Calculus of variations

Lectures 07-08
 Calculus of variations

Stationary point of a function f(x)

f(x)
f(x) has a minimum at x = x*

df/dx = 0 at x = x*

d f/dx > 0 at x= x*
X* X

f(x)
f(x) has a maximum at x = x*

df/dx = 0 at x= x*

d f/ dx < 0 at x = x*

X* X

f(x) f(x) has neither a minimum nor a maximum at x = x*

df/dx = 0 at x = x*

d f/dx = 0 at x = x*

f(x) has a point of inflection at x = x*

x* is also called stationary points of the function f(x). The function f(x) may have a minimum,
a maximum or neither of them at x = x*.
 Calculus of variations

dy
y (x) and y ' (x)= are functions of an independent variable x.
dx

y (x) represents a curve and y ' (x) represents the slope of the curve y (x).

f ( y , y ' , x) is a function of y (x), y ' (x) and the independent variable x.

The integral I [ y]=∫ f ( y , y ' , x)dx is called a functional of the function y (x)
x1
between two points x1 and x2.

The calculus of variations involves finding the actual curve y (x) passing through
two end points x1 and x2 for which the integral I [y] takes a staionary value.

Suppose the integral I[y] has a stationary value for a specific y∗(x) passing through
two points x1 and x2. Then an infinitesimal change from the curve y∗(x) changes
the integral.

y∗(x): The actual curve passing through points x1 and x2 . The integral I [y*] has a
stationary value. Y

y (x): Neighbouring curves passing through y2

y1
x1 x2 X

y (x)= y∗( x)+α η(x), where η(x 1)=η(x 2)=0. y*(x)

As α→0, y (x)→ y∗(x) for any choice of the function η(x).

For a fixed choice of the function η(x), the integral may be considered as a
function of the parameter α .
integral
The change in the action for small values of α may be written as:

I [y]
δI= ( )
∂I
∂α α=0
α=0 if S has a stationary value.

⇒ ( ∂∂αI )
α=0
=0
 Calculus of variations

For a given choice of the arbitrary function (x), the integral I [ ] is

x x
I[ ]= f (y, y , x) dx = f (y* + , y* + , x) dx, y(x) = y*(x) + (x)
x x

To find the stationary value of the integral I [ ], we set = 0.

This equation is true for an arbitrary choice of the function (x).

This equation is known as Euler-Lagrange equation.

Examples: I [y]
1

(1) Find the path y(x) for which the integral I=∫ ( y + y ' 2 )dx ,
0
y (0)=0, y (1)=2 takes a stationary value. Evaluate the integral.

1 1

I=∫ f ( y , y ' , x)dx=∫ ( y+ y ' 2)dx .

⇒ f ( y , y ' , x)=( y+ y ' 2)

Euler-lagrange equation:
∂ f d ∂f
− ( )
∂ y dx ∂ y '
=0.

∂f
∂y
=1 &
∂f
∂ y'
=2 y ' ⇒
d ∂f
( )
dx ∂ y '
=2 y ' ' .

1 1
⇒ 2 y ' '=1 ⇒ y ' =C 1+ x ⇒ y (x)=C 1 x+ x 2 +C 2.
2 4

7
As y (0)=0, C 2=0. Further, y (1)=2 ⇒ C1=
4.

7 x x2 7 x
⇒ y (x)= + & y ' (x)= +
4 4 4 2

( )( )
2 2 2
27x x 49 x 7 x 7 x x 49
f ( y , y ' , x)=( y+ y ' )= + + + + = + +
4 4 16 4 4 2 2 16

) [ ]
1

(
1 1
7 x x 2 49 7 x 2 x 3 49 x
I=∫ f ( y , y ' , x)dx=∫ + + dx= + +
0 0 2 2 16 4 6 16 0

I= [ 7 1 49 239
+ +
4 6 16
=
48 ]
.
 Calculus of variations

(2) Show that the shortest distance between two points on a plane is the distance
between these points on a straigh line passing through these points.

ds= √(dx)2 +(dy)2=dx √ 1+ y ' 2

2 2 2

L=∫ ds=∫ dx √ 1+ y ' =∫ f ( y , y ' , x)dx

1 1 1

f ( y , y ' , x)=√ 1+ y ' 2 , which does not contain y explicitly.

∂f
⇒ =0.
∂y

Euler-Lagrange equation: ( )
d ∂f ∂f
= =0.
dx ∂ y ' ∂ y

⇒ ( )
∂f
∂y'
=C , a constant

1 2 y'
⇒ =C
2 √ (1+ y ' 2)

2 2 2 2 C2
⇒ y ' =C (1+ y ' ) ⇒ y' = 2
=m2 ⇒ y ' (x)=m
1−C

⇒ y (x)=mx + y 0 .
 Calculus of variations

(3) Fermat’s principle:The actual path that light follows while travelling from a
source (point 1) to an observer (point 2) is the one for which the time of travel is
stationary (minimum !).
2 2 2
ds μ c
T =∫ dt =∫ =∫ ds , μ= is the refractive index of the medium.
1 1 v 1 c
v

ds= √(dx)2+(dy)2=dx √ 1+ y ' 2

μ(x , y)
f ( y , y ' , x)=
c
√ 1+ y ' 2

Case 1: μ( x , y)=μ0, a constant

μ0
f ( y , y ' , x)=
c
√ 1+ y '
2

⇒
∂f
∂y
=0 ⇒ ( )
∂f
=
1 2y'
∂ y ' 2c √ 1+ y ' 2
=K .

⇒ y ' 2=m2 ⇒ y ' =m ⇒ y (x)=m x+ y 0.

Case 2: μ( x , y)=μ( y). Assume, it is increasing with the height.

Time for a rays of light to leave the source point S and reach an observation point O
is stationary.