ikj . .....(15.2.1)
Proof. The result immediately follows from (15.1.5).
Property–II : R.ijk has cyclic property in its subscripts, i.e.,
R.ijk R.jki R.kij 0. .....(15.2.2)
Proof. By definition (15.1.5), we have
R.ijk k .....(15.2.3)
x j ik x ij ik j ij k
R.jki i .....(15.2.4)
x k ji x jk ji k jk i
R.kij .....(15.2.5)
xi kj x j ki kj i ki j
Adding these, we get
R.ijk R.jki R.kij 0. .....(15.2.6)
Hence the result follows.
Property–III : R.ijk vanishes on contraction in and i, i.e.,
R.jk = 0. .....(15.2.7)
Proof. In equation (15.2.5) of §15.2 contracting over and i, we get
R.jk k .....(15.2.8)
x j k x j k j j k
The last two terms cancel out as and are dummy indices, therefore using Property–4 of
Christoffel symbols, we get
R.jk j k
log g k j log g 0. .....(15.2.9)
x x x x
15.3 Covariant curvature tensor
The covariant curvature tensor is defined as
Rrijk g r R.ijk . .....(15.3.1)
335
It is fourth order covariant tensor and is also called as Riemann tensor. The symbol Rrijk is
also called Riemann’s symbol of first kind. It is an associate tensor of Riemann-Christoffel tensor.
The properties of Rrijk can easily be studies if we express it in a more suitable form. Substituting
the value of R.ijk in the above definition, we get
Rrijk g r j k
x ik x ij ik j ij k
g r
g r j
k g r
x j ik x ik x ij
g r .....(15.3.2)
k
g r g r
x ij ik
j ij
k
Using properties (15.3.1) and (15.3.2) of Christoffel symbols, the above expression reduces to
Rrijk j
ik , r rj , j, r k ij , r rk , k , r
x ik x ij
j, r k, r
ik ij
ik , r ij , r rk , r rj , r .....(15.3.3)
x j x k ij
ik
the remaining terms cancel out by suitable changes of dummy indices. It can further be simplified as
1 gir g kr gik 1 gir gir gij
Rrijk k i r 2 k j i r
2 x j x x x x x x x
g ij , rk , g ik , rj, . .....(15.3.4)
Finally, we have
1 2 g rk 2 gij 2 gik 2 g rj
Rrijk i j r k r j i k
2 x x x x x x x x
g rk , ij , rj , ik , , .....(15.3.5)
which is an important formula from the point of view of studying the properties.
15.4 Properties of covariant curvature tensor
Property–I : Rrijk is skew-symmetric in the pair of first two indices, i.e.,
Rrijk = – Rirjk (skew-symmetric property) .....(15.4.1)
The above result can easily be proved by interchanging r and i in (15.3.2) of §15.3.
Property–II : Rrijk is skew-symmetric in the pair of last two indices, i.e.,
Rrijk = – Rrikj (skew-symmetric property) .....(15.4.2)
It can also be proved by interchanging j and k in (15.3.2) of §15.3.
336
Property–III : Rrijk is symmetric in two pairs (first and last) of indices, i.e.,
Rrijk = Rjkri (symmetric property) .....(15.4.3)
If may be easily seen from (15.3.2) of §15.3 by interchanging r and j, then i and k.
Property–IV : Rrijk has cyclic property in last three indices, i.e.,
Rrijk + Rrjki + Rrkij = 0 (cyclic property) .....(15.4.4)
Proof : We have Rrijk g r R.ijk . .....(15.4.5)
Giving cyclic rotation to i, j, k and adding, we get
Rrijk + Rrjki + Rrkij g r R.ijk g r R.jki g r R.kij
g r R.ijk R.jki R.kij
= 0. [using property–II of §15.2] .....(15.4.6)
Property–V : (Bianchi’s identity).
