Geodesics Problems

Lecture 6
Formulation of
Calculus of Variations Problems
in Geometry and Mechanics
ME256 Indian Institute of Science
Va r i a t i o n a l M e t h o d s a n d S t r u c t u r a l O p t i m i z a t i o n
G. K. Ananthasuresh
Professor, Mechanical Engineering, Indian Institute of Science, Banagalore
[email protected]
Outline of the lecture
We will discuss some geometry problems that can be cast as
problems of calculus of variations.
We will also discuss the role of calculus of variations in
mechanics and structural optimization.
What we will learn:
◦ What kinds of problems belong to calculus of variations?
◦ How do we formulate calculus of variations problems?
◦ What is the connection between mechanics and calculus of variations?
◦ What is the connection between structural optimization and calculus
of variations?
◦ How does a functional look like?
ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 2

Geometry and calculus of variations
There are many problems in geometry that relate to calculus of
variations.
They pertain to minimal curves and surfaces.
Minimal curves
◦ Geodesics
◦ Maximum enclosing area for a given perimeter
◦ Chains hanging in a force field
◦ Etc.
Minimal surfaces
◦ Minimum surface of revolution
◦ Surfaces of least area enclosed by a given boundary
◦ Etc.

Mechanics and calculus of variations
There are three ways to write equations of statics Three
and dynamics. views of Statics Dynamics
mechanics
Two of these are related to calculus of variations. Force F = ma
◦ We will discuss them in this lecture and later too. Final balance
result of
Structural optimization is essentially calculus of calculus of
variations. variation!
◦ What do we want to optimize in a structure? Principle D’Lambert
◦ Stiffness, flexibility, strength, weight, cost, of virtual principle
manufacturability, natural frequency, mode shape, An work
stability, buckling loads, contact stress, etc. intermedia
te result of
◦ All of these can be posed as objective function and
calculus of
constraints in the framework of calculus of variations.
variations
We will consider a few problems and formulate Calculus Minimum Hamilton’s
them in this lecture. of potential principle
variations energy
principle

variations.
Minimal curves
◦ Geodesics
◦ Maximum enclosing area for a given perimter
◦ Etc.
Minimal surfaces
◦ Etc.

Curve of least distance between two points in a plane.
You are given two points in a flat plane. You can draw many, many curves that connect
the two points. Of all those curves, which one has the least length?
The answer is obvious: it is a straight line joining the two points.
Pretend that you do not know the answer or someone is not convinced about it.
How will you pose this as a problem whose solution gives you a convincing proof?
Here is how:
We take a small segment ds
and integrate it to get the
length of the curve y(x)
between the two given
points.

Geodesic in a plane
Geodesic:
◦ Curve of least distance between two given points.
L here is the functional. Its integrand

depends on the first derivative of y(x),
which is denoted as y’(x).
Solution in another lecture!
Observe the problem for now and
understand it.

Geodesic on a sphere
A spherical surface can be described in
parametric form by azimuthal and elevation
angles and radius R.
Then, we can write the differential quantities as…
Therefore,

Geodesic on a sphere (contd.)
Here, we describe a curve on the sphere as
Thus, the geodesic

problem on a
sphere becomes…

Geodesic on any given surface
Then, we can write the differential
quantities as…
∂x ∂x
=
dx du + dv
∂u ∂v Now, the length of a
∂y ∂y curve on the surface,
=
dy du + dv given in its
∂u ∂v parametric form,
∂z ∂z v(u), is given by
=
dz du + dv
∂u ∂v
Any surface can be
described in parametric
form using u and v

Geodesic on any surface (contd.)
This is the general form of the
geodesic problem for any surface
specified in parametric form.

Now, with a constraint.
Geodesic problems have an objective function, which
is an integral. The integral depended on the derivative
of the variable function.
Now, we will consider a problem with a constraint
that is also an integral of the variable function.
Such problems where the constraint is also an
integral, we call them isoperimetric problems.
By the way, the expressions in the integral form are
called functionals. But functionals need not be of
only integral form. More later….

Queen Dido’s “isoperimetric” problem
If someone gave you a closed loop of a chain of length
L and asked you to take as much land you can enclose
with it, as Dido, the Queen of Carthage (present day
Tunisia) did, what shape would you put that chain on
land? (provided you want to have maximum area of
land to own)
Constant perimeter and hence it is called an isoperimetric problem.

Maximum area enclosed by a curve of given perimeter.
It is convenient to use parametric
representation of a closed curve because
explicit form y(x) may need to be multi-
valued. Let t = 0 to L, be the parameter.
Let the curve be given by x(t) and y(t).
Perimeter
Enclosed area
1 dx 
L L
dy 1
Notation A= ∫0 2 ( xy − yx ) dt
∫0 2  x(t ) dt − y(t ) dt  dt =

Maximum enclosed area with a curve of given
perimeter.
L
1
Min −= A ∫ ( yx − xy ) dt
x ( t ), y ( t )
0
2
Equality -
Subject to constraine
d calculus
∫( )
L of
x 2 + y 2 dt − L =
0 variations
problem!
0
Data : L
New features in problem formulation:

1. An integral (a form of functional) type constraint exists.
2. Two variable functions, x(t) and y(t), which need to be found.
3. Maximization problem can simply be made into a minimization problem by
changing the sign.
Shape of a hanging chain
What shape does a chain held at its ends
take when left freely under gravity?
It tries to minimize its potential energy by
g coming down as much as it could.
h h
=
Min PE
y( x)
0
gy ) ds ∫ ρ gy
∫ ( ρ= 0
1 + y′2 dx Equality-
constraine
d calculus
Subject to of
variations
∫( )
h problem
1 + y′2 dx − L =
0 with one
variable
function.
Mass per unit 0
length of the Data := h, y (h) v, ρ , g
L, y (0) 0,=
chain
Chatterjee problem: maximum enclosed area of a
given perimeter with an inequality constraint
A farmer is free to choose a field with a given
length of fence bounded by a river and three
roads as shown in the figure on the left.
What should be the curve to maximize the
enclosed area?
An inequality constraint
New feature:
Posed by Prof. Anindya Chatterjee, IIT-
Kanpur
variations.
Minimal curves
◦ Geodesics
◦ Maximum enclosing area for a givenWe will consider a few of them
◦ perimeter length
◦ Etc.
Minimal surfaces
◦ Etc.

Minimum surface of revolution of a curve
Here is a problem that looks exactly
What shape? like the hanging chain problem as
far as mathematical formulation is
concerned.
So, don’t you expect the solution to
be the same as well?
L x2
= ∫=
2π y ds ∫ π + ′ 2
Min S 2 y 1 y dx
y( x)
0 x1
Subject to
∫( )
x2
Given end points (x1,y1) and ′
1 + y dx − L =
2
0
(x2,y2), find the curve which when x1
rotated about the x-axis will have
least surface of revolution. Data : =
L, x1 , y ( x1 ) y=
1 , x2 , y ( x2 ) y2
Soap films solve a calculus of variations problem!
Take an easily bendable
wire and make a loop or
even multiple loops with
it. Dip it in soap water and
watch the shape of the
soap film that forms.
Soap films want to

minimize the surface
tension and hence take up
the surface of least area as
they attach to the
boundary of the wire.
http://www.math.hmc.edu/~jacobsen/demolab/soapfilm.html

Plateau’s problem of least surface area for a given
boundary curve in 3D (simpler version)
3D curve
Surface
= surface (single-valued)
 2 
 ∂z   ∂z  
2

Min S = ∫∫ 1 +   +   dx dy
2D area =
z ( x, y )   ∂x   ∂y  
D
 
projection of
the surface, Data : D
z(x,y)
New features:
The functional can be a double-integral.
The variable function can depend on two
independent variables.

Plateau’s problem of least surface area for a given
boundary curve in 3D (more complex version)
What if the contour is
irregular and it is
multi-valued within
the projected 2D
domain D?
Posing and solving http://fathom-the-
universe.tumblr.com/post/
the problem become 55740943330/the-beauty-
difficult. of-minimal-surfaces-there-
Field’s medals have are-many
been awarded
Douglas, for
Jesse (1931). this of the problem of
"Solution
work!
Plateau". Trans. Amer. Math. Soc. (Transactions of
the American Mathematical Society, Vol. 33, No. 1)
33 (1): 263–321.