The differential property satisfied by covariant derivative states
Rrijk,p + Rrikp,j + Rripj,i = 0
or equivalently R.ijk , p R.ikp , j R.ipj ,k 0. .....(15.4.7)
Proof : The identity is proved conveniently, by choosing geodesic coordinate system with the
pole at P0, so that all the Christoffel symbols vanish at P0. We recollect that by choosing so, the first
covariant derivative of any tensor at P0 reduces to the corresponding partial derivative. For example
Ai
A,i j . .....(15.4.8)
at P0 x j
at P0
Now R.ijk k ,
x j ik x ij ik j ij k
at P0 becomes R.ijk
x j ik x k ij
2 2
and R.ijk
,p
p j
x x
p k
ik x x
.
ij
.....(15.4.9)
Similarly, after cyclic rotation to j, k, p, we get
2 2
R.ikp
,j
j k
x x
j p
ip x x
ik
.....(15.4.10)
2 2
and R.ipj
,k
x k x p
k
ij x x
j .
ip
.....(15.4.11)
Adding these three expressions (15.4.9) to (15.4.11), we get
R R R
.ijk
,p
. jkp
,j
.kpj
,i
0. .....(15.4.12)
337
Taking inner multiplication by gr and remembering that the fundamental tensor behave like a
constant is covariant differentiation, we obtain
g r R.ijk g
,p
r R.jkp g
,j
r R.ipj ,k
0.
or Rrijk,p + Rrikp,j + Rripj,k = 0. .....(15.4.13)
This is Bianchy identity.
15.5 Illustrative examples
2
Ex.1. Prove that R1212 G G for the V2 whose line element is ds2 = du2 + G2 dv2,
u 2
where G is a function of u and v.
Or
Show that the component R1212 of the curvature tensor for a V2 with metric
2
1 2 f 1 f
ds2 = dx2 + f (x, y) dy2 equals .
2
2 x 4 f x
Sol. For the metric
ds2 = du2 + G2 dv2,
we have x1 = u, x2 = v, g11 = 1, g22 = G2, g12 = 0, g21 = 0. .....(1)
1 2 g rk 2 gij 2 g rj 2 g
Since Rrijk i j r k i k r ik j
2 x x x x x x x x
g rk , ij , rj, ik ,
the value of R1212 is given by
1 2 g 2 g
R1212 0 0 2 112 1 121 g 12, 21, 11, 22,
2 x x x x
1 2
2 u 2
G 2 g11 12,1 21,1 11,1 22,1
g 22 12, 2 21, 2 11, 2 22, 2
1 G 11 22 1 g 22 1 g 22
2G g 0 g 1 1
2 u u 2 x 2 x
2
1 G 1
G 2
2 u u 4G u
G2 .
.....(2)
2
R1212 G G . .....(3)
u 2
Taking G f and u = x the alternative form may easily be obtained.
338
15.6 Contraction of Riemann-Christoffel tensor-Ricci tensor
Theorem 1. The Riemann-Christoffel tensor R.ijk (or mixed curvature tensor) can be con-
tracted in two different ways-one of these leads to a zero tensor and other to a symmetric tensor
Rij, known as Ricci-tensor.
Proof. We have by definition
R.ijk k .....(15.6.1)
x j ik x ij ik j ij k
(i) Contracting over i and , i.e., setting i = , we get
R.jk k .....(15.6.2)
x j k x j k j j k
Interchanging the dummy indices and , the last two terms cancel out. Further, using property
of Christoffel symbols
R.jk log g
j log g
x j x k x
k
x
= 0. .....(15.6.3)
(ii) Contracting over k and , i.e., setting k = , we get
R.ij .....(15.6.4)
x j i x ij i j ij
Writing, (Ricci tensor)
Rij = R.ij g r Rrij . .....(15.6.5)
The above expression becomes
Rij j
x x i
log g
x
log g
ij i j ij x
.....(15.6.6)
It may easily be observed that, by interchanging i and j in (15.6.6), that
Rij = Rji (symmetric property) .....(15.6.7)
Thus Rij is a symmetric tensor and is called Ricci tensor. It is defined by (15.6.5) and (15.6.6).
Notes :
1. The contraction over and j (j = ) does not yield any new tensor, because
R.ik Rik Rik
[using skew-symmetric property of Riemann-Christoffel tensor.]
where is Ricci tensor with negative sign.
2. If g is negative, replace log g by log g in (15.6.6).
339
15.7 Curvature invariant-Einstein space
The curvature invariant R is defined as
R g ij Rij Rii . .....(15.7.1)
A space for which at every point of it
Rij = Igij, .....(15.7.2)
where I is an invariant and it is called on Einstein space. The inner multiplication of (15.7.2) by gij and
using (15.7.1), we get
R = NI. .....(15.7.3)
Hence for an Einstein space
R
Rij gij . .....(15.7.4)
N
15.8 Einstein tensor
It is defined as
il 1 i
G ij g R jl R j , .....(15.8.1)
2
and it has a considerable importance in the theory of relativity.
Theorem 2. The divergence of Einstein tensor vanishes, i.e.,
G ij ,i 0.
Proof. We have from Bianchy identity
Rrijk , p Rrikp , j Rripj ,k 0. ......(15.8.2)
Taking inner multiplication of the above relation by gij grk, we get
g ij
g rk Rrijk g
,p
ij
g rk Rrikp g
,j
ij
g rk Rripj ,k
0. .....(15.8.3)
Using the definition of Ricci tensor and the skew-symmetric property of covariant curvature ten-
sor, we find
g R g R g R
ij
ij
,p
ij
ip
,j
rk
rp
,k
0.
or R R r 0
,p
j
p
,j
k
p
,k
or R R R
,p
j
p
,j
j
p
,j
or R , p 2R pj , j
1 1 R
or R pj , j R, p , .....(15.8.4)
2 2 x p
where R pj g ij Rip is the associate tensor of Rip.
1
Now G ij R ij Rij . .....(15.8.5)
2
340
Taking its divergence and using (15.8.4), we get
G
div G ij i
j ,i
1
R ij ,j
ij R,i
2
1 1
R, j R, j 0. .....(15.8.6)
2 2
Hence Proved.
Theorem 3. An Einstein space VN (N > 2) has constant curvature (Curvature inva-
riant R).
Proof. Taking inner multiplication of Bianchy identity by gij grk, we get
1 R
R pj , j , .....(15.8.7)
2 x p
where R pj g ij Rip .
In an Einstein space VN, we have
R
Rip gip . .....(15.8.8)
N
Taking inner multiplication of it by gij, we get
R ij
g ij Rip g gip
N
R j
or R pj
p. .....(15.8.9)
N
Now taking covariant derivative with respect to xj, we get
1 j 1
R ip , j p R, j R, p . .....(15.8.10)
N N
Hence from (15.8.7) & (15.8.8), we conclude that
1 1 R R
p 0, i.e., 0, since N 2 .
2 N x x p
or we can say that R is constant.
Thus Einstein space VN (N > 2) has constant curvature.
15.9 Riemannian curvature of a VN at a point
If Ai and Bi be any two contravariant vectors at a point of a VN, then
Rrijk Ar A j B i B k
,
grj gik g rk gij Ar A j Bi B k
is called the Riemannian curvature of the space VN associated with the vectors Ai and Bi.
It is an invariant, which is unaltered at a point, when the two vectors determining it are replaced
by any linear combination of them.
341
15.10 Illustrative examples
Ex.2. The metric of V2 formed by the surface of a sphere of radius a is
ds2 = a2 d2 + a2 sin2 d2,
1
in spherical polar coordinates. Show that the curvature of the surface of the sphere is , which
a2
is constant.
Sol. For the metric ds2 = a2 d2 + a2 sin2 d2,
we have x1 = , x2 = , g11 = a2, g22 = a2 sin2 , g12 = 0, g21 = 0. .....(1)
g11 g12
Therefore, g a 4 sin 2 . .....(2)
g 21 g 22
2
1 2 g 22 22 1 g 22
Now R1212 g
2 x1 x1 2 x1
2
1 2 2 2 1 2 2
2 2
a sin 2 2
4a sin
a sin
R1212 = a2 sin2 . .....(3)
R1212 a 2 sin 2 1
4 2 2,
g a sin a
which is curvature of the surface of the sphere.
Ex.3. Show that the Riemannian of a V2 is uniquely determined at each point, and its value
is given by
R1212
g
Sol. In a two-dimensional space, at any point of it, there exists only two independent vectors.
Therefore the Riemannian curvature of a V2 is uniquely determined at each point. In a V2 the number of
independent components of Rrijk is 1. The value of can easily be found by choosing the two vectors
whose components are (1, 0) and (0, 1) respectively. Then
R1212 A1B 2 A1B 2 R
1 2 1 2
1212 ,
g11g 22 g12 g 21 A B A B g
as A1 = 1, B2 = 1.
15.11 Flat space
A space for which the Riemannian curvature is identically zero at every point of it ( = 0), is
called a flat space.
342
Theorem 4. The necessary and sufficient condition for a space VN to be flat is that the
Riemann-Christoffel tensor be identically zero, i.e., R.ijk 0.