An optimal control problem: area maximization
problem with optimal steering
Helicopter speed speed = v0
Wind speed = w0
A surveillance helicopter travelling at constant speed (vo) under the constant wind
speed of (w0) needs to enclose maximum area by taking a closed path in a given
time T. The optimization variable is the steering angle, . The starting point is
(x0,y0).

Study this functional…
The objective functional in this problem is interesting. Its

new feature is that it is an integral but it has integrals to be
evaluated within it and those integrals have the unknown
variable function in their integrands.
The purpose of these examples is to let us appreciate the
variety of functionals. We will study the formal notion of a
functional in a later lecture.

There are three ways to write equations of statics and Three
dynamics. views of Statics Dynamics
mechanics
Two of these are related to calculus of variations.
◦ We will discuss them in this lecture and later too. Force F = ma
A result of balance
Structural optimization is essentially calculus of calculus of
variations. variation!
◦ What do we want to optimize in a structure?
◦ Stiffness, flexibility, strength, weight, cost,
manufacturability, natural frequency, mode shape, Principle D’Lambert
stability, buckling loads, contact stress, etc. of virtual principle
◦ All of these can be posed as objective function and Calculus work
constraints in the framework of calculus of variations. of
variations
We will consider a few problems and formulate them Minimum Hamilton’s
in this lecture. potential principle
energy
principle

Static equilibrium of a beam
Method 1: Force and moment balance approach
This differential equation for the small transverse

displacement w(x) of a beam under transverse load, q(x) is
derived based on moment balance at a cross-section and
the bending moment itself is computed based on force
and moment balance.

Method 2: Minimum potential energy principle
As an alternative to force/moment balance, we can simply

minimize the potential energy (PE) with respect to the
unknown variable function, w(x).
The solution to this calculus of variations problem is the
differential equation shown in the pervious slide.
Method 3: Principle of virtual work
For all kinematically admissible .
Internal virtual work = external virtual work

As the second alternative to force/moment balance, we can simply solve
this equation that is valid for any kinematically admissible function,
.
This statement is a consequence of the minimization of the potential
energy functional of the previous slide.
But this is an independent way of stating static equilibrium!
Now, we know three independent ways of writing conditions for static
equilibrium.
Method 1: Force/moment balance approach
◦ The differential equation with boundary conditions
◦ Called the strong form
Method 2: Principle of minimum potential energy (calculus of variations)

◦ All we need to know is an expression for the potential energy.
◦ The boundary conditions will emerge out of this statement.
Method 3: Principle of virtual work

◦ An intermediate result of calculus of variations
◦ Called also the weak form
◦ Notice that the highest order derivative of the unknown function is lower here as compared to
the one in the strong form.
We will discuss details of Methods 2 and 3 in later lectures.

Understand the three methods with a simple
spring.
= displacement (stretch) of the spring at equilibrium
Since there is just one
scalar variable x, it is a
finite-variable
optimization here and
NOT calculus of
variations.
Method 1 Method 2 Method 3
Force equilibrium Minimum potential energy Principle of virtual work
1 2
Min =
PE kx − Fx
x 2
Internal ∂PE Internal
force = =0 ⇒ kx =F virtual work
external ∂x = external
force virtual work
Static equilibrium of a general elastic body
Method 1
where
Force equilibrium
Method 2
Minimum potential
energy
Method 3
Principle of virtual We will
work discuss the
notation and
derivations in
later lectures.

Contact problems in elasticity: beam
= gap function
Calculus of variations problem, in the framework of minimum

potential energy principle, can easily account for contact
conditions, as shown here.
Just an inequality constraint!

Vibrating string: Hamilton’s principle
A taut vibration string with tension, T.
Length = L; mass per unit length =
Equation of motion obtained using force-balance.
Calculus of variations
statement: Hamilton’s
principle
Notice that it is not minimization or maximization; it is simply
extremization of a functional; also notice that the variable function
depends on space variable x and time variable t.
Equation of motion of a beam
Equation of motion obtained using

force-balance.
Calculus of variations statement: Hamilton’s

principle
Which function w(x,t) will extremize H, the

Hamiltonian?
There are three ways to write equations of statics Three views
and dynamics. of mechanics Statics Dynamics
Two of these are related to calculus of variations.
◦ We will discuss them in this lecture and later Force F = ma
too. A result of balance
calculus of
Structural optimization is essentially calculus of variation!
variations.
◦ What do we want to optimize in a structure? Principle D’Lambert
of virtual principle
◦ Stiffness, flexibility, strength, weight, cost, Calculus of work
manufacturability, natural frequency, mode shape, variations
stability, buckling loads, contact stress, etc.
◦ All of these can be posed as objective function and Minimum Hamilton’s
constraints in the framework of calculus of potential principle
variations. energy
principle
We will consider a few problems and formulate
them in this lecture.