Proof : Necessary condition :
Let the space VN be flat, them = 0 at every point of VN , i.e.,
Rrijk Ar Bj Bi Bk = 0,
for all vectors Ai and Bi. From it we cannot jump to the conclusion that Rrijk = 0. It must be remem-
bered that in the form Rrijk Ar Bj Bi Bk the coefficient of the product Ar Aj Bi Bk is mixed up with the
coefficients of Aj Ar Bi Bk, Ar Aj Bk Bi and Aj Ar Bk Bi, it is fact Rrijk + Rjirk + Rrkji + Rjkri which
may be obtained by interchanging the dummy indices as follows :
Rrijk Ar Aj Bi Bk = 0 .....(15.11.1)
Also Rjirk Aj Ar Bi Bk = 0 .....(15.11.2)
Rrkji Ar Aj Bk Bi = 0 .....(15.11.3)
Rjkri Aj Ar Bk Bi = 0 .....(15.11.4)
On addition, (Rrijk + Rjirk + Rrkji + Rjkri) Ar Aj Bi Bk = 0. .....(15.11.5)
This implies that for arbitrary Ai, Bi
Rrijk + Rjirk + Rrkji + Rjkri = 0.
Rrijk + Rrkji + Rrkji + Rrijk = 0 [using symmetric property]
2(Rrijk + Rrkji) = 0,
Rrijk = Rrkji. [using skew-symmetric property] .....(15.11.6)
Interchanging i, j and k cyclically in (15.11.6), we get
Rrjki = Rrijk. .....(15.11.7)
From (15.11.6) and (15.11.7), we have
Rrijk = Rrjki = Rrkij. .....(15.11.8)
Now substituting (15.11.8) in the cyclic property–IV of §15.4
Rrijk + Rrjki + Rrkij = 0. .....(15.11.9)
we find 3Rrijk = 0
Rrijk = 0. .....(15.11.10)
Since, Rrijk g r R.ijk ,
and gris an arbitrary, it implies that
R.ijk = 0. .....(15.11.11)
Sufficeint condition : Conversely, if R.ijk = 0. i.e., Rrijk = 0, then it in clear = 0.
Hence the theorem.
dx j
Note : Taking inner multiplication of Ai,j = 0 with , we get
dt
dx j
Ai, j 0
dt
343
Ai
0.
t
This shows that in a flat space the property of parallelism is independent of the choice of a curve.
Thus parallelism is an absolute property of a flat space.
15.12 Isotropic point
An isotropic point in Riemannian space is a point at which the Riemannian curvature is indepen-
dent of the vectors Ai and Bi associated to it.
It implies that [Rrijk – (grj gik – gij grk)] Ar Aj Bi Bk = 0, .....(15.12.1)
for all vectors Ai and Bi at the isotropic point.
Now we define a tensor Trijk by the equation
Trijk = Rrijk – (grj gik – gij grk). .....(15.2.2)
then (15.2.1) reduces to Trijk Ar Aj Bi Bk = 0. .....(15.12.3)
Proceeding parallel to Theorem 4, we conclude from (15.12.3) that will be true for any vectors
Ai and Bi if
Trijk + Tjirk + Trkji + Tjkri = 0. ......(15.12.4)
According to (15.12.2) we see that the tensor Trijk satisfies the same four properties as by Rrijk
viz.
Trijk = – Tirjk ; Trijk = – Trikj ; Trijk = Tjkri
and Trijk + Trjki + Trkij = 0. .....(15.12.5)
Hence repeating the same steps as in (15.12.3), replacing Rrijk by Trijk, we finally get
Trijk = 0. .....(15.12.6)
Thus from (15.12.2) to (15.12.6), it follows that at an isotropic point the Riemannian curvature
satisfies the condition
Rrijk = (grj gik – gij grk). .....(15.12.7)
Now we state an important theorem due to Schur.
Theorem 5. (Schur’s theorem).
If a Riemannian space VN (N > 2) is isotropic at each point in a region, then the Rieman-
nian curvature is constant throughout that region.
(Such a space VN is called a space of constant curvature).
Proof. We know that if the Riemannian space VN is isotropic at each point in a region, then
Rrijk = (grj gik – gij grk), .....(1)
where is the function of coordinates xi.
Taking covariant differentiation of (1) with respect to xi, we get
Rrijk,t = (grj gik – gij grk) ,t. .....(2)
since the covariant derivative of the metric tensor vanishes.
344
Permuting the subscripts j, k, t cyclically, we find
Rrikt,j = (grk git – gik grt) ,j .....(3)
and Rritj,k = (grt gij – git grj) ,k. .....(4)
Adding equations (2), (3) and (4) and using Bianchi identity, we get
(grj gik – gij grk) ,t + (grk git – gik grt) ,j + (grt gij – git grj) ,k= 0. .....(5)
Taking inner multiplication by grj gik, we find
N 2
kj kj ,t kj tk N it , j rj kj N tk ,k 0
N 2
N ,t 1 N tj , j 1 N tk ,k 0
(1 – N) [,t + ,t – N,t] = 0
(1 – N) (2 – N) ,t = 0. .....(6)
But N > 2, therefore ,t = 0, but this is simply 0. Hence it follows that is a constant.
x t
Such a VN is called space of constant curvature. Hence the theorem is proved.
15.13 Illustrative examples
Ex.4. If the metric of a two dimensional flat space is f (r) [(dx1)2 + (dx2)2],
where (r)2 = (x1)2 + (x2)2, show that f (r) = c (r)k, where c and k are constants.
Sol. We have the metric
ds2 = f (r) {(dx1)2 + (dx2)2},
therefore g11 = f (r), g22 = f (r), g12 = 0, g21 = 0. .....(1)
The only non-zero component of Rrijk in a V2 is R1212 which in the present case is given by
1 2 f 2 f 1 f 2 f 2
R1212 1 1 2 2 1 2 .....(2)
2 x x x x 2 f x x
R1212
In a V2, .
g
For a flat space = 0, i.e.
R1212 = 0.
2 2
2 f 2 f 1 f f
Hence 1 2 . .....(3)
x1 x1 x 2 x 2 f x x
But f = f (r) (r)2 = (x1)2 + (x2)2. .....(4)
Therefore, changing (3) to polar coordinates, we get
2
2 f 1 f 1 f
2
. .....(5)
r r r f r
345
Let, F = log f (r). .....(6)
Then (5) reduces to
d 2 F 1 dF
0
dr 2 r dr
d dF dF
r 0, i.e., r (constant)
dr dr dr
dF r
or , i.e., F log . .....(7)
dr r A
From (6) and (7), we conclude that f (r) = c (r)k, where c and k are constants.
Ex.5. Prove that, in space VN of constant curvature ,
(i) Rij = – (N – 1) gij, and
(ii) R = – N (N – 1) .
Deduce that a space of constant curvature is an Einstein space.
Sol. In the space of constant curvature
(i) Rrijk = (grj gik – gij grk). .....(1)
Taking inner product with grk, we get
k
grk Rrijk j gik gij N , .....(2)
Using the definition of Ricci tensor, we have
Rij = (1 – N) gij. .....(3)
(ii) Again taking the inner product of (1) by gij, we get
gij Rij = (1 – N) gij gij
R = (1 – N) N. .....(4)
From (3) and (4), it follows that
R
Rij gij . .....(5)
N
This shows that a space VN (N > 2) of constant curvature is an Einstein space.
Ex.6. In a V2, prove that
R (gij grj – gij grk) = – 2 Rrijk
and hence that Rg = – 2 R1212.
In this case, prove also that the components of Ricci tensor are proportional to the com-
ponents of metric tensor that is
g Rik = – R1212 gik.
Sol. Since a V2 is isotropic, the equation
Rrijk = (grj gik – gij grk), .....(1)
holds throughout any V2.
346
Using the method of example 5, we get
Rij = – gij, N=2 .....(2)
and R = – 2. .....(3)
From (1) and (3), – 2Rrijk = R (grj gik – gij grk)
– 2R1212 = R (g11 g22 – g122)
– 2R1212 = R.g. .....(4)
Now from equation (2), (3) and (4),
R
gRij ggij R1212 gij . .....(5)
2
This gives the required result.
15.14 Self-learning exercises
1. Define Riemann-Christoffel tensor.
2. Write the necessary and sufficeint conditions for covariant differentiation of a vector Ai to
be commutative.
3. Define covariant curvature tensor.
4. What is Ricci tensor ?
5. What is Bianchy identity ?
6. Define Einstein tensor.
7. What is the divergence of Einstein tensor ?
8. What do you mean by flat space ?
9. Write the statement of Schur’s theorem.
15.14 Summary
In this unit we have studied the commutativity of covariant differentiation of vectors and defined
Riemann-Christoffel tensor. On the basis of this we have defined covariant curvature tensor. The prop-
erties of covariant curvature tensor are also given. The contraction in Riemann-Christoffel tensor gives
Ricci tensor. Then the Einstein space has been defined. Bianchy identity and Einstein tensor have also
been studied. The divergence of Einstein tensor vanishes. A space for which Riemann curvature is identi-
cally zero at every point of it, is called flat space.