Objectives and constraints in structural optimization
Weight Dynamic response
Stiffness Contact stress
Strength Etc.
Flexibility
Cost
Any of these can be the objective
function or be part of a
Stability
constraint.
Buckling load
Variable functions, the design
Natural frequency variables, will be related shape and
Mode shape size; and topology (how many holes
are there?)
CALCULUS OF VARIATIONS AND STRUCTURAL
OPTIMIZATION
Structural optimization of a beam
Minimize the strain energy of the beam for an upper bound on the
volume of material.
The less the strain energy, the stiffer
the beam.
The breadth of the beam is the design
variable.
The displacement of the beam (w(x)) is
the state variable.
The governing equation (the
equilibrium equation) for the state
variable.
The volume constraint is an This will be the
typical structure
inequality. of any
structural
Data constitutes the known quantities. optimization
problem.

Min-max of stress: design for a strong beam
Minimize the maximum stress for an upper bound on the volume of
material.
New feature in the formulation:

The functional has another
maximization problem in it.
This is a min-max problem.
Note that minimization and
maximization of the same
quantity is with respect to two
different variables.
They are not uncommon in
structural optimization.
Electro-thermal-compliant actuator design
New features in the
formulation:
The functional is
simply one variable,
the displacement at a
point.
There are three
governing equations
pertaining to electrical,
thermal, and elastic
problems.
There are six state
variables, V, Vv, T, Tv,
u, uv.
Features of calculus of variations problems
There can be constraints which are functionals or functions.
Constraints can be equalities are inequalities.
Objective functions are always functionals.
A functional can be of many forms.
◦ Just an integral
◦ Ratio of integrals
◦ Integral with another integral inside it
◦ Maximum or a minimum of a function
◦ Etc.
You have now seen what a functional is, in many of its forms.
We will learn about them formally after a brief de tour of
theory of finite-variable optimization.

The end note
Many problems in geometry can be posed as calculus of variations
geometry and mechanics
Calculus of variations in
problems.
Curves of least length and surfaces of least area are popular.
Mechanics problems can be posed in three different ways;
Two of them are directly under the purview of calculus of variations.
Structural optimization problems are essentially calculus of variations

problems.
Integrals
Constraints can Integrals within an integral
be equalities Ratio of two integrals
and inequalities Min or max of a function
in calculus of Can depend on more than one variable
variations too. function
Can involve more than one
independent variable
Functionals can
be…
Can depend on space and time Thanks
variables

Geodesics Problems

Uploaded by

Copyright:

Available Formats

Geodesics Problems

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Geodesics Problems

Uploaded by

Copyright:

Available Formats

Lecture 6

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 2

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 3

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 4

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 5

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 6

L here is the functional. Its integrand

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 7

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 8

Here, we describe a curve on the sphere as

Thus, the geodesic

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 9

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 10

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 11

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 12

Constant perimeter and hence it is called an isoperimetric problem.

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 13

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 14

New features in problem formulation:

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 18

Soap films want to

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 20

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 21

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 22

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 23

The objective functional in this problem is interesting. Its

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 24

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 25

This differential equation for the small transverse

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 26

As an alternative to force/moment balance, we can simply

For all kinematically admissible .

Internal virtual work = external virtual work

Method 2: Principle of minimum potential energy (calculus of variations)

Method 3: Principle of virtual work

We will discuss details of Methods 2 and 3 in later lectures.

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 31

Calculus of variations problem, in the framework of minimum

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 32

Equation of motion obtained using force-balance.

Equation of motion obtained using

Calculus of variations statement: Hamilton’s

Which function w(x,t) will extremize H, the

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 35

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 37

New feature in the formulation:

ME256 / G. K. Ananthasuresh, IISc VARIATIONAL METHODS AND STRUCTURAL OPTIMIZATION 40

Structural optimization problems are essentially calculus of variations

You might also like