15.15 Answers to self-learning exercises
1. § 15.1 2. § 15.1 3. § 15.3
4. § 15.6 5. § 15.4 6. § 15.8
7. § 15.8 8. § 15.11 9. § 15.12
347
15.16 Exercises
1. Prove that
p
Bl ,mn Bl ,nm Rlmn Bp ,
p
where Bp is an arbitrary covariant tensor of rank 1 and deduce that Rlmn is a tensor..
2. Define Riemann’s symbols of first and second kind. If Bi are components of a vector, prove that
Bi, jk Bi ,kj B Rijk .
3. Show that the space of constant curvature is Einstein space.
4. For a V2 space, prove that
gRij = – gij R1212 and gR = – 2R1212.
Hence deduce that every V2 is an Einstein space.
5. Show that the number of independent components of the covariant curvature tensor in a space
of N-dimension is
1 2 2
12
N N 1 .
348
Reference Books
1. Differential Geometry
J.L. Bansal and P.R. Sharma
Jaipur Publishing House, Jaipur
2. Tensor Analysis
J.L. Bansal
Jaipur Publishing House, Jaipur
3. An Elementary Treatise on Coordinate Geometry of Three Dimensions
R.J.T. Bell
Macmillan & Co. Ltd., London
4. Tensor Analysis
S.S. Seth and A.K. Johri
Student’s Friends & Co., Agra
5. Differential Geometry
H.S. Sharma & G.C. Chadda
Student’s Friends & Co., Agra
6. Tensor Calculus & Riemannian Geometry
J.K. Goyal & K.P. Gupta
Pragati Prakashan, Meerut.
7. Introduction to Differential Geometry
T.J. Willmore
Oxford University Press, New Delhi.
349
You might also like
- CSIR NET PYQ Linear Algebra - (2011-2021)No ratings yetCSIR NET PYQ Linear Algebra - (2011-2021)34 pages
- (Universitext) Sterling K. Berberian (Auth.) - Fundamentals of Real Analysis (1999, Springer) (10.1007 - 978-1-4612-0549-4) - Libgen - LiNo ratings yet(Universitext) Sterling K. Berberian (Auth.) - Fundamentals of Real Analysis (1999, Springer) (10.1007 - 978-1-4612-0549-4) - Libgen - Li490 pages
- (Comprehensive Course in Analysis) Barry Simon-Advanced Complex Analysis - A Comprehensive Course in Analysis, Part 2B-American Mathematical Society (2015)100% (2)(Comprehensive Course in Analysis) Barry Simon-Advanced Complex Analysis - A Comprehensive Course in Analysis, Part 2B-American Mathematical Society (2015)370 pages
- UG - B.Sc. - Mathematics - 113 51 - Modern Algebra - BSC MATHS - 5665100% (1)UG - B.Sc. - Mathematics - 113 51 - Modern Algebra - BSC MATHS - 5665424 pages
- II Sem B.Sc. (NEP) Mathematics Lab Manual UOM100% (1)II Sem B.Sc. (NEP) Mathematics Lab Manual UOM48 pages
- M. Sc. I Sem. II Partial Diff. Equations AllNo ratings yetM. Sc. I Sem. II Partial Diff. Equations All317 pages
- Real Analysis Complete Notes For CSIR NET GATE IITJAM100% (1)Real Analysis Complete Notes For CSIR NET GATE IITJAM506 pages
- M - SC - (Mathematics) - 311 14 - Topology - I PDF100% (1)M - SC - (Mathematics) - 311 14 - Topology - I PDF264 pages
- Delhi University Notes 2ND Semester Real Nalysis75% (4)Delhi University Notes 2ND Semester Real Nalysis18 pages
- RMM Abstract Algebra Marathon 1 100 CompressedNo ratings yetRMM Abstract Algebra Marathon 1 100 Compressed129 pages
- Real Analysis Notes by Prof Syed Gul ShahNo ratings yetReal Analysis Notes by Prof Syed Gul Shah176 pages
- Keshawa Prasad Gupta - Measure Theory (Including Banach Space, Hilbert Space, Spectral Theory) - Krishna Prakashan Mandir (1994)100% (2)Keshawa Prasad Gupta - Measure Theory (Including Banach Space, Hilbert Space, Spectral Theory) - Krishna Prakashan Mandir (1994)445 pages
- Differential Equations Previous Years Questions100% (1)Differential Equations Previous Years Questions14 pages
- CSIR UGC NET - MATHEMATICS - FREE SOLVED PAPER VPM Class PDF0% (1)CSIR UGC NET - MATHEMATICS - FREE SOLVED PAPER VPM Class PDF46 pages
- BSC SY Real A II Riemann Integral and Improper IntegralNo ratings yetBSC SY Real A II Riemann Integral and Improper Integral109 pages
- Functional Analysis by Balmohan Vishnu Limaye100% (1)Functional Analysis by Balmohan Vishnu Limaye4 pages
- Linear Algebra Lec - 11 Iit - Jam Math Crash Course 2023No ratings yetLinear Algebra Lec - 11 Iit - Jam Math Crash Course 202339 pages
- The Use of Integral Transforms - CompressNo ratings yetThe Use of Integral Transforms - Compress1 page
- J.R. Institute of Mathematics: 1. Simple Variational ProblemsNo ratings yetJ.R. Institute of Mathematics: 1. Simple Variational Problems37 pages
- International Islamic University, Islamabad BS Mathematics Real Analysis IINo ratings yetInternational Islamic University, Islamabad BS Mathematics Real Analysis II29 pages
- Academic Syllabus For M. Sc. in Mathematics: Binod Bihari Mahto Koyalanchal University, Dhanbad WWW - Bbmku.ac - inNo ratings yetAcademic Syllabus For M. Sc. in Mathematics: Binod Bihari Mahto Koyalanchal University, Dhanbad WWW - Bbmku.ac - in28 pages
- Original Mathematics by Adrian Cox B.Sc.No ratings yetOriginal Mathematics by Adrian Cox B.Sc.70 pages
- Sequence and Series of Real Numbers M.T. NairNo ratings yetSequence and Series of Real Numbers M.T. Nair38 pages
- Positivity in Algebraic Geometry, II: Robert LazarsfeldNo ratings yetPositivity in Algebraic Geometry, II: Robert Lazarsfeld371 pages
- The Stone-Cech Compactification and The Space of UltrafiltersNo ratings yetThe Stone-Cech Compactification and The Space of Ultrafilters10 pages
- Brouwer Degree The Core of Nonlinear Analysis 1st Edition George Dinca DownloadNo ratings yetBrouwer Degree The Core of Nonlinear Analysis 1st Edition George Dinca Download51 pages
- Strakes & Spiral Staircase - Differential GeometryNo ratings yetStrakes & Spiral Staircase - Differential Geometry5 pages
- A Geometric Interpretation of The Riemann-Stieltjes IntegralNo ratings yetA Geometric Interpretation of The Riemann-Stieltjes Integral9 pages
- Derived Category of Sheaves and Verdier DualityNo ratings yetDerived Category of Sheaves and Verdier Duality72 pages
- Unit-Iii - Complex Integration: F (Z) DZ 0No ratings yetUnit-Iii - Complex Integration: F (Z) DZ 020 pages
- Willmott Fuzzy Implication in Fuzzy Databases1111No ratings yetWillmott Fuzzy Implication in Fuzzy Databases111115 pages
- Special - Theory - of - Relativity by UgarovNo ratings yetSpecial - Theory - of - Relativity by Ugarov2 pages
- Chapter 4. The First Fundamental Form (Induced Metric)No ratings yetChapter 4. The First Fundamental Form (Induced Metric)16 pages
- A Note On On Beta-Open Sets and Ideals in Topological SpacesNo ratings yetA Note On On Beta-Open Sets and Ideals in Topological Spaces5 pages
- R - Intersection of Tangent Cone - KendigNo ratings yetR - Intersection of Tangent Cone - Kendig6 pages
- Syllabus M.A. M.sc. Semester I Compulsory PapersNo ratings yetSyllabus M.A. M.sc. Semester I Compulsory Papers3 pages
- De Rham Currents in Discrete ElectromagnetismNo ratings yetDe Rham Currents in Discrete Electromagnetism7 pages
- Craven, Two-Properties-Of-Bochner-Integrals (1970)No ratings yetCraven, Two-Properties-Of-Bochner-Integrals (1970)6 pages
